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 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
6 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
7 rather than 0 to represent an unknown size. Assert that the size
8 is known when the mode isn't BLKmode.
9 (may_trap_p_1): Use -1 for unknown sizes.
10 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
12 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
13 Alan Hayward <alan.hayward@arm.com>
14 David Sherwood <david.sherwood@arm.com>
16 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
17 to poly_int64. Update commentary for positions_needed.large.
18 (read_info_type): Change offset and width from HOST_WIDE_INT
20 (set_usage_bits): Likewise.
21 (canon_address): Return the offset as a poly_int64 rather than
22 a HOST_WIDE_INT. Use strip_offset_and_add.
23 (set_all_positions_unneeded, any_positions_needed_p): Use
24 positions_needed.large to track stores with non-constant widths.
25 (all_positions_needed_p): Likewise. Take the offset and width
26 as poly_int64s rather than ints. Assert that rhs is nonnull.
27 (record_store): Cope with non-constant offsets and widths.
28 Nullify the rhs of an earlier store if we can't tell which bytes
30 (find_shift_sequence): Take the access_size and shift as poly_int64s
32 (get_stored_val): Take the read_offset and read_width as poly_int64s
33 rather than HOST_WIDE_INTs.
34 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
35 non-constant offsets and widths.
37 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
38 Alan Hayward <alan.hayward@arm.com>
39 David Sherwood <david.sherwood@arm.com>
41 * inchash.h (inchash::hash::add_poly_int): New function.
42 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
43 Use poly_int64 rather than HOST_WIDE_INT.
44 (ao_ref::max_size_known_p): New function.
45 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
46 rather than HOST_WIDE_INT.
47 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
48 to temporaries until its interface is adjusted to match.
49 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
50 (aliasing_component_refs_p, decl_refs_may_alias_p)
51 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
52 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
53 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
55 * alias.c (ao_ref_from_mem): Likewise.
56 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
57 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
58 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
59 (maybe_trim_complex_store, maybe_trim_constructor_store)
60 (live_bytes_read, dse_classify_store): Likewise.
61 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
62 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
63 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
64 (vn_reference_lookup_3): Likewise.
65 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
67 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
68 Alan Hayward <alan.hayward@arm.com>
69 David Sherwood <david.sherwood@arm.com>
71 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
72 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
73 instead of ranges_overlap_p.
75 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
76 Alan Hayward <alan.hayward@arm.com>
77 David Sherwood <david.sherwood@arm.com>
79 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
80 sizes as poly_int64s rather than HOST_WIDE_INTs.
82 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
83 Alan Hayward <alan.hayward@arm.com>
84 David Sherwood <david.sherwood@arm.com>
86 * gimple-fold.h (fold_ctor_reference): Take the offset and size
87 as poly_uint64 rather than unsigned HOST_WIDE_INT.
88 * gimple-fold.c (fold_ctor_reference): Likewise.
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 * target.def (dwarf_poly_indeterminate_value): New hook.
95 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
96 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
97 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
98 * doc/tm.texi: Regenerate.
99 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
100 offset as a poly_int64.
101 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
102 offset as a poly_int64.
103 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
104 Take the offset as a poly_int64.
105 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
106 (frame_pointer_fb_offset): Change to a poly_int64.
107 (int_loc_descriptor): Take the offset as a poly_int64. Use
108 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
109 (based_loc_descr): Take the offset as a poly_int64.
110 Use strip_offset_and_add to handle (plus X (const)).
111 Use new_reg_loc_descr instead of an open-coded version of the
112 previous implementation.
113 (mem_loc_descriptor): Handle CONST_POLY_INT.
114 (compute_frame_pointer_to_fb_displacement): Take the offset as a
115 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
117 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
118 Alan Hayward <alan.hayward@arm.com>
119 David Sherwood <david.sherwood@arm.com>
121 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
122 (gen_rtx_REG_offset): Take the offset as a poly_int64.
123 * inchash.h (inchash::hash::add_poly_hwi): New function.
124 * gengtype.c (main): Register poly_int64.
125 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
126 offset as a poly_int.
127 (reg_attr_hasher::equal): Use must_eq to compare offsets.
128 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
129 offset as a poly_int64.
130 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
131 * print-rtl.c (print_poly_int): New function.
132 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
134 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
136 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
137 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
138 (vt_get_decl_and_offset): Return the offset as a poly_int64.
139 Enforce track_offset_p for parts of a PARALLEL.
140 (vt_add_function_parameter): Use const_offset for the final
141 offset to track. Use get_tracked_reg_offset for the parts
144 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
145 Alan Hayward <alan.hayward@arm.com>
146 David Sherwood <david.sherwood@arm.com>
148 * target.def (truly_noop_truncation): Take poly_uint64s instead of
149 unsigned ints. Change default to hook_bool_puint64_puint64_true.
150 * doc/tm.texi: Regenerate.
151 * hooks.h (hook_bool_uint_uint_true): Delete.
152 (hook_bool_puint64_puint64_true): Declare.
153 * hooks.c (hook_bool_uint_uint_true): Delete.
154 (hook_bool_puint64_puint64_true): New function.
155 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
156 instead of unsigned ints.
157 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
158 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
160 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
161 Alan Hayward <alan.hayward@arm.com>
162 David Sherwood <david.sherwood@arm.com>
164 * optabs.h (expand_operand): Add an int_value field.
165 (create_expand_operand): Add an int_value parameter and use it
166 to initialize the new expand_operand field.
167 (create_integer_operand): Replace with a declaration of a function
168 that accepts poly_int64s. Move the implementation to...
169 * optabs.c (create_integer_operand): ...here.
170 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
171 the mode preserves the value of int_value, instead of calling
172 const_int_operand on the rtx. Use gen_int_mode to generate
175 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
176 Alan Hayward <alan.hayward@arm.com>
177 David Sherwood <david.sherwood@arm.com>
179 * dumpfile.h (dump_dec): Declare.
180 * dumpfile.c (dump_dec): New function.
181 * pretty-print.h (pp_wide_integer): Turn into a function and
182 declare a poly_int version.
183 * pretty-print.c (pp_wide_integer): New function for poly_ints.
185 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
186 Alan Hayward <alan.hayward@arm.com>
187 David Sherwood <david.sherwood@arm.com>
189 * doc/generic.texi (POLY_INT_CST): Document.
190 * tree.def (POLY_INT_CST): New tree code.
191 * treestruct.def (TS_POLY_INT_CST): New tree layout.
192 * tree-core.h (tree_poly_int_cst): New struct.
193 (tree_node): Add a poly_int_cst field.
194 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
195 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
196 instead of a wide_int_ref.
197 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
199 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
200 instead of an unsigned HOST_WIDE_INT.
201 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
202 (ptrdiff_tree_p): Declare.
203 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
204 extern inline implementations if the target doesn't use POLY_INT_CST.
205 (poly_int_tree_p): New function.
206 (wi::unextended_tree): New class.
207 (wi::int_traits <unextended_tree>): New override.
208 (wi::extended_tree): Add a default constructor.
209 (wi::extended_tree::get_tree): New function.
210 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
211 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
212 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
213 (wi::tree_to_poly_wide_ref): New typedefs.
214 (wi::ints_for): Provide overloads for extended_tree and
216 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
217 (wi::to_wide): New functions.
218 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
219 * tree.c (poly_int_cst_hasher): New struct.
220 (poly_int_cst_hash_table): New variable.
221 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
222 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
224 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
225 (init_ttree): Initialize poly_int_cst_hash_table.
226 (build_int_cst, build_int_cst_type, build_invariant_address): Take
227 a poly_int64 instead of a HOST_WIDE_INT.
228 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
229 instead of an unsigned HOST_WIDE_INT.
230 (wide_int_to_tree): Rename to...
231 (wide_int_to_tree_1): ...this.
232 (build_new_poly_int_cst, build_poly_int_cst): New functions.
233 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
234 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
235 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
236 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
237 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
239 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
240 (streamer_read_tree_body): Likewise.
241 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
242 (streamer_write_tree_body): Likewise.
243 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
244 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
245 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
246 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
247 * gimple-expr.h (is_gimple_constant): Likewise.
248 * gimplify.c (maybe_with_size_expr): Likewise.
249 * print-tree.c (print_node): Likewise.
250 * tree-data-ref.c (data_ref_compare_tree): Likewise.
251 * tree-pretty-print.c (dump_generic_node): Likewise.
252 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
253 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
254 * tree-vrp.c (compare_values_warnv): Likewise.
255 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
256 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
257 (force_expr_to_var_cost): Likewise.
258 * tree-ssa-loop.c (for_each_index): Likewise.
259 * fold-const.h (build_invariant_address, size_int_kind): Take a
260 poly_int64 instead of a HOST_WIDE_INT.
261 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
262 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
264 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
265 (int_const_binop_2): New function, split out from...
266 (int_const_binop_1): ...here. Handle POLY_INT_CST.
267 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
268 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
269 * gimple-ssa-strength-reduction.c (slsr_process_add)
270 (slsr_process_mul): Check for INTEGER_CSTs before using them
272 * stor-layout.c (bits_from_bytes): New function.
273 (bit_from_pos): Use it.
274 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
275 by BITS_PER_UNIT to get the TYPE_SIZE.
276 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
277 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
279 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
280 Alan Hayward <alan.hayward@arm.com>
281 David Sherwood <david.sherwood@arm.com>
283 * doc/rtl.texi (const_poly_int): Document. Also document the
284 rtl sharing behavior.
285 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
286 * rtl.h (const_poly_int_def): New struct.
287 (rtx_def::u): Add a cpi field.
288 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
289 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
290 (wi::rtx_to_poly_wide_ref): New typedef
291 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
292 (poly_int_rtx_p): New functions.
293 (trunc_int_for_mode): Declare a poly_int64 version.
294 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
295 (immed_wide_int_const): Take a poly_wide_int_ref rather than
297 (strip_offset): Declare.
298 (strip_offset_and_add): New function.
299 * rtl.def (CONST_POLY_INT): New rtx code.
300 * rtl.c (rtx_size): Handle CONST_POLY_INT.
301 (shared_const_p): Use poly_int_rtx_p.
302 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
304 (gen_int_shift_amount): Likewise.
305 * emit-rtl.c (const_poly_int_hasher): New class.
306 (const_poly_int_htab): New variable.
307 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
308 (const_poly_int_hasher::hash): New function.
309 (const_poly_int_hasher::equal): Likewise.
310 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
311 (immed_wide_int_const): Rename to...
312 (immed_wide_int_const_1): ...this and make static.
313 (immed_wide_int_const): New function, taking a poly_wide_int_ref
314 instead of a wide_int_ref.
315 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
316 (gen_lowpart_common): Handle CONST_POLY_INT.
317 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
318 * cselib.c (cselib_hash_rtx): Likewise.
319 * dwarf2out.c (const_ok_for_output_1): Likewise.
320 * expr.c (convert_modes): Likewise.
321 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
322 * rtlhash.c (add_rtx): Likewise.
323 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
324 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
325 Handle existing CONST_POLY_INT rtxes.
326 * expmed.h (expand_shift): Take a poly_int64 instead of a
328 * expmed.c (expand_shift): Likewise.
329 * rtlanal.c (strip_offset): New function.
330 (commutative_operand_precedence): Give CONST_POLY_INT the same
331 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
333 * rtl-tests.c (const_poly_int_tests): New struct.
334 (rtl_tests_c_tests): Use it.
335 * simplify-rtx.c (simplify_const_unary_operation): Handle
337 (simplify_const_binary_operation): Likewise.
338 (simplify_binary_operation_1): Fold additions of symbolic constants
340 (simplify_subreg): Handle extensions and truncations of
342 (simplify_const_poly_int_tests): New struct.
343 (simplify_rtx_c_tests): Use it.
344 * wide-int.h (storage_ref): Add default constructor.
345 (wide_int_ref_storage): Likewise.
346 (trailing_wide_ints): Use GTY((user)).
347 (trailing_wide_ints::operator[]): Add a const version.
348 (trailing_wide_ints::get_precision): New function.
349 (trailing_wide_ints::extra_size): Likewise.
351 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
352 Alan Hayward <alan.hayward@arm.com>
353 David Sherwood <david.sherwood@arm.com>
355 * emit-rtl.h (gen_int_shift_amount): Declare.
356 * emit-rtl.c (gen_int_shift_amount): New function.
357 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
359 * calls.c (shift_return_value): Likewise.
360 * cse.c (fold_rtx): Likewise.
361 * dse.c (find_shift_sequence): Likewise.
362 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
363 (expand_shift, expand_smod_pow2): Likewise.
364 * lower-subreg.c (shift_cost): Likewise.
365 * optabs.c (expand_superword_shift, expand_doubleword_mult)
366 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
367 (expand_vec_perm_var): Likewise.
368 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
369 (simplify_binary_operation_1): Likewise.
370 * combine.c (try_combine, find_split_point, force_int_to_mode)
371 (simplify_shift_const_1, simplify_shift_const): Likewise.
372 (change_zero_ext): Likewise. Use simplify_gen_binary.
374 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
376 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
378 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
380 * doc/invoke.texi (ARM Options): Document accepted extension options
381 for -march=armv8.3-a.
383 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
386 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
387 or --with-float=softfp, set the default CPU to arm10e.
389 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
391 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
392 * config/visium/predicates.md (const_shift_operand): Likewise.
393 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
394 (visium_legitimize_reload_address): Likewise.
396 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
399 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
401 2017-12-20 Jakub Jelinek <jakub@redhat.com>
404 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
406 2017-12-20 Julia Koval <julia.koval@intel.com>
408 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
409 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
410 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
411 * config.gcc: Include vpclmulqdqintrin.h.
412 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
413 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
415 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
416 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
417 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
418 * config/i386/i386.c (isa2_opts): Add -mcx16.
419 (isa_opts): Add -mpclmulqdq, remove -mcx16.
420 (ix86_option_override_internal): Move mcx16 to flags2.
421 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
422 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
423 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
424 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
425 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
426 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
427 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
428 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
429 * doc/invoke.texi: Add -mvpclmulqdq.
431 2017-12-20 Tom de Vries <tom@codesourcery.com>
434 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
435 * calls.c (rtx_for_static_chain): ... here. New function.
436 * calls.h (rtx_for_static_chain): Declare.
437 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
438 instead of targetm.calls.static_chain.
439 * df-scan.c (df_get_entry_block_def_set): Same.
441 2017-12-19 Tom de Vries <tom@codesourcery.com>
443 PR tree-optimization/83493
444 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
447 2017-12-19 Jakub Jelinek <jakub@redhat.com>
449 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
450 inform with hardcoded english plural handling.
452 2017-12-18 Jeff Law <law@redhat.com>
454 PR tree-optimization/83477
455 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
456 a non-virtual PHI, always push a new range.
458 2017-12-19 Martin Sebor <msebor@redhat.com>
461 * builtins.c (compute_objsize): Handle non-constant offsets.
463 2017-12-19 Jakub Jelinek <jakub@redhat.com>
465 PR tree-optimization/83444
466 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
467 character load case, if get_stridx on MEM_REF's operand doesn't
468 look usable, retry with get_addr_stridx.
470 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
473 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
474 (variable_tracking_main_1): Keep markers even when VTA fails.
477 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
478 even if there are markers before it.
479 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
481 2017-12-19 Jakub Jelinek <jakub@redhat.com>
483 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
484 typical order conditions.
485 * sel-sched.c (extract_new_fences_from): Likewise.
486 * config/visium/constraints.md (J, K, L): Likewise.
487 * config/visium/predicates.md (const_shift_operand): Likewise.
488 * config/visium/visium.c (visium_legitimize_address,
489 visium_legitimize_reload_address): Likewise.
490 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
491 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
492 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
493 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
494 SET_DUMP_DETAIL): Likewise.
495 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
496 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
497 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
498 avr_set_core_architecture, avr_set_current_function,
499 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
500 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
501 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
502 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
503 avr_map_decompose, avr_fold_builtin): Likewise.
504 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
505 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
506 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
507 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
508 * config/m32c/m32c.c (m32c_conditional_register_usage,
509 m32c_address_cost): Likewise.
510 * config/m32c/predicates.md (shiftcount_operand,
511 longshiftcount_operand): Likewise.
512 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
513 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
514 can_use_cdx_ldstw): Likewise.
515 * config/nios2/nios2.h (CDX_REG_P): Likewise.
516 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
518 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
519 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
520 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
521 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
522 * config/vax/vax.c (adjacent_operands_p): Likewise.
523 * config/ft32/constraints.md (L, b, KA): Likewise.
524 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
526 * cfgexpand.c (expand_stack_alignment): Likewise.
527 * gcse.c (insert_expr_in_table): Likewise.
528 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
529 * cgraphunit.c (cgraph_node::expand): Likewise.
530 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
531 * emit-rtl.c (add_insn): Likewise.
532 * input.c (dump_location_info): Likewise.
533 * passes.c (NEXT_PASS): Likewise.
534 * read-rtl-function.c (parse_note_insn_name,
535 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
537 * sched-rgn.c (sched_rgn_init): Likewise.
538 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
539 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
540 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
541 * explow.c (eliminate_constant_term): Likewise.
542 * final.c (leaf_renumber_regs_insn): Likewise.
543 * cfgrtl.c (print_rtl_with_bb): Likewise.
544 * genhooks.c (emit_init_macros): Likewise.
545 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
546 * tree-data-ref.c (conflict_fn): Likewise.
547 * selftest.c (assert_streq): Likewise.
548 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
549 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
550 fold_binary_loc, multiple_of_p): Likewise.
551 * reload.c (push_reload, find_equiv_reg): Likewise.
552 * et-forest.c (et_nca, et_below): Likewise.
553 * dbxout.c (dbxout_symbol_location): Likewise.
554 * reorg.c (relax_delay_slots): Likewise.
555 * dojump.c (do_compare_rtx_and_jump): Likewise.
556 * gengtype-parse.c (type): Likewise.
557 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
558 simplify_const_relational_operation): Likewise.
559 * reload1.c (do_output_reload): Likewise.
560 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
561 * gengtype.c (type_for_name): Likewise.
562 * gimple-ssa-sprintf.c (format_directive): Likewise.
564 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
567 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
568 accessing it. Adjust comment.
570 2017-12-19 Jakub Jelinek <jakub@redhat.com>
573 * predict.c (zero_one_minusone): New function.
574 (apply_return_prediction): Avoid return prediction for functions
575 returning only -1, 0 and 1 values, unless they only return -1 and 0
578 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
580 * config/arc/arc.c (legitimate_scaled_address_p): Clean
581 fall-through warning.
582 (arc_compute_frame_size): Remove unused variables.
583 (arc_print_operand): Fix fprintif format.
584 (arc_can_follow_jump): Clean fall-through warning.
586 2017-12-19 Marek Polacek <polacek@redhat.com>
589 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
592 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
594 * config/arc/arc.c (overriderregs): New variable.
595 (arc_override_options): Track fixed/call saved/call options.
596 (arc_conditional_register_usage): Check against overrideregs
597 variable whenever we change register properties.
599 2017-12-19 Nathan Sidwell <nathan@acm.org>
601 * opts.c (finish_options): Don't prefix dump_base_name if it
602 already contains directories.
604 2017-12-19 Martin Liska <mliska@suse.cz>
606 PR rtl-optimization/82675
607 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
608 more element in sbitmap.
610 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
612 * gimplify.c (gimplify_expr): Use error_operand_p.
614 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
617 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
620 2017-12-19 Daniel Cederman <cederman@gaisler.com>
622 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
623 the jump is to a label.
625 2017-12-19 Jakub Jelinek <jakub@redhat.com>
627 PR tree-optimization/83444
628 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
633 * ipa-inline.c (flatten_remove_node_hook): New function.
634 (ipa_inline): Keep only nodes with flatten attribute at the end of
635 the array in the order from ipa_reverse_postorder, only walk that
636 portion of array for flattening, if there is more than one such
637 node, temporarily register a removal hook and ignore removed nodes.
639 PR tree-optimization/80631
640 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
641 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
642 IFN_REDUC_MAX or IFN_REDUC_MIN.
644 2017-12-18 Jeff Law <law@redhat.com>
646 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
647 record symbolic equivalences from backedges in the CFG.
650 2017-11-19 Jeff Law <law@redhat.com>
652 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
653 of degenerates resulting from ignoring an edge.
655 2017-12-18 Martin Sebor <msebor@redhat.com>
658 PR tree-optimization/78450
659 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
660 (handle_builtin_strlen): Call it.
662 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
664 PR rtl-optimization/83424
665 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
667 2017-12-18 Marek Polacek <polacek@redhat.com>
670 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
671 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
674 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
676 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
677 (umaddsidi4, umaddsidi_split): Likewise.
679 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
681 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
684 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
687 * config/s390/s390.c (s390_option_override): Avoid strncpy.
689 2017-12-18 Richard Biener <rguenther@suse.de>
691 PR tree-optimization/81877
692 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
693 (outermost_indep_loop): Adjust.
694 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
695 (can_sm_ref_p): Adjust.
697 2017-12-18 Richard Biener <rguenther@suse.de>
700 * tree.c (array_at_struct_end_p): Return true if the underlying
701 object has space for at least one element in excess of what
702 the array domain specifies.
704 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
706 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
707 example to avoid overfull hbox.
708 * doc/invoke.texi (Option Summary): Add missing @gol.
709 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
712 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
713 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
715 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
716 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
717 (AArch64 Options): Document -mverbose-cost-dump.
718 (ARM Options): Likewise, plus -mflip-thumb.
720 2017-12-17 Martin Sebor <msebor@redhat.com>
723 * gimple-ssa-warn-restrict.c
724 (builtin_memref::offset_out_of_bounds): Correct the handling of
727 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
729 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
731 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
733 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
735 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
737 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
740 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
742 * doc/invoke.texi: Fix some typos.
744 2017-12-16 Martin Sebor <msebor@redhat.com>
746 PR tree-optimization/78918
747 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
748 * builtins.c (check_sizes): Rename...
749 (check_access): ...to this. Rename function arguments for clarity.
750 (check_memop_sizes): Adjust names.
751 (expand_builtin_memchr, expand_builtin_memcpy): Same.
752 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
753 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
754 (check_strncat_sizes, expand_builtin_strncat): Same.
755 (expand_builtin_strncpy, expand_builtin_memset): Same.
756 (expand_builtin_bzero, expand_builtin_memcmp): Same.
757 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
758 (maybe_emit_sprintf_chk_warning): Same.
759 (expand_builtin_strcpy): Adjust.
760 (expand_builtin_stpcpy): Same.
761 (expand_builtin_with_bounds): Detect out-of-bounds accesses
762 in pointer-checking forms of memcpy, memmove, and mempcpy.
763 (gcall_to_tree_minimal, max_object_size): Define new functions.
764 * builtins.h (max_object_size): Declare.
765 * calls.c (alloc_max_size): Call max_object_size instead of
766 hardcoding ssizetype limit.
767 (get_size_range): Handle new argument.
768 * calls.h (get_size_range): Add a new argument.
769 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
770 * doc/invoke.texi (-Wrestrict): Adjust, add example.
771 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
773 (gimple_fold_builtin_memory_chk): Same.
774 (gimple_fold_builtin_stxcpy_chk): New function.
775 * gimple-ssa-warn-restrict.c: New source.
776 * gimple-ssa-warn-restrict.h: New header.
777 * gimple.c (gimple_build_call_from_tree): Propagate location.
778 * passes.def (pass_warn_restrict): Add new pass.
779 * tree-pass.h (make_pass_warn_restrict): Declare.
780 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
782 (handle_builtin_strcat): Same.
783 (strlen_optimize_stmt): Rename...
784 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
785 stpncpy, strncpy, and their checking forms.
787 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
789 PR rtl-optimization/82849
790 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
791 and get_max_loop_iterations_int.
793 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
794 Alan Hayward <alan.hayward@arm.com>
795 David Sherwood <david.sherwood@arm.com>
797 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
798 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
799 size as a poly_uint64.
800 (mode_for_vector, mode_for_int_vector): Take the number of vector
801 elements as a poly_uint64.
802 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
803 size as a poly_uint64.
804 (mode_for_vector, mode_for_int_vector): Take the number of vector
805 elements as a poly_uint64.
807 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
808 Alan Hayward <alan.hayward@arm.com>
809 David Sherwood <david.sherwood@arm.com>
811 * machmode.h (MACRO_MODE): New macro.
812 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
813 * targhooks.c (default_libcall_value, default_secondary_reload)
814 (default_memory_move_cost, default_register_move_cost)
815 (default_class_max_nregs): Likewise.
817 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
818 Alan Hayward <alan.hayward@arm.com>
819 David Sherwood <david.sherwood@arm.com>
821 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
822 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
824 * genautomata.c (main): Likewise.
825 * genconditions.c (write_header): Likewise.
826 * genemit.c (main): Likewise.
827 * genextract.c (print_header): Likewise.
828 * genopinit.c (main): Likewise.
829 * genoutput.c (output_prologue): Likewise.
830 * genpeep.c (main): Likewise.
831 * genpreds.c (write_insn_preds_c): Likewise.
832 * genrecog.c (writer_header): Likewise.
833 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
834 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
835 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
836 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
837 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
838 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
839 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
840 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
841 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
842 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
843 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
844 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
845 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
846 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
847 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
848 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
849 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
850 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
851 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
852 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
853 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
854 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
855 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
856 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
857 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
858 * config/darwin.c (IN_TARGET_CODE): Likewise.
859 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
860 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
861 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
862 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
863 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
864 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
865 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
866 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
867 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
868 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
869 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
870 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
871 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
872 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
873 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
874 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
875 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
876 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
877 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
878 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
879 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
880 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
881 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
882 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
883 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
884 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
885 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
886 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
887 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
888 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
889 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
890 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
891 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
892 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
893 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
894 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
895 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
896 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
897 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
898 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
899 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
900 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
901 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
902 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
903 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
904 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
905 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
906 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
907 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
908 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
909 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
910 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
911 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
912 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
913 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
914 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
915 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
916 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
917 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
918 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
919 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
920 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
921 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
922 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
923 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
924 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
925 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
926 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
927 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
928 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
929 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
930 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
931 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
932 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
933 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
934 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
935 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
936 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
937 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
938 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
939 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
940 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
941 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
942 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
943 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
944 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
945 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
946 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
947 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
948 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
949 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
950 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
951 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
952 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
953 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
954 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
955 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
956 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
957 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
958 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
959 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
960 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
961 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
962 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
963 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
964 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
966 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
967 Alan Hayward <alan.hayward@arm.com>
968 David Sherwood <david.sherwood@arm.com>
970 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
973 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
974 Alan Hayward <alan.hayward@arm.com>
975 David Sherwood <david.sherwood@arm.com>
977 * doc/generic.texi (VEC_SERIES_EXPR): Document.
978 * doc/md.texi (vec_series@var{m}): Document.
979 * tree.def (VEC_SERIES_EXPR): New tree code.
980 * tree.h (build_vec_series): Declare.
981 * tree.c (build_vec_series): New function.
982 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
983 * tree-pretty-print.c (dump_generic_node): Likewise.
984 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
985 * tree-inline.c (estimate_operator_cost): Likewise.
986 * expr.c (expand_expr_real_2): Likewise.
987 * optabs-tree.c (optab_for_tree_code): Likewise.
988 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
989 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
990 * expmed.c (make_tree): Handle VEC_SERIES.
991 * optabs.def (vec_series_optab): New optab.
992 * optabs.h (expand_vec_series_expr): Declare.
993 * optabs.c (expand_vec_series_expr): New function.
994 * tree-vect-generic.c (expand_vector_operations_1): Check that
995 the operands also have vector type.
997 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
998 Alan Hayward <alan.hawyard@arm.com>
999 David Sherwood <david.sherwood@arm.com>
1001 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1002 (VEC_COND_EXPR): Add missing @tindex.
1003 * doc/md.texi (vec_duplicate@var{m}): Document.
1004 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1005 * tree.c (build_vector_from_val): Add stubbed-out handling of
1006 variable-length vectors, using VEC_DUPLICATE_EXPR.
1007 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1008 * cfgexpand.c (expand_debug_expr): Likewise.
1009 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1010 * tree-inline.c (estimate_operator_cost): Likewise.
1011 * tree-pretty-print.c (dump_generic_node): Likewise.
1012 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1013 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1014 (test_vec_duplicate_folding): New function.
1015 (fold_const_c_tests): Call it.
1016 * optabs.def (vec_duplicate_optab): New optab.
1017 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1018 * optabs.h (expand_vector_broadcast): Declare.
1019 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1020 vec_duplicate_optab.
1021 * expr.c (store_constructor): Try using vec_duplicate_optab for
1023 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1025 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1028 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1029 div/mod latencies a bit.
1031 2017-12-15 Jeff Law <law@redhat.com>
1033 PR tree-optimization/36550
1034 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1035 (mark_threaded_blocks): Rewrite code to avoid block copying when
1036 optimizing for size. Don't pessimize blocks which will be
1037 copied, but all the statements will be dead.
1039 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1041 PR tree-optimization/81165
1042 * tree-ssa-threadupdate.c (uses_in_bb): New.
1043 (estimate_threading_killed_stmts): New.
1044 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
1045 * tree-ssa-threadedge.c
1046 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
1049 2017-12-15 Jeff Law <law@redhat.com>
1051 PR tree-optimization/83410
1052 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
1053 threads when parallelizing loops.
1055 2017-12-15 Jakub Jelinek <jakub@redhat.com>
1057 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
1059 * config/alpha/alpha.c (vms_attribute_table): Swap
1060 affects_type_identity and handler fields, adjust comments.
1061 * config/mips/mips.c (mips_attribute_table): Likewise.
1062 * config/visium/visium.c (visium_attribute_table): Likewise.
1063 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
1064 * config/microblaze/microblaze.c (microblaze_attribute_table):
1066 * config/spu/spu.c (spu_attribute_table): Likewise.
1067 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
1068 * config/arc/arc.c (arc_attribute_table): Likewise.
1069 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
1070 * config/v850/v850.c (v850_handle_interrupt_attribute,
1071 v850_handle_data_area_attribute): Formatting fixes.
1072 (v850_attribute_table): Swap affects_type_identity and handler
1073 fields, adjust comments.
1074 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
1075 * config/arm/arm.c (arm_attribute_table): Likewise.
1076 * config/avr/avr.c (avr_attribute_table): Likewise.
1077 * config/s390/s390.c (s390_attribute_table): Likewise.
1078 * config/sh/sh.c (sh_attribute_table): Likewise.
1079 * config/i386/i386.c (ix86_handle_cconv_attribute,
1080 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
1081 (ix86_attribute_table): Swap affects_type_identity and handler
1082 fields, adjust comments.
1083 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1084 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
1085 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
1086 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
1087 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
1088 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1089 * config/rx/rx.c (rx_attribute_table): Likewise.
1090 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
1091 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
1092 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
1093 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
1094 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1095 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
1096 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
1097 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
1098 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1099 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1100 * doc/plugins.texi (user_attr): Likewise. Add NULL for
1102 * attribs.c (empty_attribute_table): Swap affects_type_identity and
1104 (register_scoped_attributes, decl_attributes): Formatting fixes.
1106 PR tree-optimization/83269
1107 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
1108 subtraction in arg0's type if type is signed and arg0 is unsigned.
1112 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
1113 convert? on @0 instead of convert. Check type of @1, not @0.
1114 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
1115 convert? on @0 instead of convert. Check type of @1, not @0.
1116 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
1117 only optimize if either both @1 and @2 types are narrower
1118 precision, or both are wider or equal precision, and in the former
1119 case only if both have undefined overflow.
1121 2017-12-15 Richard Biener <rguenther@suse.de>
1124 * internal-fn.def (IFN_NOP): Add.
1125 * internal-fn.c (expand_NOP): Do nothing.
1126 * lto-streamer-in.c (input_function): Instead of removing
1127 sanitizer calls replace them with IFN_NOP calls.
1129 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
1130 Alan Hayward <alan.hayward@arm.com>
1131 David Sherwood <david.sherwood@arm.com>
1133 * dse.c (store_info, read_info_type): Replace begin and end with
1135 (print_range): New function.
1136 (set_all_positions_unneeded, any_positions_needed_p)
1137 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
1139 (record_store): Likewise. Optimize the case in which all positions
1141 (get_stored_val): Replace read_begin and read_end with read_offset
1143 (replace_read): Update call accordingly.
1145 2017-12-15 Bin Cheng <bin.cheng@arm.com>
1147 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
1148 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
1149 (loop_cand::loop_cand): Initialize above members.
1150 (loop_cand::supported_operations): Delete.
1151 (loop_cand::can_interchange_p): Inline above function.
1152 (loop_cand::classify_simple_reduction): Record number of constant
1153 initialized simple reductions.
1154 (should_interchange_loops): New parameters. Check stmt cost of loops
1156 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
1157 Update call to should_interchange_loops.
1158 (should_interchange_loop_nest): Update call to
1159 should_interchange_loops.
1161 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
1164 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
1165 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
1166 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
1168 2017-12-15 Julia Koval <julia.koval@intel.com>
1170 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
1171 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
1172 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
1173 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
1174 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
1176 2017-12-15 Julia Koval <julia.koval@intel.com>
1178 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
1179 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
1180 * config/i386/sse.md (vaesenc_<mode>): New pattern.
1181 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
1182 _mm_aesenc_epi128): New intrinsics.
1184 2017-12-15 Julia Koval <julia.koval@intel.com>
1186 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
1187 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
1188 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
1189 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
1190 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
1192 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1194 * gimple-ssa-strength-reduction.c (analyze_increments):
1195 Distinguish replacement costs for constant strides from those for
1198 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1200 * var-tracking.c (variable_tracking_main_1): Formatting fix.
1202 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
1204 * doc/invoke.texi: Document -Wcast-function-type.
1205 * recog.h (stored_funcptr): Change signature.
1206 * tree-dump.c (dump_node): Avoid warning.
1207 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
1209 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
1212 * gimple-fold.c (get_range_strlen): Add the handling of non-member
1215 2017-12-14 David Malcolm <dmalcolm@redhat.com>
1217 PR tree-optimization/83312
1218 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
1219 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
1220 "val" param, and to cope with arbitrary basic blocks.
1221 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
1222 handle NULL_TREE for "val", dropping "bb" param.
1223 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
1224 drop "bb" param. Handle NULL_TREE for "val".
1225 (find_case_label_for_value): Make "switch_stmt" param const.
1226 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
1228 (vrp_prop::check_all_array_refs): Reimplement as...
1229 (check_array_bounds_dom_walker::before_dom_children): ...this new
1230 vfunc. Replace linear search through BB block list, excluding
1231 those with non-executable in-edges via dominator walk.
1233 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1235 * config/arm/arm.opt (mverbose-cost-dump): New option.
1236 * config/arm/arm.c (arm_rtx_costs): Use it.
1238 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
1241 * reload1.c (emit_input_reload_insns): Skip debug markers.
1243 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
1245 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
1249 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
1250 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
1251 * config/sh/sh.c: Skip debug insns besides notes.
1252 * config/sh/sh.md: Likewise.
1253 * config/sh/sh_treg_combine.cc: Likewise.
1254 * config/sh/sync.md: Likewise.
1256 2017-12-14 Tom de Vries <tom@codesourcery.com>
1258 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1259 Add item for weak_undefined.
1261 2017-12-14 Richard Biener <rguenther@suse.de>
1263 PR tree-optimization/67842
1264 * sese.h (bb_in_region): Remove #if 0'ed code.
1266 2017-12-14 Richard Biener <rguenther@suse.de>
1268 PR tree-optimization/83326
1269 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1270 may_be_zero parameter and handle it by not marking the first
1271 peeled copy as not exiting the loop.
1272 (try_peel_loop): Likewise.
1273 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
1274 to handle the case of constant or zero iterations and perform
1275 loop header copying on-the-fly.
1277 2017-12-14 Richard Biener <rguenther@suse.de>
1279 PR tree-optimization/83418
1280 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
1281 Instead of asserting we don't get unfolded comparisons deal with
1284 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1287 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
1288 basic blocks. Assert debug bind insns don't appear outside of bbs,
1289 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
1292 PR tree-optimization/83198
1293 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
1294 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
1295 value if arg is a REAL_CST with incompatible type.
1297 2017-12-14 Sudakshina Das <sudi.das@arm.com>
1298 Bin Cheng <bin.cheng@arm.com>
1301 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
1303 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
1306 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1308 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
1309 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
1311 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
1312 Alan Hayward <alan.hayward@arm.com>
1313 David Sherwood <david.sherwood@arm.com>
1315 * poly-int.h: New file.
1316 * poly-int-types.h: Likewise.
1317 * coretypes.h: Include them.
1318 (POLY_INT_CONVERSION): Define.
1319 * target.def (estimated_poly_value): New hook.
1320 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
1321 * doc/tm.texi: Regenerate.
1322 * doc/poly-int.texi: New file.
1323 * doc/gccint.texi: Include it.
1324 * doc/rtl.texi: Describe restrictions on subreg modes.
1325 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
1326 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
1327 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
1328 * targhooks.h (default_estimated_poly_value): Declare.
1329 * targhooks.c (default_estimated_poly_value): New function.
1330 * target.h (estimated_poly_value): Likewise.
1331 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
1332 (wi::unary_traits): Delete.
1333 (wi::binary_traits::signed_shift_result_type): Define for
1334 offset_int << HOST_WIDE_INT, etc.
1335 (generic_wide_int::operator <<=): Define for all types and use
1336 wi::lshift instead of <<.
1337 (wi::hwi_with_prec): Add a default constructor.
1338 (wi::ints_for): New class.
1339 (operator <<): Define for all wide-int types.
1340 (operator /): New function.
1341 (operator %): Likewise.
1342 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
1343 (ASSERT_MAYBE_NE_AT): New macros.
1345 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
1346 Dominik Vogt <vogt@linux.vnet.ibm.com>
1349 * emit-rtl.c (init_emit): Remove ??? comment.
1350 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1351 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
1352 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
1353 alignment of 3 virtual registers to BITS_PER_WORD.
1355 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
1357 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
1359 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
1360 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
1362 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1365 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
1367 2017-12-13 Jakub Jelinek <jakub@redhat.com>
1369 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
1370 with the exception of debug begin stmt markers appear before
1374 * final.c (rest_of_handle_final): Call variable_tracking_main only
1375 if !flag_var_tracking.
1377 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1378 Jakub Jelinek <jakub@redhat.com>
1382 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
1383 labels debug stmts that can only appear after labels.
1385 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
1387 PR rtl-optimization/82398
1388 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
1389 EXPR_USEFULNESS in priority comparison.
1391 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
1393 PR rtl-optimization/83393
1394 * combine.c (move_deaths): If reg_stat points to a too new insn in
1395 last_death, do not use it: find the proper insn instead.
1397 2017-12-12 Jeff Law <law@redhat.com>
1399 PR tree-optimization/83298
1400 PR tree-optimization/83362
1401 PR tree-optimization/83383
1402 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
1403 push_value_range a public interface. Add new argument to
1404 record_ranges_from_stmt.
1405 * gimple-ssa-evrp-analyze.c
1406 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
1407 Update comments. Handle recording temporary equivalences.
1408 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
1409 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
1410 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
1411 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
1412 gimple-ssa-evrp-analyze.h.
1413 (record_temporary_equivalences_from_phis): Add new argument. When
1414 the PHI arg is an SSA_NAME, set the result's range to the range
1416 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
1417 from statements too.
1418 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
1419 Pass it down to children as needed.
1420 (thread_outgoing_edges): Likewise.
1421 (thread_across_edge): Likewise. Push/pop range state as needed.
1422 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
1424 2017-12-12 Julia Koval <julia.koval@intel.com>
1426 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
1427 (PTA_CANNONLAKE): Remove PTA_CLWB.
1429 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1432 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
1433 (vcondv2div2df): Likewise.
1434 (vconduv2dfv2di): Likewise.
1436 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1438 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
1439 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
1440 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
1441 ATTR_MATHFN_FPROUNDING_ERRNO.
1443 2017-12-12 Richard Biener <rguenther@suse.de>
1445 PR tree-optimization/83385
1446 * graphite-scop-detection.c (get_order, order): Remove.
1447 (bb_to_rpo): New global.
1449 (build_scops): Sort pbbs in RPO order.
1451 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
1453 * combine.c (simplify_set): Do not transform subregs to zero_extends
1454 if the destination mode is a vector mode.
1456 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1458 PR tree-optimization/83359
1459 * tree-cfg.h (fold_loop_internal_call): Declare.
1460 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
1461 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
1462 (find_loop_dist_alias): New function.
1463 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
1464 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
1467 PR tree-optimization/80631
1468 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
1469 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
1470 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
1471 hardcoding zero as the value if COND_EXPR is never true. For
1472 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
1473 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
1474 hardcoding MAX_EXPR as the reduction operation.
1475 (is_nonwrapping_integer_induction): Allow negative step.
1476 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
1477 vect_create_epilog_for_reduction, if no value is suitable, don't
1478 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
1480 2017-12-12 Richard Biener <rguenther@suse.de>
1482 PR tree-optimization/81889
1483 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
1484 range info from the non-wrapping IV instead of just the range
1487 2017-12-12 Julia Koval <julia.koval@intel.com>
1489 * config.gcc: Add vaesintrin.h.
1490 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1492 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1493 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1495 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1496 * config/i386/immintrin.h: Include vaesintrin.h.
1497 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1498 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1499 _mm_aesdec_epi128): New intrinsics.
1501 2017-12-12 Julia Koval <julia.koval@intel.com>
1503 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
1504 OPTION_MASK_ISA_VAES_UNSET): New.
1505 (ix86_handle_option): Handle -mvaes.
1506 * config/i386/cpuid.h: Define bit_VAES.
1507 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
1508 * config/i386/i386-c.c (__VAES__): New.
1509 * config/i386/i386.c (ix86_target_string): Add -mvaes.
1510 (ix86_valid_target_attribute_inner_p): Ditto.
1511 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
1512 * config/i386/i386.opt: Add -mvaes.
1513 * doc/invoke.texi: Ditto.
1515 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
1517 * debug.h (gcc_debug_hooks): Add inline_entry.
1518 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
1519 * debug.c (do_nothing_debug_hooks): Likewise.
1520 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1521 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
1522 (dwarf2_lineno_debug_hooks): Likewise.
1524 * common.opt (gstatement-frontiers): New, setting
1525 debug_nonbind_markers_p.
1526 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
1527 * toplev.c (process_options): Autodetect value for debug statement
1529 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
1530 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
1532 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
1533 markers. Integrate source bind into debug stmt expand loop.
1534 (pass_expand::execute): Check debug marker limit. Avoid deep
1535 TER and expand debug locations for debug bind insns only.
1536 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
1538 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
1539 * final.c (reemit_insn_block_notes): Take current block from
1540 nonbind markers. Declare note where it's first set.
1541 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
1542 begin stmt markers if enabled.
1543 (notice_source_line): Handle nonbind markers. Fail if their
1544 location is unknown or that of builtins.
1545 (rest_of_handle_final): Convert begin stmt markers to notes if
1546 var-tracking didn't run.
1547 (rest_of_clean_state): Skip begin stmt markers.
1548 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
1550 * function.c (allocate_struct_function): Set begin_stmt_markers.
1551 * function.h (struct function): Add debug_marker_count counter
1552 and debug_nonbind_markers flag.
1553 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
1554 * gimple-low.c (lower_function_body): Adjust
1555 debug_nonbind_markers.
1556 (lower_stmt): Drop or skip gimple debug stmts.
1557 (lower_try_catch): Skip debug stmts.
1558 * gimple.c (gimple_build_debug_begin_stmt): New.
1559 (gimple_copy): Increment debug_marker_count if copying one.
1560 * gimple.h (gimple_build_debug_begin_stmt): Declare.
1561 * gimplify.c (rexpr_location): New.
1562 (rexpr_has_location): New.
1563 (warn_switch_unreachable_r): Handle gimple debug stmts.
1564 (shortcut_cond_r): Call expr_location.
1566 (find_goto_label): New.
1567 (shortcut_cond_expr): Call expr_has_location, expr_location, and
1569 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
1571 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
1572 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
1573 (LANG_HOOKS_INITIALIZER): ... this.
1574 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
1575 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
1577 (update_ebb_live_info): Skip debug insn markers.
1578 * lra.c (debug_insn_static_data): Rename to...
1579 (debug_bind_static_data): ... this.
1580 (debug_marker_static_data): New.
1581 (lra_set_insn_recog_data): Select one of the above depending
1583 (lra_update_isn_regno_info): Don't assume debug insns have
1585 (push_insns): Skip debug insns.
1586 * lto-streamer-in.c (input_function): Drop debug stmts
1587 depending on active options. Adjust debug_nonbind_markers.
1588 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
1589 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
1590 begin stmt marker notes.
1591 (print_insn): Likewise.
1592 * recog.c (extract_insn): Recognize rtl for debug markers.
1593 * rtl.def (DEBUG_MARKER): New.
1594 * tree-inline.c: Include params.h.
1595 (remap_gimple_stmt): Handle nonbind markers.
1596 (maybe_move_debug_stmts_to_successors): Likewise.
1597 (copy_debug_stmt): Likewise.
1598 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
1599 markers regardless of no side effects.
1600 (tsi_link_before): Don't update container's side effects when adding
1601 a begin stmt marker.
1602 (tsi_link_after): Likewise.
1603 (expr_first): Skip begin stmt markers.
1604 (expr_last): Likewise.
1605 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
1606 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1607 Disregard nonbind markers.
1608 * tree.c (make_node_stat): Don't set side effects for begin stmt
1610 (build1_stat): Likewise.
1611 * tree.def (DEBUG_BEGIN_STMT): New.
1612 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
1613 * var-tracking.c (delete_debug_insns): Renamed to...
1614 (delete_vta_debug_insns): ... this.
1615 (reemit_marker_as_note): New.
1616 (vt_initialize): Reemit markers.
1617 (delete_vta_debug_insns): Likewise.
1618 (vt_debug_insns_local): Reemit or delete markers.
1619 (variable_tracking_main_1): Likewise.
1620 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
1621 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
1622 (gimple_debug_nonbind_marker_p): New.
1623 (gimple_build_debug_bind): Adjust.
1624 (gimple_build_debug_begin_stmt): New.
1625 * doc/invoke.texi (max-debug-marker-count): New param.
1626 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
1627 (debug_parameter_ref, debug_marker): New.
1628 (NOTE_INSN_BEGIN_STMT): New.
1629 (DEBUG_INSN): Describe begin stmt markers.
1631 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
1632 without debug insns, we wouldn't, but clean up debug insns
1633 after a control flow insn nevertheless.
1635 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
1636 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
1637 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
1638 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
1639 (rtl_tidy_fallthru_edge): Likewise.
1640 (rtl_verify_fallthru): Likewise.
1641 (rtl_verify_bb_layout): Likewise.
1642 (skip_insns_after_block): Likewise.
1643 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
1644 * dwarf2out.c: Include print-rtl.h.
1645 (dwarf2out_next_real_insn): New.
1646 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
1647 Dump debug binds in asm comments.
1648 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
1649 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
1650 callers to use gsi_start_nondebug_bb instead.
1651 (gsi_after_labels): Skip gimple debug stmts.
1652 (gsi_start_nondebug): New.
1653 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
1654 (proper_loop_form_for_interchange): Adjust.
1655 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
1656 * gimple.h (gimple_seq_last_nondebug_stmt): New.
1657 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
1658 (collect_fallthrough_labels): Likewise.
1659 (should_warn_for_implicit_fallthrough): Likewise.
1660 (warn_implicit_fallthrough_r): Likewise.
1661 (expand_FALLTHROUGH_r): Likewise.
1662 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
1663 (graphite_copy_stmts_from_block): Skip nonbind markers.
1664 * haifa-sched.c (sched_extend_bb): Skip debug insns.
1665 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
1666 * jump.c (clean_barriers): Skip debug insns.
1667 * omp-expand.c (expand_parallel_call): Skip debug insns.
1668 (expand_task_call): Likewise.
1669 (remove_exit_barrier): Likewise.
1670 (expand_omp_taskreg): Likewise.
1671 (expand_omp_for_init_counts): Likewise.
1672 (expand_omp_for_generic): Likewise.
1673 (expand_omp_for_static_nochunk): Likewise.
1674 (expand_omp_for_static_chunk): Likewise.
1675 (expand_omp_simd): Likewise.
1676 (expand_omp_taskloop_for_outer): Likewise.
1677 (expand_omp_taskloop_for_inner): Likewise.
1678 (expand_oacc_for): Likewise.
1679 (expand_omp_sections): Likewise.
1680 (expand_omp_single): Likewise.
1681 (expand_omp_synch): Likewise.
1682 (expand_omp_atomic_load): Likewise.
1683 (expand_omp_atomic_store): Likewise.
1684 (expand_omp_atomic_fetch_op): Likewise.
1685 (expand_omp_atomic_pipeline): Likewise.
1686 (expand_omp_atomic_mutex): Likewise.
1687 (expand_omp_target): Likewise.
1688 (grid_expand_omp_for_loop): Likewise.
1689 (grid_expand_target_grid_body): Likewise.
1690 (build_omp_regions_1): Likewise.
1691 * omp-low.c (check_combined_parallel): Skip debug stmts.
1692 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
1693 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
1694 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
1696 * tree-cfg.c (make_blobs_1): Skip debug stmts.
1697 (make_edges): Likewise.
1698 (cleanup_dead_labels): Likewise.
1699 (gimple_can_merge_blocks_p): Likewise.
1700 (stmt_starts_bb_p): Likewise.
1701 (gimple_block_label): Likewise.
1702 (gimple_redirect_edge_and_branch): Likewise.
1703 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
1705 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
1707 * tree-pretty-print (print_declaration): Omit initializer in slim
1709 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
1711 (eliminate_unnecessary_stmts): Stabilize block removal order.
1712 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
1713 * var-tracking.c (get_first_insn): New.
1714 (vt_emit_notes): Call it.
1715 (vt_initialize): Walk any insns before the first BB.
1716 (delete_debug_insns): Likewise.
1718 * gimple.h (enum gimple_debug_subcode): Add
1719 GIMPLE_DEBUG_BEGIN_STMT.
1720 (gimple_debug_begin_stmt_p): New.
1721 (gimple_debug_nonbind_marker_p): New.
1722 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
1723 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
1724 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
1725 * insn-notes.def (BEGIN_STMT): New.
1726 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
1727 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
1728 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
1729 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
1730 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
1731 (INSN_DEBUG_MARKER_KIND): New.
1732 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
1733 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
1734 (INSN_VAR_LOCATION_PTR): New.
1735 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
1737 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
1738 deps only if debug bind insns are enabled.
1739 (pass_expand::execute): Avoid deep TER and expand
1740 debug locations for debug bind insns only.
1741 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
1742 debug stmts special handling down to debug bind stmts.
1743 * combine.c (try_combine): Narrow debug insns special handling
1744 down to debug bind insns.
1745 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
1746 Narrow debug insns preexisting special handling down to debug
1748 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
1749 handling down to debug bind insns.
1750 * function.c (instantiate_virtual_regs): Skip debug markers,
1751 adjust handling of debug binds.
1752 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
1753 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
1754 * haifa-sched.c (schedule_insn): Narrow special handling of debug
1755 insns to debug bind insns.
1756 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
1757 special handling of debug stmts to debug bind stmts.
1758 * ipa-split.c (split_function): Likewise.
1759 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
1760 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
1762 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
1763 * regrename.c (build_def_use): Likewise.
1764 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
1765 (pass_cprop_hardreg): Narrow special casing of debug insns to
1767 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
1768 * reload1.c (reload): Likewise.
1769 * sese.c (sese_insert_phis_for_liveouts): Narrow special
1770 casing of debug stmts to debug bind stmts.
1771 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
1772 * ssa-iterators.h (num_imm_uses): Likewise.
1773 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
1774 debug stmts to debug bind stmts.
1775 * tree-inline.c (tree_function_versioning): Narrow special casing
1776 of debug stmts to debug bind stmts.
1777 * tree-loop-distribution.c (generate_loops_for_partition):
1778 Narrow special casing of debug stmts to debug bind stmts.
1779 * tree-sra.c (analyze_access_subtree): Narrow special casing
1780 of debug stmts to debug bind stmts.
1781 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
1782 stmts to debug bind stmts.
1783 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
1784 casing of debug stmts to debug bind stmts.
1785 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
1786 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
1787 casing of debug stmts to debug bind stmts.
1788 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1790 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
1791 debug stmts to debug bind stmts.
1792 (gimple_replace_ssa_lhs): Likewise.
1793 (insert_debug_temp_for_var_def): Likewise.
1794 (insert_debug_temps_for_defs): Likewise.
1795 (reset_debug_uses): Likewise.
1796 * tree-ssanames.c (release_ssa_name_fn): Likewise.
1797 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
1798 (adjust_debug_stmts): Likewise.
1799 (adjust_phi_and_debug_stmts): Likewise.
1800 (vect_do_peeling): Likewise.
1801 * tree-vect-loop.c (vect_transform_loop): Likewise.
1802 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
1803 * var-tracking.c (adjust_mems): Narrow special casing of debug
1804 insns to debug bind insns.
1805 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
1806 (compute_bb_dataflow, vt_find_locations): Likewise.
1807 (vt_expand_loc, emit_notes_for_changes): Likewise.
1808 (vt_init_cfa_base): Likewise.
1809 (vt_emit_notes): Likewise.
1810 (vt_initialize): Likewise.
1811 (vt_finalize): Likewise.
1813 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
1814 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
1815 (next_nonnote_nondebug_insn_bb): New.
1816 (prev_nonnote_nondebug_insn_bb): New.
1817 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
1818 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
1819 (prev_nonnote_nondebug_insn_bb): Declare.
1820 (next_nonnote_nondebug_insn_bb): Declare.
1821 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
1822 * cfgrtl.c (get_last_bb_insn): Likewise.
1823 * lra.c (push_insns): Likewise.
1825 2017-12-11 David Malcolm <dmalcolm@redhat.com>
1828 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
1830 * selftest.c (selftest::test_runner::test_runner): New ctor.
1831 (selftest::test_runner::~test_runner): New dtor.
1832 * selftest.h (class selftest::test_runner): New class.
1834 2017-12-11 Carl Love <cel@us.ibm.com>
1836 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
1837 vec_extract_fp32_from_shortl]): Add #defines.
1838 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
1839 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
1840 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
1841 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
1842 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
1843 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
1844 * doc/extend.texi: Add documentation for the added builtins.
1846 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
1848 PR rtl-optimization/80693
1849 PR rtl-optimization/81019
1850 PR rtl-optimization/81020
1851 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
1852 are not mentioned in i3. Place the REG_UNUSED note on i2,
1853 possibly modified to REG_DEAD, if it did not originate in i3.
1855 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1857 * recog.c (store_data_bypass_p_1): New function.
1858 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
1859 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
1861 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
1863 PR rtl-optimization/83361
1864 * ifcvt.c (if_convert): Call fixup_partitions.
1866 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1868 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1869 early folding of splat_u{8,16,32}.
1871 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1873 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
1874 output_operand_lossage first argument with capital letter.
1875 (aarch64_override_options): Don't start error and sorry first argument
1876 with capital letter.
1878 2017-12-11 Andi Kleen <ak@linux.intel.com>
1880 PR gcov-profile/83355
1881 * auto-profile.c (string_table::get_index_by_decl): Don't
1882 recurse when abstract origin points to itself.
1884 2017-12-11 Bin Cheng <bin.cheng@arm.com>
1886 PR tree-optimization/83320
1887 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
1888 (prune_datarefs_not_in_loop): Ditto.
1890 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
1892 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
1894 2017-12-10 Jakub Jelinek <jakub@redhat.com>
1896 PR tree-optimization/83337
1897 * gimple-loop-interchange.cc (compute_access_stride): Handle
1898 bitfield DRs properly.
1900 2017-12-09 Jakub Jelinek <jakub@redhat.com>
1902 PR tree-optimization/83338
1903 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
1904 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
1905 vector of unsigned integers to vector of signed integers.
1907 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
1909 PR rtl-optimization/83317
1910 * lra-constraints.c (process_address_1): Add insn code check.
1912 2017-12-08 Michael Matz <matz@suse.de>
1914 Fix PR tree-optimization/83323
1915 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
1916 head-controlled loops and loop BBs.
1917 * common.opt (funroll-and-jam): Remove, instead ...
1918 (floop-unroll-and-jam): ... reuse this option.
1919 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
1920 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
1921 (-floop-unroll-and-jam): ... this option.
1923 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1925 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
1926 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
1927 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
1929 2017-12-08 Richard Biener <rguenther@suse.de>
1932 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
1933 handle accesses outside of zero-sized vars.
1935 2017-12-08 Martin Jambor <mjambor@suse.cz>
1937 PR tree-optimization/83141
1938 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
1939 test for MEM_REFs implicitely changing types with padding. Remove
1941 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
1943 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1945 * config/arc/arc.c (arc_attribute_table): Add exclusions to
1947 * config/avr/avr.c (avr_attribute_table): Likewise.
1948 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1949 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1950 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1951 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
1952 of struct attribute_spec.
1953 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1955 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
1958 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
1960 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
1963 * profile-count.c (profile_count::from_gcov_type): Move from
1964 profile-count.h; handle overflow.
1965 * profile-count.h (profile_count::from_gcov_type): Move offline.
1967 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
1969 PR rtl-optimization/83304
1970 * combine.c (move_deaths): If we do not know where a register died,
1973 2017-12-08 Richard Biener <rguenther@suse.de>
1975 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1976 Provide -fopt-info-loop feedback when we interchange in a nest.
1978 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1980 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
1981 for armv6 ARM CPU IDs.
1983 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1985 * common/config/arm/arm-common.c: Include <algorithm>.
1986 (INCLUDE_VECTOR): Define.
1987 (compare_opt_names): New function.
1988 (arm_rewrite_selected_arch): Only strip out extensions that can be
1989 expressed through -mfpu. Sort the remaining extensions
1992 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1994 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
1995 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
1996 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
1997 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
1999 2017-06-08 Tristan Gingold <gindold@adacore.com>
2002 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2003 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2005 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2007 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2008 of struct attribute_spec.
2010 2017-12-08 Julia Koval <julia.koval@intel.com>
2012 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2013 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2014 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2015 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2016 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2017 _mm_maskz_dpwssds_epi32): Ditto.
2019 2017-12-08 Richard Biener <rguenther@suse.de>
2021 PR tree-optimization/81303
2022 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2023 conditions try to create a comparison vector type matching
2024 the data vector type.
2025 (vectorizable_condition): Adjust.
2026 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2027 Leave invariant conditions alone in case we can vectorize those.
2029 2017-12-08 Julia Koval <julia.koval@intel.com>
2031 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2032 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2033 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2034 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2035 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2037 2017-12-08 Julia Koval <julia.koval@intel.com>
2039 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2040 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2041 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
2042 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
2043 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
2044 _mm_maskz_dpbusds_epi32): New intrinsics.
2046 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2048 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
2050 (ld<bh>io_signed): Likewise.
2051 (st<bhw>io): Likewise.
2052 * config/nios2/predicates.md (ldstio_memory_operand): Allow
2053 SMALL_INT12 constant integer operand.
2055 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2057 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
2059 (nios2_large_constant_allowed): ...this. Adjust uses.
2060 (nios2_plus_symbolic_constant_p): Rename to...
2061 (nios2_plus_large_constant_p): ...this. Adjust uses.
2062 (nios2_legitimate_address_p): Correct CONST_INT handling.
2063 (nios2_symbolic_memory_operand_p): Rename to...
2064 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
2065 (nios2_large_constant_p): Check for large constant integers too.
2066 (nios2_split_large_constant): Handle constant integers.
2067 (nios2_split_symbolic_memory_operand): Rename to...
2068 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
2069 (nios2_legitimize_constant_address): Handle constant integers.
2070 (r0rel_constant_p): Handle small constant integers.
2071 (nios2_print_operand_address): Handle r0-relative integer addresses.
2072 * config/nios2/nios2-protos.h: Adjust for renamed functions.
2073 * config/nios2/nios2.md: Adjust for renamed functions.
2075 2017-12-07 Andrew Waterman <andrew@sifive.com>
2077 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
2078 (TARGET_HAVE_SRODATA_SECTION): New define.
2079 (riscv_select_section): New function.
2081 2017-12-08 Joseph Myers <joseph@codesourcery.com>
2082 Alexander Monakov <amonakov@ispras.ru>
2083 Jakub Jelinek <jakub@redhat.com>
2086 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
2088 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2090 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
2091 a bool success value. Don't call output_operand_lossage here.
2092 (aarch64_print_ldpstp_address): Return a bool success value.
2093 (aarch64_print_operand_address): Call output_addr_const if
2094 aarch64_print_address_internal fails.
2095 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
2096 'y'; call output_operand_lossage instead. Call output_operand_lossage
2097 if aarch64_print_ldpstp_address fails.
2099 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2101 * tree-vector-builder.h
2102 (tree_vector_builder::binary_encoded_nelts): Declare.
2103 * tree-vector-builder.c
2104 (tree_vector_builder::binary_encoded_nelts): New function.
2105 * fold-const.c (negate_expr_p): Likewise.
2106 (operand_equal_p, fold_checksum_tree): Likewise.
2107 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
2108 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
2109 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
2110 (uniform_vector_p): Likewise.
2111 * varasm.c (const_hash_1, compare_constant): Likewise.
2112 * tree-ssa-ccp.c: Include tree-vector-builder.h.
2113 (valid_lattice_transition): Operate directly on the VECTOR_CST
2115 * ipa-icf.c: Include tree-vector-builder.h.
2116 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
2117 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
2119 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2121 * tree.c (build_vector): Delete.
2122 * tree.h (build_vector): Make static and move into the self-testing
2125 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2127 * vector-builder.h (vector_builder::derived): New const overload.
2128 (vector_builder::elt): New function.
2129 * tree-vector-builder.h (tree_vector_builder::type): New function.
2130 (tree_vector_builder::apply_step): Declare.
2131 * tree-vector-builder.c (tree_vector_builder::apply_step): New
2133 * gimple-fold.h (tree_vector_builder): Declare.
2134 (gimple_build_vector): Take a tree_vector_builder instead of a
2135 type and vector of elements.
2136 * gimple-fold.c (gimple_build_vector): Likewise.
2137 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
2139 (get_initial_defs_for_reduction): Likewise.
2140 (vectorizable_induction): Likewise.
2142 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2144 * tree-vector-builder.h
2145 (tree_vector_builder::new_binary_operation): Declare.
2146 * tree-vector-builder.c
2147 (tree_vector_builder::new_binary_operation): New function.
2148 * fold-const.c (fold_relational_const): Use it.
2149 (const_binop): Likewise. Check that both input vectors have
2150 the same number of elements, thus excluding things like WIDEN_SUM.
2151 Check whether it is possible to operate directly on the encodings
2154 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2156 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
2157 new_unary_operation, operating only on the encoded elements.
2158 (const_unop): Likewise.
2159 (exact_inverse): Likewise.
2160 (distributes_over_addition_p): New function.
2161 (const_binop): Use tree_vector_builder and new_unary_operation
2162 for combinations of VECTOR_CST and INTEGER_CST. Operate only
2163 on the encoded elements unless the encoding is strided and the
2164 operation does not distribute over addition.
2165 (fold_convert_const): Use tree_vector_builder and
2166 new_unary_operation. Operate only on the encoded elements
2167 for truncating integer conversions, or for non-stepped encodings.
2169 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2171 * config/sparc/sparc.c: Include tree-vector-builder.h.
2172 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
2173 * expmed.c: Include tree-vector-builder.h.
2174 (make_tree): Use tree_vector_builder instead of build_vector.
2175 * fold-const.c: Include tree-vector-builder.h.
2176 (const_binop): Use tree_vector_builder instead of build_vector.
2177 (const_unop): Likewise.
2178 (native_interpret_vector): Likewise.
2179 (fold_vec_perm): Likewise.
2180 (fold_ternary_loc): Likewise.
2181 * gimple-fold.c: Include tree-vector-builder.h.
2182 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
2184 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
2185 (simplify_vector_constructor): Use tree_vector_builder instead
2187 * tree-vect-generic.c: Include tree-vector-builder.h.
2188 (add_rshift): Use tree_vector_builder instead of build_vector.
2189 (expand_vector_divmod): Likewise.
2190 (optimize_vector_constructor): Likewise.
2191 * tree-vect-loop.c: Include tree-vector-builder.h.
2192 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
2193 of build_vector. Explicitly use a stepped encoding for
2195 * tree-vect-slp.c: Include tree-vector-builder.h.
2196 (vect_get_constant_vectors): Use tree_vector_builder instead
2198 (vect_transform_slp_perm_load): Likewise.
2199 (vect_schedule_slp_instance): Likewise.
2200 * tree-vect-stmts.c: Include tree-vector-builder.h.
2201 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
2202 (vect_gen_perm_mask_any): Likewise.
2203 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
2204 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
2206 (build_vector_from_val): Likewise. Explicitly use a duplicate
2209 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
2211 * doc/generic.texi (VECTOR_CST): Describe new representation of
2213 * vector-builder.h: New file.
2214 * tree-vector-builder.h: Likewise.
2215 * tree-vector-builder.c: Likewise.
2216 * Makefile.in (OBJS): Add tree-vector-builder.o.
2217 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
2218 * tree-core.h (tree_base): Add a vector_cst field to the u union.
2219 (tree_vector): Change the number of elements to
2220 vector_cst_encoded_nelts.
2221 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
2222 (VECTOR_CST_ELTS): Delete.
2223 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
2224 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
2225 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
2226 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
2227 (VECTOR_CST_ENCODED_ELT): Likewise.
2228 (vector_cst_encoded_nelts): New function.
2229 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2230 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2231 (vector_cst_int_elt, vector_cst_elt): Declare.
2232 * tree.c: Include tree-vector-builder.h.
2233 (tree_code_size): Abort if passed VECTOR_CST.
2234 (tree_size): Update for new VECTOR_CST layout.
2235 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2236 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2237 (build_vector): Use tree_vector_builder.
2238 (vector_cst_int_elt, vector_cst_elt): New functions.
2239 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
2240 encoded elements and then create the vector in the canonical form.
2241 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
2242 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
2243 (tree_c_tests): Call test_vector_cst_patterns.
2244 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
2246 (hash_tree): Likewise.
2247 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
2248 (streamer_write_tree_header): Likewise.
2249 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
2250 (streamer_alloc_tree): Likewise. Update call to make_vector.
2251 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
2253 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2255 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
2256 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
2257 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
2259 2017-12-07 Bin Cheng <bin.cheng@arm.com>
2260 Richard Biener <rguenther@suse.de>
2262 PR tree-optimization/81303
2263 * Makefile.in (gimple-loop-interchange.o): New object file.
2264 * common.opt (floop-interchange): Reuse the option from graphite.
2265 * doc/invoke.texi (-floop-interchange): Ditto. New document for
2266 -floop-interchange and mention it for -O3.
2267 * opts.c (default_options_table): Enable -floop-interchange at -O3.
2268 * gimple-loop-interchange.cc: New file.
2269 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
2270 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
2271 * passes.def (pass_linterchange): New pass.
2272 * timevar.def (TV_LINTERCHANGE): New time var.
2273 * tree-pass.h (make_pass_linterchange): New declaration.
2274 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
2275 interchange. Record IV before/after increment in new parameters.
2276 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
2277 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
2279 (check_reduction_path): ...New function here.
2280 * tree-vectorizer.h (check_reduction_path): New declaration.
2282 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
2285 PR rtl-optimization/80818
2286 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
2287 always early clobbered.
2288 * lra-lives.c (process_bb_lives): Check input hard regs for early
2289 clobbered non-operand hard reg.
2291 2017-12-07 Jakub Jelinek <jakub@redhat.com>
2294 * tree-cfg.c (verify_gimple_assign_binary): Don't require
2295 types_compatible_p, just that TYPE_MODE is the same.
2297 2017-12-07 Martin Sebor <msebor@redhat.com>
2300 * attribs.c (empty_attribute_table): Initialize new member of
2301 struct attribute_spec.
2302 (decl_attributes): Add argument. Handle mutually exclusive
2303 combinations of attributes.
2304 (selftests::test_attribute_exclusions): New function.
2305 (selftests::attribute_c_tests): Ditto.
2306 * attribs.h (decl_attributes): Add default argument.
2307 * selftest.h (attribute_c_tests): Declare.
2308 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
2309 * tree-core.h (attribute_spec::exclusions, exclude): New type and
2311 * doc/extend.texi (Common Function Attributes): Update const and pure.
2312 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
2313 of struct attribute_spec.
2314 * config/arc/arc.c (arc_attribute_table): Same.
2315 * config/arm/arm.c (arm_attribute_table): Same.
2316 * config/avr/avr.c ( avr_attribute_table): Same.
2317 * config/bfin/bfin.c (bfin_attribute_table): Same.
2318 * config/cr16/cr16.c (cr16_attribute_table): Same.
2319 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
2320 * config/h8300/h8300.c (h8300_attribute_table): Same.
2321 * config/i386/i386.c (ix86_attribute_table): Same.
2322 * config/ia64/ia64.c (ia64_attribute_table): Same.
2323 * config/m32c/m32c.c (m32c_attribute_table): Same.
2324 * config/m32r/m32r.c (m32r_attribute_table): Same.
2325 * config/m68k/m68k.c (m68k_attribute_table): Same.
2326 * config/mcore/mcore.c (mcore_attribute_table): Same.
2327 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
2328 * config/mips/mips.c (mips_attribute_table): Same.
2329 * config/msp430/msp430.c (msp430_attribute_table): Same.
2330 * config/nds32/nds32.c (nds32_attribute_table): Same.
2331 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
2332 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
2333 * config/rl78/rl78.c (rl78__attribute_table): Same.
2334 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
2335 * onfig/rx/rx.c (rx_attribute_table): Same.
2336 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
2337 * config/sh/sh.c (sh_attribute_table): Same.
2338 * config/sparc/sparc.c (sparc_attribute_table): Same.
2339 * config/spu/spu.c (spu_attribute_table): Same.
2340 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
2341 * config/v850/v850.c (v850_attribute_table): Same.
2342 * config/visium/visium.c (visium_attribute_table): Same.
2344 2017-12-07 Tamar Christina <tamar.christina@arm.com>
2347 * config/arm/arm.c (INCLUDE_STRING): Define.
2348 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
2349 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
2352 2017-12-07 Michael Matz <matz@suse.de>
2354 Add unroll and jam pass
2356 * gimple-loop-jam.c: New file.
2357 * Makefile.in (OBJS): Add gimple-loop-jam.o.
2358 * common.opt (funroll-and-jam): New option.
2359 * opts.c (default_options_table): Add unroll-and-jam at -O3.
2360 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
2361 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
2362 * passes.def: Add pass_loop_jam.
2363 * timevar.def (TV_LOOP_JAM): Add.
2364 * tree-pass.h (make_pass_loop_jam): Declare.
2365 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
2366 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
2367 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
2368 to flow_loop_tree_node_add.
2369 (duplicate_subloops, copy_loops_to): Append to sibling list.
2370 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
2371 * doc/invoke.texi (-funroll-and-jam): Document new option.
2372 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
2374 2017-12-07 Richard Biener <rguenther@suse.de>
2376 PR tree-optimization/83296
2377 PR tree-optimization/67769
2378 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
2379 flow sensitive info in an unrelated BB.
2380 (value_replacement): Use reset_flow_sensitive_info.
2381 (minmax_replacement): Reset flow sensitive info on the def
2382 we move. Do not reset flow sensitive info in the whole BB
2383 we move the stmt to.
2384 (abs_replacement): Likewise.
2386 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2389 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2390 rs6000_cpu to the given -mcpu=, or to the default processor.
2392 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2394 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
2395 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
2396 instead of rs6000_cpu_attr.
2397 (rs6000_adjust_cost): Ditto.
2398 (is_microcoded_insn): Ditto.
2399 (rs6000_adjust_priority): Ditto.
2400 (rs6000_issue_rate): Ditto.
2401 (rs6000_use_sched_lookahead): Ditto.
2402 (rs6000_use_sched_lookahead_guard): Ditto.
2403 (rs6000_sched_reorder): Ditto.
2404 (force_new_group): Ditto.
2405 * config/rs6000/rs6000.md (cpu attribute): Ditto.
2406 (group_ending_nop): Ditto.
2408 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2410 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
2411 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
2412 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
2413 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
2415 (rs6000_reassociation_width): Ditto.
2416 (rs6000_emit_epilogue): Ditto.
2417 (rs6000_adjust_cost): Ditto.
2418 (is_microcoded_insn): Ditto.
2419 (is_cracked_insn): Ditto.
2420 (rs6000_adjust_priority): Ditto.
2421 (rs6000_sched_reorder): Ditto.
2422 (rs6000_sched_reorder2): Ditto.
2423 (insn_must_be_first_in_group): Ditto.
2424 (insn_must_be_last_in_group): Ditto.
2425 (rs6000_register_move_cost): Ditto.
2426 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
2429 2017-12-07 Julia Koval <julia.koval@intel.com>
2431 * config.gcc: Add vaesintrin.h.
2432 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2434 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2435 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2437 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2438 * config/i386/immintrin.h: Include vaesintrin.h.
2439 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2440 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2441 _mm_aesdec_epi128): New intrinsics.
2443 2017-12-06 David Malcolm <dmalcolm@redhat.com>
2445 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
2446 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
2447 c-family/c-spellcheck.cc.
2448 (best_macro_match::best_macro_match): Likewise.
2450 (struct edit_distance_traits<cpp_hashnode *>): Move to
2451 c-family/c-spellcheck.h.
2452 (class best_macro_match): Likewise.
2454 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2456 PR tree-optimization/83293
2457 * gimple-ssa-strength-reduction.c (insert_initializers): Use
2458 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
2459 might insert into empty bb.
2462 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
2463 simplify for plus with :c added, and pointer_plus without that.
2464 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
2465 with undefined overflow and the conversion is not widening,
2466 perform negation in utype and only convert to type afterwards.
2467 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
2468 simplify for plus with :c added, and pointer_plus without that.
2469 If type is integral with undefined overflow and the conversion is
2470 not widening, perform minus in utype and only convert to type
2471 afterwards. Move the last pointer_diff_expr simplify into the
2474 2017-12-06 Martin Sebor <msebor@redhat.com>
2476 PR tree-optimization/82646
2477 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
2478 strncpy, not maxlen.
2480 2017-12-06 Martin Sebor <msebor@redhat.com>
2482 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
2485 PR tree-optimization/83075
2486 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
2487 strncat/strncpy don't change length of source string.
2489 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
2492 2017-11-29 Martin Aberg <maberg@gaisler.com>
2494 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2495 to prevent b2bst errata sequence.
2496 (sqrtdf2_fix): Likewise.
2498 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2500 PR tree-optimization/81945
2501 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
2502 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
2503 to dest_cfun has orig_loop_num set, either remap it to the new
2504 loop number if the loop got moved too, or clear it.
2506 2017-12-05 Steve Ellcey <sellcey@cavium.com>
2508 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
2510 (thunderx2t99_nothing): New insn reservation.
2511 (thunderx2t99_mrs): New insn reservation.
2512 (thunderx2t99_multiple): New insn reservation.
2513 (thunderx2t99_alu_basi): Add bfx to reservation.
2514 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
2516 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2519 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
2521 2017-12-05 Bin Cheng <bin.cheng@arm.com>
2523 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
2524 tree-ssa-pre.c::remove_dead_inserted_code.
2525 * tree-ssa-dce.h: New file.
2526 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
2527 (remove_dead_inserted_code): Move and rename to function
2528 tree-ssa-dce.c::simple_dce_from_worklist.
2529 (pass_pre::execute): Update use.
2531 2017-12-05 Richard Biener <rguenther@suse.de>
2533 PR tree-optimization/83277
2534 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
2535 to code-gen liveout vars.
2537 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
2539 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
2540 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
2541 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
2544 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
2546 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
2548 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
2549 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
2552 2017-12-05 Richard Biener <rguenther@suse.de>
2554 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
2555 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
2556 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
2557 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
2559 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2561 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
2563 2017-12-05 Martin Liska <mliska@suse.cz>
2564 Jakub Jelinek <jakub@redhat.com>
2566 * doc/invoke.texi: Document the options.
2567 * flag-types.h (enum sanitize_code): Add
2568 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2569 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
2570 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2571 * opts.c: Define new sanitizer options.
2572 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
2573 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
2575 2017-12-05 Julia Koval <julia.koval@intel.com>
2577 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
2578 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
2579 (ix86_handle_option): Handle -mavx512vnni.
2580 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
2581 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2582 * config/i386/i386-c (__AVX512VNNI__): New.
2583 * config/i386/i386.c (ix86_target_string): Handle new option.
2584 (ix86_valid_target_attribute_inner_p): Handle new option.
2585 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
2586 * config/i386/i386.opt (mavx512vnni): New option.
2588 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2591 * athlon.md: Disable for generic.
2592 * haswell.md: Enable for generic.
2593 * i386.c (ix86_sched_init_global): Add core hooks for generic.
2594 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
2596 (ix86_adjust_cost): Move generic to haswell path.
2598 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
2600 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
2601 instead of MEM_P in a couple more places. Fix formatting issues.
2603 2017-12-04 Jim Wilson <jimw@sifive.com>
2605 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
2606 instead of GP_REG_LAST-1.
2607 (riscv_adjust_libcall_cfi_prologue): Likewise.
2608 (riscv_adjust_libcall_cri_epilogue): Likewise.
2609 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
2612 2017-12-04 Luis Machado <luis.machado@linaro.org>
2614 * ipa-pure-const.c (check_decl): Add missing newline.
2615 (state_from_flags): Likewise.
2617 2017-12-04 Jeff Law <law@redhat.com>
2619 PR tree-optimizatin/78496
2620 * gimple-ssa-evrp-analyze.h
2621 (evrp_range_analyzer::get_vr_values): Simplify.
2622 * gimple-ssa-evrp-analyze.c: Corresponding changes.
2623 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
2624 and gimple-ssa-evrp-analyze.h.
2625 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
2626 (simplify_stmt_for_jump_threading): Copy a blob of code from
2627 tree-vrp.c to use ranges to simplify statements.
2628 (dom_opt_dom_walker::before_dom_children): Call
2629 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
2630 (dom_opt_dom_walker::after_dom_children): Similarly for
2631 evrp_range_analyzer::leave.
2632 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
2635 * gimple-ssa-evrp-analyze.c
2636 (evrp_range_analyzer::extract_range_from_stmt): Always use
2637 vr_values::update_value_range so preexisting range info is
2638 medged with new range info, even if the new range is VR_VARYING.
2640 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2642 * combine.c: Adjust comment.
2643 (use_crosses_set_p): Delete.
2644 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
2645 (try_combine): Ditto.
2647 2017-12-04 Richard Biener <rguenther@suse.de>
2649 PR tree-optimization/83255
2650 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
2651 Re-add zero-iteration check.
2653 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2655 PR rtl-optimization/83245
2656 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
2657 hard registers as earlyclobber, also if not in an asm.
2659 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2663 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2666 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2667 rs6000_cpu based on cpu_index, not tune_index.
2669 2017-12-04 Richard Biener <rguenther@suse.de>
2671 PR tree-optimization/83238
2672 * graphite-scop-detection.c (scop_detection::merge_sese): Make
2673 code match comment, rejecting invalid SESE regions.
2675 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
2677 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
2678 require base operand is a REG_POINTER prior to reload on targets
2679 with non-equivalent space registers.
2681 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2683 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
2684 (update_specialized_profile): Fix updating of counts.
2685 (perhaps_add_new_callers): Likewise.
2687 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2690 * x86-tune.def: Remove obsolette FIXMEs.
2691 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
2692 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
2693 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
2695 (X86_TUNE_PAD_RETURNS): Disable for generic.
2697 2017-12-02 Jakub Jelinek <jakub@redhat.com>
2699 PR tree-optimization/83170
2700 PR tree-optimization/83241
2701 * gimple-ssa-store-merging.c
2702 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
2703 gimple_vuse (ins_stmt) in case it has changed.
2704 (imm_store_chain_info::output_merged_store): Likewise.
2706 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
2710 * tree-cfg.c (pass_warn_function_return::execute): Handle
2711 __builtin_ubsan_handle_missing_return like __builtin_unreachable
2712 with BUILTINS_LOCATION.
2716 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
2717 is BLKmode for vector field with vector raw mode, use TYPE_MODE
2718 instead of DECL_MODE.
2720 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
2721 last argument to rtx pointer.
2722 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
2723 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
2724 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
2725 depending on the chosen ISAs.
2726 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
2727 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
2728 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
2730 * config/i386/sse.md (mov<mode>_internal): Likewise.
2731 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
2733 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2735 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
2737 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2739 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
2740 parameters from prototype.
2741 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
2742 parameters. Don't print a comment.
2743 (emit_fusion_gpr_load): Adjust.
2744 (emit_fusion_load_store): Adjust.
2745 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
2746 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
2747 comment on the second line.
2749 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2752 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2753 rs6000_cpu based on cpu_index, not tune_index.
2755 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2757 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
2758 which_alternative instead of which_alternative + 1.
2759 (output_asm_insn): Print an extra tab if the template is short.
2761 2017-12-01 Jim Wilson <jimw@sifive.com>
2763 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
2765 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
2766 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
2768 * doc/tm.texi: Regenerate.
2770 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2773 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
2774 whether we can allocate pseudos before trying to fix an address.
2775 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
2776 memory address is indexed or indirect.
2777 (floatuns_<mode>si2_hw2): Likewise.
2779 2017-12-01 Jason Merrill <jason@redhat.com>
2781 * Makefile.in (TAGS): Add c-family/*.cc.
2783 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
2785 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
2786 (call_value_insn): Likewise.
2787 (sibcall_insn): Likewise.
2788 (sibcall_value_insn): Likewise.
2789 (movsi_aarch64): Likewise.
2790 (movdi_aarch64): Likewise.
2791 (add_losym_): Likewise.
2792 (ldr_got_small_): Likewise.
2793 (ldr_got_small_sidi): Likewise.
2794 (ldr_got_small_28k_): Likewise.
2795 (ldr_got_small_28k_sidi): Likewise.
2796 * config/aarch64/aarch64.c (aarch64_print_address_internal):
2797 Move output_addr_const to symbolic case. Add error check.
2799 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2801 * config/s390/predicates.md (plus16_Q_operand): New predicate.
2802 * config/s390/s390.md: Disable MVC merging peephole if it would
2803 disable operand forwarding.
2804 (new peephole2): Split MVCs if it would turn them into up to 2
2807 2017-12-01 Richard Biener <rguenther@suse.de>
2809 PR tree-optimization/83232
2810 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
2811 detection of same access. Instead of breaking the group here
2812 do not consider the duplicate. Add comment explaining real fix.
2814 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
2816 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
2818 2017-12-01 Sudakshina Das <sudi.das@arm.com>
2820 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
2822 2017-12-01 Jakub Jelinek <jakub@redhat.com>
2824 * function.h (struct function): Remove cilk_frame_decl,
2825 is_cilk_function and calls_cilk_spawn fields.
2826 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
2828 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
2829 cilk_elemental field.
2830 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
2831 * target.def: Adjust comment.
2832 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
2833 Don't test cilk_elemental.
2835 PR tree-optimization/83233
2836 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
2837 bswap_stat name for the struct.
2840 * tree.h (SWITCH_BREAK_LABEL_P): Define.
2841 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
2842 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
2843 SWITCH_BREAK_LABEL_P set on the label.
2844 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
2845 added for default case if it was missing and not all cases covered.
2846 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
2847 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
2849 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
2850 to avoid -Wimplicit-fallthrough warning.
2851 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
2852 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
2854 PR tree-optimization/83221
2855 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
2857 (init_reassoc): Formatting fix.
2860 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
2861 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
2865 * tree-cfg.c: Include asan.h.
2866 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
2868 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2870 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
2872 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2874 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
2875 vpcmpeqd instruction.
2877 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2879 * config/i386/i386.c (standard_sse_constant_opcode): Fix
2880 registers type for 128bit mode.
2882 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2884 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
2886 * gengtype-state.c (read_state_pair): Likewise.
2887 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
2888 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
2889 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
2890 * ifcvt.c (noce_try_store_flag_constants): Likewise.
2891 * tree-ssa-ccp.c (ccp_finalize): Likewise.
2892 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
2893 * builtins.c (fold_builtin_3): Likewise.
2894 * graphite-scop-detection.c
2895 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
2896 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
2899 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
2902 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
2903 parameter. Return true if ignore_decl_rtl_set_p is true and other
2904 conditions are satisfied.
2905 * asan.h (asan_protect_global): Add new parameter.
2906 * varasm.c (categorize_decl_for_section): Pass true as second parameter
2907 to asan_protect_global calls.
2909 2017-11-30 Jim Wilson <jimw@sifive.com>
2911 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
2912 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
2913 -mstrict-align, add info on default value. Delete redundant lines for
2914 -mabi. Add missing -mexplicit-relocs docs.
2916 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2918 * config/arc/arc.md (trap): New pattern.
2920 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2922 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
2923 instruction to end into a delay slot.
2924 * config/arc/arc.md (cond_delay_insn): Check if the instruction
2925 can be placed into a delay slot against reg_note.
2927 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2929 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
2930 labels number of usages.
2932 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2934 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
2936 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
2938 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2941 * internal-fn.c (expand_mul_overflow): Optimize unsigned
2942 multiplication by power of 2 constant into two shifts + comparison.
2944 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
2947 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
2949 2017-11-30 Richard Biener <rguenther@suse.de>
2951 PR tree-optimization/83202
2952 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2953 allow_peel argument and guard peeling.
2954 (canonicalize_loop_induction_variables): Likewise.
2955 (canonicalize_induction_variables): Pass false.
2956 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
2957 peeling from cunrolli.
2959 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2961 * combine.c (try_combine): Print a message to dump file whenever
2962 I0, I1, or I2 cannot be combined into I3.
2964 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2966 PR rtl-optimization/83156
2967 PR rtl-optimization/82621
2968 * combine.c (try_combine): Don't split an I2 if one of the dests is
2969 set again before I3. Allow unused dests.
2971 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2973 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
2975 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
2977 PR rtl-optimization/80818
2978 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
2979 recursively. Use insn code for clobber.
2980 (lra_set_insn_recog_data): Pass the new arg to
2981 collect_non_operand_hard_regs.
2982 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
2984 (lra_update_insn_regno_info): Pass insn to
2985 add_regs_to_insn_regno_info.
2987 2017-11-29 Jim Wilson <jimw@sifive.com>
2988 Andrew Waterman <andrew@sifive.com>
2990 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
2991 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
2992 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
2993 use SINGLE_SHIFT_COST.
2994 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
2995 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
2997 2017-11-29 Julia Koval <julia.koval@intel.com>
2999 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3000 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3001 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3002 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3003 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3004 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3005 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3006 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3007 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3008 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3009 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3010 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3011 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3012 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3013 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3014 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3015 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3016 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3017 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3018 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3019 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3020 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3021 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3022 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3023 __builtin_ia32_vpshldv_v2di_mask,
3024 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3025 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3026 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3028 2017-11-29 Julia Koval <julia.koval@intel.com>
3030 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3031 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3032 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3033 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3034 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3035 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3036 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3037 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3038 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3039 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3040 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3041 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
3042 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
3043 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
3044 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
3045 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
3046 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
3047 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
3048 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3049 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
3050 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
3052 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3054 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
3055 movsi_pic_gotdata_op instruction as a load for the UT699 errata
3058 2017-11-29 Martin Aberg <maberg@gaisler.com>
3060 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3061 to prevent b2bst errata sequence.
3062 (sqrtdf2_fix): Likewise.
3064 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3066 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
3067 (div_sqrt_insn_p): New function.
3068 (sparc_do_work_around_errata): Insert NOP instructions to
3069 prevent sequences that could trigger the TN-0013 errata for
3070 certain LEON3 processors.
3071 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
3072 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
3073 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
3074 (in_branch_delay): Prevent div and sqrt in delay slot if
3076 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
3078 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3080 * config/sparc/sparc.c (atomic_insn_p): New function.
3081 (sparc_do_work_around_errata): Insert NOP instructions to
3082 prevent sequences that could trigger the TN-0010 errata for
3084 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
3085 instruction referable in atomic_insns_p.
3087 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3089 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
3090 (atomic_compare_and_swap_leon3_1): Likewise.
3093 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3095 * config/sparc/sparc.c (fpop_insn_p): New function.
3096 (sparc_do_work_around_errata): Insert NOP instructions to
3097 prevent sequences that could trigger the TN-0012 errata for
3099 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
3100 * config/sparc/sparc.md (fix_gr712rc): New attribute.
3101 (in_branch_annul_delay): Prevent floating-point instructions
3102 in delay slot of annulled integer branch.
3104 2017-11-29 Richard Biener <rguenther@suse.de>
3106 PR tree-optimization/83202
3107 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
3109 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
3110 nodes vectorized to the same stmts multiple times.
3111 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
3112 (vect_analyze_slp_instance): Adjust.
3113 (scalar_stmts_to_slp_tree_map_t): New typedef.
3114 (vect_schedule_slp_instance): Add a map recording the SLP node
3115 representing the vectorized stmts for a set of scalar stmts.
3116 Avoid code-generating redundancies.
3117 (vect_schedule_slp): Allocate map and pass it down.
3119 2017-11-29 Nathan Sidwell <nathan@acm.org>
3122 * tree.c (build_complex_type): Fix canonicalization. Only fill in
3125 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
3127 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
3128 cases for printing LDP/STP memory addresses.
3129 (aarch64_print_address_internal): Renamed from
3130 aarch64_print_operand_address, added parameter, add Pmode check.
3131 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
3132 (aarch64_print_operand_address): Indirect to
3133 aarch64_print_address_internal.
3134 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
3135 'y' operand output specifier.
3137 2017-11-29 Jakub Jelinek <jakub@redhat.com>
3140 * tree.c (build_simple_mem_ref_loc): Handle
3141 get_addr_base_and_unit_offset returning a MEM_REF.
3144 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
3147 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
3150 2017-11-28 David Malcolm <dmalcolm@redhat.com>
3152 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
3154 (layout::move_to_column): Likewise.
3156 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3159 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
3160 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
3162 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3163 Martin Jambor <mjambor@suse.cz>
3166 * tree.h (expr_type_first_operand_type_p): Declare
3167 * tree.c (expr_type_first_operand_type_p): New function.
3168 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
3169 (ipa_value_from_jfunc): Adjust declaration.
3170 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
3171 Use it as result type for arithmetics, unless it is NULL in which case
3172 be more conservative.
3173 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
3174 ipa_get_jf_pass_through_result.
3175 (propagate_vals_across_pass_through): Likewise.
3176 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
3177 is to propagate_vals_across_pass_through.
3178 (propagate_constants_across_call): Pass PARM_TYPE to
3179 propagate_scalar_across_jump_function.
3180 (find_more_scalar_values_for_callers_subset): Pass parameter type to
3181 ipa_value_from_jfunc.
3182 (cgraph_edge_brings_all_scalars_for_node): Likewise.
3183 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
3184 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
3185 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
3186 target_type, pass it to ipa_value_from_jfunc.
3187 (update_indirect_edges_after_inlining): Pass parameter type to
3188 try_make_edge_direct_simple_call.
3190 2017-11-28 Jeff Law <law@redhat.com>
3192 * gimple-ssa-evrp-analyze.c
3193 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
3194 refine ranges if scev_initialized_p returns true.
3195 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
3197 2017-11-28 Julia Koval <julia.koval@intel.com>
3199 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
3200 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
3201 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
3202 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
3203 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
3204 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
3205 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
3206 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
3207 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
3208 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
3209 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
3210 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
3211 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
3212 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
3213 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
3214 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
3215 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
3216 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
3217 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
3218 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
3219 __builtin_ia32_vpshrd_v2di_mask): New builtins.
3220 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
3222 2017-11-28 Julia Koval <julia.koval@intel.com>
3224 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
3225 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
3226 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
3227 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
3228 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
3229 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
3230 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
3231 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
3232 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
3233 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
3234 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
3235 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
3236 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
3237 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
3238 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
3239 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
3240 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
3241 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
3242 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
3243 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
3244 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
3245 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
3246 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
3247 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
3248 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
3249 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
3250 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
3251 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
3252 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
3253 __builtin_ia32_vpshld_v2di_mask): New builtins.
3254 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3255 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
3257 2017-11-28 Richard Biener <rguenther@suse.de>
3259 PR tree-optimization/80776
3260 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
3262 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
3264 (evrp_range_analyzer::record_ranges_from_incoming_edges):
3265 If the incoming edge is an effective fallthru because the other
3266 edge only reaches a __builtin_unreachable () then record ranges
3267 derived from the controlling condition in SSA info.
3268 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
3269 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3271 2017-11-28 Olivier Hainque <hainque@adacore.com>
3273 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
3274 on mingw build hosts.
3276 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3279 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
3281 2017-11-28 Richard Biener <rguenther@suse.de>
3284 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
3285 copies generated from memcpy use a character array as reference
3288 2017-11-28 Julia Koval <julia.koval@intel.com>
3289 Sebastian Peryt <sebastian.peryt@intel.com>
3291 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
3292 c-family/c-cilkplus.o, c-family/array-notation-common.o,
3293 cilk-common.o, cilk.h, cilk-common.c): Remove.
3295 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
3296 * builtins.c (is_builtin_name): Remove cilkplus condition.
3297 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
3298 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
3299 cilk-builtins.def, cilkplus.def): Remove.
3300 * cif-code.def (CILK_SPAWN): Remove.
3301 * cilk-builtins.def: Delete.
3302 * cilk-common.c: Ditto.
3304 * cilkplus.def: Ditto.
3305 * config/darwin.h (fcilkplus): Delete.
3306 * cppbuiltin.c: Ditto.
3307 * doc/extend.texi: Remove cilkplus doc.
3308 * doc/generic.texi: Ditto.
3309 * doc/invoke.texi: Ditto.
3310 * doc/passes.texi: Ditto.
3311 * gcc.c (fcilkplus): Remove.
3312 * gengtype.c (cilk.h): Remove.
3313 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
3315 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
3317 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
3319 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
3320 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
3321 cilkplus conditions.
3322 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
3323 inline_read_section): Ditto.
3324 * ipa-inline-analysis.c (cilk.h): Remove.
3325 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
3326 * lto-wrapper.c (merge_and_complain, append_compiler_options,
3327 append_linker_options): Remove condition for fcilkplus.
3328 * lto/lto-lang.c (cilk.h): Remove.
3329 (lto_init): Remove condition for fcilkplus.
3330 * omp-expand.c (expand_cilk_for_call): Delete.
3331 (expand_omp_taskreg, expand_omp_for_static_chunk,
3332 expand_omp_for): Remove cilkplus
3334 (expand_cilk_for): Delete.
3335 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
3336 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
3337 execute_lower_omp, diagnose_sb_0): Ditto.
3338 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
3339 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3340 * tree-nested.c: Ditto.
3341 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
3342 (dump_generic_node): Ditto.
3343 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3344 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
3346 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
3348 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3350 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
3351 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
3352 for vector mode and !TARGET_SIMD.
3354 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3356 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
3358 * tree.h (SWITCH_LABELS): Remove.
3359 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
3360 assert SWITCH_BODY is non-NULL.
3361 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
3363 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
3365 PR tree-optimization/80788
3366 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
3367 has TREE_OVERFLOW set, call drop_tree_overflow.
3369 2017-11-28 Richard Biener <rguenther@suse.de>
3371 PR tree-optimization/83158
3372 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
3374 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
3377 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
3378 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
3380 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
3382 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
3383 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
3385 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3388 * varasm.c (compare_constant): Take the mode of the constants into
3389 account when comparing floating point constants.
3391 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
3393 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
3394 from explicit instantiation of debug_helper.
3395 * vec.h (DEFINE_DEBUG_VEC): Ditto.
3397 2017-11-27 Richard Biener <rguenther@suse.de>
3399 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
3402 2017-11-27 Richard Biener <rguenther@suse.de>
3404 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
3405 (type_hash_canon): Also clear int_cst_hash_table entry for
3407 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
3409 2017-11-27 Tamar Christina <tamar.christina@arm.com>
3411 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3412 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3413 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3415 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3417 * hash-map.h (gt_cleare_cache): Avoid UB.
3419 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
3421 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
3422 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
3423 (decide_unroll_runtime_iterations): Likewise.
3424 (decide_unroll_stupid): Likewise.
3426 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3429 * config/i386/i386.md: Add a loop with incssp.
3431 2017-11-27 Martin Jambor <mjambor@suse.cz>
3433 PR tree-optimization/81248
3434 * tree-sra.c (splice_param_accesses): Remove size check.
3435 (decide_one_param_reduction): Fix size check.
3436 * gimple-pretty-print.c (dump_profile): Silence warning.
3437 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
3439 2017-11-27 Jakub Jelinek <jakub@redhat.com>
3442 * dbxout.c (lastlineno): New variable.
3443 (dbx_debug_hooks): Use dbxout_switch_text_section as
3444 switch_text_section debug hook.
3445 (dbxout_function_end): Switch to current_function_section
3446 rather than function_section. If crtl->has_bb_partition,
3447 output just one N_FUN, depending on in_cold_section_p.
3448 (dbxout_source_line): Remember last lineno in lastlineno.
3449 (dbxout_switch_text_section): New function.
3450 (dbxout_function_decl): Adjust dbxout_block caller.
3451 (dbx_block_with_cold_children): New function.
3452 (dbxout_block): Return true if any LBRAC/RBRAC have been
3453 emitted. Use dbx_block_with_cold_children at depth == 0
3454 in second partition. Add PARENT_BLOCKNUM argument, pass
3455 it optionally adjusted to children. Output LBRAC/RBRAC
3456 around recursive call only if the block is in the current
3457 partition, if not and anything was output, emit empty
3459 * final.c (final_scan_insn): Compute cold_function_name
3460 before calling switch_text_section debug hook. Call
3461 that hook even if dwarf2out_do_frame if not emitting
3465 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
3466 TREE_READONLY decls.
3468 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3470 PR rtl-optimization/82488
3471 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
3473 2017-11-26 Julia Koval <julia.koval@intel.com>
3475 * config/i386/i386.c (processor_target_table): Add skylake_cost for
3477 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
3480 2017-11-26 Julia Koval <julia.koval@intel.com>
3482 * config/i386/driver-i386.c (host_detect_local_cpu):
3483 Detect skylake-avx512.
3485 2017-11-26 Julia Koval <julia.koval@intel.com>
3487 * config.gcc: Add -march=cannonlake.
3488 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
3489 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
3490 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
3491 (PTA_CANNONLAKE): New.
3492 (processor_target_table): Add cannonlake.
3493 (ix86_option_override_internal): Ditto.
3494 (fold_builtin_cpu): Ditto.
3495 (get_builtin_code_for_version): Handle cannonlake.
3496 (M_INTEL_COREI7_CANNONLAKE): New.
3497 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
3498 * doc/invoke.texi: Add -march=cannonlake.
3500 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
3502 * plugin.c (add_new_plugin): Use platform-specific library extensions.
3503 (try_init_one_plugin): Alternative implementation for MinGW.
3504 * Makefile.in (plugin_implib): New.
3505 (gengtype-lex.c): Fix broken AIX workaround.
3506 * configure: Regenerate.
3507 * doc/plugins.texi: Document support for MinGW.
3509 2017-11-25 Jakub Jelinek <jakub@redhat.com>
3511 PR rtl-optimization/81553
3512 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
3513 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
3514 is a shift where C1 has different mode than the whole shift, use C1's
3515 mode for MULT rather than the shift's mode.
3518 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
3519 builtins not enabled in the currently selected ISA.
3521 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
3523 PR tree-optimization/71026
3524 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
3525 (insert_reciprocals): Change to insert reciprocals before a division
3526 by a square and to insert the square of a reciprocal.
3527 (execute_cse_reciprocals_1): Change to consider division by a square.
3528 (register_division_in): Add importance parameter.
3530 2017-11-24 Richard Biener <rguenther@suse.de>
3532 PR tree-optimization/82402
3533 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
3534 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
3536 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
3538 * match.pd (0-ptr): New transformation.
3540 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
3543 * ipa-inline.c (inline_small_functions): Set current badnes correctly
3544 when skipping checking.
3546 2017-11-24 Richard Biener <rguenther@suse.de>
3548 PR tree-optimization/83128
3549 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
3550 (vn_reference_lookup_3): Likewise.
3552 2017-11-24 Jakub Jelinek <jakub@redhat.com>
3555 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
3556 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
3559 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
3560 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
3561 check for the latter with gimple_call_builtin_p. Do not handle
3562 BUILT_IN_STPNCPY_CHK which is not a pass through call.
3564 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
3566 * config/arm/arm_neon.h: Fix pragma GCC push_options before
3569 2017-11-23 Julia Koval <julia.koval@intel.com>
3571 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
3572 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
3573 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
3574 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
3575 _mm512_maskz_expandloadu_epi16): New intrinsics.
3576 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
3577 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
3578 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
3579 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
3580 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
3581 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
3582 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
3583 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
3584 _mm256_maskz_expandloadu_epi8): New intrinsics.
3585 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
3586 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
3587 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
3588 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
3589 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
3590 * config/i386/sse.md (VI248_VLBW): New iterator.
3591 (expand<mode>_mask, expand<mode>_maskz): New patterns.
3593 2017-11-23 Julia Koval <julia.koval@intel.com>
3595 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
3596 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
3597 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
3598 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
3599 _mm512_mask_compressstoreu_epi16): New.
3600 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
3601 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
3602 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
3603 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
3604 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
3605 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
3606 _mm256_mask_compressstoreu_epi8): New.
3607 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
3608 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
3609 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
3610 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
3611 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
3612 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
3613 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
3614 __builtin_ia32_compresshi128_mask,
3615 __builtin_ia32_compressstoreuqi512_mask,
3616 __builtin_ia32_compressstoreuhi512_mask,
3617 __builtin_ia32_compressstoreuqi256_mask,
3618 __builtin_ia32_compressstoreuqi128_mask,
3619 __builtin_ia32_compressstoreuhi256_mask,
3620 __builtin_ia32_compressstoreuhi128_mask): New builtins.
3621 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
3623 (ix86_expand_special_args_builtin): Handle new types.
3624 (s4fma_expand): Handle new builtin array.
3625 * config/i386/immintrin.h: Include new headers.
3626 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
3627 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
3629 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3631 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
3632 fixes. Declare temp and g variables at the top in order to avoid
3633 {} in most of the cases.
3635 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
3637 * match.pd (ptr-0): New transformation.
3639 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3641 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
3642 (struct addr_mode_cost_table): New.
3643 (struct tune_params): Add field addr_mode_costs.
3644 * config/arm/arm.c (generic_addr_mode_costs): New.
3645 (arm_slowmul_tune): Initialise addr_mode_costs field.
3646 (arm_fastmul_tune): Likewise.
3647 (arm_strongarm_tune): Likewise.
3648 (arm_xscale_tune): Likewise.
3649 (arm_9e_tune): Likewise.
3650 (arm_marvell_pj4_tune): Likewise.
3651 (arm_v6t2_tune): Likewise.
3652 (arm_cortex_tune): Likewise.
3653 (arm_cortex_a8_tune): Likewise.
3654 (arm_cortex_a7_tune): Likewise.
3655 (arm_cortex_a15_tune): Likewise.
3656 (arm_cortex_a35_tune): Likewise.
3657 (arm_cortex_a53_tune): Likewise.
3658 (arm_cortex_a57_tune): Likewise.
3659 (arm_exynosm1_tune): Likewise.
3660 (arm_xgene1_tune): Likewise.
3661 (arm_cortex_a5_tune): Likewise.
3662 (arm_cortex_a9_tune): Likewise.
3663 (arm_cortex_a12_tune): Likewise.
3664 (arm_cortex_a73_tune): Likewise.
3665 (arm_v7m_tune): Likewise.
3666 (arm_cortex_m7_tune): Likewise.
3667 (arm_v6m_tune): Likewise.
3668 (arm_fa726te_tune): Likewise.
3669 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
3671 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3673 * config/arm/arm.c (arm_mem_costs): New function.
3674 (arm_rtx_costs_internal): Use arm_mem_costs.
3676 2017-11-23 Mark Wielaard <mark@klomp.org>
3678 * dwarf2out.c (init_sections_and_labels): Use generation to create
3679 unique ranges_section_label and ranges_base_label. Return generation.
3680 (output_rnglists): Add generation argument. Use generation to create
3681 unique ranges labels.
3682 (dwarf2out_finish): Get generation from init_sections_and_labels
3683 and pass generation to output_rnglists.
3685 2017-11-23 Mike Stump <mikestump@comcast.net>
3686 Eric Botcazou <ebotcazou@adacore.com>
3688 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
3689 * cfgloop.h (struct loop): Add unroll field.
3690 * function.h (struct function): Add has_unroll bitfield.
3691 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
3692 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
3693 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
3694 (pass_rtl_unroll_loops::gate): Likewise.
3695 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
3696 for which loop->unroll==1.
3697 (decide_unroll_constant_iterations): Use note for consistency and
3698 take loop->unroll into account. Return early if loop->unroll is set.
3699 Fix thinko in existing test.
3700 (decide_unroll_runtime_iterations): Use note for consistency and
3701 take loop->unroll into account.
3702 (decide_unroll_stupid): Likewise.
3703 * lto-streamer-in.c (input_cfg): Read loop->unroll.
3704 * lto-streamer-out.c (output_cfg): Write loop->unroll.
3705 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
3707 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
3708 (print_loop): Print loop->unroll if set.
3709 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
3710 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
3711 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
3713 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
3714 loop->unroll is set and smaller than the trip count. Otherwise bypass
3715 entirely the heuristics if loop->unroll is set. Remove dead note.
3716 Fix off-by-one bug in other note.
3717 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
3718 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
3720 (tree_unroll_loops_completely): Make static.
3721 (pass_complete_unroll::execute): Use correct type for variable.
3722 (pass_complete_unrolli::execute): Fix formatting.
3723 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
3725 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
3727 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
3728 enable when TARGET_PREFER_AVX128 is set.
3730 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3732 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
3734 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
3735 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
3737 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3739 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
3740 frequencies. Fix estimation of aggregate parameters.
3742 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3744 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
3747 2017-11-23 Tom de Vries <tom@codesourcery.com>
3749 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
3751 2017-11-23 Tom de Vries <tom@codesourcery.com>
3753 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
3756 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
3759 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
3760 sibcall_value_pcrel_fdpic): Use local variable instead of
3762 (calli_tbr_rel): Add missing operand 2.
3763 (call_valuei_tbr_rel): Add missing operand 3.
3765 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3768 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
3769 bitpos/bitsize covering the whole destination, use store_expr only if
3770 the complex mode is the same. Otherwise, use expand_normal and if
3771 it returns CONCAT, subreg each part separately instead of trying to
3772 subreg the whole result.
3774 2017-11-23 Richard Biener <rguenther@suse.de>
3776 PR tree-optimization/23094
3777 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
3778 come here from walking over backedges in the first iteration.
3779 (vn_reference_lookup_3): Skip clobbers that store the same value.
3781 2017-11-23 Richard Biener <rguenther@suse.de>
3783 PR tree-optimization/81403
3784 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
3785 a block we need a leader relative to.
3786 (phi_translate_1): For nary processing require a leader from
3787 get_representative_for given we run expression simplification
3788 using match-and-simplify. Remove previous fix.
3790 2017-11-22 Jeff Law <law@redhat.com>
3792 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3793 Use new method allocate_value_range rather than accessing the
3794 vrp_value_range_pool data member directly.
3795 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
3796 to use extract_range_from_stmt method to avoid need for
3797 extract_range_from_assignment method.
3798 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
3799 method rather than setting values_propgated data member directly.
3800 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
3801 and values propagated data members and extract_range_from_assignment
3802 method. Reorder private data members to conform to standards.
3803 Add new methods set_lattice_propagation_complete and
3804 allocate_value_range.
3806 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
3808 PR rtl-optimization/83030
3809 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
3810 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
3811 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
3812 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
3813 CROSSING_JUMP_P flag is already set before setting it.
3814 * cfgrtl.c (fixup_partition_crossing): Likewise.
3815 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
3818 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3820 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
3821 Handle the case where both arguments are using gen_const_vec_series.
3823 2017-11-22 David Malcolm <dmalcolm@redhat.com>
3826 * pretty-print.c (pp_format): Move quoting implementation to
3827 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
3828 to pass address of "quote" local.
3829 (pp_begin_quote): New function.
3830 (pp_end_quote): New function.
3831 * pretty-print.h (printer_fn): Convert penultimate param from bool
3833 (pp_begin_quote): New decl.
3834 (pp_end_quote): New decl.
3835 * tree-diagnostic.c (default_tree_printer): Convert penultimate
3836 param from bool to bool *.
3837 * tree-diagnostic.h (default_tree_printer): Likewise.
3839 2017-11-22 Jeff Law <law@redhat.com>
3841 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
3842 Initialize vr_values.
3843 (evrp_range_analyzer::try_find_new_range): Call methods attached to
3844 vr_values via vr_values class instance rather than delegators.
3845 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
3846 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3847 (evrp_range_analyzer::push_value_range): Likewise.
3848 (evrp_range_analyzer::pop_value_range): Likewise.
3849 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
3850 most delegators. Those remaining are exposed as public interfaces.
3851 Make vr_values a pointer and private.
3852 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
3854 (evrp_range_analyzer::get_vr_value): New method.
3855 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
3856 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
3857 (class evrp_dom_walker): Attach evrp_folder class, initialize
3858 it in the ctor. Remove temporary delegators.
3859 (evrp_dom_walker::before_dom_children): Call methods in attached
3860 evrp_range_analyzer class via class instance pointer. Use
3861 free value_range_constant_singleton to remove need for
3862 op_with_constant_singleton_value delegator method. Do not
3863 create a vrp_prop class instance for every call! Narrow
3864 scope of a couple variables.
3865 (evrp_dom_walker::cleanup): Call methods in attached
3866 evrp_range_analyzer class via class instance pointer.
3867 * vr-values.h (class vr_values): Privatize many methods and
3870 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3872 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
3873 * cfgexpand.c (expand_debug_expr): Remove handling for them.
3874 * expr.c (expand_expr_real_2): Likewise.
3875 * fold-const.c (const_unop): Likewise.
3876 * optabs-tree.c (optab_for_tree_code): Likewise.
3877 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3878 * tree-inline.c (estimate_operator_cost): Likewise.
3879 * tree-pretty-print.c (dump_generic_node): Likewise.
3880 (op_code_prio): Likewise.
3881 (op_symbol_code): Likewise.
3882 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
3883 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
3884 * internal-fn.c (direct_internal_fn_optab): New function.
3885 (direct_internal_fn_array, direct_internal_fn_supported_p
3886 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
3887 * fold-const-call.c (fold_const_reduction): New function.
3888 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
3890 * tree-vect-loop.c: Include internal-fn.h.
3891 (reduction_code_for_scalar_code): Rename to...
3892 (reduction_fn_for_scalar_code): ...this and return an internal
3894 (vect_model_reduction_cost): Take an internal_fn rather than
3896 (vect_create_epilog_for_reduction): Likewise. Build calls rather
3898 (vectorizable_reduction): Use internal functions rather than tree
3899 codes for the reduction operation. Update calls to the functions
3901 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
3902 Use calls to internal functions rather than REDUC tree codes.
3903 * config/aarch64/aarch64-simd.md: Update comment accordingly.
3905 2017-11-22 Olivier Hainque <hainque@adacore.com>
3907 * config/vxworks.c (vxworks_override_options): Pick default
3908 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
3909 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
3910 DWARF_GNAT_ENCODINGS_DEFAULT.
3911 * config/vxworksae.h: Likewise.
3913 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
3915 PR tree-optimization/83104
3916 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
3919 2017-11-22 Marek Polacek <polacek@redhat.com>
3920 H.J. Lu <hongjiu.lu@intel.com>
3921 Jason Merrill <jason@redhat.com>
3926 * calls.c (initialize_argument_information): Call
3927 warn_parameter_passing_abi target hook.
3928 (store_one_arg): Use 0 for empty record size. Don't push 0 size
3929 argument onto stack.
3930 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
3931 * common.opt: Update -fabi-version description.
3932 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
3933 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
3935 (ix86_is_empty_record): New function.
3936 (ix86_warn_parameter_passing_abi): New function.
3937 (TARGET_EMPTY_RECORD_P): Redefine.
3938 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
3939 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
3940 * doc/tm.texi: Regenerated.
3941 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
3942 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
3943 * dwarf2out.c (get_ultimate_context): Move to tree.c.
3944 * explow.c (hard_function_value): Call arg_int_size_in_bytes
3945 instead of int_size_in_bytes.
3946 * expr.c (copy_blkmode_to_reg): Likewise.
3947 * function.c (aggregate_value_p): Return 0 for empty types.
3948 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
3950 (locate_and_pad_parm): Call arg size_in_bytes instead
3952 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
3953 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
3954 * target.def (empty_record_p, warn_parameter_passing_abi): New target
3956 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
3957 (std_gimplify_va_arg_expr): Skip empty records. Call
3958 arg_size_in_bytes instead size_in_bytes.
3959 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
3960 * tree-core.h (tree_type_common): Add empty_flag.
3961 (tree_decl_common): Update comments.
3962 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
3964 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3965 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
3967 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3968 * tree.c (default_is_empty_type): New function.
3969 (default_is_empty_record): New function.
3970 (arg_int_size_in_bytes): New function.
3971 (arg_size_in_bytes): New function.
3972 (get_ultimate_context): New function.
3973 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
3974 TRANSLATION_UNIT_WARN_EMPTY_P.
3975 (default_is_empty_record, arg_int_size_in_bytes,
3976 arg_size_in_bytes, get_ultimate_context): Declare.
3978 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3980 * config/arm/arm.c (cmse_clear_registers): New function.
3981 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
3982 code by call to cmse_clear_registers.
3983 (cmse_nonsecure_entry_clear_before_return): Likewise.
3985 2017-11-22 Tamar Christina <tamar.christina@arm.com>
3987 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
3988 (vdot_s32, vdotq_s32): New.
3989 (vdot_lane_u32, vdotq_lane_u32): New.
3990 (vdot_lane_s32, vdotq_lane_s32): New.
3993 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3996 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
3997 for unsigned values with fewer HWIs than the precision.
3998 (test_overflow): New function.
3999 (wide_int_cc_tests): Call it.
4001 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4003 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4004 has the same mode class as Pmode.
4006 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4008 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4009 padding_bits_to_clear_ptr.
4010 (cmse_nonsecure_entry_clear_before_return): Likewise.
4012 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4014 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4015 auto_sbitap instead of integer bitfield to control register needing
4018 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4020 PR tree-optimization/83044
4021 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4022 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4023 that need the upper bound. Subtract offset from
4024 get_addr_base_and_unit_offset only if positive and subtract it
4025 before division by eltsize rather than after it.
4028 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4029 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4030 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4034 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4035 expand_binop with *mul_widen_optab, make sure at least one of the
4036 operands doesn't have VOIDmode.
4039 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4041 PR rtl-optimization/82044
4042 PR tree-optimization/82042
4043 * dse.c (record_store): Check for overflow.
4044 (check_mem_read_rtx): Properly check for overflow if width == -1, call
4045 add_wild_read instead of clear_rhs_from_active_local_stores on
4046 overflow and log it into dump_file.
4048 2017-11-22 Richard Biener <rguenther@suse.de>
4050 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
4051 fake edges to exit when looking for a place to insert.
4052 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
4054 (insert_into_preds_of_block): Commit edge insertion immediately,
4055 assert that doesn't require new BBs.
4056 (fini_pre): Release expressions.
4057 (pass_pre::execute): Shuffle things around a bit, if the fn
4058 is too large do not compute AVAIL either as this is really the
4061 2017-11-22 Richard Biener <rguenther@suse.de>
4063 PR tree-optimization/83089
4064 * tree-if-conv.c (pass_if_conversion::execute): If anything
4065 changed reset SCEV and free the number of iteration estimates.
4067 2017-11-21 Martin Sebor <msebor@redhat.com>
4069 PR tree-optimization/82945
4070 * calls.h (warn_nonstring_bound): Remove unused function.
4072 2017-11-21 Martin Sebor <msebor@redhat.com>
4074 PR tree-optimization/82945
4075 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
4076 * calls.h (maybe_warn_nonstring_arg): Declare new function.
4077 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
4079 (initialize_argument_information): Call maybe_warn_nonstring_arg.
4080 * calls.h (get_attr_nonstring_decl): Declare new function.
4081 * doc/extend.texi (attribute nonstring): Update.
4082 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
4083 get_attr_nonstring_decl and handle it.
4084 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
4085 detection of nul-termination.
4086 (strlen_to_stridx): Change to a pointer.
4087 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
4088 (pass_strlen::execute): Same.
4090 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
4092 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
4093 for the new option -mprefer-vector-width=[none|128|256|512].
4094 * config/i386/i386.c (ix86_target_string): remove old style options
4095 -mprefer-avx256 and make -mprefer-avx128 as alias.
4096 (ix86_option_override_internal): Apply defaults for the
4097 -mprefer-vector-width=[128|256] option.
4098 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
4099 Implement macros to work with -mprefer-vector-width=.
4100 * config/i386/i386.opt: Implemented option
4101 -mprefer-vector-width=[none|128|256|512].
4102 * doc/invoke.texi: Documentation for
4103 -mprefer-vector-width=[none|128|256|512].
4105 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
4107 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
4109 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
4111 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
4113 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
4115 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
4117 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
4118 * expr.c (expand_expr_real_2): Likewise.
4119 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
4120 fold_binary_loc): Likewise.
4121 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
4122 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
4123 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
4124 MINUS_EXPR transformations.
4125 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
4126 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
4127 * tree-inline.c (estimate_operator_cost): Likewise.
4128 * tree-pretty-print.c (dump_generic_node, op_code_prio,
4129 op_symbol_code): Likewise.
4130 * tree-vect-stmts.c (vectorizable_operation): Likewise.
4131 * vr-values.c (extract_range_from_binary_expr): Likewise.
4132 * varasm.c (initializer_constant_valid_p_1): Likewise.
4133 * tree.def: New tree code POINTER_DIFF_EXPR.
4135 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
4137 * config/i386/i386.md (*bswap<mode>2_movbe): Add
4138 integer suffix to movbe mnemonic.
4139 (*bswaphi2_movbe): Ditto.
4140 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
4142 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4145 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
4146 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
4147 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
4149 2017-11-21 Martin Liska <mliska@suse.cz>
4151 * tree-inline.c (expand_call_inline): Remove not needed
4154 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
4155 Jakub Jelinek <jakub@redhat.com>
4158 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
4159 Remove static keyword from f variable.
4161 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4163 PR tree-optimization/83086
4164 * gimple-ssa-store-merging.c
4165 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
4166 rather than n.base_addr.
4168 2017-11-21 Martin Liska <mliska@suse.cz>
4170 PR rtl-optimization/82044
4171 PR tree-optimization/82042
4172 * dse.c (check_mem_read_rtx): Check for overflow.
4174 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4176 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
4179 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4181 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
4182 and its corresponding call arg location note.
4184 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4185 Andrew Burgess <andrew.burgess@embecosm.com>
4187 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
4189 (arc_return_slot_offset): Likewise.
4190 (arc_eh_return_address_location): New declaration.
4191 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
4192 (MUST_SAVE_REGISTER): Add exception handler case.
4193 (MUST_SAVE_RETURN_ADDR): Likewise.
4194 (arc_frame_pointer_required): Likewise.
4195 (arc_frame_pointer_needed): New function.
4196 (arc_compute_frame_size): Changed.
4197 (arc_expand_prologue): Likewise.
4198 (arc_expand_epilogue): Likewise.
4199 (arc_initial_elimination_offset): Likewise.
4200 (arc_return_slot_offset): Delete.
4201 (arc_eh_return_address_location): New function.
4202 (arc_builtin_setjmp_frame_value): Likewise.
4203 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
4204 (EH_RETURN_STACKADJ_RTX): Define.
4205 (EH_RETURN_HANDLER_RTX): Likewise.
4206 * config/arc/arc.md (eh_return): Delete.
4208 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4210 * print-tree.h (debug_vec_tree): Remove prototype.
4211 * gdbinit.in (pvt): Remove macro.
4213 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4215 PR tree-optimization/83047
4216 * gimple-ssa-store-merging.c
4217 (imm_store_chain_info::output_merged_store): If the loads with the
4218 same vuse are in different basic blocks, for load_gsi pick a load
4219 location that is dominated by the other loads.
4222 * config/i386/i386.c (ix86_memmodel_check): Start
4223 -Winvalid-memory-model diagnostics with lowercase letter.
4226 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
4227 set in_cold_section_p to the partition containing loc_list->first.
4228 When seeing loc_list->last_before_switch node, update secname and
4229 perform range_across_switch second partition handling only after that.
4232 * run-rtl-passes.c: Include debug.h.
4233 (run_rtl_passes): Call debug_hooks->assembly_start.
4234 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
4238 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
4239 OPTAB_DIRECT in calls to expand_simple_binop.
4241 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4244 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
4246 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4249 * spellcheck.h (best_match::blithely_get_best_candidate): New
4252 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4254 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
4255 with lower case letter.
4257 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
4259 * config/i386/i386.md (bswaphi2): New expander.
4260 (*bswaphi2_movbe): New insn pattern.
4261 (bswaphi -> rorhi pepehole2): New peephole pattern.
4263 2017-11-20 Jeff Law <law@redhat.com>
4265 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
4266 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
4267 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
4268 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
4269 gimple-ssa-evrp-analyze.h.
4271 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
4272 set BB_VISITED here.
4273 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
4275 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4277 * tree-predcom.c: Add general comment on Store-Store chains.
4278 (split_data_refs_to_components): Postpone clearing eliminate_store_p
4280 (get_chain_last_ref_at): Rename into...
4281 (get_chain_last_write_at): ...this.
4282 (get_chain_last_write_before_load): New function.
4283 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
4284 CT_STORE_STORE when write reference is added.
4285 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
4286 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
4287 (initialize_root_vars_store_elim_1): Ditto.
4288 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
4289 definition is created.
4290 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
4291 chain by replacing it with dominant stored value.
4293 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4295 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
4297 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
4299 * vr-values.c (extract_range_from_binary_expr): Use a full range
4302 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
4304 * config/arm/arm.md (R4_REGNUM): Define constant.
4305 (nonsecure_call_internal): Remove r4 clobber.
4306 (nonsecure_call_value_internal): Likewise.
4307 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
4308 clobber and resequence match_operands.
4309 (nonsecure_call_value_reg_thumb1_v5): Likewise.
4310 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
4311 (nonsecure_call_value_reg_thumb2): Likewise.
4313 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4315 PR tree-optimization/78821
4316 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
4317 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
4318 to the address of the base rather than the base itself.
4319 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
4320 (find_bswap_or_nop_finalize): New function.
4321 (find_bswap_or_nop): Use it.
4322 (bswap_replace): Return a tree rather than bool, change first
4323 argument from gimple * to gimple_stmt_iterator, allow inserting
4324 into an empty sequence, allow ins_stmt to be NULL - then emit
4325 all stmts into gsi. Fix up MEM_REF address gimplification.
4326 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
4327 (struct store_immediate_info): Add N and INS_STMT non-static
4329 (store_immediate_info::store_immediate_info): Initialize them
4330 from newly added ctor args.
4331 (merged_store_group::apply_stores): Formatting fixes. Sort by
4333 (stmts_may_clobber_ref_p): For stores call also
4334 refs_anti_dependent_p.
4335 (gather_bswap_load_refs): New function.
4336 (imm_store_chain_info::try_coalesce_bswap): New method.
4337 (imm_store_chain_info::coalesce_immediate_stores): Use it.
4338 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
4339 (imm_store_chain_info::output_merged_store): Fail if number of
4340 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
4341 and NOP_EXPR rhs_code.
4342 (pass_store_merging::process_store): Compute n and ins_stmt, if
4343 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
4344 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
4346 (pass_store_merging::execute): Calculate dominators.
4348 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
4349 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
4350 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
4351 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4352 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
4353 class pass_optimize_bswap, bswap_replace,
4354 pass_optimize_bswap::execute): Moved to ...
4355 * gimple-ssa-store-merging.c: ... this file.
4356 Include optabs-tree.h.
4357 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
4358 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4359 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
4360 anonymous namespace, remove static keywords.
4361 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
4362 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
4364 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
4367 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
4369 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
4371 * vec.h (debug_helper): New function.
4372 (DEFINE_DEBUG_VEC): New macro.
4373 * hash-set.h (debug_helper): New function.
4374 (DEFINE_DEBUG_HASH_SET): New macro.
4375 * cfg.c (debug_slim (edge)): New function.
4376 Call DEFINE_DEBUG_VEC for edges.
4377 Call DEFINE_DEBUG_HASH_SET for edges.
4378 * cfghooks.c (debug_slim (basic_block)): New function.
4379 Call DEFINE_DEBUG_VEC for basic blocks.
4380 Call DEFINE_DEBUG_HASH_SET for basic blocks.
4381 * print-tree.c (debug_slim): New function to handle trees.
4382 Call DEFINE_DEBUG_VEC for trees.
4383 Call DEFINE_DEBUG_HASH_SET for trees.
4384 (debug (vec<tree, va_gc>) &): Remove.
4385 (debug (<vec<tree, va_gc>) *): Remove.
4386 * print-rtl.c (debug_slim): New function to handle const_rtx.
4387 Call DEFINE_DEBUG_VEC for rtx_def.
4388 Call DEFINE_DEBUG_VEC for rtx_insn.
4389 Call DEFINE_DEBUG_HASH_SET for rtx_def.
4390 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
4391 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
4392 (debug (vec<rtx_insn *> *ptr): Remove.
4393 (debug_insn_vector): Remove.
4394 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
4396 2017-11-20 Tom de Vries <tom@codesourcery.com>
4398 PR rtl-optimization/82020
4399 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
4400 IF_THEN_ELSE condition.
4402 2017-11-19 Jeff Law <law@redhat.com>
4404 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
4405 of degenerates resulting from ignoring an edge.
4407 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4410 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
4412 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4415 * profile-count.c (profile_count::to_sreal_scale): Fix return value
4416 for uninitialied counts.
4418 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4421 * tree-inline.c (estimate_num_insns): Set to 1 at least.
4423 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4426 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
4428 2017-11-19 Tom de Vries <tom@codesourcery.com>
4430 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
4433 2017-11-19 Tom de Vries <tom@codesourcery.com>
4435 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
4438 2017-11-19 Tom de Vries <tom@codesourcery.com>
4440 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4443 2017-11-19 Tom de Vries <tom@codesourcery.com>
4445 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
4447 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
4449 2017-11-19 Tom de Vries <tom@codesourcery.com>
4451 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
4453 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
4454 semicolon after MCORE_EXPORT_NAME call.
4456 2017-11-19 Tom de Vries <tom@codesourcery.com>
4459 * vmsdbgout.c (vmsdbgout_early_finish): New function.
4460 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
4462 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
4464 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
4465 (cgraph_edge::clone): Cleanup updating of profile.
4466 * ipa-cp.c (update_profiling_info): Likewise.
4467 * ipa-inline-transform.c (inline_transform): Likewise.
4468 * ipa-inline.c (inline_small_functions): Add missing space to dump.
4469 * ipa-split.c (execute_split_functions): Do not split when function
4471 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
4472 * profile-count.c (profile_count::dump): Add global0.
4473 (profile_count::to_cgraph_frequency): Do not ICE when entry is
4475 (profile_count::to_sreal_scale): Likewise.
4476 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
4477 (profile_count::combine_with_ipa_count): New function.
4478 * profile-count.h (profile_guessed_global0adjusted): New.
4479 (profile_count::adjusted_zero): New.
4480 (profile_count::global0adjusted): New.
4481 (profile_count::combine_with_ipa_count): New.
4482 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
4483 correct profile of return block of split functions.
4484 (copy_cfg_body): Remove unused profile_count.
4485 (copy_body): Likewise.
4486 (expand_call_inline): Update.
4487 (tree_function_versioning): Update.
4489 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
4491 * hash-set.h (hash_set::empty): New.
4492 * tree-ssa-threadbackward.h: Delete.
4493 * tree-ssa-threadbackward.c (class thread_jumps): New.
4494 Move max_threaded_paths into class.
4495 (fsm_find_thread_path): Remove arguments that are now in class.
4496 (profitable_jump_thread_path): Rename to...
4497 (thread_jumps::profitable_jump_thread_path): ...this.
4498 (convert_and_register_jump_thread_path): Rename to...
4499 (thread_jumps::convert_and_register_current_path): ...this.
4500 (check_subpath_and_update_thread_path): Rename to...
4501 (thread_jumps::check_subpath_and_update_thread_path): ...this.
4502 (register_jump_thread_path_if_profitable): Rename to...
4503 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
4504 (handle_phi): Rename to...
4505 (thread_jumps::handle_phi): ...this.
4506 (handle_assignment): Rename to...
4507 (thread_jumps::handle_assignment): ...this.
4508 (fsm_find_control_statement_thread_paths): Rename to...
4509 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
4510 (find_jump_threads_backwards): Rename to...
4511 (thread_jumps::find_jump_threads_backwards): ...this.
4512 Initialize path local data.
4513 (pass_thread_jumps::execute): Call find_jump_threads_backwards
4514 from within thread_jumps class.
4515 (pass_early_thread_jumps::execute): Same.
4517 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4519 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
4521 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4523 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
4524 to canonical location.
4526 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4529 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
4531 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
4533 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
4535 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
4536 rcrt1.o%s/grcrt1.o%s for -static-pie.
4538 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4540 * i386.c (ix86_multiplication_cost, ix86_division_cost,
4541 ix86_shift_rotate_cost): Break out from ...
4542 (ix86_rtx_costs): ... here.
4543 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
4546 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4548 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
4549 when entry block was promoted unlikely.
4550 (estimate_bb_frequencies): Increase frequency scale.
4551 * profile-count.h (profile_count): Export precision info.
4553 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4555 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
4556 disturbin profile of entry block.
4558 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4560 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
4563 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4565 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
4566 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
4568 2017-11-17 Jeff Law <law@redhat.com>
4570 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
4571 from evrp_dom_walker class. Various methods moved into new class.
4572 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
4573 (evrp_range_analyzer::enter): New method.
4574 (evrp_range_analyzer::leave): New method.
4575 (evrp_dom_walker): Remove delegators no longer needed by this class.
4576 Replace vr_values data member with evrp_range_analyzer
4578 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
4579 method extracted from evrp_dom_walker::before_dom_children.
4580 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
4581 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
4583 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
4584 Add private copy constructor and move assignment operators.
4585 Privatize methods and class data where trivially possible.
4586 (evrp_dom_walker::cleanup): New function, extracted from
4587 execute_early_vrp. Simplify access to class data.
4589 * vr-values.h (get_output_for_vrp): Prototype.
4590 * vr-values.c (get_output_for_vrp): New function extracted from
4591 vrp_visit_assignment_or_call and extract_range_from_stmt.
4592 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
4594 2017-11-17 Luis Machado <luis.machado@linaro.org>
4596 * config/aarch64/aarch64.c
4597 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
4598 (qdf24xx_tunings) <autoprefetcher_model>: Set to
4599 tune_params::AUTOPREFETCHER_WEAK.
4601 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4604 * config/arm/arm.c (arm_valid_target_attribute_rec):
4605 Parse "arch=" and "+<ext>".
4606 (arm_valid_target_attribute_tree): Re-init global options.
4607 (arm_option_override): Make non-static.
4608 (arm_options_perform_arch_sanity_checks): Make errors fatal.
4609 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
4610 (__ARM_FEATURE_CRC32): Support undef.
4611 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
4612 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
4614 2017-11-17 David Malcolm <dmalcolm@redhat.com>
4616 * gdbinit.in (break-on-diagnostic): New command.
4618 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4620 * config/i386/i386.c (ix86_expand_epilogue): Change simple
4621 return to indirect jump for EH return if control-flow
4622 protection is enabled. Change explicit 'false' argument in
4623 pro_epilogue_adjust_stack with a value of
4625 * config/i386/i386.md (simple_return_indirect_internal):
4626 Remove SImode restriction to support 64-bit.
4628 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4630 * combine.c (added_notes_insn): New.
4631 (try_combine): Handle added_notes_insn like added_links_insn.
4632 Rewrite return value code.
4633 (distribute_notes): Set added_notes_insn to the earliest insn we added
4636 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4638 PR rtl-optimization/82621
4639 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
4640 dest of one of those SETs is unused.
4642 2017-11-17 Richard Biener <rguenther@suse.de>
4645 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
4646 * tree-pretty-print.c (dump_generic_node): Handle
4647 annot_expr_parallel_kind.
4648 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
4649 * gimplify.c (gimple_boolify): Likewise.
4651 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4653 * config.gcc (extra_headers): Add cet.h for x86 targets.
4654 * config/i386/cet.h: New file.
4655 * doc/install.texi: Add --enable-cet/--disable-cet.
4657 2017-11-17 Richard Biener <rguenther@suse.de>
4659 PR tree-optimization/83017
4660 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
4661 (gen_parallel_loop): Properly count iterations.
4662 (parallelize_loops): Handle loop->can_be_parallel independent
4663 of flag_loop_parallelize_all. Make static profitability test match
4665 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
4666 * invoke.texi (parloops-min-per-thread): Document.
4668 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
4670 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
4671 upstreaming review comments.
4673 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4675 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
4676 with fast unaligned access.
4677 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
4679 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
4681 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
4682 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
4684 2017-11-17 Richard Biener <rguenther@suse.de>
4686 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
4687 folding of references.
4689 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
4692 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
4693 of replacing call to strncmp with corresponding call to strcmp when
4696 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
4698 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
4699 option prefer-avx256 for skylake-avx512 configuration.
4700 * config/i386/i386.c (ix86_option_override_internal): Ditto.
4701 (get_builtin_code_for_version): Ditto.
4703 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4704 Monk Chiang <sh.chiang04@gmail.com>
4706 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
4707 (FIXED_REGISTERS): Reserve more register numbers.
4708 (CALL_USED_REGISTERS): Likewise.
4709 (REG_ALLOC_ORDER): Likewise.
4710 (REG_CLASS_CONTENTS): Likewise.
4711 (REGISTER_NAMES): Likewise.
4713 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4714 Kito Cheng <kito.cheng@gmail.com>
4716 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
4718 * config/nds32/iterators.md: Add vector mode iterators and attributes.
4720 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
4722 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
4723 parameter list for vec_splats.
4725 2017-11-16 Joseph Myers <joseph@codesourcery.com>
4727 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
4729 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
4731 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
4734 * doc/install.texi: Document incompatibility of Perl >=5.6.26
4735 with the required version of automake 1.11.6.
4737 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
4739 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
4741 (DU_C2_3_power9): Correct reservation combinations.
4742 (FP_DIV_power9, VEC_DIV_power9): New.
4743 (power9-alu): Split out rotate/shift...
4744 (power9-rot): ...to here, correct dispatch resource.
4745 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
4747 (power9-fp): Correct latency.
4748 (power9-sdiv): Add div/sqrt resource.
4749 (power9-ddiv): Correct latency, add div/sqrt resource.
4750 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
4751 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
4753 (power9-qpdiv, power9-qpmul): Adjust resource usage.
4755 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4757 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
4758 switch statement mapping KF built-ins to TF built-ins if we don't
4759 have the proper ISA 3.0 assembler support.
4761 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4763 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
4764 (lower_emutls_function_body): Do not compute it.
4766 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4768 * ipa-split.c (split_bb_info): Turn time to sreal.
4769 (split_point): Likewise.
4770 (dump_split_point): Likewise.
4771 (fine_split_points): Likewise.
4772 (execute_split_functions): Only zero split_bbs; turn time to sreals.
4774 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4776 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
4779 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4781 * predict.c (combine_predictions_for_bb): Preserve zero predicted
4783 (expensive_function_p): Remove useless assert.
4784 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
4786 2017-11-16 Martin Sebor <msebor@redhat.com>
4788 PR tree-optimization/82588
4789 PR tree-optimization/82583
4790 * tree-vrp.c (check_array_ref): Handle flexible array members,
4791 string literals, and inner indices.
4792 (search_for_addr_array): Add detail to diagnostics.
4794 2017-11-16 Nathan Sidwell <nathan@acm.org>
4798 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
4799 (SET_DECL_ASSEMBLER_NAME): Forward to
4800 overwrite_decl_assembler_name.
4801 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
4802 (overwrite_decl_assembler_name): Declare.
4803 * tree.c (overwrite_decl_assembler_name): New.
4804 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
4805 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
4806 (LANG_HOOKS_INITIALIZER): Add it.
4807 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
4808 * langhooks.c (lhd_set_decl_assembler_name): Use
4809 SET_DECL_ASSEMBLER_NAME.
4810 (lhd_overwrite_decl_assembler_name): Default implementation.
4812 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4813 Jackson Woodruff <jackson.woodruff@arm.com>
4815 PR tree-optimization/71026
4816 * match.pd: Canonicalize constant multiplies in division.
4818 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4820 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
4821 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
4822 Remove OPT_fomit_frame_pointer entry.
4823 * common/config/arc/arc-common.c: Likewise.
4824 * common/config/arm/arm-common.c: Likewise.
4825 * common/config/avr/avr-common.c: Likewise.
4826 * common/config/c6x/c6x-common.c: Likewise.
4827 * common/config/cr16/cr16-common.c: Likewise.
4828 * common/config/cris/cris-common.c: Likewise.
4829 * common/config/epiphany/epiphany-common.c: Likewise.
4830 * common/config/fr30/fr30-common.c: Likewise.
4831 * common/config/frv/frv-common.c: Likewise.
4832 * common/config/ia64/ia64-common.c: Likewise.
4833 * common/config/iq2000/iq2000-common.c: Likewise.
4834 * common/config/lm32/lm32-common.c: Likewise.
4835 * common/config/m32r/m32r-common.c: Likewise.
4836 * common/config/mcore/mcore-common.c: Likewise.
4837 * common/config/microblaze/microblaze-common.c: Likewise.
4838 * common/config/mips/mips-common.c: Likewise.
4839 * common/config/mmix/mmix-common.c: Likewise.
4840 * common/config/mn10300/mn10300-common.c: Likewise.
4841 * common/config/nios2/nios2-common.c: Likewise.
4842 * common/config/pa/pa-common.c: Likewise.
4843 * common/config/pdp11/pdp11-common.c: Likewise.
4844 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
4845 * common/config/riscv/riscv-common.c: Likewise.
4846 * common/config/rs6000/rs6000-common.c: Likewise.
4847 * common/config/rx/rx-common.c: Likewise.
4848 * common/config/s390/s390-common.c: Likewise.
4849 * common/config/sh/sh-common.c: Likewise.
4850 * common/config/sparc/sparc-common.c: Likewise.
4851 * common/config/tilegx/tilegx-common.c: Likewise.
4852 * common/config/tilepro/tilepro-common.c: Likewise.
4853 * common/config/v850/v850-common.c: Likewise.
4854 * common/config/visium/visium-common.c: Likewise.
4855 * common/config/xstormy16/xstormy16-common.c: Likewise.
4856 * common/config/xtensa/xtensa-common.c: Likewise.
4857 * invoke.texi (-fomit-frame-pointer): Update documentation.
4859 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4861 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
4863 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4865 * cfg.c (scale_bbs_frequencies_int,
4866 cale_bbs_frequencies_gcov_type): Remove.
4867 * cfg.h (scale_bbs_frequencies_int,
4868 cale_bbs_frequencies_gcov_type): Remove.
4870 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4872 * tree-ssa-loop-manip.c
4873 (scale_dominated_blocks_in_loop): Update to profile counts.
4874 (tree_transform_and_unroll_loop): Likewise.
4876 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4878 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
4879 scale_bbs_frequencies_int.
4881 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4883 * final.c (compute_alignments): Use counts rather than frequencies.
4885 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4887 * cfgloopanal.c: Include sreal.h
4888 (average_num_loop_insns): Use counts and sreal for accounting.
4890 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4892 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
4895 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4896 Kito Cheng <kito.cheng@gmail.com>
4898 * config/nds32/constraints.md: Provide more constraints.
4899 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
4900 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
4901 support constraints usage.
4903 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4905 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
4907 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4908 Kito Cheng <kito.cheng@gmail.com>
4910 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
4911 * config/nds32/nds32.opt: Refine the layout.
4912 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
4913 TARGET_EXT_STRING): Support new options.
4914 * config/nds32/nds32.h: Likewise.
4915 * config/nds32/nds32.md: Likewise.
4916 * config/nds32/nds32-predicates.c: Likewise.
4917 * config/nds32/constraints.md: Likewise.
4918 * common/config/nds32/nds32-common.c: Likewise.
4920 2017-11-16 Julia Koval <julia.koval@intel.com>
4923 * config/i386/gfniintrin.h: Add sse check.
4924 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
4926 2017-11-16 Julia Koval <julia.koval@intel.com>
4928 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
4929 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
4930 (ix86_handle_option): Handle -mavx512vbmi2.
4931 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
4932 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4933 * config/i386/i386-c.c (__AVX512VBMI2__): New.
4934 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
4935 (ix86_valid_target_attribute_inner_p): Ditto.
4936 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
4937 * config/i386/i386.opt (mavx512vbmi2): New option.
4938 * doc/invoke.texi: Add new option.
4940 2017-11-16 Julia Koval <julia.koval@intel.com>
4942 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
4943 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
4944 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
4945 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
4946 _mm512_gf2p8mul_epi8): New intrinsics.
4947 * config/i386/i386-builtin-types.def
4948 (V64QI_FTYPE_V64QI_V64QI): New type.
4949 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
4950 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
4951 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
4952 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
4953 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
4954 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4956 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
4958 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
4959 explicitly as a stream of bytes.
4961 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4963 * config/rs6000/altivec.h (vec_xst_be): New #define.
4964 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
4965 and externalize from *altivec_vperm_<mode>_internal.
4966 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
4968 (XL_BE_V8HI): Likewise.
4969 (XL_BE_V4SI): Likewise.
4970 (XL_BE_V4SI): Likewise.
4971 (XL_BE_V2DI): Likewise.
4972 (XL_BE_V4SF): Likewise.
4973 (XL_BE_V2DF): Likewise.
4974 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
4975 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
4976 all array entries with these keys: VSX_BUILTIN_VEC_XL,
4977 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
4978 VSX_BUILTIN_VEC_XST_BE.
4979 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
4980 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
4982 (altivec_init_builtins): Replace conditional calls to def_builtin
4983 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
4984 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
4985 with unconditional calls. Remove calls to def_builtin for
4986 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
4987 __builtin_vec_xst_be.
4988 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
4989 to define_expand, and add alternate RTL generation for P8.
4990 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
4991 vsx_ld_elemrev_v8hi.
4992 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
4993 add alternate RTL generation for P8.
4994 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
4995 vsx_ld_elemrev_v16qi.
4996 (vsx_st_elemrev_v8hi): Convert define_insn
4997 to define_expand, and add alternate RTL generation for P8.
4998 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
4999 vsx_st_elemrev_v8hi.
5000 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5001 add alternate RTL generation for P8.
5002 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5003 vsx_st_elemrev_v16qi.
5005 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5008 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5009 TARGET_AVX512ER check.
5010 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5011 neither -mzeroupper nor -mno-zeroupper is used and
5012 TARGET_EMIT_VZEROUPPER is set.
5013 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5014 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5016 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5018 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5019 folding of vector compares.
5020 (fold_build_vec_cmp): New helper function.
5021 (fold_compare_helper): New helper function.
5022 (builtin_function_type): Add compare builtins to the list of functions
5023 having unsigned arguments. Cosmetic updates to comment indentation.
5024 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5025 the not+eq combination.
5027 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5029 PR tree-optimization/82726
5030 PR tree-optimization/70754
5031 * tree-predcom.c (order_drefs_by_pos): New function.
5032 (combine_chains): Move code setting has_max_use_after to...
5033 (try_combine_chains): ...here. New parameter. Sort combined chains
5034 according to position information.
5035 (tree_predictive_commoning_loop): Update call to above function.
5036 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5038 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5040 PR tree-optimization/82726
5042 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5044 PR tree-optimization/70754
5045 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5046 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5047 combined stmt before it if not NULL.
5048 (combine_chains): Process refs reversely and compute dominance point
5052 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5054 PR tree-optimization/79663
5055 * tree-predcom.c (combine_chains): Process refs in reverse order
5056 only for ZERO length chains, and add explaining comment.
5058 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5060 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
5061 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
5062 * doc/invoke.texi (ARM Options): Add armv8.3-a.
5064 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5066 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
5068 2017-11-15 Martin Liska <mliska@suse.cz>
5070 * tree-cfg.c (pass_warn_function_return::execute):
5071 Compare warn_return_type for greater than zero.
5073 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
5077 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
5078 to return true on Xeon and not on Xeon Phi.
5079 (ix86_check_avx256_register): Changed to ...
5080 (ix86_check_avx_upper_register): ... this. Add extra check for
5081 VALID_AVX512F_REG_OR_XI_MODE.
5082 (ix86_avx_u128_mode_needed): Changed
5083 ix86_check_avx256_register to ix86_check_avx_upper_register.
5084 (ix86_check_avx256_stores): Changed to ...
5085 (ix86_check_avx_upper_stores): ... this. Changed
5086 ix86_check_avx256_register to ix86_check_avx_upper_register.
5087 (ix86_avx_u128_mode_after): Changed
5088 avx_reg256_found to avx_upper_reg_found. Changed
5089 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
5090 (ix86_avx_u128_mode_entry): Changed
5091 ix86_check_avx256_register to ix86_check_avx_upper_register.
5092 (ix86_avx_u128_mode_exit): Ditto.
5093 * config/i386/i386.h: (host_detect_local_cpu): New define.
5095 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5097 * config/arm/xgene1.md (xgene1): Split into automatons
5098 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
5099 (xgene1_f_load): Adjust reservations and/or types.
5100 (xgene1_f_store): Likewise.
5101 (xgene1_load_pair): Likewise.
5102 (xgene1_store_pair): Likewise.
5103 (xgene1_fp_load1): Likewise.
5104 (xgene1_load1): Likewise.
5105 (xgene1_store1): Likewise.
5106 (xgene1_move): Likewise.
5107 (xgene1_alu): Likewise.
5108 (xgene1_simd): Likewise.
5109 (xgene1_bfm): Likewise.
5110 (xgene1_neon_load1): Likewise.
5111 (xgene1_neon_store1): Likewise.
5112 (xgene1_neon_logic): Likewise.
5113 (xgene1_neon_st1): Likewise.
5114 (xgene1_neon_ld1r): Likewise.
5115 (xgene1_alu_cond): Added.
5116 (xgene1_shift_reg): Likwise.
5117 (xgene1_bfx): Likewise.
5118 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
5120 2017-11-15 Jakub Jelinek <jakub@redhat.com>
5123 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
5125 (can_widen_mult_without_libcall): New function.
5126 (expand_mul_overflow): If only checking unsigned mul overflow,
5127 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
5128 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
5129 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
5130 (expand_DIVMOD): Formatting fix.
5131 * expmed.h (expand_mult): Add NO_LIBCALL argument.
5132 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
5133 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
5135 PR tree-optimization/82977
5136 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
5137 constructed temporary to strlen_to_stridx.put.
5139 2017-11-15 Martin Liska <mliska@suse.cz>
5141 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
5142 * configure: Regenerate.
5144 2017-11-15 Martin Liska <mliska@suse.cz>
5147 * config/sh/sh-mem.cc: Use proper probability for
5150 2017-11-14 Jeff Law <law@redhat.com>
5152 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
5153 the red zone for stack_clash_protection_final_dynamic_probe targets
5154 when the total dynamic stack size is zero bytes.
5156 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
5157 blocks is post order.
5159 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
5161 * dumpfile.h (TDF_COMPARE_DEBUG): New.
5162 * final.c (rest_of_clean_state): Set it for the
5163 -fcompare-debug dump.
5164 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
5165 class when TDF_COMPARE_DEBUG is set.
5167 * dwarf2out.c (gen_producer_string): Discard
5170 2017-11-15 Joseph Myers <joseph@codesourcery.com>
5173 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
5174 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
5175 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
5176 (__TGMATH_REAL_2_3): Remove macros.
5177 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
5178 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
5180 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
5182 (remquo): Define using __TGMATH_REAL_3.
5184 2017-11-14 Jeff Law <law@redhat.com>
5186 * vr-values.c: New file with contents extracted from tree-vrp.c.
5187 * Makefile.in (OBJS): Add vr-values.o
5188 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
5189 (set_value_range, set_and_canonicalize_value_range): Likewise.
5190 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
5191 (value_range_constant_singleton, symbolic_range_p): Likewise.
5192 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
5193 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
5194 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
5195 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
5196 (find_case_label_range, find_case_label_index): Likewise.
5197 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
5198 (range_int_cst_singleton_p, value_inside_range): Likewise.
5199 (get_single_symbol): Likewise.
5200 (switch_update): Move structure definition here.
5201 (to_remove_edges, to_update_switch_stmts): Provide externs.
5202 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
5203 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
5204 (vrp_val_is_min, set_value_range): Likewise.
5205 (set_and_canonicalize_value_range, copy_value_range): Likewise.
5206 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
5207 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
5208 (range_is_nonnull, range_int_cst_p): Likewwise.
5209 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
5210 (get_single_symbol, operand_less_p): Likewise
5211 (compare_values_warnv, compare_values): Likewise.
5212 (value_inside_range, value_range_constant_singleton): Likewise.
5213 (zero_nonzero_bitgs_from_vr): Likewise.
5214 (extract_range_from_binary_expr_1): Likewise.
5215 (overflow_comparison_p): Likewise.
5216 (to_remove_edges, to_update_switch_stmts): Likewise.
5217 (find_case_label-index, find_case_label_range): Likewise.
5218 (switch_update, set_value_range_to_nonnegative): Remove.
5219 (set_value_range_to_truthvalue): Likewise.
5220 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
5221 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
5222 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
5223 (find_case_label_ranges, test_for_singularity): Likewise.
5224 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
5225 (x_vr_values): Move to its remaining use site.
5227 2017-11-10 Jeff Law <law@redhat.com>
5229 * vr-values.h (VR_INITIALIZER): Move #define here.
5230 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
5231 * Makefile.in (OBJS): Add tree-evrp.o
5232 * tree-vrp.h (assert_info): Move structure definition here.
5233 (set_value_range_to_varying): Prototype.
5234 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5235 (infer_value_range, register_edge_assert_for): Likewise.
5236 (stmt_interesting_for_vrp): Likewise.
5237 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
5238 (set_value_range_to_varying): No longer static.
5239 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5240 (infer_value_range, register_edge_assert_for): Likewise.
5242 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5244 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
5245 generate the XXBRD instruction.
5247 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
5248 (rs6000_builtin_type_compatible): Treat _Float128 and long double
5249 as being compatible if -mabi=ieeelongdouble.
5250 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
5251 to setup float128 built-ins with hardware support.
5252 (BU_FLOAT128_HW_2): Likewise.
5253 (BU_FLOAT128_HW_3): Likewise.
5254 (BU_FLOAT128_HW_VSX_1): Likewise.
5255 (BU_FLOAT128_HW_VSX_2): Likewise.
5256 (scalar_extract_expq): Change float128 built-in functions to
5257 accommodate having both KFmode and TFmode functions. Use the
5258 KFmode variant as the default.
5259 (scalar_extract_sigq): Likewise.
5260 (scalar_test_neg_qp): Likewise.
5261 (scalar_insert_exp_q): Likewise.
5262 (scalar_insert_exp_qp): Likewise.
5263 (scalar_test_data_class_qp): Likewise.
5264 (sqrtf128_round_to_odd): Delete processing the round to odd
5265 built-in functions as special built-in functions, and define them
5266 as float128 built-ins. Use the KFmode variant as the default.
5267 (truncf128_round_to_odd): Likewise.
5268 (addf128_round_to_odd): Likewise.
5269 (subf128_round_to_odd): Likewise.
5270 (mulf128_round_to_odd): Likewise.
5271 (divf128_round_to_odd): Likewise.
5272 (fmaf128_round_to_odd): Likewise.
5273 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5274 support for KFmode and TFmode xststdcqp calls.
5275 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
5276 point, switch the built-in handlers for the get/set float128
5277 exponent, get float128 mantissa, float128 test built-ins, and the
5278 float128 round to odd built-in functions. Eliminate creating the
5279 float128 round to odd built-in functions as special built-ins.
5280 (rs6000_init_builtins): Eliminate special creation of the float128
5281 round to odd built-in functions.
5282 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
5283 function insns to support both TFmode and KFmode variants.
5284 (xsxsigqp_<mode>): Likewise.
5285 (xsiexpqpf_<mode>): Likewise.
5286 (xsiexpqp_<mode>): Likewise.
5287 (xststdcqp_<mode>): Likewise.
5288 (xststdcnegqp_<mode>): Likewise.
5289 (xststdcqp_<mode>): Likewise.
5291 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5293 * tree-ssa-threadupdate.c (compute_path_counts): Remove
5294 unused path_in_freq_ptr parameter.
5295 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
5297 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5299 * ipa-inline.c (edge_badness): Dump sreal frequency.
5300 (compute_inlined_call_time): Match natural implementaiton ...
5301 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
5302 forgotten division by CGRAPH_FREQ_BASE.
5304 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5306 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
5307 Solaris 11. Update comment.
5308 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
5310 * config/sol2.h (STARTFILE_SPEC): Likewise.
5311 * configure: Regenerate.
5313 2017-11-14 Carl Love <cel@us.ibm.com>
5315 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
5316 le_ and be_ prefixes to swap* variables. Remove
5317 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
5320 2017-11-14 Jason Merrill <jason@redhat.com>
5322 Support GTY((cache)) on hash_map.
5323 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
5324 (ggc_cache_remove): Override it instead of ggc_mx.
5325 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
5326 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
5327 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
5328 (simple_cache_map_traits): Override maybe_mx.
5329 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
5330 (hash_map): Friend gt_cleare_cache.
5331 (gt_cleare_cache): New.
5332 * tree.h (tree_cache_traits): New hash_map traits class.
5333 (tree_cache_map): New typedef.
5335 2017-11-14 Richard Biener <rguenther@suse.de>
5337 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
5338 paramter and handling.
5339 (cleanup_control_flow_bb): Likewise.
5340 (cleanup_control_flow_pre): New helper performing a DFS walk
5341 to call cleanup_control_flow_bb in PRE order.
5342 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
5343 via cleanup_control_flow_pre.
5345 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
5347 * config/aarch64/aarch64-simd.md
5348 (aarch64_simd_bsl<mode>_internal): Remove DImode.
5349 (*aarch64_simd_bsl<mode>_alt): Likewise.
5350 (aarch64_simd_bsldi_internal): New.
5351 (aarch64_simd_bsldi_alt): Likewise.
5353 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5355 * tracer.c (better_p): Do not compare frequencies.
5356 * reg-stack.c (better_edge): Likewise.
5357 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
5360 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5362 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
5363 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
5364 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
5365 * lto-streamer-in.c (input_function): Use update_max_bb_count.
5366 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
5367 * predict.c (maybe_hot_frequency_p): Inline to ...
5368 (maybe_hot_count_p): ... here; rewrite to counts.
5369 (counts_to_freqs): Rename to ...
5370 (update_max_bb_count): ... this one.
5371 (expensive_function_p): Use counts.
5372 (estimate_bb_frequencies): Update.
5373 (rebuild_frequencies): Update.
5374 * predict.h (counts_to_freqs): Rename to ...
5375 (update_max_bb_count): ... this one.
5376 * profile.c (compute_branch_probabilities): Add debug info
5377 * tree-inline.c (expand_call_inline): Update debug info.
5378 (optimize_inline_calls): Use update_max_bb_count..
5379 (tree_function_versioning): Use update_max_bb_count..
5380 * value-prof.c (gimple_value_profile_transformations):
5381 Do not use update_max_bb_count.
5383 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5385 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
5386 always use frequencies.
5388 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5390 * bb-reorder.c: Remove frequencies from comments.
5391 (better_edge_p): Use profile counts.
5392 (find_traces): Dump profile counts.
5393 (rotate_loop): Use profile counts.
5394 (find_traces_1_round): Likewise.
5395 (connect_better_edge_p): Use counts instead of probabilities for
5397 (copy_bb_p): Drop early check for non-0 frequency.
5398 (sanitize_hot_paths): Update comments.
5400 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5402 * ipa-split.c (struct split_point): Add count.
5403 (consider_split): Do not compute incoming frequency; compute incoming
5404 count and store it to split_point.
5405 (split_function): Set count of the call to split part correctly.
5407 2017-11-13 Carl Love <cel@us.ibm.com>
5409 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
5411 2017-11-13 Tom Tromey <tom@tromey.com>
5413 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
5415 2017-11-13 Carl Love <cel@us.ibm.com>
5417 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
5418 Add support for builtins:
5419 unsigned int vec_first_{,miss}_match_{,or_eos}index,
5420 vector {un,}signed {char,int,short},
5421 vector {un,}signed {char,int,short}) arguments.
5422 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
5423 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
5424 Add BU_P9V_AV_2 expansions for the builtins.
5425 * config/rs6000/altivec.h (vec_first_match_index,
5426 vec_first_mismatch_index, vec_first_match_or_eos_index,
5427 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
5428 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
5429 new extern declaration.
5430 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
5431 * config/rs6000/vsx.md (first_match_index_<mode>,
5432 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
5433 first_mismatch_or_eos_index_<mode>): Add define expand.
5434 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
5435 * doc/extend.texi: Update the built-in documenation file for the new
5438 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5440 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
5441 into the min/max operations for _Float<N> and _Float<N>X types.
5443 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5446 * dwarf2out.c (do_eh_frame): New static variable.
5447 (dwarf2out_begin_prologue): Set it.
5448 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
5450 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5452 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
5454 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
5456 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5459 * debug.h (dwarf2out_do_eh_frame): Declare.
5460 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
5461 (dwarf2out_do_frame): Use it.
5462 (dwarf2out_do_cfi_asm): Likewise.
5463 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
5464 (dwarf2out_assembly_start): Likewise.
5465 (dwarf2out_begin_prologue): Fix comment.
5466 * toplev.c (compile_file): Always call dwarf2out_frame_finish
5467 if the target needs either debug or unwind DWARF2 info.
5468 * lto-opts.c (lto_write_options): Do not save -fexceptions,
5469 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
5470 -ftrapping-math, -ftrapv and -fwrapv.
5472 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5474 * cgraph.c (cgraph_edge::sreal_frequency): New function.
5475 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
5476 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
5477 (estimate_edge_size_and_time): Likewise.
5478 (ipa_merge_fn_summary_after_inlining): Likewise.
5479 * ipa-inline.c (cgraph_freq_base_rec): Remove.
5480 (compute_uninlined_call_time): Use sreal_frequency.
5481 (compute_inlined_call_time): Likewise.
5482 (ipa_inline): Do not initialize cgraph_freq_base_rec.
5483 * profile-count.c: Include sreal.h.
5484 (profile_count::to_sreal_scale): New.
5485 * profile-count.h: Forward declare sreal.
5486 (profile_count::to_sreal_scale): Declare.
5488 2017-11-13 Nathan Sidwell <nathan@acm.org>
5490 * diagnostic.c (maybe_line_and_column): New.
5491 (diagnostic_get_location_text): Use it.
5492 (diagnostic_report_current_module): Likewise.
5493 (test_diagnostic_get_location_text): Add tests.
5495 2017-11-13 Luis Machado <luis.machado@linaro.org>
5497 * doc/md.texi (Specifying processor pipeline description): Fix
5498 incorrect latency for the div instruction example.
5500 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5502 PR tree-optimization/78821
5503 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
5504 that bit_not_p is the same.
5505 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
5506 (split_group): Count precisely bit_not_p bits in each statement.
5507 (invert_op): New function.
5508 (imm_store_chain_info::output_merged_store): Use invert_op to
5509 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
5510 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
5512 2017-11-13 Martin Liska <mliska@suse.cz>
5514 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
5515 (struct source_info): Likewise.
5516 (add_branch_counts): Likewise.
5517 (add_line_counts): Likewise.
5518 (function_summary): Likewise.
5519 (output_intermediate_line): Likewise.
5520 (generate_results): Likewise.
5522 2017-11-13 Martin Liska <mliska@suse.cz>
5524 * gcov.c (struct block_info): Remove typedef for block_t.
5525 (struct line_info): Likewise.
5526 (line_info::has_block): Likewise.
5527 (EXIT_BLOCK): Likewise.
5528 (unblock): Likewise.
5529 (circuit): Likewise.
5530 (get_cycles_count): Likewise.
5531 (process_file): Likewise.
5532 (read_graph_file): Likewise.
5533 (solve_flow_graph): Likewise.
5534 (find_exception_blocks): Likewise.
5535 (add_line_counts): Likewise.
5536 (accumulate_line_info): Likewise.
5537 (output_line_details): Likewise.
5539 2017-11-13 Martin Liska <mliska@suse.cz>
5541 * gcov.c (struct arc_info): Remove typedef for arc_t.
5542 (struct line_info): Likewise.
5543 (add_branch_counts): Likewise.
5544 (output_branch_count): Likewise.
5545 (function_info::~function_info): Likewise.
5546 (circuit): Likewise.
5547 (output_intermediate_line): Likewise.
5548 (read_graph_file): Likewise.
5549 (solve_flow_graph): Likewise.
5550 (find_exception_blocks): Likewise.
5551 (add_line_counts): Likewise.
5552 (accumulate_line_info): Likewise.
5553 (output_line_details): Likewise.
5554 (output_function_details): Likewise.
5556 2017-11-13 Martin Liska <mliska@suse.cz>
5558 * gcov.c (struct function_info): Remove typedef for function_t.
5559 (struct source_info): Likewise.
5560 (source_info::get_functions_at_location): Likewise.
5561 (solve_flow_graph): Likewise.
5562 (find_exception_blocks): Likewise.
5563 (add_line_counts): Likewise.
5564 (output_intermediate_file): Likewise.
5565 (process_file): Likewise.
5566 (generate_results): Likewise.
5567 (release_structures): Likewise.
5568 (read_graph_file): Likewise.
5569 (read_count_file): Likewise.
5570 (accumulate_line_counts): Likewise.
5571 (output_lines): Likewise.
5573 2017-11-13 Martin Liska <mliska@suse.cz>
5575 * gcov.c (function_info::function_info): Remove num_counts
5576 and add vector<gcov_type>.
5577 (function_info::~function_info): Use the vector.
5578 (process_file): Likewise.
5579 (read_graph_file): Likewise.
5580 (read_count_file): Likewise.
5581 (solve_flow_graph): Likewise.
5583 2017-11-13 Martin Liska <mliska@suse.cz>
5585 * gcov.c (function_info::is_artificial): New function.
5586 (process_file): Erase all artificial early.
5587 (generate_results): Skip as all artificial are already
5590 2017-11-13 Martin Liska <mliska@suse.cz>
5592 * gcov.c (read_graph_file): Store to global vector of functions.
5593 (read_count_file): Iterate the vector.
5594 (process_file): Likewise.
5595 (generate_results): Likewise.
5596 (release_structures): Likewise.
5598 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5600 PR tree-optimization/82954
5601 * gimple-ssa-store-merging.c
5602 (imm_store_chain_info::coalesce_immediate_stores): If
5603 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
5605 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
5607 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
5608 Upddate call to ENDIAN_LANE_N.
5609 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
5610 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
5611 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
5612 and use aarch64_endian_lane_rtx.
5613 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
5615 2017-11-12 Tom de Vries <tom@codesourcery.com>
5617 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
5619 2017-11-12 Tom de Vries <tom@codesourcery.com>
5621 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
5623 2017-11-12 Tom de Vries <tom@codesourcery.com>
5625 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
5627 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5628 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5629 * defaults.h (ASM_OUTPUT_LABELREF): Same.
5631 2017-11-11 Martin Sebor <msebor@redhat.com>
5634 * doc/extend.texi (attribute nonstring): Remove spurious argument.
5637 * prefic.c (translate_name): Replace strncpy with memcpy to
5638 avoid -Wstringop-truncation.
5640 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5642 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
5644 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5646 * predict.c (maybe_hot_frequency_p): Do not use cfun.
5648 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5650 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
5653 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
5655 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
5656 enable generating XXBRH if the value is in a vector register.
5657 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
5658 value is in a vector register.
5659 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
5660 register to register bswap64's instead of doing the GPR sequence
5661 used on previous machines.
5662 (bswapdi2_xxbrd): New insn.
5663 (bswapdi2_reg): Disallow on ISA 3.0.
5664 (register to register bswap64 splitter): Do not split the insn on
5665 ISA 3.0 systems that use XXBRD.
5667 2017-11-10 Martin Sebor <msebor@redhat.com>
5670 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
5672 (find_subframework_file): Same.
5674 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5676 * auto-profile.c (afdo_indirect_call): Drop frequency.
5677 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
5678 (cgraph_node::create_edge): Drop frequency argument.
5679 (cgraph_node::create_indirect_edge): Drop frequency argument.
5680 (cgraph_edge::make_speculative): Drop frequency arguments.
5681 (cgraph_edge::resolve_speculation): Do not update frequencies
5682 (cgraph_edge::dump_edge_flags): Do not dump frequency.
5683 (cgraph_node::dump): Check consistency in IPA mode.
5684 (cgraph_edge::maybe_hot_p): Use IPA counter.
5685 (cgraph_edge::verify_count_and_frequency): Rename to ...
5686 (cgraph_edge::verify_count): ... this one; drop frequency checking.
5687 (cgraph_node::verify_node): Update.
5688 * cgraph.h (struct cgraph_edge): Drop frequency.
5689 (cgraph_edge::frequency): New function.
5690 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
5692 (cgraph_edge::rebuild_edges): Likewise.
5693 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
5694 (duplicate_thunk_for_node): Do not pass frequency.
5695 (cgraph_node::create_clone): Scale only counts.
5696 (cgraph_node::create_virtual_clone): Do not pass frequency.
5697 (cgraph_node::create_edge_including_clones): Do not pass frequency.
5698 (cgraph_node::create_version_clone): Do not pass frequency.
5699 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
5700 (cgraph_node::expand_thunk): Do not pass frequency.
5701 (cgraph_node::create_wrapper): Do not pass frequency.
5702 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
5704 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
5705 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
5706 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
5707 (gather_caller_stats): Use frequency function.
5708 (ipcp_cloning_candidate_p): Use frequency function.
5709 (ipcp_propagate_stage): Use frequency function.
5710 (get_info_about_necessary_edges): Use frequency function.
5711 (update_profiling_info): Update only IPA profile.
5712 (update_specialized_profile): Use frequency functoin.
5713 (perhaps_add_new_callers): Update only IPA profile.
5714 * ipa-devirt.c (ipa_devirt): Use IPA profile.
5715 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
5716 (dump_ipa_call_summary): Use frequency function.
5717 (estimate_edge_size_and_time): Use frequency function.
5718 (ipa_merge_fn_summary_after_inlining): Use frequency function.
5719 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
5720 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
5721 (update_noncloned_counts): ... ths one; scale counts only.
5722 (clone_inlined_nodes): Do not scale frequency.
5723 (inline_call): Do not pass frequency.
5724 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
5725 (compute_inlined_call_time): Use IPA profile.
5726 (want_inline_small_function_p): Use IPA profile.
5727 (want_inline_self_recursive_call_p): Use IPA profile.
5728 (edge_badness): Use IPA profile.
5729 (lookup_recursive_calls): Use IPA profile.
5730 (recursive_inlining): Do not pass frequency.
5731 (resolve_noninline_speculation): Do not update frequency.
5732 (inline_small_functions): Collect max of IPA profile.
5733 (dump_overall_stats): Dump IPA porfile.
5734 (dump_inline_stats): Dump IPA porfile.
5735 (ipa_inline): Collect IPA stats.
5736 * ipa-inline.h (clone_inlined_nodes): Update prototype.
5737 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
5738 (ipa_propagate_frequency): Use frequency function.
5739 (ipa_profile): Cleanup.
5740 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
5741 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
5742 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
5743 (input_node): Do not stream frequency.
5744 (input_edge): Do not stream frequency.
5745 (merge_profile_summaries): Scale only IPA profiles.
5746 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
5747 * predict.c (drop_profile): Do not recompute frequency.
5748 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
5749 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
5750 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
5751 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
5752 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
5753 * tree-inline.c (copy_bb): Do not scale frequency.
5754 (expand_call_inline): Do not scale frequency.
5755 (tree_function_versioning): Do not scale frequency.
5756 * ubsan.c (ubsan_create_edge): Do not pass frequency.
5758 2017-11-10 Julia Koval <julia.koval@intel.com>
5760 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
5761 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
5762 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
5763 (_mm256_maskz_gf2p8affine_epi64_epi8)
5764 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
5765 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
5766 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
5767 (__builtin_ia32_vgf2p8affineqb_v32qi)
5768 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
5769 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
5771 2017-11-10 Tamar Christina <tamar.christina@arm.com>
5775 (arm_option_override): Refactor.
5776 (arm_option_reconfigure_globals): New.
5777 (arm_options_perform_arch_sanity_checks): New.
5778 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
5780 (arm_options_perform_arch_sanity_checks): Likewise
5782 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
5784 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
5785 (power9-qpmul): New.
5786 * rs6000/rs6000.md ("type" attr): Add qmul.
5787 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
5788 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
5789 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
5791 2017-11-10 Martin Sebor <msebor@redhat.com>
5794 * builtins.c (compute_objsize): Handle arrays that
5795 compute_builtin_object_size likes to fail for. Make extern.
5796 * builtins.h (compute_objsize): Declare.
5797 (check_strncpy_sizes): New function.
5798 (expand_builtin_strncpy): Call check_strncpy_sizes.
5799 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
5800 -Wstringop-truncation.
5801 (gimple_fold_builtin_strncat): Same.
5802 * gimple.c (gimple_build_call_from_tree): Set call location.
5803 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
5804 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
5805 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
5806 (handle_builtin_strlen): Use strlen_to_stridx.
5807 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
5809 Use strlen_to_stridx.
5810 (pass_strlen::execute): Release strlen_to_stridx.
5811 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
5812 (-Wstringop-truncation): Document new option.
5814 2017-11-10 Martin Liska <mliska@suse.cz>
5816 PR gcov-profile/82702
5817 * gcov.c (main): Handle intermediate files in a different
5819 (get_gcov_intermediate_filename): New function.
5820 (output_gcov_file): Remove support of intermediate files.
5821 (generate_results): Allocate intermediate file.
5822 (release_structures): Clean-up properly fn_end.
5823 (output_intermediate_file): Start iterating with line 1.
5825 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5827 PR tree-optimization/82929
5828 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5829 ops_swapped_p non-static data member.
5830 (store_immediate_info::store_immediate_info): Clear it.
5831 (imm_store_chain_info::coalesce_immediate_stores): If swapping
5832 ops set ops_swapped_p.
5833 (count_multiple_uses): Handle ops_swapped_p.
5835 2017-11-10 Martin Liska <mliska@suse.cz>
5837 * coverage.c (coverage_init): Stream information about
5838 support of has_unexecuted_blocks.
5839 * doc/gcov.texi: Document that.
5840 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
5841 * gcov.c (read_graph_file): Likewise.
5842 (output_line_beginning): Fix a small issue with
5845 2017-11-10 Bin Cheng <bin.cheng@arm.com>
5847 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
5848 reference of trivial component.
5850 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5853 * gimple-ssa-store-merging.c
5854 (pass_store_merging::terminate_all_aliasing_chains): For
5855 gimple_store_p stmts also call refs_output_dependent_p.
5857 PR rtl-optimization/82913
5858 * compare-elim.c (try_merge_compare): Punt if def_insn is not
5861 2017-11-09 Jeff Law <law@redhat.com>
5863 * vr-values.h: New file with vr_values class.
5864 * tree-vrp.c: Include vr-values.h
5865 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
5866 data objects into the vr_values class.
5867 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
5868 (get_value_range): Make it a member function within vr_values class.
5869 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
5870 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
5871 (op_with_constant_singleton_value_range): Likewise.
5872 (extract_range_for_var_from_comparison_expr): Likewise.
5873 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
5874 (extract_range_from_binary_expr): Likewise.
5875 (extract_range_from_unary_expr): Likewise.
5876 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
5877 (check_for_binary_op_overflow, extract_range_basic): Likewise.
5878 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
5879 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
5880 (compare_name_with_value, compare_names): Likewise.
5881 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
5882 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
5883 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
5884 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
5885 (extract_range_from_phi_node): Likewise.
5886 (simplify_truth_ops_using_ranges): Likewise.
5887 (simplify_div_or_mod_using_ranges): Likewise.
5888 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
5889 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
5890 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
5891 (simplify_float_conversion_using_ranges): Likewise.
5892 (simplify_internal_call_using_ranges): Likewise.
5893 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
5894 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
5895 poitner via x_vr_values for calls into gimple folder.
5896 (vrp_initialize_lattice): Make this the vr_values ctor.
5897 (vrp_free_lattice): Make this the vr_values dtor.
5898 (set_vr_value): New function.
5899 (class vrp_prop): Add vr_values data member. Add various member
5900 functions as well as member functions that delegate to vr_values.
5901 (check_array_ref): Make a member function within vrp_prop class.
5902 (search_for_addr_array, vrp_initialize): Likewise.
5903 (vrp_finalize): Likewise. Revamp to avoid direct access to
5904 vr_value, values_propagated, etc.
5905 (check_array_bounds): Extract vrp_prop class instance pointer from
5906 walk info structure. Use it to call member functions.
5907 (check_all_array_refs): Make a member function within vrp_prop class.
5908 Smuggle class instance pointer via walk info structure.
5909 (x_vr_values): New local static.
5910 (vrp_valueize): Use x_vr_values to get class instance.
5911 (vr_valueize_1): Likewise.
5912 (class vrp_folder): Add vr_values data member. Add various member
5913 functions as well as member functions that delegate to vr_values.
5914 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
5915 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
5916 class instance from vr_values. Use it to call member functions.
5917 (vrp_dom_walker): Add vr_values data member.
5918 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
5919 instance via x_vr_values.
5920 (identify_jump_threads): Accept vr_values as argument. Store
5921 it into the walker structure.
5922 (evrp_dom_walker): Add vr_values class data member. Add various
5924 (evrp_dom_walker::try_find_new_range): Use vr_values data
5925 member to access the memory allocator.
5926 (evrp_dom_walker::before_dom_children): Store vr_values class
5927 instance into the vrp_folder class.
5928 (evrp_dom_walker::push_value_range): Rework to avoid direct
5929 access to num_vr_values and vr_value.
5930 (evrp_dom_walker::pop_value_range): Likewise.
5931 (execute_early_vrp): Remove call to vrp_initialize_lattice.
5932 Use vr_values to get to dump_all_value_ranges member function.
5933 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
5934 and simplify_cond_using_ranges_2 via vrp_prop class instance.
5935 Pass vr_values class instance down to identify_jump_threads.
5936 Remove call to vrp_free_lattice.
5937 (debug_all_value_ranges): Remove.
5939 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
5940 (vrp_folder): Likewise.
5942 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
5943 Get it from the existing bitmap instead.
5944 (vrp_intersect_ranges_1): Likewise.
5946 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5948 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5950 (store_immediate_info::store_immediate_info): Add bitnotp argument,
5951 set bit_not_p to it.
5952 (imm_store_chain_info::coalesce_immediate_stores): Break group
5953 if bit_not_p is different.
5954 (count_multiple_uses, split_group,
5955 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
5956 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
5957 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
5958 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
5959 to store_immediate_info ctor.
5961 2017-11-09 Jim Wilson <jimw@sifive.com>
5963 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
5964 (scan_prog_file): Likewise.
5966 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5968 * bb-reorder.c (max_entry_frequency): Remove.
5969 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
5970 connect_traces, push_to_next_round_p): Remove prototypes.
5971 (find_traces_1_round): Use counts only.
5972 (push_to_next_round_p): Likewise.
5973 (find_traces): Likewise.
5974 (rotate_loop): Likewise.
5975 (find_traces_1_round): Likewise.
5976 (connect_traces): Likewise.
5977 (edge_order): Likewise.
5979 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
5981 * config/arm/arm.c (output_return_instruction): Add comments to
5982 indicate requirement for cmse_nonsecure_entry return to account
5983 for the size of clearing instruction output here.
5984 (thumb_exit): Likewise.
5985 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
5986 return in hardfloat mode.
5988 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
5990 * config/rs6000/rs6000.c (machine_function): Add a bool,
5991 "toc_is_wrapped_separately".
5992 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
5993 if it wasn't explicitly set or unset, we are optimizing for speed, and
5994 doing separate shrink-wrapping.
5995 (rs6000_get_separate_components): Enable the TOC component if
5996 saving the TOC register in the prologue.
5997 (rs6000_components_for_bb): Handle the TOC component.
5998 (rs6000_emit_prologue_components): Store the TOC register where needed.
5999 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6000 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6002 2017-11-09 Martin Jambor <mjambor@suse.cz>
6004 * ipa-param-manipulation.c: New file.
6005 * ipa-param-manipulation.h: Likewise.
6006 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6007 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6008 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6009 (ipa_parm_adjustment): Likewise.
6010 (ipa_parm_adjustment_vec): Likewise.
6011 (ipa_get_vector_of_formal_parms): Moved declaration to
6012 ipa-param-manipulation.h.
6013 (ipa_get_vector_of_formal_parm_types): Likewise.
6014 (ipa_modify_formal_parameters): Likewise.
6015 (ipa_modify_call_arguments): Likewise.
6016 (ipa_combine_adjustments): Likewise.
6017 (ipa_dump_param_adjustments): Likewise.
6018 (ipa_modify_expr): Likewise.
6019 (ipa_get_adjustment_candidate): Likewise.
6020 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6021 ipa-param-manipulation.c.
6022 (ipa_get_vector_of_formal_parm_types): Likewise.
6023 (ipa_modify_formal_parameters): Likewise.
6024 (ipa_modify_call_arguments): Likewise.
6025 (ipa_modify_expr): Likewise.
6026 (get_ssa_base_param): Likewise.
6027 (ipa_get_adjustment_candidate): Likewise.
6028 (index_in_adjustments_multiple_times_p): Likewise.
6029 (ipa_combine_adjustments): Likewise.
6030 (ipa_dump_param_adjustments): Likewise.
6031 * tree-sra.c: Also include ipa-param-manipulation.h
6032 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6035 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6036 Alan Hayward <alan.hayward@arm.com>
6037 David Sherwood <david.sherwood@arm.com>
6039 * doc/sourcebuild.texi (vect_masked_store): Document.
6041 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6042 Alan Hayward <alan.hayward@arm.com>
6043 David Sherwood <david.sherwood@arm.com>
6045 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
6047 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6048 Alan Hayward <alan.hayward@arm.com>
6049 David Sherwood <david.sherwood@arm.com>
6051 * doc/sourcebuild.texi (vect_variable_length): Document.
6053 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6054 Alan Hayward <alan.hayward@arm.com>
6055 David Sherwood <david.sherwood@arm.com>
6057 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
6059 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6060 Alan Hayward <alan.hayward@arm.com>
6061 David Sherwood <david.sherwood@arm.com>
6063 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
6065 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6066 Alan Hayward <alan.hayward@arm.com>
6067 David Sherwood <david.sherwood@arm.com>
6069 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
6070 previously undocumented selectors.
6071 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
6073 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6075 * doc/rtl.texi (const_vector): Say that elements can be
6076 const_wide_ints too.
6077 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
6078 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
6079 (gen_vec_duplicate): Use it instead of CONSTANT_P.
6080 * optabs.c (expand_vector_broadcast): Likewise.
6082 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6083 Alan Hayward <alan.hayward@arm.com>
6084 David Sherwood <david.sherwood@arm.com>
6086 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
6087 scaled index even if the unscaled address was invalid.
6088 Don't increase the complexity of using a scale in that case.
6090 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6091 Alan Hayward <alan.hayward@arm.com>
6092 David Sherwood <david.sherwood@arm.com>
6094 * doc/rtl.texi: Rewrite the subreg rules so that they partition
6095 the inner register into REGMODE_NATURAL_SIZE bytes rather than
6096 UNITS_PER_WORD bytes.
6097 * emit-rtl.c (validate_subreg): Divide subregs into blocks
6098 based on REGMODE_NATURAL_SIZE of the inner mode.
6099 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
6100 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
6101 * expmed.c (lowpart_bit_field_p): Divide the value up into
6102 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
6103 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
6104 whether something is likely to occupy more than one register.
6106 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6109 * ipa-inline-transform.c (update_noncloned_frequencies): Use
6110 profile_count::adjust_for_ipa_scaling.
6111 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
6112 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
6114 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
6116 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6118 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
6119 (split_group): Add total_orig and total_new arguments, estimate the
6120 number of statements related to the store group without store merging
6121 and with store merging.
6122 (imm_store_chain_info::output_merged_store): Adjust split_group
6123 callers, punt if estimated number of statements with store merging
6124 is not smaller than estimated number of statements without it.
6126 (handled_load): Remove has_single_use checks.
6127 (pass_store_merging::process_store): Likewise.
6129 2017-11-09 Richard Biener <rguenther@suse.de>
6131 PR tree-optimization/82902
6132 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
6134 2017-11-09 Martin Liska <mliska@suse.cz>
6137 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
6140 2017-11-09 Martin Liska <mliska@suse.cz>
6142 PR tree-optimization/82669
6143 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
6145 2017-11-09 Martin Liska <mliska@suse.cz>
6147 PR gcov-profile/48463
6148 * coverage.c (coverage_begin_function): Output also end locus
6149 of a function and information whether the function is
6151 * gcov-dump.c (tag_function): Parse and print the information.
6152 * gcov.c (INCLUDE_MAP): Add include.
6153 (INCLUDE_SET): Likewise.
6154 (struct line_info): Move earlier in the source file because
6155 of vector<line_info> in function_info structure.
6156 (line_info::line_info): Likewise.
6157 (line_info::has_block): Likewise.
6158 (struct source_info): Add new member index.
6159 (source_info::get_functions_at_location): New function.
6160 (function_info::group_line_p): New function.
6161 (output_intermediate_line): New function.
6162 (output_intermediate_file): Use the mentioned function.
6163 (struct function_start): New.
6164 (struct function_start_pair_hash): Likewise.
6165 (process_file): Add code that identifies group functions.
6166 Assign lines either to global or function scope.
6167 (generate_results): Skip artificial functions.
6168 (find_source): Assign index for each source file.
6169 (read_graph_file): Read new flag artificial and end_line.
6170 (add_line_counts): Assign it either to global of function scope.
6171 (accumulate_line_counts): Isolate core of the function to
6172 accumulate_line_info and call it for both function and global
6174 (accumulate_line_info): New function.
6175 (output_line_beginning): Fix GNU coding style.
6176 (print_source_line): New function.
6177 (output_line_details): Likewise.
6178 (output_function_details): Likewise.
6179 (output_lines): Iterate both source (global) scope and function
6181 (struct function_line_start_cmp): New class.
6182 * doc/gcov.texi: Reflect changes in documentation.
6184 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6187 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
6188 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
6189 and similarly for not instead of neg.
6191 2017-11-08 Andi Kleen <ak@linux.intel.com>
6193 * config/i386/i386.opt: Add -mforce-indirect-call.
6194 * config/i386/predicates.md: Check for flag_force_indirect_call.
6195 * doc/invoke.texi: Document -mforce-indirect-call
6197 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
6199 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
6201 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
6202 config/riscv/riscv.c (predict.h): New include.
6203 (riscv_slow_unaligned_access_p): No longer static.
6204 (riscv_block_move_straight): Add require.
6205 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6207 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6210 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
6211 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
6212 nonimmediate_operand predicate for operand 1 instead of
6215 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6217 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
6219 * config/aarch64/constraints.md (Uml): New constraint.
6220 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
6223 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6225 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
6226 of two vec_duplicates into a vec_concat.
6228 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6230 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6231 Simplify vec_merge of vec_duplicate and vec_concat.
6232 * config/aarch64/constraints.md (Utq): New constraint.
6233 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
6236 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6238 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6239 Simplify vec_merge of vec_duplicate and const_vector.
6240 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
6242 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
6243 mode iterator. Update predicate on operand 1 to
6244 handle non-const_vec constants. Delete constraints.
6245 (*aarch64_combinez_be<mode>): Likewise for operand 2.
6247 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6249 PR tree-optimization/78821
6250 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
6252 (store_operand_info::store_operand_info): Initialize it to false.
6253 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
6254 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
6255 store in the group, and if chain_info is non-NULL, to ignore altogether
6257 (compatible_load_p): Fail if bit_not_p does not match.
6258 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
6259 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
6260 (pass_store_merging::process_store): Adjust
6261 terminate_all_aliasing_chains calls to pass NULL in all current spots,
6262 call terminate_all_aliasing_chains newly when adding a store into
6263 a chain with non-NULL chain_info.
6265 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6267 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
6269 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6271 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6273 (aarch64_layout_frame): Initialise emit_frame_chain.
6274 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
6275 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
6277 2017-11-08 Martin Liska <mliska@suse.cz>
6279 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
6280 of gimple_call_internal_p.
6282 2017-11-07 Tom de Vries <tom@codesourcery.com>
6284 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
6286 2017-11-07 Tom de Vries <tom@codesourcery.com>
6288 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6291 2017-11-08 Martin Liska <mliska@suse.cz>
6294 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
6296 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6298 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
6300 2017-11-07 Jakub Jelinek <jakub@redhat.com>
6303 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
6304 (*cmp<mode>_ccz_1): New insn with $k alternative.
6307 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
6308 RTX_COMM_COMPARE as commutative as well.
6309 (ix86_binary_operator_ok): Formatting fix.
6310 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
6311 *<code><mode>3<mask_name><round_saeonly_name>,
6312 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
6313 *<s>mul<mode>3_highpart<mask_name>,
6314 *vec_widen_umult_even_v16si<mask_name>,
6315 *vec_widen_umult_even_v8si<mask_name>,
6316 *vec_widen_umult_even_v4si<mask_name>,
6317 *vec_widen_smult_even_v16si<mask_name>,
6318 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
6319 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
6320 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
6321 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
6322 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
6323 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
6324 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
6325 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
6326 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
6327 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
6328 ix86_binary_operator_ok. Formatting fixes.
6329 (*<plusminus_insn><mode>3<mask_name><round_name>,
6330 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
6333 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
6335 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
6336 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
6337 eq and ne if TARGET_ISEL.
6338 (cmp): New code_iterator.
6339 (UNS, UNSU_, UNSIK): New code_attrs.
6340 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
6341 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
6343 ("eq<mode>3"): ... this.
6344 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
6346 ("ne<mode>3"): ... this.
6348 2017-11-07 Julia Koval <julia.koval@intel.com>
6351 * common/config/i386/i386-common.c
6352 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
6353 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
6354 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
6355 * config/i386/i386.opt: Ditto.
6356 * config/i386/i386.c (ix86_target_string): Ditto.
6357 (ix86_option_override_internal): Ditto.
6358 (ix86_init_mpx_builtins): Move MPX to args2.
6359 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
6360 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6361 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6362 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6363 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6364 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6365 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
6367 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
6370 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
6371 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
6372 (zero-extendsidi peephole2): Remove peephole.
6374 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6377 * stor-layout.c: Include attribs.h.
6378 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
6379 explicit lookup of "aligned" attribute.
6381 2017-11-07 Andrew Waterman <andrew@sifive.com>
6383 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
6384 (riscv_expand_block_move): Likewise.
6385 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
6387 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
6388 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
6389 * config/riscv/riscv.c (riscv_block_move_straight): New function.
6390 (riscv_adjust_block_mem): Likewise.
6391 (riscv_block_move_loop): Likewise.
6392 (riscv_expand_block_move): Likewise.
6393 * config/riscv/riscv.md (movmemsi): New pattern.
6395 2017-11-07 Michael Clark <michaeljclark@mac.com>
6397 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
6398 (MUSL_DYNAMIC_LINKER): Likewise.
6400 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
6402 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
6403 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
6406 2017-11-07 Richard Biener <rguenther@suse.de>
6408 * match.pd: Fix build.
6410 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
6411 Jackson Woodruff <jackson.woodruff@arm.com>
6413 PR tree-optimization/71026
6414 * match.pd: Canonicalize negate in division.
6416 2017-11-07 Sudakshina Das <sudi.das@arm.com>
6419 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
6421 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6423 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
6424 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
6426 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6428 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
6429 non-scalar integral types.
6430 * match.pd (negate_expr_p): Handle MINUS_EXPR.
6431 (-(A-B), -(~A)): New transformations.
6433 2017-11-07 Tom de Vries <tom@codesourcery.com>
6435 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
6436 semicolon after "do {} while (0)".
6437 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6438 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6439 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6440 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6441 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6443 2017-11-07 Tom de Vries <tom@codesourcery.com>
6445 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
6446 after "do {} while (0)".
6447 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6448 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6449 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6450 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6451 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6453 2017-11-07 Tom de Vries <tom@codesourcery.com>
6456 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
6458 (arm_rtx_costs_internal): Add missing semicolon after
6459 HANDLE_NARROW_SHIFT_ARITH call.
6461 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6463 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6464 disable isel if it was not set explicitly.
6466 2017-11-06 James Bowman <james.bowman@ftdichip.com>
6468 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
6469 (add_type_attribute) likewise.
6471 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
6473 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
6474 of struct ix86_frame.
6475 (ix86_initial_elimination_offset): Likewise.
6476 (ix86_expand_split_stack_prologue): Likewise.
6478 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
6480 * tree-vrp.h (enum value_range_type): Update stale comment.
6482 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6483 Alan Hayward <alan.hayward@arm.com>
6484 David Sherwood <david.sherwood@arm.com>
6486 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
6487 (aarch64_expand_vec_perm_const): Take the number of units too.
6488 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
6489 (aarch64_expand_vec_perm_const): Likewise.
6490 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
6491 (vec_perm<mode>): Update accordingly.
6493 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6494 Alan Hayward <alan.hayward@arm.com>
6495 David Sherwood <david.sherwood@arm.com>
6497 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
6498 Take the number of units too.
6499 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
6500 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
6501 but check for a vector mode before rather than after the call.
6502 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6503 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
6504 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
6505 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
6506 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
6507 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
6508 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
6509 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
6510 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
6511 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
6512 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
6513 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
6514 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
6515 (aarch64_sqdmull2_n<mode>): Update accordingly.
6517 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6518 Alan Hayward <alan.hayward@arm.com>
6519 David Sherwood <david.sherwood@arm.com>
6521 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
6522 the number of units too.
6523 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
6524 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
6525 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
6526 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
6527 (vec_store_lanesxi<mode>): Update accordingly.
6529 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6530 Alan Hayward <alan.hayward@arm.com>
6531 David Sherwood <david.sherwood@arm.com>
6533 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
6534 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
6535 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
6536 of units rather than the mode.
6537 * config/aarch64/iterators.md (nunits): New mode attribute.
6538 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
6539 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
6540 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
6541 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
6542 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
6543 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
6544 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
6545 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
6546 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
6547 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
6548 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6549 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
6550 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
6551 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
6552 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
6553 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
6554 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
6555 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
6556 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
6557 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
6558 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
6559 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
6560 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
6561 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6562 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
6563 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
6564 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6565 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
6566 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
6567 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
6568 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
6569 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
6570 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
6571 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
6572 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
6573 (aarch64_simd_vec_setv2di): Likewise.
6575 2017-11-06 Carl Love <cel@us.ibm.com>
6577 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
6579 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
6580 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
6582 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
6584 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
6585 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
6586 Add power 8 macro expansions.
6587 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
6588 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
6589 power 8 instructions. (VSX_XXBR): Add iterator.
6591 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
6593 * config/arm/arm.md (predicable_short_it): Change default to "no",
6594 improve documentation, remove uses that are identical to the default.
6595 (enabled_for_depr_it): Rename to enabled_for_short_it.
6596 * gcc/config/arm/arm-fixed.md (predicable_short_it):
6597 Remove default uses.
6598 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
6599 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
6600 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
6601 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
6603 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
6606 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
6607 float128 helper macros, which are no longer used after deleting
6608 the old 'q' built-in functions, and moving the round to odd
6609 built-in functions to being special built-in functions.
6610 (BU_FLOAT128_2): Likewise.
6611 (BU_FLOAT128_1_HW): Likewise.
6612 (BU_FLOAT128_2_HW): Likewise.
6613 (BU_FLOAT128_3_HW): Likewise.
6614 (FABSQ): Delete old 'q' built-in functions.
6615 (COPYSIGNQ): Likewise.
6616 (SQRTF128_ODD): Move round to odd built-in functions to be
6617 special built-in functions, so that we can handle
6618 -mabi=ieeelongdouble.
6619 (TRUNCF128_ODD): Likewise.
6620 (ADDF128_ODD): Likewise.
6621 (SUBF128_ODD): Likewise.
6622 (MULF128_ODD): Likewise.
6623 (DIVF128_ODD): Likewise.
6624 (FMAF128_ODD): Likewise.
6625 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
6626 built-in names to 'f128'.
6627 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
6628 old 'q' built-in functions, as the machine independent code for
6629 'f128' built-in functions handles this.
6630 (rs6000_expand_builtin): Add expansion for float128 round to odd
6631 functions, keying off on -mabi=ieeelongdouble of whether to use
6632 the KFmode or TFmode variant.
6633 (rs6000_init_builtins): Initialize the _Float128 round to odd
6635 * doc/extend.texi (PowerPC Built-in Functions): Document the old
6636 _Float128 'q' built-in functions are now mapped into the new
6637 'f128' built-in functions.
6639 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
6641 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
6642 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
6644 2017-11-06 David Malcolm <dmalcolm@redhat.com>
6647 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
6648 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
6649 * toplev.c: Include "ipa-fnsummary.h".
6650 (toplev::finalize): Call ipa_fnsummary_c_finalize.
6652 2017-11-06 Jakub Jelinek <jakub@redhat.com>
6654 PR tree-optimization/82838
6655 * gimple-ssa-store-merging.c
6656 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
6657 on a separate gimple_seq which is then appended to seq.
6659 2017-11-06 Jeff Law <law@redhat.com>
6662 * config/i386/i386.c (PROBE_INTERVAL): Remove.
6663 (get_probe_interval): New functions.
6664 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
6665 (ix86_adjust_stack_and_probe): Likewise.
6666 (output_adjust_stack_and_probe): Likewise.
6667 (ix86_emit_probe_stack_range): Likewise.
6668 (ix86_expand_prologue): Likewise.
6670 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6672 PR tree-optimization/82816
6673 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
6674 if the modes of the two types are the same.
6675 (convert_plusminus_to_widen): Likewise.
6677 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6679 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
6681 (usadv16qi): New define_expand.
6682 (usadv8hi): New define_expand.
6684 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
6687 * ipa-inline-transform.c (update_noncloned_frequencies): Always
6689 (inline_transform): Likewise.
6690 * predict.c (counts_to_freqs): Remove useless conditional.
6691 * profile-count.h (profile_count::apply_scale): Move sanity check.
6692 * tree-inline.c (copy_bb): Always scale.
6693 (copy_cfg_body): Likewise.
6695 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
6698 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
6700 (*cmp_ite0): Add enabled_for_depr_it attribute.
6701 (*cmp_ite1): Likewise.
6703 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6705 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
6708 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6710 * tree-vrp.c (vrp_int_const_binop): Return true on success and
6711 return the value by pointer.
6712 (extract_range_from_multiplicative_op_1): Update accordingly.
6713 Return as soon as an operation fails.
6715 2017-11-05 Tom de Vries <tom@codesourcery.com>
6718 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
6719 (DEF_SANITIZER_BUILTIN): ... here.
6720 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
6721 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
6723 2017-11-05 Tom de Vries <tom@codesourcery.com>
6726 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
6728 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
6729 ASM_OUTPUT_BEFORE_CASE_LABEL call.
6730 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
6732 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6733 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6734 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6736 2017-11-05 Tom de Vries <tom@codesourcery.com>
6739 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
6742 2017-11-04 Michael Clark <michaeljclark@mac.com>
6744 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
6745 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
6747 (*addsi3_extended): Likewise.
6748 (*addsi3_extended2): Likewise.
6749 (<optab>si3): Likewise.
6750 (<optab>di3): Likewise.
6751 (<optab><mode>3): Likewise.
6752 (<*optabe>si3_internal): Likewise.
6753 (zero_extendqi<SUPERQI:mode>2): Likewise.
6754 (*add<mode>hi3): Likewise.
6755 (*xor<mode>hi3): Likewise.
6756 (<optab>di3): Likewise.
6757 (*<optab>si3_extend): Likewise.
6758 (*sge<u>_<X:mode><GPR:mode>): Likewise.
6759 (*slt<u>_<X:mode><GPR:mode>): Likewise.
6760 (*sle<u>_<X:mode><GPR:mode>): Likewise.
6762 2017-11-04 Andrew Waterman <andrew@sifive.com>
6764 * config/riscv/riscv.c (riscv_option_override): Conditionally set
6765 TARGET_STRICT_ALIGN based upon -mtune argument.
6767 2017-11-04 Andrew Waterman <andrew@sifive.com>
6769 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
6771 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
6773 * config/i386/i386.c (choose_basereg): Use optional scratch
6774 register and add assertion.
6775 (x86_emit_outlined_ms2sysv_save): Use scratch register when
6776 needed, and don't allocate stack.
6777 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
6778 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
6779 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
6781 2017-11-03 Jeff Law <law@redhat.com>
6783 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
6784 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
6785 to probe at the start of a noreturn function.
6787 2017-11-03 Jakub Jelinek <jakub@redhat.com>
6789 PR tree-optimization/78821
6790 * gimple-ssa-store-merging.c: Update the file comment.
6791 (MAX_STORE_ALIAS_CHECKS): Define.
6792 (struct store_operand_info): New type.
6793 (store_operand_info::store_operand_info): New constructor.
6794 (struct store_immediate_info): Add rhs_code and ops data members.
6795 (store_immediate_info::store_immediate_info): Add rhscode, op0r
6796 and op1r arguments to the ctor, initialize corresponding data members.
6797 (struct merged_store_group): Add load_align_base and load_align
6799 (merged_store_group::merged_store_group): Initialize them.
6800 (merged_store_group::do_merge): Update them.
6801 (merged_store_group::apply_stores): Pick the constant for
6802 encode_tree_to_bitpos from one of the two operands, or skip
6803 encode_tree_to_bitpos if neither operand is a constant.
6804 (class pass_store_merging): Add process_store method decl. Remove
6805 bool argument from terminate_all_aliasing_chains method decl.
6806 (pass_store_merging::terminate_all_aliasing_chains): Remove
6807 var_offset_p argument and corresponding handling.
6808 (stmts_may_clobber_ref_p): New function.
6809 (compatible_load_p): New function.
6810 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
6811 if there is overlap and rhs_code is not INTEGER_CST. For
6812 non-overlapping stores terminate group if rhs is not mergeable.
6813 (get_alias_type_for_stmts): Change first argument from
6814 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
6815 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
6816 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
6818 (get_location_for_stmts): Change first argument from
6819 auto_vec<gimple *> & to vec<gimple *> &.
6820 (struct split_store): Remove orig_stmts data member, add orig_stores.
6821 (split_store::split_store): Create orig_stores rather than orig_stmts.
6822 (find_constituent_stmts): Renamed to ...
6823 (find_constituent_stores): ... this. Change second argument from
6824 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
6825 to info structures rather than the statements.
6826 (split_group): Rename ALLOW_UNALIGNED argument to
6827 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
6828 it. Adjust find_constituent_stores caller.
6829 (imm_store_chain_info::output_merged_store): Handle rhs_code other
6830 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
6831 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
6832 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
6833 (mem_valid_for_store_merging): New function.
6834 (handled_load): New function.
6835 (pass_store_merging::process_store): New method.
6836 (pass_store_merging::execute): Use process_store method. Adjust
6837 terminate_all_aliasing_chains caller.
6839 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6841 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
6842 Return true for more constants, symbols and label references.
6843 (aarch64_valid_floating_const): Remove unused function.
6845 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6848 * config/aarch64/aarch64.c (aarch64_layout_frame):
6849 Undo forcing of LR at bottom of frame.
6851 2017-11-03 Jeff Law <law@redhat.com>
6854 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
6855 for int_registers_saved.
6857 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
6858 extracted from tree-ssa-dom.c.
6859 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
6860 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
6861 (record_equivalences_from_incoming_edge): Add additional argument
6862 to single_pred_edge_ignoring_loop_edges call.
6863 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
6864 (uncprop_dom_walker::before_dom_children): Add additional argument
6865 to single_pred_edge_ignoring_loop_edges call.
6866 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
6867 single_pred_edge_ignoring_loop_edges rather than open coding.
6868 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
6870 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
6872 * match.pd (-(-A)): Rewrite.
6874 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6876 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
6877 (rs6000_emit_int_cmove): New declaration.
6878 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
6879 (rs6000_emit_sISEL): Delete.
6880 (rs6000_emit_int_cmove): Make non-static.
6881 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
6882 instead of rs6000_emit_sISEL.
6884 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
6886 * asan.c (create_cond_insert_point): Maintain profile.
6887 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6889 * basic-block.h (struct basic_block_def): Remove frequency.
6890 (EDGE_FREQUENCY): Use to_frequency
6891 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
6893 (find_traces): Update to use to_frequency.
6894 (find_traces_1_round): Likewise; use only IPA counts.
6895 (bb_to_key): Likewise.
6896 (connect_traces): Use IPA counts only.
6897 (copy_bb_p): Update to use to_frequency.
6898 (fix_up_crossing_landing_pad): Likewise.
6899 (sanitize_hot_paths): Likewise.
6900 * bt-load.c (basic_block_freq): Likewise.
6901 * cfg.c (init_flow): Set count_max to uninitialized.
6902 (check_bb_profile): Remove frequencies; check counts.
6903 (dump_bb_info): Do not dump frequencies.
6904 (update_bb_profile_for_threading): Update counts only.
6905 (scale_bbs_frequencies_int): Likewise.
6906 (MAX_SAFE_MULTIPLIER): Remove.
6907 (scale_bbs_frequencies_gcov_type): Update counts only.
6908 (scale_bbs_frequencies_profile_count): Update counts only.
6909 (scale_bbs_frequencies): Update counts only.
6910 * cfg.h (struct control_flow_graph): Add count-max.
6911 (update_bb_profile_for_threading): Update prototype.
6912 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
6913 (find_many_sub_basic_blocks): Likewise.
6914 * cfgcleanup.c (try_forward_edges): Likewise.
6915 (try_crossjump_to_edge): Likewise.
6916 * cfgexpand.c (expand_gimple_cond): Likewise.
6917 (expand_gimple_tailcall): Likewise.
6918 (construct_init_block): Likewise.
6919 (construct_exit_block): Likewise.
6920 * cfghooks.c (verify_flow_info): Check consistency of counts.
6921 (dump_bb_for_graph): Do not dump frequencies.
6922 (split_block_1): Do not update frequencies.
6923 (split_edge): Do not update frequencies.
6924 (make_forwarder_block): Do not update frequencies.
6925 (duplicate_block): Do not update frequencies.
6926 (account_profile_record): Do not update frequencies.
6927 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
6928 for global heuristics.
6929 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
6930 (expected_loop_iterations_unbounded): Use counts only.
6931 * cfgloopmanip.c (scale_loop_profile): Simplify.
6932 (create_empty_loop_on_edge): Simplify
6934 (duplicate_loop_to_header_edge): Simplify
6935 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
6936 (update_br_prob_note): Take care of removing note when profile
6938 (relink_block_chain): Do not dump frequency.
6939 (rtl_account_profile_record): Use to_frequency.
6940 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
6941 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
6942 (cgraph_update_edges_for_call_stmt_node): Likewise.
6943 (cgraph_edge::verify_count_and_frequency): Update.
6944 (cgraph_node::verify_node): Temporarily disable frequency verification.
6945 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
6946 to_cgraph_frequency.
6947 (cgraph_edge::rebuild_edges): Convert to ipa counts.
6948 * cgraphunit.c (init_lowered_empty_function): Do not initialize
6950 (cgraph_node::expand_thunk): Update profile.
6951 * except.c (dw2_build_landing_pads): Do not update frequency.
6952 * final.c (compute_alignments): Use to_frequency.
6953 (dump_basic_block_info): Do not dump frequency.
6954 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
6955 (dump_gimple_bb_header): Do not dump frequency.
6956 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
6958 * gimple-streamer-in.c (input_bb): Do not stream frequency.
6959 * gimple-streamer-out.c (output_bb): Do not stream frequency.
6960 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
6961 (init_before_recovery): Do not update frequency.
6962 (sched_create_recovery_edges): Do not update frequency.
6963 * hsa-gen.c (convert_switch_statements): Do not update frequency.
6964 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
6965 (ipa_cp_c_finalize): Set max_count to uninitialized.
6966 * ipa-fnsummary.c (get_minimal_bb): Use counts.
6967 (param_change_prob): Use counts.
6968 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
6970 * ipa-split.c (consider_split): Use to_frequency.
6971 (split_function): Use to_frequency.
6972 * ira-build.c (loop_compare_func): Likewise.
6973 (mark_loops_for_removal): Likewise.
6974 (mark_all_loops_for_removal): Likewise.
6975 * loop-doloop.c (doloop_modify): Do not update frequency.
6976 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
6978 * lto-streamer-in.c (input_function): Update count_max.
6979 * omp-expand.c (expand_omp_taskreg): Update count_max.
6980 * omp-simd-clone.c (simd_clone_adjust): Update profile.
6981 * predict.c (maybe_hot_frequency_p): Use to_frequency.
6982 (maybe_hot_count_p): Use ipa counts only.
6983 (maybe_hot_bb_p): Simplify.
6984 (maybe_hot_edge_p): Simplify.
6985 (probably_never_executed): Do not take frequency argument.
6986 (probably_never_executed_bb_p): Do not pass frequency.
6987 (probably_never_executed_edge_p): Likewise.
6988 (combine_predictions_for_bb): Check that profile is nonzero.
6989 (propagate_freq): Do not set frequency.
6990 (drop_profile): Simplify.
6991 (counts_to_freqs): Simplify.
6992 (expensive_function_p): Use to_frequency.
6993 (propagate_unlikely_bbs_forward): Simplify.
6994 (determine_unlikely_bbs): Simplify.
6995 (estimate_bb_frequencies): Add hack to silence graphite issues.
6996 (compute_function_frequency): Use ipa counts.
6997 (pass_profile::execute): Update.
6998 (rebuild_frequencies): Use counts only.
6999 (force_edge_cold): Use counts only.
7000 * profile-count.c (profile_count::dump): Dump new count types.
7001 (profile_count::differs_from_p): Check compatiblity.
7002 (profile_count::to_frequency): New function.
7003 (profile_count::to_cgraph_frequency): New function.
7004 * profile-count.h (struct function): Declare.
7005 (enum profile_quality): Add profile_guessed_local and
7006 profile_guessed_global0.
7007 (class profile_proability): Decrease number of bits to 29;
7008 update from_reg_br_prob_note and to_reg_br_prob_note.
7009 (class profile_count: Update comment; decrease number of bits
7010 to 61. Check compatibility.
7011 (profile_count::compatible_p): New private member function.
7012 (profile_count::ipa_p): New member function.
7013 (profile_count::operator<): Handle global zero correctly.
7014 (profile_count::operator>): Handle global zero correctly.
7015 (profile_count::operator<=): Handle global zero correctly.
7016 (profile_count::operator>=): Handle global zero correctly.
7017 (profile_count::nonzero_p): New member function.
7018 (profile_count::force_nonzero): New member function.
7019 (profile_count::max): New member function.
7020 (profile_count::apply_scale): Handle IPA scalling.
7021 (profile_count::guessed_local): New member function.
7022 (profile_count::global0): New member function.
7023 (profile_count::ipa): New member function.
7024 (profile_count::to_frequency): Declare.
7025 (profile_count::to_cgraph_frequency): Declare.
7026 * profile.c (OVERLAP_BASE): Delete.
7027 (compute_frequency_overlap): Delete.
7028 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7029 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7030 * sched-ebb.c (rank): Use counts only.
7031 * shrink-wrap.c (handle_simple_exit): Use counts only.
7032 (try_shrink_wrapping): Use counts only.
7033 (place_prologue_for_one_component): Use counts only.
7034 * tracer.c (find_best_predecessor): Use to_frequency.
7035 (find_trace): Use to_frequency.
7036 (tail_duplicate): Use to_frequency.
7037 * trans-mem.c (expand_transaction): Do not update frequency.
7038 * tree-call-cdce.c: Do not update frequency.
7039 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7040 (gimple_merge_blocks): Likewise.
7041 (gimple_split_edge): Likewise.
7042 (gimple_duplicate_sese_region): Likewise.
7043 (gimple_duplicate_sese_tail): Likewise.
7044 (move_sese_region_to_fn): Likewise.
7045 (gimple_account_profile_record): Likewise.
7046 (insert_cond_bb): Likewise.
7047 * tree-complex.c (expand_complex_div_wide): Likewise.
7048 * tree-eh.c (lower_resx): Update profile.
7049 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
7051 (initialize_cfun): Do not initialize frequencies
7052 (freqs_to_counts): Delete.
7053 (copy_cfg_body): Ignore count parameter.
7054 (copy_body): Update.
7055 (expand_call_inline): Update count_max.
7056 (optimize_inline_calls): Update count_max.
7057 (tree_function_versioning): Update count_max.
7058 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
7059 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
7061 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
7062 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
7063 (try_peel_loop): Likewise.
7064 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
7066 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
7067 (tree_transform_and_unroll_loop): Do not use frequencies
7068 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
7069 Use reliable prediction only.
7070 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
7071 * tree-ssa-sink.c (select_best_block): Use to_frequency.
7072 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
7073 probability scaling.
7074 * tree-ssa-threadupdate.c (create_block_for_threading): Do
7075 not update frequency
7076 (any_remaining_duplicated_blocks): Likewise.
7077 (update_profile): Likewise.
7078 (estimated_freqs_path): Delete.
7079 (freqs_to_counts_path): Delete.
7080 (clear_counts_path): Delete.
7081 (ssa_fix_duplicate_block_edges): Likewise.
7082 (duplicate_thread_path): Likewise.
7083 * tree-switch-conversion.c (gen_inbound_check): Use counts.
7084 * tree-tailcall.c (decrease_profile): Do not update frequency.
7085 (eliminate_tail_call): Likewise.
7086 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
7087 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
7088 (optimize_mask_stores): Likewise.
7089 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7090 * ubsan.c (ubsan_expand_null_ifn): Update profile.
7091 (ubsan_expand_ptr_ifn): Update profile.
7092 * value-prof.c (gimple_ic): Simplify.
7093 * value-prof.h (gimple_ic): Update prototype.
7094 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
7095 * ipa-inline.c (compute_uninlined_call_time): Be sure that
7097 (want_inline_self_recursive_call_p): Likewise.
7098 (resolve_noninline_speculation): Only cummulate defined counts.
7099 (inline_small_functions): Use nonzero_p.
7100 (ipa_inline): Do not access freed node.
7102 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7104 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7105 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
7107 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
7109 * config/riscv/riscv.c (riscv_legitimize_move): Handle
7110 non-legitimate address.
7112 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7114 * config/rs6000/rs6000.md (*lt0_disi): Delete.
7115 (*lt0_<mode>di, *lt0_<mode>si): New.
7117 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7119 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
7120 TARGET_PAIRED_FLOAT.
7122 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
7123 Jim Wilson <jim.wilson@linaro.org>
7125 * config/aarch64/aarch64-cores.def (saphira): New CPU.
7126 * config/aarch64/aarch64-tune.md: Regenerated.
7127 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
7128 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
7130 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
7132 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
7133 (arc_expand_prologue): Restore blink for millicode.
7134 * config/arc/linux.h (LINK_EH_SPEC): Defined.
7136 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
7139 * config/i386/i386.c (ix86_vector_duplicate_value): Use
7140 gen_vec_duplicate after forcing the scalar into a register.
7142 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
7144 * combine (try_combine): Print the insns input to try_combine to the
7147 2017-11-02 Steve Ellcey <sellcey@cavium.com>
7150 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
7151 Remove second argument from aarch64_process_target_attr call.
7152 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
7154 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
7156 (aarch64_handle_attr_arch): Remove second argument.
7157 (aarch64_handle_attr_cpu): Ditto.
7158 (aarch64_handle_attr_tune): Ditto.
7159 (aarch64_handle_attr_isa_flags): Ditto.
7160 (aarch64_process_one_target_attr): Ditto.
7161 (aarch64_process_target_attr): Ditto.
7162 (aarch64_option_valid_attribute_p): Remove second argument.
7163 on aarch64_process_target_attr call.
7165 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7167 * diagnostic.c: Include "selftest-diagnostic.h".
7168 (selftest::assert_location_text): New function.
7169 (selftest::test_diagnostic_get_location_text): New function.
7170 (selftest::diagnostic_c_tests): Call it.
7172 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7174 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
7175 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
7176 (class selftest::test_diagnostic_context): Move to...
7177 * selftest-diagnostic.c: New file.
7178 * selftest-diagnostic.h: New file.
7180 2017-11-02 James Bowman <james.bowman@ftdichip.com>
7182 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
7183 offset range for FT32B.
7184 * config/ft32/ft32.h: option "mcompress" enables relaxation.
7185 * config/ft32/ft32.md: Add TARGET_NOPM.
7186 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
7187 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
7189 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
7191 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
7193 2017-11-02 Jeff Law <law@redhat.com>
7195 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
7196 virtual keyword on FINAL OVERRIDE members.
7198 * tree-ssa-propagate.h (ssa_propagation_engine): Group
7199 virtuals together. Add virtual destructor.
7200 (substitute_and_fold_engine): Similarly.
7202 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
7204 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
7206 2017-11-02 Richard Biener <rguenther@suse.de>
7208 PR tree-optimization/82795
7209 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
7211 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7213 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
7215 (gcc_GAS_CHECK_FEATURE): Remove.
7216 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
7217 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
7218 * configure: Regenerate.
7220 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
7222 * config/arc/arc.c (hwloop_optimize): Account for empty
7225 2017-11-02 Richard Biener <rguenther@suse.de>
7228 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
7229 Truncate ARRAY_REF index and element size.
7231 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7233 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
7235 2017-11-01 Jeff Law <law@redhat.com>
7237 * tree-ssa-ccp.c (ccp_folder): New class derived from
7238 substitute_and_fold_engine.
7239 (ccp_folder::get_value): New member function.
7240 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
7241 (ccp_fold_stmt): Remove prototype.
7242 (ccp_finalize): Call substitute_and_fold from the ccp_class.
7243 * tree-ssa-copy.c (copy_folder): New class derived from
7244 substitute_and_fold_engine.
7245 (copy_folder::get_value): Renamed from get_value.
7246 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
7247 * tree-vrp.c (vrp_folder): New class derived from
7248 substitute_and_fold_engine.
7249 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
7250 (vrp_folder::get_value): New member function.
7251 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
7252 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
7253 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
7254 provide a class interface to folder/substitute routines.
7255 (ssa_prop_fold_stmt_fn): Remove typedef.
7256 (ssa_prop_get_value_fn): Likewise.
7257 (subsitute_and_fold): Remove prototype.
7258 (replace_uses_in): Likewise.
7259 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
7260 Renamed from replace_uses_in. Call the virtual member function
7261 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
7262 (substitute_and_fold_dom_walker): Remove initialization of
7263 data member entries for calbacks. Add substitute_and_fold_engine
7264 member and initialize it.
7265 (substitute_and_fold_dom_walker::before_dom_children0: Use the
7266 member functions for get_value, replace_phi_args_in c
7267 replace_uses_in, and fold_stmt calls.
7268 (substitute_and_fold_engine::substitute_and_fold): Renamed from
7269 substitute_and_fold. Remove assert. Update ctor call.
7271 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
7272 (ssa_prop_visit_phi_fn): Likewise.
7273 (class ssa_propagation_engine): New class to provide an interface
7275 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
7277 (ssa_prop_visit_phi): Likewise.
7278 (ssa_propagation_engine::simulate_stmt): Moved into class.
7279 Call visit_phi/visit_stmt from the class rather than via
7280 file scoped static variables.
7281 (ssa_propagation_engine::simulate_block): Moved into class.
7282 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
7283 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
7284 set file scoped statics for the visit_stmt/visit_phi callbacks.
7285 * tree-complex.c (complex_propagate): New class derived from
7286 ssa_propagation_engine.
7287 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
7288 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
7289 (tree_lower_complex): Call ssa_propagate via the complex_propagate
7291 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
7292 ssa_propagation_engine.
7293 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
7294 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
7295 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
7296 * tree-ssa-copy.c (copy_prop): New class derived from
7297 ssa_propagation_engine.
7298 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
7299 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
7300 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
7301 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
7302 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
7303 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
7304 (execute_vrp): Call ssa_propagate from the vrp_prop class.
7306 2017-11-01 Jakub Jelinek <jakub@redhat.com>
7308 PR rtl-optimization/82778
7309 PR rtl-optimization/82597
7310 * compare-elim.c (struct comparison): Add in_a_setter field.
7311 (find_comparison_dom_walker::before_dom_children): Remove killed
7312 bitmap and df_simulate_find_defs call, instead walk the defs.
7313 Compute last_setter and initialize in_a_setter. Merge definitions
7314 with first initialization for a few variables.
7315 (try_validate_parallel): Use insn_invalid_p instead of
7316 recog_memoized. Return insn rather than just the pattern.
7317 (try_merge_compare): Fix up comment. Don't uselessly test if
7318 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
7320 (execute_compare_elim_after_reload): Remove df_chain_add_problem
7323 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7324 Alan Hayward <alan.hayward@arm.com>
7325 David Sherwood <david.sherwood@arm.com>
7327 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
7328 aarch64_hard_regno_nregs to get the number of registers
7331 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7332 Alan Hayward <alan.hayward@arm.com>
7333 David Sherwood <david.sherwood@arm.com>
7335 * config/aarch64/constraints.md (Upl): Rename to...
7337 * config/aarch64/aarch64.md
7338 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
7341 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7342 Alan Hayward <alan.hayward@arm.com>
7343 David Sherwood <david.sherwood@arm.com>
7345 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
7346 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
7349 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7350 Alan Hayward <alan.hayward@arm.com>
7351 David Sherwood <david.sherwood@arm.com>
7353 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
7354 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
7355 (aarch64_evpc_dup): Generate rtl direcly, rather than using
7357 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
7358 of a single element.
7359 * config/aarch64/iterators.md: Add a comment above the permute
7360 unspecs to say that they are generated directly by
7361 aarch64_expand_vec_perm_const.
7362 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
7364 2017-11-01 Nathan Sidwell <nathan@acm.org>
7366 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
7368 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7370 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
7371 and medany code models, and describe what they do.
7373 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7375 Revert accidental duplicate:
7377 * combine.c (can_change_dest_mode): Reject changes in
7378 REGMODE_NATURAL_SIZE.
7380 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
7382 PR rtl-optimization/64682
7383 PR rtl-optimization/69567
7384 PR rtl-optimization/69737
7385 PR rtl-optimization/82683
7386 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
7387 register mentioned in the note, drop the note, unless it came from I3,
7388 in which case it should go to I3 again.
7390 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7392 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
7393 and return false if not.
7394 (clear_bytes_written_by, live_bytes_read): Update accordingly.
7396 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7398 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
7399 range is known to be empty.
7401 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7402 Alan Hayward <alan.hayward@arm.com>
7403 David Sherwood <david.sherwood@arm.com>
7405 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
7406 and CONST_VECTOR_NUNITS instead of computing the number of units from
7407 the byte sizes of the vector and element.
7408 (simplify_binary_operation_1): Likewise.
7409 (simplify_const_binary_operation): Likewise.
7410 (simplify_ternary_operation): Likewise.
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 * var-tracking.c (INT_MEM_OFFSET): Replace with...
7417 (int_mem_offset): ...this new function.
7418 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
7419 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
7420 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
7423 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7424 Alan Hayward <alan.hayward@arm.com>
7425 David Sherwood <david.sherwood@arm.com>
7427 * lower-subreg.c (interesting_mode_p): New function.
7428 (compute_costs, find_decomposable_subregs, decompose_register)
7429 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
7430 (resolve_clobber, dump_choices): Use it.
7432 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7433 Alan Hayward <alan.hayward@arm.com>
7434 David Sherwood <david.sherwood@arm.com>
7436 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
7443 truncations. Check the precision rather than the size.
7445 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7446 Alan Hayward <alan.hayward@arm.com>
7447 David Sherwood <david.sherwood@arm.com>
7449 * machmode.h (is_narrower_int_mode): New function
7450 * optabs.c (expand_float, expand_fix): Use it.
7451 * dwarf2out.c (rotate_loc_descriptor): Likewise.
7453 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7454 Alan Hayward <alan.hayward@arm.com>
7455 David Sherwood <david.sherwood@arm.com>
7457 * rtl.h (narrower_subreg_mode): New function.
7458 * ira-color.c (update_costs_from_allocno): Use it.
7460 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7461 Alan Hayward <alan.hayward@arm.com>
7462 David Sherwood <david.sherwood@arm.com>
7464 * optabs-query.h (convert_optab_p): New function, split out from...
7465 (convert_optab_handler): ...here.
7466 (widening_optab_handler): Delete.
7467 (find_widening_optab_handler): Remove permit_non_widening parameter.
7468 (find_widening_optab_handler_and_mode): Likewise. Provide an
7469 override that operates on mode class wrappers.
7470 * optabs-query.c (widening_optab_handler): Delete.
7471 (find_widening_optab_handler_and_mode): Remove permit_non_widening
7472 parameter. Assert that the two modes are the same class and that
7473 the "from" mode is narrower than the "to" mode. Use
7474 convert_optab_handler instead of widening_optab_handler.
7475 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
7476 instead of widening_optab_handler.
7477 * expr.c (expand_expr_real_2): Update calls to
7478 find_widening_optab_handler.
7479 * optabs.c (expand_widen_pattern_expr): Likewise.
7480 (expand_binop_directly): Take the insn_code as a parameter.
7481 (expand_binop): Only call find_widening_optab_handler for
7482 conversion optabs; use optab_handler otherwise. Update calls
7483 to find_widening_optab_handler and expand_binop_directly.
7484 Use convert_optab_handler instead of widening_optab_handler.
7485 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
7486 find_widening_optab_handler and use scalar_mode rather than
7488 (convert_plusminus_to_widen): Likewise.
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 * machmode.h (fixed_size_mode): New class.
7495 * rtl.h (get_pool_mode): Return fixed_size_mode.
7496 * gengtype.c (main): Add fixed_size_mode.
7497 * target.def (get_raw_result_mode): Return a fixed_size_mode.
7498 (get_raw_arg_mode): Likewise.
7499 * doc/tm.texi: Regenerate.
7500 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
7501 * targhooks.c (default_get_reg_raw_mode): Likewise.
7502 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
7503 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
7504 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
7505 (msp430_get_raw_result_mode): Likewise.
7506 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
7507 * dbxout.c (dbxout_parms): Require fixed-size modes.
7508 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
7509 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7510 * omp-low.c (lower_oacc_reductions): Likewise.
7511 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
7512 (simplify_subreg): Update accordingly.
7513 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
7514 (force_const_mem): Update accordingly. Return NULL_RTX for modes
7515 that aren't fixed-size.
7516 (get_pool_mode): Return a fixed_size_mode.
7517 (output_constant_pool_2): Take a fixed_size_mode.
7519 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7520 Alan Hayward <alan.hayward@arm.com>
7521 David Sherwood <david.sherwood@arm.com>
7523 * doc/rtl.texi (vec_series): Document.
7524 (const): Say that the operand can be a vec_series.
7525 * rtl.def (VEC_SERIES): New rtx code.
7526 * rtl.h (const_vec_series_p_1): Declare.
7527 (const_vec_series_p): New function.
7528 * emit-rtl.h (gen_const_vec_series): Declare.
7529 (gen_vec_series): Likewise.
7530 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
7531 (gen_vec_series): Likewise.
7532 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
7533 * simplify-rtx.c (simplify_unary_operation): Handle negations
7535 (simplify_binary_operation_series): New function.
7536 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
7537 (test_vector_ops_series): New function.
7538 (test_vector_ops): Call it.
7539 * config/powerpcspe/altivec.md (altivec_lvsl): Use
7540 gen_const_vec_series.
7541 (altivec_lvsr): Likewise.
7542 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
7544 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7545 Alan Hayward <alan.hayward@arm.com>
7546 David Sherwood <david.sherwood@arm.com>
7548 * doc/rtl.texi (const): Update description of address constants.
7549 Say that vector constants are allowed too.
7550 * common.md (E, F): Use CONSTANT_P instead of checking for
7552 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
7553 checking for CONST_VECTOR.
7554 * expmed.c (make_tree): Use build_vector_from_val for a CONST
7556 * expr.c (expand_expr_real_2): Check for vector modes instead
7557 of checking for CONST_VECTOR.
7558 * rtl.h (const_vec_p): New function.
7559 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
7560 (unwrap_const_vec_duplicate): Handle them here too.
7562 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7563 David Malcolm <dmalcolm@redhat.com>
7564 Alan Hayward <alan.hayward@arm.com>
7565 David Sherwood <david.sherwood@arm.com>
7567 * rtl.h (vec_duplicate_p): New function.
7568 * selftest-rtl.c (assert_rtx_eq_at): New function.
7569 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
7570 (assert_rtx_eq_at): Declare.
7571 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
7572 * selftest-run-tests.c (selftest::run_tests): Call it.
7573 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
7574 (simplify_unary_operation_1): Recursively handle vector duplicates.
7575 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
7577 (simplify_subreg): Handle subregs of vector duplicates.
7578 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
7579 (selftest::simplify_rtx_c_tests): New functions.
7581 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7582 Alan Hayward <alan.hayward@arm.com>
7583 David Sherwood <david.sherwood@arm.com>
7585 * emit-rtl.h (gen_const_vec_duplicate): Declare.
7586 (gen_vec_duplicate): Likewise.
7587 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
7589 (gen_const_vector): ...here.
7590 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
7591 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
7592 whose elements are all equal.
7593 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
7594 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
7595 (simplify_relational_operation): Likewise.
7596 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
7598 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
7599 (aarch64_expand_vector_init): Likewise.
7600 * config/arm/arm.c (neon_vdup_constant): Likewise.
7601 (neon_expand_vector_init): Likewise.
7602 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
7603 (arm_block_set_unaligned_vect): Likewise.
7604 (arm_block_set_aligned_vect): Likewise.
7605 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
7606 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
7607 (expand_vec_perm_even_odd_pack): Likewise.
7608 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
7609 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
7610 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
7611 gen_const_vec_duplicate.
7612 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
7613 * config/mips/mips.c (mips_gen_const_int_vector): Use
7614 gen_const_vec_duplicate.
7615 (mips_expand_vector_init): Use CONST0_RTX.
7616 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
7617 (define_split): Use gen_const_vec_duplicate.
7618 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
7619 (define_split): Use gen_const_vec_duplicate.
7620 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
7621 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
7622 * config/spu/spu.c (spu_const): Likewise.
7624 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7625 Alan Hayward <alan.hayward@arm.com>
7626 David Sherwood <david.sherwood@arm.com>
7628 * combine.c (can_change_dest_mode): Reject changes in
7629 REGMODE_NATURAL_SIZE.
7631 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
7633 * configure.ac (--enable-libssp): New.
7634 (gcc_cv_libc_provides_ssp): Check for explicit setting before
7635 trying to determine target-specific default. Adjust indentation.
7636 * configure: Regenerated.
7637 * doc/install.texi (Configuration): Expand --disable-libssp
7640 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
7642 config/i386/i386.c (ix86_expand_epilogue): Correct stack
7645 2017-10-31 Martin Jambor <mjambor@suse.cz>
7648 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
7650 2017-10-31 David Malcolm <dmalcolm@redhat.com>
7652 * auto-profile.c (autofdo_source_profile::read): Use
7653 UNKNOWN_LOCATION rather than 0.
7654 * diagnostic-core.h (warning_at_rich_loc): Rename to...
7655 (warning_at): ...this overload.
7656 (warning_at_rich_loc_n): Rename to...
7657 (warning_n): ...this overload.
7658 (error_at_rich_loc): Rename to...
7659 (error_at): ...this overload.
7660 (pedwarn_at_rich_loc): Rename to...
7661 (pedwarn): ...this overload.
7662 (permerror_at_rich_loc): Rename to...
7663 (permerror): ...this overload.
7664 (inform_at_rich_loc): Rename to...
7665 (inform): ...this overload.
7666 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
7667 (diagnostic_n_impl_richloc): Rename to...
7668 (diagnostic_n_impl): ...this rich_location *-based decl.
7669 (inform_at_rich_loc): Rename to...
7670 (inform): ...this, and add an assertion.
7671 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
7672 (warning_at_rich_loc): Rename to...
7673 (warning_at): ...this, and add an assertion.
7674 (warning_at_rich_loc_n): Rename to...
7675 (warning_n): ...this, and add an assertion.
7676 (warning_n): Update location_t-based implementation for removal of
7677 location_t-based diagnostic_n_impl.
7678 (pedwarn_at_rich_loc): Rename to...
7679 (pedwarn): ...this, and add an assertion.
7680 (permerror_at_rich_loc): Rename to...
7681 (permerror): ...this, and add an assertion.
7682 (error_n): Update for removal of location_t-based diagnostic_n_impl.
7683 (error_at_rich_loc): Rename to...
7684 (error_at): ...this, and add an assertion.
7685 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
7686 (driver::do_spec_on_infiles): Likewise.
7687 * substring-locations.c (format_warning_va): Update for renaming
7688 of inform_at_rich_loc.
7690 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
7692 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
7693 _Float<N>X built-in functions so that the variant without the
7694 "__builtin_" prefix is only enabled for the GNU C and Objective C
7695 languages when they are in non-strict ANSI/ISO mode.
7696 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
7697 * target.def (floatn_builtin_p): Add a target hook to control
7698 whether _Float<N> and _Float<N>X built-in functions without the
7699 "__builtin_" prefix are enabled, and return true for C and
7700 Objective C in the default hook. Include langhooks.h in
7702 * targhooks.h (default_floatn_builtin_p): Likewise.
7703 * targhooks.c (default_floatn_builtin_p): Likewise.
7704 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
7705 floatn_builtin_p target hook.
7706 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
7708 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
7709 Eric Botcazou <ebotcazou@adacore.com>
7711 PR rtl-optimization/81803
7712 * lra-constraints.c (curr_insn_transform): Also reload the whole
7713 register for a strict subreg no wider than a word if this is for
7714 a WORD_REGISTER_OPERATIONS target.
7716 2017-10-31 Jason Merrill <jason@redhat.com>
7718 * gdbinit.in: Skip over inlines from timevar.h.
7720 2017-10-31 Martin Liska <mliska@suse.cz>
7722 * doc/gcov.texi: Document new option.
7723 * gcov.c (print_usage): Likewise print it.
7724 (process_args): Support the argument.
7725 (format_count): New function.
7726 (format_gcov): Use the function.
7728 2017-10-31 Martin Liska <mliska@suse.cz>
7730 * gcov.c (struct name_map): do not use typedef.
7731 Define operator== and operator<.
7732 (name_search): Remove.
7733 (name_sort): Remove.
7734 (main): Do not allocate names.
7735 (process_file): Add vertical space.
7736 (generate_results): Use std::find.
7737 (release_structures): Do not release memory.
7738 (find_source): Use std::find.
7740 2017-10-31 Martin Liska <mliska@suse.cz>
7742 * gcov.c (struct line_info): Remove it's typedef.
7743 (line_info::line_info): Add proper ctor.
7744 (line_info::has_block): Do not use a typedef.
7745 (struct source_info): Do not use typedef.
7746 (circuit): Likewise.
7747 (get_cycles_count): Likewise.
7748 (output_intermediate_file): Iterate via vector iterator.
7749 (add_line_counts): Use std::vector methods.
7750 (accumulate_line_counts): Likewise.
7751 (output_lines): Likewise.
7753 2017-10-31 Martin Liska <mliska@suse.cz>
7755 * gcov.c (struct source_info): Remove typedef.
7756 (source_info::source_info): Add proper ctor.
7757 (accumulate_line_counts): Use struct, not it's typedef.
7758 (output_gcov_file): Likewise.
7759 (output_lines): Likewise.
7760 (main): Do not allocate an array.
7761 (output_intermediate_file): Use size of vector container.
7762 (process_file): Resize the vector.
7763 (generate_results): Do not preallocate, use newly added vector
7765 (release_structures): Do not release sources.
7766 (find_source): Use vector methods.
7767 (add_line_counts): Do not use typedef.
7769 2017-10-31 Martin Liska <mliska@suse.cz>
7771 * doc/gcov.texi: Document that.
7772 * gcov.c (add_line_counts): Mark lines with a non-executed
7774 (output_line_beginning): Handle such lines.
7775 (output_lines): Pass new argument.
7776 (output_intermediate_file): Print it in intermediate format.
7778 2017-10-31 Martin Liska <mliska@suse.cz>
7780 * color-macros.h: New file.
7781 * diagnostic-color.c: Factor out color related to macros to
7783 * doc/gcov.texi: Document -k option.
7784 * gcov.c (INCLUDE_STRING): Include string.h.
7785 (print_usage): Add -k option.
7786 (process_args): Parse it.
7787 (pad_count_string): New function.
7788 (output_line_beginning): Likewise.
7789 (DEFAULT_LINE_START): New macro.
7790 (output_lines): Support color output.
7792 2017-10-31 Martin Liska <mliska@suse.cz>
7794 PR gcov-profile/82633
7795 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
7796 their interaction with GCOV infrastructure.
7797 * configure.ac: Add -fkeep-{inline,static}-functions to
7799 * configure: Regenerate.
7801 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
7804 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
7806 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
7809 * config/rs6000/rs6000.md (allocate_stack): Force update interval
7810 into a register if it does not fit into an immediate offset field.
7812 2017-10-31 Olivier Hainque <hainque@adacore.com>
7814 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
7816 2017-10-31 Julia Koval <julia.koval@intel.com>
7818 * config.gcc: Add gfniintrin.h.
7819 * config/i386/gfniintrin.h: New.
7820 * config/i386/i386-builtin-types.def
7821 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7822 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7823 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7824 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7825 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7826 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
7827 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
7828 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
7829 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
7830 V64QI_FTYPE_V64QI_V64QI_INT): New types.
7831 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
7832 * config/i386/immintrin.h: Include gfniintrin.h.
7833 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
7835 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
7837 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
7839 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7841 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
7842 (arm_ashldi3_1bit): Remove pattern.
7843 (ashrdi3): Remove shift by 1 expansion.
7844 (arm_ashrdi3_1bit): Remove pattern.
7845 (lshrdi3): Remove shift by 1 expansion.
7846 (arm_lshrdi3_1bit): Remove pattern.
7847 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
7848 cost of ashldi3 by 1.
7849 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
7850 (<shift>di3_neon): Likewise.
7852 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7854 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
7855 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
7856 and (*aarch64_simd_mov<VQ:mode>).
7857 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
7858 pattern alternative.
7859 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
7860 attributes to match pattern alternative.
7862 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
7864 * config.gcc (powerpc*-*-*): Add emmintrin.h.
7865 * config/rs6000/emmintrin.h: New file.
7866 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
7868 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7870 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
7871 * (movdi_vfp_cortexa8): Remove pattern.
7873 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7875 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
7877 (Specific, alpha*-dec-osf5.1): Remove.
7878 (Specific, mips-sgi-irix5): Remove.
7879 (Specific, mips-sgi-irix6): Remove.
7881 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7884 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
7885 arguments to clear_bit_region_be.
7887 2017-10-30 Jim Wilson <wilson@tuliptree.org>
7889 * gimplify.c: Include memmodel.h.
7891 2017-10-30 Martin Jambor <mjambor@suse.cz>
7893 * omp-grid.c (grid_attempt_target_gridification): Also insert a
7894 condition whether loop should be executed at all.
7896 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
7898 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7899 gimple folding of vec_madd() intrinsics.
7900 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
7901 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7902 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
7904 2017-10-30 Richard Biener <rguenther@suse.de>
7906 PR tree-optimization/82762
7908 2017-10-23 Richard Biener <rguenther@suse.de>
7910 PR tree-optimization/82129
7912 2017-08-01 Richard Biener <rguenther@suse.de>
7914 PR tree-optimization/81181
7915 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7916 (compute_antic): ... end of iteration here.
7918 2017-10-30 Joseph Myers <joseph@codesourcery.com>
7920 * doc/invoke.texi (C Dialect Options): Document -std=c17,
7921 -std=iso9899:2017 and -std=gnu17.
7922 * doc/standards.texi (C Language): Document C17 support.
7923 * doc/cpp.texi (Overview): Mention -std=c17.
7924 (Standard Predefined Macros): Document C11 and C17 values of
7925 __STDC_VERSION__. Do not refer to C99 support as incomplete.
7926 * doc/extend.texi (Inline): Do not list individual options for
7927 standards newer than C99.
7928 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
7930 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
7933 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
7935 * asan.c (asan_finish_file): Align asan globals array by shadow
7938 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7941 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
7942 (struct store_immediate_info): Add bitregion_start and bitregion_end
7944 (store_immediate_info::store_immediate_info): Add brs and bre
7945 arguments and initialize bitregion_{start,end} from those.
7946 (struct merged_store_group): Add bitregion_start, bitregion_end,
7947 align_base and mask fields. Drop unnecessary struct keyword from
7948 struct store_immediate_info. Add do_merge method.
7949 (clear_bit_region_be): Use memset instead of loop storing zeros.
7950 (merged_store_group::do_merge): New method.
7951 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
7952 stores as long as the surrounding bitregions have no gaps.
7953 (merged_store_group::merge_overlapping): Use do_merge.
7954 (merged_store_group::apply_stores): Test that bitregion_{start,end}
7955 is byte aligned, rather than requiring that start and width are
7956 byte aligned. Drop unnecessary struct keyword from
7957 struct store_immediate_info. Allocate and populate also mask array.
7958 Make start of the arrays relative to bitregion_start rather than
7959 start and size them according to bitregion_{end,start} difference.
7960 (struct imm_store_chain_info): Drop unnecessary struct keyword from
7961 struct store_immediate_info.
7962 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
7963 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
7964 struct keyword from struct store_immediate_info.
7965 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
7966 between stores as long as the surrounding bitregions have no gaps.
7968 (struct split_store): Add orig non-static data member.
7969 (split_store::split_store): Initialize orig to false.
7970 (find_constituent_stmts): Return store_immediate_info *, non-NULL
7971 if there is exactly a single original stmt. Change stmts argument
7972 to pointer from reference, if NULL, don't push anything to it. Add
7973 first argument, use it to optimize skipping over orig stmts that
7974 are known to be before bitpos already. Simplify.
7975 (split_group): Return unsigned int count how many stores are or
7976 would be needed rather than a bool. Add allow_unaligned argument.
7977 Change split_stores argument from reference to pointer, if NULL,
7978 only do a dry run computing how many stores would be produced.
7979 Rewritten algorithm to use both alignment and misalign if
7980 !allow_unaligned and handle bitfield stores with gaps.
7981 (imm_store_chain_info::output_merged_store): Set start_byte_pos
7982 from bitregion_start instead of start. Compute allow_unaligned
7983 here, if true, do 2 split_group dry runs to compute which one
7984 produces fewer stores and prefer aligned if equal. Punt if
7985 new count is bigger or equal than original before emitting any
7986 statements, rather than during that. Remove no longer needed
7987 new_ssa_names tracking. Replace num_stmts with
7988 split_stores.length (). Use 32-bit stack allocated entries
7989 in split_stores auto_vec. Try to reuse original store lhs/rhs1
7990 if possible. Handle bitfields with gaps.
7991 (pass_store_merging::execute): Ignore bitsize == 0 stores.
7992 Compute bitregion_{start,end} for the stores and construct
7993 store_immediate_info with that. Formatting fixes.
7995 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
7998 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
7999 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8001 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8003 * gimplify.c: Include tm_p.h.
8005 * common.opt (gcoff): Re-add as ignored option.
8006 (gcoff1, gcoff2, gcoff3): Likewise.
8008 * Makefile.in (OBJS): Delete sdbout.o.
8009 (GTFILES): Delete $(srcdir)/sdbout.c.
8010 * debug.h: Delete sdb_debug_hooks.
8011 * final.c: Delete sdbout.h include.
8012 (final_scan_insn): Delete SDB_DEBUG check.
8013 (rest_of_clean_state): Likewise.
8014 * output.h: Delete sdb_begin_function_line.
8017 * toplev.c: Delete sdbout.h include.
8018 (process_options): Delete SDB_DEBUG check.
8019 * tree-core.h (tree_type_common): Delete pointer field of
8021 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8022 TYPE_SYMTAB_POINTER.
8023 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8024 (TYPE_SYMTAB_IS_POINTER): Delete.
8025 (TYPE_SYMTAB_IS_DIE): Renumber.
8026 * xcoffout.c: Refer to former sdbout.c file.
8027 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8029 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8030 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8031 (Debugging Options): Delete -gcoff.
8032 (-gstabs): Delete SDB reference.
8034 (-gcoff@var{level}): Delete.
8035 * doc/passes.texi (Debugging information output): Delete SDB and
8036 sdbout.c references.
8037 * doc/tm.texi: Regenerate.
8038 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8039 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8041 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
8042 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
8043 (SDB_DEBUGGING_INFO): Delete.
8044 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
8045 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
8046 * target.def (output_source_filename): Delete COFF reference.
8048 * common.opt (gcoff): Delete.
8049 (gxcoff+): Update Negative chain.
8050 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
8052 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
8053 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
8054 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
8056 (expand_function_start): Change sdb reference to past tense.
8057 (expand_function_end): Change sdb reference to past tense.
8058 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
8059 * opts.c (debug_type_names): Delete coff entry.
8060 (common_handle_option): Delete OPT_gcoff case.
8061 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
8063 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
8064 * config/cris/cris.h: Delete SDB reference in comment.
8065 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
8066 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
8067 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
8068 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
8070 (ix86_expand_prologue): Likewise.
8071 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
8072 * config/ia64/ia64.h: Likewise.
8073 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
8074 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
8076 * config/mmix/mmix.h: Likewise.
8077 * config/nds32/nds32.c: Likewise.
8078 * config/stormy/storym16.h: Likewise.
8079 * config/visium/visium.h: Likewise.
8080 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
8082 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
8084 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
8085 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
8086 FRAME_POINTER_REGNUM point at high end of local var area.
8088 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
8090 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
8091 Move comment around. Do not reset best_edge for a copiable
8092 destination if the copy would cause a partition change.
8093 (better_edge_p): Remove redundant check.
8095 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8097 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
8099 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
8101 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
8102 for math functions that have _Float<N> and _Float<N>X variants.
8103 (mathfn_built_in_2): Add support for math functions that have
8104 _Float<N> and _Float<N>X variants.
8105 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
8106 (expand_builtin_mathfn_ternary): Add support for fma with
8107 _Float<N> and _Float<N>X variants.
8108 (expand_builtin): Likewise.
8109 (fold_builtin_3): Likewise.
8110 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
8111 create math function _Float<N> and _Float<N>X variants as external
8113 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
8114 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
8115 the __builtin_ prefix and if not strict ansi, without the prefix.
8116 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
8117 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
8118 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
8119 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
8120 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
8121 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
8122 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
8123 function signatures for fma _Float<N> and _Float<N>X variants.
8124 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
8125 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
8126 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
8127 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
8128 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
8129 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
8130 * gencfn-macros.c (print_case_cfn): Add support for math functions
8131 that have _Float<N> and _Float<N>X variants.
8132 (print_define_operator_list): Likewise.
8133 (fltfn_suffixes): Likewise.
8135 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
8136 for math functions that have _Float<N> and _Float<N>X variants.
8137 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
8138 and _Float<N>X variants.
8139 (COPYSIGN): Likewise.
8142 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
8143 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
8145 (integer_valued_read_call_p): Likewise.
8146 * fold-const-call.c (fold_const_call_ss): Likewise.
8147 (fold_const_call_sss): Add support for copysign, fmin, and fmax
8148 _Float<N> and _Float<N>X variants.
8149 (fold_const_call_ssss): Add support for fma _Float<N> and
8150 _Float<N>X variants.
8151 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
8152 support for copysign and fma _Float<N> and _Float<N>X variants.
8153 (backprop::process_builtin_call_use): Likewise.
8154 * tree-call-cdce.c (can_test_argument_range); Add support for
8155 sqrt _Float<N> and _Float<N>X variants.
8156 (edom_only_function): Likewise.
8157 (get_no_error_domain): Likewise.
8158 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
8159 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
8160 copysign _Float<N> and _Float<N>X variants.
8161 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
8162 handled by machine independent code.
8163 (FMAF128): Likewise.
8164 * doc/cpp.texi (Common Predefined Macros): Document defining
8165 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
8166 fma _Float<N> and _Float<N>X variants.
8168 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8171 * config/i386/i386-modes.def (CCFPU): Remove definition.
8172 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
8173 (ix86_cc_modes_compatible): Ditto.
8174 (ix86_expand_carry_flag_compare): Ditto.
8175 (ix86_expand_int_movcc): Ditto.
8176 (ix86_expand_int_addcc): Ditto.
8177 (ix86_reverse_condition): Ditto.
8178 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
8179 Return true/false for unordered/ordered fp comparisons.
8180 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
8181 (ix86_prepare_fp_compare_args): Update for rename.
8182 (ix86_expand_fp_compare): Update for rename. Generate unordered
8183 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
8184 (ix86_expand_sse_compare_and_jump): Ditto.
8185 * config/i386/predicates.md (fcmov_comparison_operator):
8186 Remove CCFPU mode handling.
8187 (ix86_comparison_operator): Ditto.
8188 (ix86_carry_flag_operator): Ditto.
8189 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
8190 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
8191 (*cmpu<mode>_cc_i387): Ditto.
8192 (FPCMP): Remove mode iterator.
8193 (unord): Remove mode attribute.
8194 (unord_subst): New define_subst transformation
8195 (unord): New define_subst attribute.
8197 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
8198 (*cmpi<unord>xf_i387): Ditto.
8199 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
8200 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
8201 using unord_subst transformation.
8202 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
8203 (round_saeonly): Also handle CCFP mode.
8204 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
8205 Remove UNSPEC_SAHF unspec handling.
8207 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8209 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
8211 2017-10-27 Jeff Law <law@redhat.com>
8213 * gimple-ssa-sprintf.c: Include domwalk.h.
8214 (class sprintf_dom_walker): New class, derived from dom_walker.
8215 (sprintf_dom_walker::before_dom_children): New function.
8216 (struct call_info): Moved into sprintf_dom_walker class
8217 (compute_formath_length, handle_gimple_call): Likewise.
8218 (sprintf_length::execute): Call the dominator walker rather
8219 than walking the statements.
8221 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
8222 gimple statement locations.
8223 (check_array_bounds): Corresponding changes. Get the statement's
8224 location directly from wi->stmt.
8226 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
8229 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
8231 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8233 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
8234 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
8236 2017-10-27 Jakub Jelinek <jakub@redhat.com>
8239 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
8240 * config/i386/i386.c (maybe_get_pool_constant): Removed.
8241 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
8242 maybe_get_pool_constant.
8243 * config/i386/predicates.md (zero_extended_scalar_load_operand):
8246 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8248 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
8249 2.26 caveat. Update gas and gld versions.
8250 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
8253 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8255 * cgraph.h (set_malloc_flag): Declare.
8256 * cgraph.c (set_malloc_flag_1): New function.
8257 (set_malloc_flag): Likewise.
8258 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
8259 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
8261 (read_ipa_call_summary): Add support for reading is_return_callee.
8262 (write_ipa_call_summary): Stream is_return_callee.
8263 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
8264 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
8265 ipa-prop.h, ipa-fnsummary.h.
8266 (pure_const_names): Change to static.
8267 (malloc_state_e): Define.
8268 (malloc_state_names): Define.
8269 (funct_state_d): Add field malloc_state.
8270 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
8271 (check_retval_uses): New function.
8272 (malloc_candidate_p): Likewise.
8273 (analyze_function): Add support for malloc attribute.
8274 (pure_const_write_summary): Stream malloc_state.
8275 (pure_const_read_summary): Add support for reading malloc_state.
8276 (dump_malloc_lattice): New function.
8277 (propagate_malloc): New function.
8278 (warn_function_malloc): New function.
8279 (ipa_pure_const::execute): Call propagate_malloc and
8280 ipa_free_fn_summary.
8281 (pass_local_pure_const::execute): Add support for malloc attribute.
8282 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
8283 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
8285 2017-10-27 Martin Liska <mliska@suse.cz>
8287 PR gcov-profile/82457
8288 * doc/invoke.texi: Document that one needs a non-strict ISO mode
8289 for fork-like functions to be properly instrumented.
8291 2017-10-27 Richard Biener <rguenther@suse.de>
8294 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
8295 info when we redirected EH.
8297 2017-10-26 Michael Collison <michael.collison@arm.com>
8299 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
8301 (<optab>_trunchf<GPI:mode>2: New pattern.
8302 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
8303 * config/aarch64/iterators.md (wv): New mode attribute.
8304 (vf, VF): New mode attributes.
8305 (vgp, VGP): New mode attributes.
8306 (s): Update attribute with SImode and DImode prefixes.
8308 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8310 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
8311 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
8312 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
8313 (nios2_option_overide): Initialize it. Don't allow R0-relative
8314 addressing with PIC.
8315 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
8316 (nios2_symbolic_constant_p): Likewise.
8317 (nios2_legitimate_address_p): Likewise.
8318 (nios2_r0rel_section_name_p): New.
8319 (nios2_symbol_ref_in_r0rel_data_p): New.
8320 (nios2_emit_move_sequence): Handle r0rel_constant_p.
8321 (r0rel_constant_p): New.
8322 (nios2_print_operand_address): Handle r0rel_constant_p.
8323 (nios2_cdx_narrow_form_p): Likewise.
8324 * config/nios2/nios2.opt (mr0rel-sec=): New option.
8325 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
8326 (Nios II Options): Document -mr0rel-sec.
8328 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8330 * config/nios2/nios2.c: Include xregex.h.
8331 (nios2_gprel_sec_regex): New.
8332 (nios2_option_overide): Initialize it. Don't allow GP-relative
8333 addressing with PIC.
8334 (nios2_small_section_name_p): Check for regex match.
8335 * config/nios2/nios2.opt (mgprel-sec=): New option.
8336 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
8337 (Nios II Options): Document -mgprel-sec.
8339 2017-10-26 Jim Wilson <wilson@tuliptree.org>
8341 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
8343 2017-10-26 Tom de Vries <tom@codesourcery.com>
8345 PR tree-optimization/82707
8346 * gimple.c (gimple_copy): Fix unsharing of
8347 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
8349 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
8351 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
8352 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
8353 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
8354 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
8355 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
8356 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
8357 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
8358 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
8359 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
8360 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
8361 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
8362 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
8363 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
8364 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
8365 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
8366 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
8367 _mm512_mask_cmpunord_ps_mask): New intrinsics.
8369 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8371 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
8373 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
8374 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
8375 warning to rs6000.c. Remove the Undocumented flag, since it has
8377 (-mabi=ibmlongdouble): Likewise.
8378 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
8379 already set, set the default format for long double.
8380 (rs6000_debug_reg_global): Print whether long double is IBM or
8382 (rs6000_option_override_internal): Rework setting long double
8383 format. Only warn if the user is changing the long double default
8384 and they did not use -Wno-psabi.
8385 * doc/invoke.texi (PowerPC options): Update the documentation for
8386 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
8388 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8389 Alan Hayward <alan.hayward@arm.com>
8390 David Sherwood <david.sherwood@arm.com>
8392 * rtl.h (wider_subreg_mode): New function.
8393 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
8394 rather than an unsigned int *.
8395 * ira-color.c (regno_max_ref_width): Replace with...
8396 (regno_max_ref_mode): ...this new variable.
8397 (coalesced_pseudo_reg_slot_compare): Update accordingly.
8398 Use wider_subreg_mode.
8399 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
8400 rather than an unsigned int *.
8401 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
8402 (process_alt_operands): Likewise.
8403 (invariant_p): Likewise.
8404 * lra-spills.c (assign_mem_slot): Likewise.
8405 (add_pseudo_to_slot): Likewise.
8406 * lra.c (collect_non_operand_hard_regs): Likewise.
8407 (add_regs_to_insn_regno_info): Likewise.
8408 * reload1.c (regno_max_ref_width): Replace with...
8409 (regno_max_ref_mode): ...this new variable.
8410 (reload): Update accordingly. Update call to
8411 ira_sort_regnos_for_alter_reg.
8412 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
8413 (init_eliminable_invariants): Update to use regno_max_ref_mode.
8414 (scan_paradoxical_subregs): Likewise.
8416 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8418 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
8419 (aarch64_frame): Add emit_frame_chain boolean.
8420 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8421 Move eh_return case to aarch64_layout_frame.
8422 (aarch64_layout_frame): Initialize emit_frame_chain.
8423 (aarch64_expand_prologue): Use emit_frame_chain.
8425 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8427 * config/aarch64/aarch64.c (aarch64_layout_frame):
8428 Ensure LR is always stored at the bottom of the callee-saves.
8429 Remove rarely used frame layout which saves callee-saves at top of
8430 frame, so the store of LR can be used as a valid probe in all cases.
8432 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8434 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
8435 Improve unaligned TImode/TFmode base/offset split.
8437 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8438 Alan Hayward <alan.hayward@arm.com>
8439 David Sherwood <david.sherwood@arm.com>
8441 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
8442 * combine.c (find_single_use_1): Likewise.
8443 (expand_field_assignment): Likewise.
8444 (move_deaths): Likewise.
8445 * lra-constraints.c (simplify_operand_subreg): Likewise.
8446 (curr_insn_transform): Likewise.
8447 * lra.c (collect_non_operand_hard_regs): Likewise.
8448 (add_regs_to_insn_regno_info): Likewise.
8449 * rtlanal.c (reg_referenced_p): Likewise.
8450 (covers_regno_no_parallel_p): Likewise.
8452 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8454 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
8455 Don't print any bits outside the precision of the value.
8456 * wide-int.cc (test_printing): Add some new tests.
8458 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8460 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
8461 supports -xbrace_comment option.
8462 * configure: Regenerate.
8463 * config.in: Regenerate.
8464 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
8465 (ASM_CPU_SPEC): Use it.
8467 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8469 * target.def (static_rtx_alignment): New hook.
8470 * targhooks.h (default_static_rtx_alignment): Declare.
8471 * targhooks.c (default_static_rtx_alignment): New function.
8472 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
8473 * doc/tm.texi: Regenerate.
8474 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
8475 instead of targetm.constant_alignment. Remove call to
8477 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8478 (cris_preferred_mininum_alignment): New function, split out from...
8479 (cris_constant_alignment): ...here.
8480 (cris_static_rtx_alignment): New function.
8481 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
8483 (ix86_constant_alignment): ...here.
8484 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8485 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8486 (mmix_static_rtx_alignment): New function.
8487 * config/spu/spu.c (spu_static_rtx_alignment): New function.
8488 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8490 2017-10-26 Tamar Christina <tamar.christina@arm.com>
8493 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
8494 Add flag_trapping_math and flag_fp_int_builtin_inexact.
8496 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
8498 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
8499 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
8500 mark as a sign-extending load.
8501 (local_pic_load_u): Define.
8503 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
8506 * fold-const.c (operand_equal_for_comparison_p): Also return true
8507 if ARG0 is a simple variant of ARG1 with narrower precision.
8508 (fold_ternary_loc): Always pass unstripped operands to the predicate.
8510 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
8512 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
8514 * i386.h (processor_costs): Add gather_static, gather_per_elt,
8515 scatter_static, scatter_per_elt.
8516 * x86-tune-costs.h: Add new cost entries.
8518 2017-10-25 Richard Biener <rguenther@suse.de>
8520 * tree-ssa-sccvn.h (vn_eliminate): Declare.
8521 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
8522 class pass_fre): Move to ...
8523 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
8524 class pass_fre): ... here and adjust for statistics.
8526 2017-10-25 Jakub Jelinek <jakub@redhat.com>
8529 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
8530 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
8531 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
8532 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
8535 2017-10-25 Richard Biener <rguenther@suse.de>
8537 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
8538 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
8539 eliminate_push_avail, eliminate_insert): Move inside...
8540 (class eliminate_dom_walker): ... this class in preparation
8542 (fini_eliminate): Remove by merging with ...
8543 (eliminate): ... this function. Adjust for class changes.
8544 (pass_pre::execute): Remove fini_eliminate call.
8545 (pass_fre::execute): Likewise.
8547 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8550 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
8551 (VPERMI2, VPERMI2I): New mode iterators.
8552 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
8553 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
8555 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
8556 mode iterator. Remove 3 old define_insn patterns.
8557 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
8558 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
8559 VPERMI2 mode iterator, remove the other two expanders.
8560 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
8561 to use VPERMI2 mode iterator, add another alternative for vpermi2*
8562 instructions, remove the other two patterns.
8563 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
8564 mode iterator, remove the other two patterns.
8565 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
8566 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
8567 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
8568 and adjust argument order accordingly.
8569 (ix86_expand_vec_perm): Adjust caller.
8570 (expand_vec_perm_1): Likewise.
8571 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
8572 (expand_vec_perm_vpermt2_vpshub2): ... this.
8573 (ix86_expand_vec_perm_const_1): Adjust caller.
8574 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
8577 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
8578 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
8579 (vec_shl_<mode>): Remove unused expander.
8580 (avx512bw_<shift_insn><mode>3): New define_insn.
8581 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
8582 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
8584 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
8587 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
8590 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8592 PR rtl-optimization/82396
8593 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
8594 (autopref_multipass_init): Simplify initialization.
8595 (autopref_rank_data): Simplify sort order.
8596 * gcc/sched-int.h (autopref_multipass_data_): Remove
8597 multi_mem_insn_p, min_offset and max_offset.
8599 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8602 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8603 Check special value of flag_omit_frame_pointer.
8604 (aarch64_can_eliminate): Likewise.
8605 (aarch64_override_options_after_change_1): Simplify handling of
8606 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
8608 2017-10-24 Richard Biener <rguenther@suse.de>
8610 PR tree-optimization/82697
8611 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
8612 zero for conditional load and unconditional store.
8614 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8616 * doc/install.texi: Document bootstrap-cet.
8618 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8621 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
8622 ENDBR instruction at function entrance if function is only
8625 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8628 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
8629 patterns to better describe from which operation the CF is computed.
8630 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
8631 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
8632 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
8633 is 0, use _0 suffixed expanders instead of emitting a comparison
8636 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
8638 * config/i386/i386.md(*movsf_internal, *movdf_internal):
8639 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
8641 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
8644 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
8645 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
8646 * loop-iv.c (iv_get_reaching_def): Likewise.
8647 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
8648 variable is promoted and the partition contains undefined values.
8650 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8652 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
8654 (nios2_address_cost): Define.
8655 (nios2_legitimize_address): Recognize (exp + constant) directly.
8656 (TARGET_ADDRESS_COST): Define.
8658 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8660 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
8661 (nios2_symbolic_memory_operand_p): Declare.
8662 (nios2_split_large_constant): Declare.
8663 (nios2_split_symbolic_memory_operand): Declare.
8664 * config/nios2/nios2.c: Adjust includes.
8665 (nios2_symbolic_constant_allowed): New.
8666 (nios2_symbolic_constant_p): New.
8667 (nios2_plus_symbolic_constant_p): New.
8668 (nios2_valid_addr_expr_p): Recognize addresses involving
8670 (nios2_legitimate_address_p): Likewise, also LO_SUM.
8671 (nios2_symbolic_memory_operand_p): New.
8672 (nios2_large_constant_p): New.
8673 (nios2_split_large_constant): New.
8674 (nios2_split_plus_large_constant): New.
8675 (nios2_split_symbolic_memory_operand): New.
8676 (nios2_legitimize_address): Code refactoring. Handle addresses
8677 involving symbolic constants.
8678 (nios2_emit_move_sequence): Likewise.
8679 (nios2_print_operand): Improve error output.
8680 (nios2_print_operand_address): Handle LO_SUM.
8681 (nios2_cdx_narrow_form_p): Likewise.
8682 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
8683 operands involving symbolic constants.
8684 (movhi_internal, movsi_internal): Likewise.
8685 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
8686 (extendhisi2, extendqi<mode>2): Likewise.
8688 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8690 * tree-pass.h (PROP_rtl_split_insns): Define.
8691 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
8693 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8695 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
8697 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8700 * target.def (const_not_ok_for_debug_p): Default to
8701 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
8702 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
8703 * targhooks.c (default_const_not_ok_for_debug_p): New function.
8704 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
8705 which targetm.const_not_ok_for_debug_p returned true.
8706 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
8708 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
8710 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
8711 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
8712 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
8714 (ix86_const_not_ok_for_debug_p): New function.
8715 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
8716 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
8718 2017-10-23 David Malcolm <dmalcolm@redhat.com>
8721 * system.h: Conditionally include "unique-ptr.h" if
8722 INCLUDE_UNIQUE_PTR is defined.
8723 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
8724 of defining INCLUDE_UNIQUE_PTR before including "system.h".
8726 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
8728 * config/rl78/rl78.md: New define_expand "subdi3".
8730 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
8733 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
8734 DF_REF_INSN if DF_REF_INSN_INFO is false.
8736 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
8738 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
8739 xmm_move instead of sse_move.
8740 (sse_store_index): New function.
8741 (ix86_register_move_cost): Be more sensible about mismatch stall;
8742 model AVX moves correctly; make difference between sse->integer and
8744 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
8745 moves; make difference between SSE and AVX.
8746 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
8747 and zmm_move. Increase size of sse load and store tables;
8748 add unaligned load and store tables; add ssemmx_to_integer.
8749 * x86-tune-costs.h: Update all entries according to real
8750 move latencies from Agner Fog's manual and chip documentation.
8752 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8755 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
8756 * config/i386/constraints.md (Wf): New constraint.
8757 * config/i386/i386.md (UNSPEC_SBB): New unspec.
8758 (cmp<dwi>_doubleword): Removed.
8759 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
8760 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
8761 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
8762 expand with cmp<dwi>_doubleword. For LTU and GEU use
8763 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
8765 * common.opt (gcolumn-info): Enable by default.
8766 * doc/invoke.texi (gcolumn-info): Document new default.
8768 2017-10-23 Richard Biener <rguenther@suse.de>
8770 PR tree-optimization/82672
8771 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
8772 Fold the stmt if we propagated into it.
8774 2017-10-23 Richard Biener <rguenther@suse.de>
8776 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
8777 (bitmap_remove_expr_from_set): ... this. All callers call this
8778 for non-constant values.
8779 (bitmap_set_subtract): Rename to...
8780 (bitmap_set_subtract_expressions): ... this. Adjust and
8782 (bitmap_set_contains_value): Remove superfluous check.
8783 (bitmap_set_replace_value): Inline into single caller ...
8784 (bitmap_value_replace_in_set): ... here and simplify.
8785 (dependent_clean): Merge into ...
8786 (clean): ... this using an overload. Adjust.
8787 (prune_clobbered_mems): Adjust.
8788 (compute_antic_aux): Likewise.
8789 (compute_partial_antic_aux): Likewise.
8791 2017-10-23 Richard Biener <rguenther@suse.de>
8793 PR tree-optimization/82129
8795 2017-08-01 Richard Biener <rguenther@suse.de>
8797 PR tree-optimization/81181
8798 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8799 (compute_antic): ... end of iteration here.
8801 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8803 * target.def (starting_frame_offset): New hook.
8804 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
8805 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
8806 * doc/tm.texi.in: Regenerate.
8807 * hooks.h (hook_hwi_void_0): Declare.
8808 * hooks.c (hook_hwi_void_0): New function.
8809 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
8810 STARTING_FRAME_OFFSET.
8811 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
8812 * reload1.c (reload): Likewise.
8813 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
8814 instead of STARTING_FRAME_OFFSET.
8815 * function.c (try_fit_stack_local): Likewise.
8816 (assign_stack_local_1): Likewise
8817 (instantiate_virtual_regs): Likewise.
8818 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
8819 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
8820 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
8821 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
8822 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
8823 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
8824 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
8825 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
8826 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
8827 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
8828 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
8829 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
8830 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
8831 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
8832 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
8833 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
8834 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
8835 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
8836 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
8837 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
8838 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
8839 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
8840 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
8841 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
8842 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
8843 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
8844 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
8845 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
8846 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
8847 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
8848 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
8849 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
8850 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
8851 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
8852 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
8853 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
8854 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
8855 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
8856 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
8857 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
8858 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
8859 * config/avr/avr.c (avr_starting_frame_offset): Make static and
8860 return a HOST_WIDE_INT.
8861 (avr_builtin_setjmp_frame_value): Use it instead of
8862 STARTING_FRAME_OFFSET.
8863 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8864 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
8865 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
8867 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8868 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
8869 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
8870 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8871 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
8872 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
8873 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8874 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
8875 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
8876 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8877 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
8878 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
8880 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8881 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
8882 * config/mips/mips.c (mips_compute_frame_info): Refer to
8883 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8884 (mips_starting_frame_offset): New function.
8885 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8886 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
8887 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
8888 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
8889 and return a HOST_WIDE_INT.
8890 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8891 (mmix_initial_elimination_offset): Refer to
8892 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8893 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
8894 * config/pa/pa.c (pa_starting_frame_offset): New function.
8895 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
8896 (pa_expand_prologue): Likewise.
8897 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8898 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
8899 !FRAME_GROWS_DOWNWARD handling to...
8900 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8901 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
8902 !FRAME_GROWS_DOWNWARD handling to...
8903 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8904 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
8905 !FRAME_GROWS_DOWNWARD handling to...
8906 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8907 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
8909 (rs6000_starting_frame_offset): New function.
8910 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
8911 !FRAME_GROWS_DOWNWARD handling to...
8912 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8913 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
8914 !FRAME_GROWS_DOWNWARD handling to...
8915 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8916 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
8917 !FRAME_GROWS_DOWNWARD handling to...
8918 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8919 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
8920 (rs6000_starting_frame_offset): New function.
8921 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
8922 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
8923 * config/vax/vax.c (vax_starting_frame_offset): New function.
8924 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
8925 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8926 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
8927 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
8928 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8929 * system.h (STARTING_FRAME_OFFSET): Poison.
8931 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8933 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8934 SCALAR_TYPE_MODE instead of TYPE_MODE.
8936 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8937 Alan Hayward <alan.hayward@arm.com>
8938 David Sherwood <david.sherwood@arm.com>
8940 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
8942 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8943 Alan Hayward <alan.hayward@arm.com>
8944 David Sherwood <david.sherwood@arm.com>
8946 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
8948 2017-10-23 Richard Biener <rguenther@suse.de>
8950 PR tree-optimization/82129
8951 * tree-ssa-pre.c (bitmap_set_and): Remove.
8952 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
8953 canonicalizing expressions in the set to those with lowest
8954 ID rather than taking that from the first edge.
8956 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8958 * combine.c (rtx_equal_for_field_assignment_p): Use
8959 byte_lowpart_offset.
8961 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8962 Alan Hayward <alan.hayward@arm.com>
8963 David Sherwood <david.sherwood@arm.com>
8965 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
8966 to a SUBREG_PROMOTED_VAR.
8968 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8969 Alan Hayward <alan.hayward@arm.com>
8970 David Sherwood <david.sherwood@arm.com>
8972 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
8973 (expand_debug_source_expr): Likewise.
8974 * combine.c (combine_simplify_rtx): Likewise.
8975 * cse.c (fold_rtx): Likewise.
8976 * optabs.c (expand_float): Likewise.
8977 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8978 (simplify_binary_operation_1): Likewise.
8980 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8981 Alan Hayward <alan.hayward@arm.com>
8982 David Sherwood <david.sherwood@arm.com>
8984 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
8985 (record_promoted_value): Likewise.
8986 * expr.c (expand_expr_real_2): Likewise.
8987 * ree.c (update_reg_equal_equiv_notes): Likewise.
8988 (combine_set_extension): Likewise.
8989 * rtlanal.c (low_bitmask_len): Likewise.
8990 * simplify-rtx.c (neg_const_int): Likewise.
8991 (simplify_binary_operation_1): Likewise.
8993 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8994 Alan Hayward <alan.hayward@arm.com>
8995 David Sherwood <david.sherwood@arm.com>
8997 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
8998 * regcprop.c (maybe_mode_change): Likewise.
8999 * reload1.c (alter_reg): Likewise.
9001 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9003 * inchash.h (inchash::hash::add_wide_int): New function.
9004 * lto-streamer-out.c (hash_tree): Use it.
9006 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9008 * inchash.h (inchash::hash::add_wide_int): Rename to...
9009 (inchash::hash::add_hwi): ...this.
9010 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9011 (polymorphic_call_target_hasher::hash): Likewise.
9012 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9013 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9014 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9015 * lto-streamer-out.c (hash_tree): Likewise.
9016 * optc-save-gen.awk: Likewise.
9017 * tree.c (add_expr): Likewise.
9019 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9022 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9023 for ordered inequality comparisons even with TARGET_IEEE_FP.
9025 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9028 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9029 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9030 Expand with cmp<dwi>_doubleword.
9032 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9034 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9035 List CET intrinsics.
9036 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9037 specific to -fcf-protection option.
9039 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9041 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
9042 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
9043 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9044 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
9045 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
9046 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
9047 (extra_objs): Add cet.o for Linux/x86 targets.
9048 (tmake_file): Add i386/t-cet for Linux/x86 targets.
9049 * config/i386/cet.c: New file.
9050 * config/i386/cetintrin.h: Likewise.
9051 * config/i386/t-cet: Likewise.
9052 * config/i386/cpuid.h (bit_SHSTK): New.
9053 (bit_IBT): Likewise.
9054 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
9055 pass IBT and SHSTK bits.
9056 * config/i386/i386-builtin-types.def
9057 (VOID_FTYPE_UNSIGNED_PVOID): New.
9058 (VOID_FTYPE_UINT64_PVOID): Likewise.
9059 * config/i386/i386-builtin.def: Add CET intrinsics.
9060 * config/i386/i386-c.c (ix86_target_macros_internal): Add
9061 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
9062 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
9063 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
9065 * config/i386/i386.c (rest_of_insert_endbranch): New.
9066 (pass_data_insert_endbranch): Likewise.
9067 (pass_insert_endbranch): Likewise.
9068 (make_pass_insert_endbranch): Likewise.
9069 (ix86_notrack_prefixed_insn_p): Likewise.
9070 (ix86_target_string): Add -mibt, -mshstk flags.
9071 (ix86_option_override_internal): Add flag_cf_protection
9073 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
9074 (ix86_print_operand): Add 'notrack' prefix output.
9075 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
9076 (ix86_expand_builtin): Expand CET intrinsics.
9077 (x86_output_mi_thunk): Add 'endbranch' instruction.
9078 * config/i386/i386.h (TARGET_IBT): New.
9079 (TARGET_IBT_P): Likewise.
9080 (TARGET_SHSTK): Likewise.
9081 (TARGET_SHSTK_P): Likewise.
9082 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
9083 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
9084 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
9085 (builtin_setjmp_setup): New pattern.
9086 (builtin_longjmp): Likewise.
9087 (rdssp<mode>): Likewise.
9088 (incssp<mode>): Likewise.
9089 (saveprevssp): Likewise.
9090 (rstorssp): Likewise.
9091 (wrss<mode>): Likewise.
9092 (wruss<mode>): Likewise.
9093 (setssbsy): Likewise.
9094 (clrssbsy): Likewise.
9095 (nop_endbr): Likewise.
9096 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
9098 * config/i386/immintrin.h: Include <cetintrin.h>.
9099 * config/i386/linux-common.h
9100 (file_end_indicate_exec_stack_and_cet): New prototype.
9101 (TARGET_ASM_FILE_END): New.
9103 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9105 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
9106 latencies instead of having separate table; make difference between
9107 integer and float costs.
9108 * i386.h (processor_costs): Remove scalar_stmt_cost,
9109 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
9110 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
9112 * x86-tune-costs.h: Remove entries which has been removed in
9113 procesor_costs from all tables; make cond_taken_branch_cost
9114 and cond_not_taken_branch_cost COST_N_INSNS based.
9116 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9118 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
9120 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9122 * config/i386/i386.md (isa): Remove fma_avx512f.
9123 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
9124 <avx512>_fmadd_<mode>_mask3<round_name>,
9125 <avx512>_fmsub_<mode>_mask<round_name>,
9126 <avx512>_fmsub_<mode>_mask3<round_name>,
9127 <avx512>_fnmadd_<mode>_mask<round_name>,
9128 <avx512>_fnmadd_<mode>_mask3<round_name>,
9129 <avx512>_fnmsub_<mode>_mask<round_name>,
9130 <avx512>_fnmsub_<mode>_mask3<round_name>,
9131 <avx512>_fmaddsub_<mode>_mask<round_name>,
9132 <avx512>_fmaddsub_<mode>_mask3<round_name>,
9133 <avx512>_fmsubadd_<mode>_mask<round_name>,
9134 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
9135 (*vec_widen_umult_even_v16si<mask_name>,
9136 *vec_widen_smult_even_v16si<mask_name>): Likewise.
9137 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
9139 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9141 * extend.texi: Add 'nocf_check' documentation.
9142 * gimple.texi: Add second parameter to
9143 gimple_build_call_from_tree.
9144 * invoke.texi: Add -fcf-protection documentation.
9145 * rtl.texi: Add REG_CALL_NOTRACK documenation.
9147 2017-10-20 Richard Biener <rguenther@suse.de>
9149 PR tree-optimization/82473
9150 * tree-vect-loop.c (vectorizable_reduction): Properly get at
9151 the largest input type.
9153 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9155 * c-attribs.c (handle_nocf_check_attribute): New function.
9156 (c_common_attribute_table): Add 'nocf_check' handling.
9157 * gimple-parser.c: Add second argument NULL to
9158 gimple_build_call_from_tree.
9159 * attrib.c (comp_type_attributes): Check nocf_check attribute.
9160 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
9162 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
9163 * common.opt: Add fcf-protection flag.
9164 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
9165 * flag-types.h: Add enum cf_protection_level.
9166 * gimple.c (gimple_build_call_from_tree): Add second parameter.
9167 Add 'nocf_check' attribute propagation to gimple call.
9168 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
9169 (gimple_build_call_from_tree): Update prototype.
9170 (gimple_call_nocf_check_p): New function.
9171 (gimple_call_set_nocf_check): Likewise.
9172 * gimplify.c: Add second argument to gimple_build_call_from_tree.
9173 * ipa-icf.c: Add nocf_check attribute in statement hash.
9174 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
9175 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
9176 * toplev.c (process_options): Add flag_cf_protection handling.
9178 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9180 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
9182 2017-10-20 Richard Biener <rguenther@suse.de>
9184 PR tree-optimization/82603
9185 * tree-if-conv.c (predicate_mem_writes): Make sure to only
9186 remove false predicated stores.
9188 2017-10-20 Richard Biener <rguenther@suse.de>
9190 * graphite-isl-ast-to-gimple.c
9191 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
9192 Remove return value and simplify, dump copied stmt after lhs
9194 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
9195 Reduce dump verbosity.
9196 (gsi_insert_earliest): Likewise.
9197 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
9198 * graphite.c (print_global_statistics): Adjust dumping.
9199 (print_graphite_scop_statistics): Likewise.
9200 (print_graphite_statistics): Do not dump loops here.
9201 (graphite_transform_loops): But here.
9203 2017-10-20 Nicolas Roche <roche@adacore.com>
9205 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
9206 * configure: Regenerate.
9208 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9211 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
9212 functions when optimizing replace GIMPLE_RETURN stmts with
9213 calls to __builtin_unreachable ().
9216 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
9217 for -fsanitize=thread link of executables.
9218 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
9219 link of executables.
9222 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
9224 (<shift_insn><mode>3<mask_name>): Change the last of the 3
9225 define_insns for logical vector shifts to use VI248_AVX512BW
9226 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
9227 condition, useless isa and prefix attributes. Change the first
9228 2 of these define_insns to ...
9229 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
9230 define_insn for avx512vl.
9231 (<shift_insn><mode>3): ... and this, new define_insn without
9232 masking for non-avx512vl.
9235 * config/i386/sse.md (*andnot<mode>3,
9236 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
9237 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
9238 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
9239 not applied use empty suffix even for TARGET_AVX512VL.
9240 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
9241 is applied, supply evex,evex or evex,evex,evex instead of just
9244 2017-10-20 Julia Koval <julia.koval@intel.com>
9246 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
9247 (OPTION_MASK_ISA_GFNI_UNSET): New.
9248 (ix86_handle_option): Handle OPT_mgfni.
9249 * config/i386/cpuid.h (bit_GFNI): New.
9250 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
9251 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
9252 * config/i386/i386.c (ix86_target_string): Add -mgfni.
9253 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
9254 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
9255 * config/i386/i386.opt: Add mgfni.
9257 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
9259 * config/msp430/msp430.c (msp430_option_override): Disable
9260 -fdelete-null-pointer-checks.
9261 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
9263 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9265 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
9266 of x87 and SSE instructions.
9268 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9270 * asan.c (create_cond_insert_point): Do not update edge count.
9271 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
9272 (afdo_propagate_circuit): Likewise.
9273 (afdo_calculate_branch_prob): Likewise.
9274 (afdo_annotate_cfg): Likewise.
9275 * basic-block.h (struct edge_def): Remove count.
9276 (edge_def::count): New accessor.
9277 * bb-reorder.c (rotate_loop): Update.
9278 (find_traces_1_round): Update.
9279 (connect_traces): Update.
9280 (sanitize_hot_paths): Update.
9281 * cfg.c (unchecked_make_edge): Update.
9282 (make_single_succ_edge): Update.
9283 (check_bb_profile): Update.
9284 (dump_edge_info): Update.
9285 (update_bb_profile_for_threading): Update.
9286 (scale_bbs_frequencies_int): Update.
9287 (scale_bbs_frequencies_gcov_type): Update.
9288 (scale_bbs_frequencies_profile_count): Update.
9289 (scale_bbs_frequencies): Update.
9290 * cfganal.c (connect_infinite_loops_to_exit): Update.
9291 * cfgbuild.c (compute_outgoing_frequencies): Update.
9292 (find_many_sub_basic_blocks): Update.
9293 * cfgcleanup.c (try_forward_edges): Update.
9294 (try_crossjump_to_edge): Update
9295 * cfgexpand.c (expand_gimple_cond): Update
9296 (expand_gimple_tailcall): Update
9297 (construct_exit_block): Update
9298 * cfghooks.c (verify_flow_info): Update
9299 (redirect_edge_succ_nodup): Update
9300 (split_edge): Update
9301 (make_forwarder_block): Update
9302 (duplicate_block): Update
9303 (account_profile_record): Update
9304 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
9305 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
9306 * cfgloopmanip.c (scale_loop_profile): Update.
9308 (lv_adjust_loop_entry_edge): Update.
9309 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9310 (force_nonfallthru_and_redirect): Update.
9311 (purge_dead_edges): Update.
9312 (rtl_flow_call_edges_add): Update.
9313 * cgraphunit.c (init_lowered_empty_function): Update.
9314 (cgraph_node::expand_thunk): Update.
9315 * gimple-pretty-print.c (dump_probability): Update.
9316 (dump_edge_probability): Update.
9317 * gimple-ssa-isolate-paths.c (isolate_path): Update.
9318 * haifa-sched.c (sched_create_recovery_edges): Update.
9319 * hsa-gen.c (convert_switch_statements): Update.
9320 * ifcvt.c (dead_or_predicable): Update.
9321 * ipa-inline-transform.c (inline_transform): Update.
9322 * ipa-split.c (split_function): Update.
9323 * ipa-utils.c (ipa_merge_profiles): Update.
9324 * loop-doloop.c (add_test): Update.
9325 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
9326 * lto-streamer-in.c (input_cfg): Update.
9327 (input_function): Update.
9328 * lto-streamer-out.c (output_cfg): Update.
9329 * modulo-sched.c (sms_schedule): Update.
9330 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
9331 * predict.c (maybe_hot_edge_p): Update.
9332 (unlikely_executed_edge_p): Update.
9333 (probably_never_executed_edge_p): Update.
9334 (dump_prediction): Update.
9335 (drop_profile): Update.
9336 (propagate_unlikely_bbs_forward): Update.
9337 (determine_unlikely_bbs): Update.
9338 (force_edge_cold): Update.
9339 * profile.c (compute_branch_probabilities): Update.
9340 * reg-stack.c (better_edge): Update.
9341 * shrink-wrap.c (handle_simple_exit): Update.
9342 * tracer.c (better_p): Update.
9343 * trans-mem.c (expand_transaction): Update.
9344 (split_bb_make_tm_edge): Update.
9345 * tree-call-cdce.c: Update.
9346 * tree-cfg.c (gimple_find_sub_bbs): Update.
9347 (gimple_split_edge): Update.
9348 (gimple_duplicate_sese_region): Update.
9349 (gimple_duplicate_sese_tail): Update.
9350 (gimple_flow_call_edges_add): Update.
9351 (insert_cond_bb): Update.
9352 (execute_fixup_cfg): Update.
9353 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
9354 * tree-complex.c (expand_complex_div_wide): Update.
9355 * tree-eh.c (lower_resx): Update.
9356 (unsplit_eh): Update.
9357 (cleanup_empty_eh_move_lp): Update.
9358 * tree-inline.c (copy_edges_for_bb): Update.
9359 (freqs_to_counts): Update.
9360 (copy_cfg_body): Update.
9361 * tree-ssa-dce.c (remove_dead_stmt): Update.
9362 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9363 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9364 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9365 (unloop_loops): Update.
9366 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9367 * tree-ssa-loop-split.c (connect_loops): Update.
9368 (split_loop): Update.
9369 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
9370 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9371 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9372 * tree-ssa-reassoc.c (branch_fixup): Update.
9373 * tree-ssa-tail-merge.c (replace_block_by): Update.
9374 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9375 (compute_path_counts): Update.
9376 (update_profile): Update.
9377 (recompute_probabilities): Update.
9378 (update_joiner_offpath_counts): Update.
9379 (estimated_freqs_path): Update.
9380 (freqs_to_counts_path): Update.
9381 (clear_counts_path): Update.
9382 (ssa_fix_duplicate_block_edges): Update.
9383 (duplicate_thread_path): Update.
9384 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9385 (case_bit_test_cmp): Update.
9386 (collect_switch_conv_info): Update.
9387 (gen_inbound_check): Update.
9388 (do_jump_if_equal): Update.
9389 (emit_cmp_and_jump_insns): Update.
9390 * tree-tailcall.c (decrease_profile): Update.
9391 (eliminate_tail_call): Update.
9392 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9393 (vect_do_peeling): Update.
9394 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9395 * ubsan.c (ubsan_expand_null_ifn): Update.
9396 (ubsan_expand_ptr_ifn): Update.
9397 * value-prof.c (gimple_divmod_fixed_value): Update.
9398 (gimple_mod_pow2): Update.
9399 (gimple_mod_subtract): Update.
9400 (gimple_ic): Update.
9401 (gimple_stringop_fixed_value): Update.
9403 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9406 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
9408 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
9410 PR rtl-optimization/82395
9411 * ira-color.c (allocno_priority_compare_func): Fix comparison step
9412 based on non_spilled_static_chain_regno_p.
9414 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9416 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
9417 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
9418 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
9420 2017-10-19 Martin Sebor <msebor@redhat.com>
9422 PR tree-optimization/82596
9423 * tree.c (array_at_struct_end_p): Handle STRING_CST.
9425 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9427 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
9428 (get_mem_refs_of_builtin_call): Likewise.
9429 * builtins.c (expand_builtin_apply): Adjust call to
9430 allocate_dynamic_stack_space.
9431 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
9432 the third argument to allocate_dynamic_stack_space, otherwise -1.
9433 (expand_builtin): Deal with all alloca variants.
9434 (is_inexpensive_builtin): Likewise.
9435 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
9436 * calls.c (special_function_p): Deal with all alloca variants.
9437 (initialize_argument_information): Adjust call to
9438 allocate_dynamic_stack_space.
9439 (expand_call): Likewise.
9440 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
9441 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
9442 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
9443 use it for the stack usage computation.
9444 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
9445 * function.c (gimplify_parameters): Call build_alloca_call_expr.
9446 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
9447 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
9448 (in_loop_p): Remove first argument and useless check.
9449 (pass_walloca::execute): Remove useless test and adjust call to above.
9450 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
9451 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
9452 (gimplify_call_expr): Deal with all alloca variants.
9453 * hsa-gen.c (gen_hsa_alloca): Likewise.
9454 (gen_hsa_insns_for_call): Likewise.
9455 * ipa-pure-const.c (special_builtin_state): Likewise.
9456 * tree-chkp.c (chkp_build_returned_bound): Likewise.
9457 * tree-object-size.c (alloc_object_size): Likewise.
9458 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
9459 (call_may_clobber_ref_p_1): Likewise.
9460 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
9461 (ccp_fold_stmt): Likewise.
9462 (optimize_stack_restore): Likewise.
9463 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9464 (mark_all_reaching_defs_necessary_1): Likewise.
9465 (propagate_necessity): Likewise.
9466 (eliminate_unnecessary_stmts): Likewise.
9467 * tree.c (build_common_builtin_nodes): Build
9468 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
9469 (build_alloca_call_expr): New function.
9470 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
9471 (CASE_BUILT_IN_ALLOCA): Likewise.
9472 (build_alloca_call_expr): Declare.
9473 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
9475 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9478 * dwarf2out.c (new_die_raw): New static inline function.
9479 (new_die): Use it to create the DIE.
9480 (add_AT_external_die_ref): Likewise.
9481 (clone_die): Likewise.
9482 (clone_as_declaration): Likewise.
9483 (dwarf2out_vms_debug_main_pointer): Likewise.
9484 (base_type_die): Likewise. Remove early return for corner cases.
9485 Do not call add_pubtype on the DIE here.
9486 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
9487 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
9488 typedefs for base types with DW_AT_endianity. Make sure a DIE with
9489 native order exists for base types, attach the DIE manually and call
9490 add_pubtype on it. Do not equate a reverse order DIE to the type.
9492 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
9494 * config/arm/arm.c (align_ok_ldrd_strd): New function.
9495 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
9497 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
9499 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9501 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
9502 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
9503 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
9504 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
9505 * opts.c (sanitizer_opts): Add builtin.
9506 * ubsan.c (instrument_builtin): New function.
9507 (pass_ubsan::execute): Call it.
9508 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
9509 * doc/invoke.texi: Document -fsanitize=builtin.
9511 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
9512 builtins, store max (log2 (align), 0) into uchar field instead of
9513 align into uptr field.
9514 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
9515 store uchar 0 field instead of uptr 0 field.
9516 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
9517 instead of passing one address of struct with 2 locations pass
9518 two addresses of structs with 1 location each.
9519 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
9520 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
9521 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
9522 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
9523 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
9524 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
9525 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
9526 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
9528 2017-10-19 Martin Liska <mliska@suse.cz>
9531 * file-find.c (remove_prefix): Remove.
9532 * file-find.h (remove_prefix): Likewise.
9533 * gcc-ar.c: Remove smartness of lookup.
9535 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
9537 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
9538 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
9539 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
9541 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9544 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
9545 (setcc + and to xor + setcc): New peephole2.
9547 2017-10-19 Tom de Vries <tom@codesourcery.com>
9549 * doc/sourcebuild.texi (Test Directives, Variants of
9550 dg-require-support): Add dg-require-stack-size.
9552 2017-10-19 Martin Liska <mliska@suse.cz>
9555 * gimplify.c (gimplify_decl_expr): Do not instrument variables
9556 that have a large alignment.
9557 (gimplify_target_expr): Likewise.
9559 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
9561 PR rtl-optimization/82602
9562 * ira.c (rtx_moveable_p): Return false for volatile asm.
9564 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
9567 * config/i386/i386-modes.def (CCGZ): New CC mode.
9568 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
9569 * config/i386/predicates.md (ix86_comparison_operator):
9571 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9572 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
9573 with double-word subtraction.
9574 (put_condition_code): Handle CCGZmode.
9576 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
9578 * wide-int.cc (debug (const wide_int &)): New.
9579 (debug (const wide_int *)): New.
9580 (debug (const widest_int &)): New.
9581 (debug (const widest_int *)): New.
9583 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
9586 * lra-constraints.c (curr_insn_transform): Use non-input operand
9587 instead of output one for matched reload.
9589 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9591 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
9592 (tree-ssa-loop-ivopts.h): New header file.
9593 (struct builtin_info): New fields.
9594 (classify_builtin_1): Compute and record base and offset parts for
9595 memset builtin partition by calling strip_offset.
9596 (offset_cmp, fuse_memset_builtins): New functions.
9597 (finalize_partitions): Fuse adjacent memset partitions by calling
9599 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
9600 Expose the interface.
9601 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
9603 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9605 PR tree-optimization/82574
9606 * tree-loop-distribution.c (find_single_drs): New parameter. Check
9607 that data reference must be executed exactly once per iteration
9608 against the outermost loop in nest.
9609 (classify_partition): Update call to above function.
9611 2017-10-18 Richard Biener <rguenther@suse.de>
9613 PR tree-optimization/82591
9614 * graphite.c (graphite_transform_loops): Move code gen message
9616 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9617 Here. Handle scop_to_isl_ast failing.
9618 (scop_to_isl_ast): Limit the number of ISL operations.
9620 2017-10-18 Richard Biener <rguenther@suse.de>
9622 * graphite-isl-ast-to-gimple.c
9623 (translate_isl_ast_to_gimple::set_rename): Simplify.
9624 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
9625 (graphite_copy_stmts_from_block): ... here.
9626 (copy_bb_and_scalar_dependences): Simplify.
9627 (add_parameters_to_ivs_params): Canonicalize.
9628 (generate_entry_out_of_ssa_copies): Simplify.
9629 * graphite-sese-to-poly.c (extract_affine_name): Simplify
9630 by passing in ISL dimension.
9631 (parameter_index_in_region_1): Rename to ...
9632 (parameter_index_in_region): ... this.
9633 (extract_affine): Adjust assert, pass down parameter index.
9634 (add_param_constraints): Use range-info when available.
9635 (build_scop_context): Adjust.
9636 * sese.c (new_sese_info): Adjust.
9637 (free_sese_info): Likewise.
9638 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
9639 Remove unused typedefs.
9640 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
9642 2017-10-18 Martin Liska <mliska@suse.cz>
9644 * combine.c (simplify_compare_const): Add gcc_fallthrough.
9646 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9648 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
9649 (s390_sched_init): Do not reset s390_sched_state if we entered the
9650 current basic block via a fallthru edge and all others are unlikely.
9652 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9654 * config/s390/s390.c (NUM_SIDES): New variable.
9655 (LONGRUNNING_THRESHOLD): New variable.
9656 (LATENCY_FACTOR): New variable.
9657 (s390_sched_score): Decrease score for long-running instructions on
9659 (s390_sched_variable_issue): Perform bookkeeping for long-running
9662 2017-10-18 Richard Biener <rguenther@suse.de>
9664 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
9665 Simplify with removal of the parameter rename map.
9666 (set_rename): Likewise.
9667 (should_copy_to_new_region): Likewise.
9668 (graphite_copy_stmts_from_block): Likewise.
9669 (copy_bb_and_scalar_dependences): Remove initialization of
9670 unused copied_bb_map.
9672 (copy_internal_parameters): Likewise.
9673 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
9674 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
9675 Use INTEGRAL_TYPE_P.
9676 (parameter_index_in_region_1): Rename to ...
9677 (assign_parameter_index_in_region): ... this. Assert we have
9678 a parameter we handle.
9679 (scan_tree_for_params): Adjust.
9680 * sese.h (parameter_rename_map_t): Remove.
9681 (struct sese_info_t): Remove unused parameter_rename_map and
9682 copied_bb_map members.
9683 * sese.c (new_sese_info): Adjust.
9684 (free_sese_info): Likewise.
9686 2017-10-18 Martin Liska <mliska@suse.cz>
9689 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
9690 on an abnormal edge.
9692 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
9694 * doc/invoke.texi (ffunction-sections and fdata-sections):
9697 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9699 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
9700 the use statement can throw internally.
9702 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9704 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
9705 any RTX present on the RHS of a SET.
9706 * compare-elim.c (try_eliminate_compare): Restore comment.
9708 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9710 * langhooks.h (struct lang_hooks): Document that tree_size langhook
9711 may be also called on tcc_type nodes.
9712 * langhooks.c (lhd_tree_size): Likewise.
9714 2017-10-17 David Malcolm <dmalcolm@redhat.com>
9716 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
9717 format_warning_at_substring.
9718 (maybe_warn): Convert source_range * param to a location_t. Pass
9719 UNKNOWN_LOCATION rather than NULL to fmtwarn.
9720 (format_directive): Remove code to extract source_ranges and
9721 source_range * in favor of just a location_t.
9722 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
9724 * substring-locations.c (format_warning_va): Convert
9725 source_range * param to a location_t.
9726 (format_warning_at_substring): Likewise.
9727 * substring-locations.h (format_warning_va): Likewise.
9728 (format_warning_at_substring): Likewise.
9730 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
9732 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
9734 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
9735 and scatter/gather ops.
9737 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
9738 vec_gather_load and vec_scatter_store.
9739 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
9740 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
9741 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
9742 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
9743 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
9744 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
9746 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
9748 * reg-stack.c (compare_for_stack_reg): Add bool argument.
9749 Detect FTST instruction and handle its register pops. Only pop
9750 second operand if can_pop_second_op is true.
9751 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
9752 set can_pop_second_op to false in the compare_for_stack_reg call.
9754 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
9755 output_fp_compare for stack register operands.
9756 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
9757 instructions here. Do not emit stack register pops here. Assert
9758 that FCOMPP pops next to top stack register. Rewrite function.
9760 2017-10-17 Nathan Sidwell <nathan@acm.org>
9763 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
9764 use DECL_ASSEMBLER_NAME_RAW.
9767 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
9770 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
9771 Wilco Dijkstra <wilco.dijkstra@arm.com>
9773 * builtins.c (expand_builtin_update_setjmp_buf): Add a
9774 converstion to Pmode from the buf_addr.
9776 2017-10-17 Richard Biener <rguenther@suse.de>
9778 * graphite-dependences.c (scop_get_reads_and_writes): Change
9779 output parameters to references.
9781 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
9783 PR 71026/tree-optimization
9784 * fold-const.c (distribute_real_division): Removed.
9785 (fold_binary_loc): Remove calls to distribute_real_divison.
9787 2017-10-17 Richard Biener <rguenther@suse.de>
9789 * graphite-scop-detection.c
9790 (scop_detection::stmt_has_simple_data_refs_p): Always use
9791 the full nest as region.
9792 (try_generate_gimple_bb): Likewise.
9793 * sese.c (scalar_evolution_in_region): Simplify now that
9794 SCEV can handle instantiation in regions.
9795 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
9796 in the non-loop part of a function if requested.
9798 2017-10-17 Richard Biener <rguenther@suse.de>
9800 PR tree-optimization/82563
9801 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
9803 (graphite_regenerate_ast_isl): Call it.
9804 * graphite-scop-detection.c (build_scops): Remove entry edge split.
9806 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9808 PR tree-optimization/82549
9809 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
9810 Formatting fixes. Instead of calling make_bit_field_ref with negative
9813 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
9815 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
9816 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
9817 _mm_maskz_reduce_ss): New.
9818 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
9819 __builtin_ia32_reducess_mask): Ditto..
9820 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
9821 * config/i386/sse.md (reduces<mode>): Renamed to ...
9822 (reduces<mode><mask_scalar_name>): ... this.
9823 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
9825 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
9826 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
9828 2017-10-16 David Malcolm <dmalcolm@redhat.com>
9830 * Makefile.in (OBJS): Add unique-ptr-tests.o.
9831 * selftest-run-tests.c (selftest::run_tests): Call
9832 selftest::unique_ptr_tests_cc_tests.
9833 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
9834 * unique-ptr-tests.cc: New file.
9836 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
9839 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9841 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9842 (make_hard_regno_born, make_hard_regno_dead): Update
9843 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
9845 2017-10-16 Jeff Law <law@redhat.com>
9847 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9849 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
9851 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
9853 2017-10-16 Olivier Hainque <hainque@adacore.com>
9855 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
9856 with_cpu if we were configured for an e500v2 target cpu name.
9858 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
9860 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
9861 * doc/invoke.texi: Document +nodsp as a valid extension for
9864 2017-10-16 Martin Liska <mliska@suse.cz>
9866 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
9867 (test_set_range): Likewise.
9868 (test_range_functions): Rename to ...
9869 (test_bit_in_range): ... this.
9870 (sbitmap_c_tests): Add new test.
9872 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9874 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
9876 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
9877 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
9879 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9881 * config/aarch64/aarch64-builtins.c
9882 (aarch64_types_quadopu_lane_qualifiers): New.
9883 (TYPES_QUADOPU_LANE): New.
9884 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
9885 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
9886 (aarch64_<sur>dot_laneq<vsi2qi>): New.
9887 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
9888 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
9889 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
9890 (Vdottype, DOTPROD): New.
9891 (sur): Add SDOT and UDOT.
9893 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9895 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
9896 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
9897 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
9899 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
9900 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
9901 Enable TARGET_DOTPROD.
9902 (cortex-a75.cortex-a55): Likewise.
9903 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
9905 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9907 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
9908 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
9910 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
9912 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
9913 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
9914 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
9915 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
9916 * config/arm/types.md (neon_dot, neon_dot_q): New.
9917 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
9919 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9921 * config/arm/arm.h (TARGET_DOTPROD): New.
9922 * config/arm/arm.c (arm_arch_dotprod): New.
9923 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
9924 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
9925 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
9926 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
9927 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
9928 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
9929 * doc/invoke.texi (armv8.2-a): Document dotprod
9931 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9933 * i386.c (ix86_vec_cost): New function.
9934 (ix86_rtx_costs): Handle vector operations better.
9935 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
9936 * x86-tune-costs.h: Add new costs to all tables.
9938 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9940 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
9942 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
9943 divsd, sqrtss and sqrtsd
9944 * x86-tune-costs.h: Add new entries to all costs.
9945 (znver1_cost): Fix to match real instruction latencies.
9947 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9948 Michael Collison <michael.collison@arm.com>
9950 * compare-elim.c: Include emit-rtl.h.
9951 (can_merge_compare_into_arith): New function.
9952 (try_validate_parallel): Likewise.
9953 (try_merge_compare): Likewise.
9954 (try_eliminate_compare): Call the above when no previous clobber
9956 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
9959 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9963 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
9964 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
9968 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
9969 to be any operand_equal_p operands. For & (B - 1) require
9970 B to be power of 2. Recognize
9971 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
9973 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9976 * optabs.c (expand_memory_blockage): Fix call of
9977 targetm.have_memory_blockage.
9979 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9982 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
9983 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
9984 objects to extra_objs instead of overwriting it.
9986 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9988 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
9989 Use any_fp_register_operand as operand[3] predicate. Simplify
9990 equality test for operands[2] and operands[4] memory location.
9991 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
9992 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
9993 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
9994 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
9995 any_fp_register_operand as operand[1] predicate. Simplify
9996 equality test for operands[0] and operands[3] memory location.
9997 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
9998 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
9999 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10001 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10003 * target-insns.def: Add memory_blockage.
10004 * optabs.c (expand_memory_blockage): New function.
10005 (expand_asm_memory_barrier): Rename ...
10006 (expand_asm_memory_blockage): ... to this.
10007 (expand_mem_thread_fence): Call expand_memory_blockage
10008 instead of expand_asm_memory_barrier.
10009 (expand_mem_singnal_fence): Ditto.
10010 (expand_atomic_load): Ditto.
10011 (expand_atomic_store): Ditto.
10012 * doc/md.texi (Standard Pattern Names For Generation):
10013 Document memory_blockage instruction pattern.
10015 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10017 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10018 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10019 * config/rl78/rl78.md: New define_expand "adddi3".
10021 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10023 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10026 2017-10-13 Jeff Law <law@redhat.com>
10028 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10030 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10033 * internal-fn.c (expand_mul_overflow): If both operands have
10034 the same highpart of -1 or 0 and the topmost bit of lowpart
10035 is different, overflow is if res <= 0 rather than res < 0.
10037 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10039 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10040 TARGET_P9_VECTOR code for unaligned_load case.
10042 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10044 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
10046 2017-10-13 Nathan Sidwell <nathan@acm.org>
10048 * tree-core.h (tree_contains_struct): Make bool.
10049 * tree.c (tree_contains_struct): Likewise.
10050 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
10051 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
10052 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
10053 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
10054 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
10056 2017-10-13 Richard Biener <rguenther@suse.de>
10058 * graphite-isl-ast-to-gimple.c
10059 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
10060 parameters and dominance check.
10061 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
10062 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
10063 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
10064 Do not update SSA form here or do intermediate IL verification.
10065 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
10066 (graphite_initialize): Remove check on the number of loops in
10067 the function and inline into graphite_transform_loops.
10068 (graphite_finalize): Inline into graphite_transform_loops.
10069 (graphite_transform_loops): Perform SSA update and IL verification
10071 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
10073 2017-10-13 Richard Biener <rguenther@suse.de>
10075 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
10076 graphite_expression_type_precision): Avoid global constructor
10078 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
10079 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
10080 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
10081 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
10082 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
10083 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
10085 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
10088 * config/i386/i386.h (ix86_red_zone_size): New.
10089 * config/i386/i386.md (push peephole2s): Replace
10090 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
10092 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10093 Alan Hayward <alan.hayward@arm.com>
10094 David Sherwood <david.sherwood@arm.com>
10096 * combine.c (can_change_dest_mode): Reject changes in
10097 REGMODE_NATURAL_SIZE.
10099 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10100 Alan Hayward <alan.hayward@arm.com>
10101 David Sherwood <david.sherwood@arm.com>
10103 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
10104 (expand_debug_source_expr): Likewise.
10105 * combine.c (combine_simplify_rtx): Likewise.
10106 * cse.c (fold_rtx): Likewise.
10107 * fwprop.c (canonicalize_address): Likewise.
10108 * targhooks.c (default_shift_truncation_mask): Likewise.
10110 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10111 Alan Hayward <alan.hayward@arm.com>
10112 David Sherwood <david.sherwood@arm.com>
10114 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
10115 (widened_mode): Likewise.
10116 (expand_unop): Likewise.
10117 * ree.c (transform_ifelse): Likewise.
10118 (merge_def_and_ext): Likewise.
10119 (combine_reaching_defs): Likewise.
10120 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10122 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10123 Alan Hayward <alan.hayward@arm.com>
10124 David Sherwood <david.sherwood@arm.com>
10126 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
10127 * combine.c (gen_lowpart_for_combine): Likewise.
10128 * dwarf2out.c (rtl_for_decl_location): Likewise.
10129 * final.c (alter_subreg): Likewise.
10130 * rtlhooks.c (gen_lowpart_general): Likewise.
10131 (gen_lowpart_if_possible): Likewise.
10133 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10134 Alan Hayward <alan.hayward@arm.com>
10135 David Sherwood <david.sherwood@arm.com>
10137 * calls.c (expand_call): Use subreg_lowpart_offset.
10138 * cse.c (cse_insn): Likewise.
10139 * regcprop.c (copy_value): Likewise.
10140 (copyprop_hardreg_forward_1): Likewise.
10142 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10145 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
10146 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
10147 =Q constraints to +Q and into insn condition add check
10148 that operands[0] and operands[1] are equal.
10149 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
10150 =Q constraints to +Q and into insn condition add check
10151 that operands[0] is equal to either operands[1] or operands[2].
10154 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
10155 instead of handling MINUS_EXPR twice (once for each argument),
10156 canonicalize operand order and handle just once, use rtype where
10157 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
10160 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
10161 any values of __C while still being pattern recognizable as a simple
10162 rotate instruction.
10164 2017-10-13 Richard Biener <rguenther@suse.de>
10166 PR tree-optimization/82451
10168 2017-10-02 Richard Biener <rguenther@suse.de>
10170 PR tree-optimization/82355
10171 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10172 a mapping for the enclosing loop but avoid generating one for
10173 the loop tree root.
10174 (copy_bb_and_scalar_dependences): Remove premature codegen
10175 error on PHIs in blocks duplicated into multiple places.
10176 * graphite-scop-detection.c
10177 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10178 in the region use it as loop and nest to analyze the DR in.
10179 (try_generate_gimple_bb): Likewise.
10180 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10181 (add_loop_constraints): For blocks in a loop not in the region
10182 create a dimension with a single iteration.
10183 * sese.h (gbb_loop_at_index): Remove assert.
10185 * cfgloop.c (loop_preheader_edge): For the loop tree root
10186 return the single successor of the entry block.
10187 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10188 Reset the SCEV hashtable and niters.
10189 * graphite-scop-detection.c
10190 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
10191 assert that we only have POLYNOMIAL_CHREC that vary in loops
10192 contained in the region.
10193 (scop_detection::graphite_can_represent_expr): Adjust.
10194 (scop_detection::stmt_has_simple_data_refs_p): For loops
10195 not in the region set loop to NULL. The nest is now the
10196 entry edge to the region.
10197 (try_generate_gimple_bb): Likewise.
10198 * sese.c (scalar_evolution_in_region): Adjust for
10199 instantiate_scev change.
10200 * tree-data-ref.h (graphite_find_data_references_in_stmt):
10201 Make nest parameter the edge into the region.
10202 (create_data_ref): Likewise.
10203 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
10204 entry edge into a region and adjust instantiate_scev calls.
10205 (create_data_ref): Likewise.
10206 (graphite_find_data_references_in_stmt): Likewise.
10207 (find_data_references_in_stmt): Pass the loop preheader edge
10208 from the nest argument.
10209 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
10210 parameter the edge into the region.
10211 (instantiate_parameters): Use the loop preheader edge as entry.
10212 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
10214 (get_instantiated_value_entry): Make instantiate_below parameter
10215 the edge into the region.
10216 (instantiate_scev_name): Likewise. Adjust dominance checks,
10217 when we cannot use loop-based instantiation instantiate by
10218 walking use-def chains.
10219 (instantiate_scev_poly): Adjust.
10220 (instantiate_scev_binary): Likewise.
10221 (instantiate_scev_convert): Likewise.
10222 (instantiate_scev_not): Likewise.
10223 (instantiate_array_ref): Remove.
10224 (instantiate_scev_3): Likewise.
10225 (instantiate_scev_2): Likewise.
10226 (instantiate_scev_1): Likewise.
10227 (instantiate_scev_r): Do not blindly handle N-operand trees.
10228 Do not instantiate array-refs. Handle all constants and invariants.
10229 (instantiate_scev): Make instantiate_below parameter
10230 the edge into the region.
10231 (resolve_mixers): Use the loop preheader edge for the region
10232 parameter to instantiate_scev_r.
10233 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
10235 2017-10-13 Richard Biener <rguenther@suse.de>
10237 PR tree-optimization/82525
10238 * graphite-isl-ast-to-gimple.c
10239 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
10241 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
10242 Fail code generation when we cannot represent the isl integer.
10243 (binary_op_to_tree): Elide modulo operations that are no-ops
10244 in the type we code generate. Remove now superfluous code
10247 2017-10-13 Richard Biener <rguenther@suse.de>
10249 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
10250 (scop_detection::harmful_loop_in_region): Remove premature
10251 IV type restriction.
10252 (scop_detection::graphite_can_represent_scev): We can handle
10253 pointer IVs just fine.
10255 2017-10-13 Alan Modra <amodra@gmail.com>
10257 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
10258 "Clobbers and Scratch Registers". Add paragraph on
10259 alternative to clobbers for scratch registers and OpenBLAS
10262 2017-10-13 Alan Modra <amodra@gmail.com>
10264 * doc/extend.texi (Clobbers): Correct vax example. Delete old
10265 example of a memory input for a string of known length. Move
10266 commentary out of table. Add a number of new examples
10267 covering array memory inputs.
10269 2017-10-12 Martin Liska <mliska@suse.cz>
10271 PR tree-optimization/82493
10272 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
10273 (test_range_functions): New function.
10274 (sbitmap_c_tests): Likewise.
10275 * selftest-run-tests.c (selftest::run_tests): Run new tests.
10276 * selftest.h (sbitmap_c_tests): New function.
10278 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10280 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10282 * config/rs6000/amo.h: Fix spacing issue.
10284 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10287 * config/i386/i386.md (*ashl<mode>3_mask_1,
10288 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
10289 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
10292 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10294 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
10295 (profile_probability): Set max_probability
10296 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
10298 * profile-count.c (profile_probability::differs_from_p): Do not
10299 rely on max_probaiblity == 10000
10301 2017-10-12 Jeff Law <law@redhat.com>
10303 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
10306 2017-10-12 Martin Sebor <msebor@redhat.com>
10310 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
10311 (handle_alias_pairs): Call it.
10312 * common.opt (-Wattribute-alias): New option.
10313 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
10314 * doc/invoke.texi (-Wattribute-alias): Document.
10316 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
10319 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10321 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10323 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10324 (make_hard_regno_born, make_hard_regno_dead): Update
10325 bb_killed_pseudos and bb_gen_pseudos.
10327 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10329 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
10331 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
10333 * config/alpha/alpha.c (alpha_split_conditional_move):
10334 Use std::swap instead of manually swapping.
10335 (alpha_stdarg_optimize_hook): Ditto.
10336 (alpha_canonicalize_comparison): Ditto.
10338 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10340 * tree-loop-distribution.c (struct builtin_info): New struct.
10341 (struct partition): Refactor fields into struct builtin_info.
10342 (partition_free): Free struct builtin_info.
10343 (build_size_arg_loc, build_addr_arg_loc): Delete.
10344 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
10345 information from struct builtin_info.
10346 (find_single_drs): New function refactored from classify_partition.
10347 Also moved builtin validity checks to this function.
10348 (compute_access_range, alloc_builtin): New functions.
10349 (classify_builtin_st, classify_builtin_ldst): New functions.
10350 (classify_partition): Refactor code into functions find_single_drs,
10351 classify_builtin_st and classify_builtin_ldst.
10352 (distribute_loop): Don't do runtime alias check when distributing
10354 (find_seed_stmts_for_distribution): New function.
10355 (pass_loop_distribution::execute): Refactor code finding seed
10356 stmts into above function. Support distribution for the innermost
10357 two-level loop nest. Adjust dump information.
10359 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10361 * tree-loop-distribution.c: Adjust the general comment.
10362 (NUM_PARTITION_THRESHOLD): New macro.
10363 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
10364 (classify_partition): Skip builtin pattern of loop nest's inner loop.
10365 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
10366 in call to build_partition_graph.
10367 (finalize_partitions): New parameter. Make loop distribution more
10368 conservative by fusing more partitions.
10369 (distribute_loop): Don't do runtime alias check in case of loop nest
10371 (find_seed_stmts_for_distribution): New function.
10372 (prepare_perfect_loop_nest): New function.
10373 (pass_loop_distribution::execute): Refactor code finding seed stmts
10374 and loop nest into above functions. Support loop nest distribution.
10375 Adjust dump information accordingly.
10377 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10379 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
10380 and set PTYPE_SEQUENTIAL for merged partition.
10382 2017-10-12 Richard Biener <rguenther@suse.de>
10384 PR tree-optimization/69728
10386 2017-09-19 Richard Biener <rguenther@suse.de>
10388 PR tree-optimization/69728
10389 * graphite-sese-to-poly.c (schedule_error): New global.
10390 (add_loop_schedule): Handle empty domain by failing the
10392 (build_original_schedule): Handle schedule_error.
10394 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
10395 domain by returning an unchanged schedule.
10397 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10399 * genrecog.c (validate_pattern): For VEC_SELECT verify that
10400 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
10402 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
10404 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
10407 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10410 * expr.c (store_field): Don't optimize away bitsize == 0 store
10411 from CALL_EXPR with addressable return type.
10413 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10415 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
10416 * config/rs6000/rs6000.md (sel): Delete mode attribute.
10417 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
10418 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
10419 TARGET_ISEL instead of TARGET_ISEL<sel>.
10421 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
10423 * config/rs6000/rs6000.c
10424 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
10426 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10428 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
10430 (reg_or_cint_operand): Fix comment.
10431 (reg_or_zero_operand): New predicate.
10432 * config/rs6000/rs6000-protos.h (output_isel): Delete.
10433 * config/rs6000/rs6000.c (output_isel): Delete.
10434 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
10435 instead of reg_or_cint_operand. Output instruction directly (not via
10437 (isel_unsigned_<mode>): Ditto.
10438 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
10439 gpc_reg_operand. Add an instruction alternative for this. Output
10440 instruction directly.
10441 (*isel_reversed_unsigned_<mode>): Ditto.
10443 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10445 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
10446 (TARGET_CANONICALIZE_COMPARISON): Define.
10448 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
10451 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10452 Check whether the dest is REG before adding REG_EQUIV note.
10454 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10457 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10459 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10460 (make_hard_regno_born, make_hard_regno_dead): Update
10461 bb_killed_pseudos and bb_gen_pseudos.
10463 2017-10-11 Nathan Sidwell <nathan@acm.org>
10465 * incpath.h (enum incpath_kind): Name enum, prefix values.
10466 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
10467 * incpath.c (heads, tails): Use INC_MAX.
10468 (add_env_var_paths, add_standard_paths): Use incpath_kind.
10469 (merge_include_chains, split_quote_chain,
10470 register_include_chains): Update incpath_kind names.
10471 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
10472 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
10474 (add_framework_path, darwin_register_objc_includes): Likewise.
10475 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
10477 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10479 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
10480 Do not use float_operator operator predicate.
10481 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
10482 * config/i386/predicates.md (float_operator): Remove predicate.
10484 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10486 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
10487 (*jccxf_i387): Ditto.
10488 (*jcc<mode>_i387): Ditto.
10489 (*jccu<mode>_i387): Ditto.
10490 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
10491 (*jcc_*_i387 splitters): Remove.
10492 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
10493 * config/i386/i386.c (ix86_split_fp_branch): Remove.
10494 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
10497 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10499 * profile-count.h (slow_safe_scale_64bit): New function.
10500 (safe_scale_64bit): New inline.
10501 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
10502 * profile-count.c: Include wide-int.h
10503 (slow_safe_scale_64bit): New.
10505 2017-10-11 Nathan Sidwell <nathan@acm.org>
10507 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
10508 HAS_DECL_ASSEMBLER_NAME_P.
10509 * gimple-expr.c (gimple_decl_printable_name: Check
10510 HAS_DECL_ASSEMBLER_NAME_P too.
10511 * ipa-utils.h (type_in_anonymous_namespace_p): Check
10512 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
10513 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
10514 * passes.c (rest_of_decl_compilation): Check
10515 HAS_DECL_ASSEMBLER_NAME_P too.
10516 * recog.c (verify_changes): Likewise.
10517 * tree-pretty-print.c (dump_decl_name): Likewise.
10518 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
10520 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
10521 (SET_DECL_ASSEMBLER_NAME): Use it.
10522 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
10523 (COPY_DECL_ASSEMBLER_NAME): Likewise.
10524 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
10526 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10528 * config.gcc (i386, x86_64): Add extra objects.
10529 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
10530 (ix86_min_insn_size): Declare.
10531 (ix86_issue_rate): Declare.
10532 (ix86_adjust_cost): Declare.
10533 (ia32_multipass_dfa_lookahead): Declare.
10534 (ix86_macro_fusion_p): Declare.
10535 (ix86_macro_fusion_pair_p): Declare.
10536 (ix86_bd_has_dispatch): Declare.
10537 (ix86_bd_do_dispatch): Declare.
10538 (ix86_core2i7_init_hooks): Declare.
10539 (ix86_atom_sched_reorder): Declare.
10540 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
10541 (COSTS_N_BYTES): Move to x86-tune-costs.h.
10542 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
10543 (rip_relative_addr_p): Rename to ...
10544 (ix86_rip_relative_addr_p): ... this one; export.
10545 (memory_address_length): Update.
10546 (ix86_issue_rate): Move to x86-tune-sched.c.
10547 (ix86_flags_dependent): Move to x86-tune-sched.c.
10548 (ix86_agi_dependent): Move to x86-tune-sched.c.
10549 (exact_dependency_1): Move to x86-tune-sched.c.
10550 (exact_store_load_dependency): Move to x86-tune-sched.c.
10551 (ix86_adjust_cost): Move to x86-tune-sched.c.
10552 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
10553 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
10554 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
10555 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
10556 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
10557 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
10558 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
10559 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
10560 (min_insn_size): Rename to ...
10561 (ix86_min_insn_size): ... this one; export.
10562 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
10563 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
10564 (core2i7_first_cycle_multipass_backtrack): Move to
10565 x86-tune-sched-core.c.
10566 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
10567 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
10568 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
10569 (ix86_avoid_jump_mispredicts): Update.
10570 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
10571 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
10572 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
10573 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
10574 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
10575 (MAX_INSN): Move to ix86-tune-sched-bd.c.
10576 (MAX_IMM): Move to ix86-tune-sched-bd.c.
10577 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
10578 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
10579 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
10580 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
10581 (MAX_STORE): Move to ix86-tune-sched-bd.c.
10582 (BIG): Move to ix86-tune-sched-bd.c.
10583 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
10584 (enum insn_path): Move to ix86-tune-sched-bd.c.
10585 (get_mem_group): Move to ix86-tune-sched-bd.c.
10586 (is_cmp): Move to ix86-tune-sched-bd.c.
10587 (dispatch_violation): Move to ix86-tune-sched-bd.c.
10588 (is_branch): Move to ix86-tune-sched-bd.c.
10589 (is_prefetch): Move to ix86-tune-sched-bd.c.
10590 (init_window): Move to ix86-tune-sched-bd.c.
10591 (allocate_window): Move to ix86-tune-sched-bd.c.
10592 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
10593 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
10594 (process_end_window): Move to ix86-tune-sched-bd.c.
10595 (allocate_next_window): Move to ix86-tune-sched-bd.c.
10596 (find_constant): Move to ix86-tune-sched-bd.c.
10597 (get_num_immediates): Move to ix86-tune-sched-bd.c.
10598 (has_immediate): Move to ix86-tune-sched-bd.c.
10599 (get_insn_path): Move to ix86-tune-sched-bd.c.
10600 (get_insn_group): Move to ix86-tune-sched-bd.c.
10601 (count_num_restricted): Move to ix86-tune-sched-bd.c.
10602 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
10603 (add_insn_window): Move to ix86-tune-sched-bd.c.
10604 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
10605 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
10606 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
10607 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
10608 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
10609 (do_dispatch): Move to ix86-tune-sched-bd.c.
10610 (has_dispatch): Move to ix86-tune-sched-bd.c.
10611 * config/i386/t-i386: Add new object files.
10612 * config/i386/x86-tune-costs.h: New file.
10613 * config/i386/x86-tune-sched-atom.c: New file.
10614 * config/i386/x86-tune-sched-bd.c: New file.
10615 * config/i386/x86-tune-sched-core.c: New file.
10616 * config/i386/x86-tune-sched.c: New file.
10618 2017-10-11 Liu Hao <lh_mouse@126.com>
10620 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
10621 the generic version below instead.
10622 (should_colorize): Recognize Windows consoles as terminals
10624 * pretty-print.c [__MINGW32__] (write_all): New function.
10625 [__MINGW32__] (find_esc_head): Likewise.
10626 [__MINGW32__] (find_esc_terminator): Likewise.
10627 [__MINGW32__] (eat_esc_sequence): Likewise.
10628 [__MINGW32__] (mingw_ansi_fputs): New function that handles
10630 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
10633 2017-10-11 Richard Biener <rguenther@suse.de>
10635 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
10636 Properly call analyze_scalar_evolution with the loop of the stmt.
10638 2017-10-11 Richard Biener <rguenther@suse.de>
10640 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
10641 * tree-core.h (tree_base): Add chrec_var union member.
10642 * tree.h (CHREC_VAR): Remove.
10643 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
10644 * tree-chrec.h (build_polynomial_chrec): Adjust.
10645 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
10646 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
10648 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
10650 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
10651 * match.pd: ... here.
10652 ((T) X == (T) Y): Relax condition.
10654 2017-10-11 Bin Cheng <bin.cheng@arm.com>
10656 PR tree-optimization/82472
10657 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
10659 (break_alias_scc_partitions): Update postorder number.
10661 2017-10-11 Martin Liska <mliska@suse.cz>
10664 * opts.c (parse_no_sanitize_attribute): Do not use error_value
10666 * opts.h (parse_no_sanitize_attribute): Remove last argument.
10668 2017-10-11 Martin Liska <mliska@suse.cz>
10670 * print-rtl.c (print_insn): Move declaration of idbuf
10671 to same scope as name.
10673 2017-10-11 Martin Liska <mliska@suse.cz>
10678 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10679 volatile arguments.
10681 2017-10-11 Martin Liska <mliska@suse.cz>
10684 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10685 volatile arguments.
10687 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
10689 * config.gcc (default_gnu_indirect_function): Default to yes for
10690 arm*-*-linux* with glibc.
10692 2017-10-11 Richard Biener <rguenther@suse.de>
10694 * tree-scalar-evolution.c (get_scalar_evolution): Handle
10695 default-defs and types we do not want to analyze.
10696 (interpret_loop_phi): Replace unreachable code with an assert.
10697 (compute_scalar_evolution_in_loop): Remove and inline ...
10698 (analyze_scalar_evolution_1): ... here, replacing condition with
10699 what makes the intent clearer. Remove handling of cases
10700 get_scalar_evolution now handles.
10702 2017-10-10 Jim Wilson <wilson@tuliptree.org>
10704 PR rtl-optimization/81434
10705 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
10706 comment for main loop. In sched_group_found if, also add checks for
10707 pass and min_cost_group.
10709 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
10711 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
10712 (rs6000_insn_cost): New function.
10713 * config/rs6000/rs6000.md (cost): New attribute.
10715 2017-10-10 Jakub Jelinek <jakub@redhat.com>
10716 H.J. Lu <hongjiu.lu@intel.com>
10720 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
10721 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
10722 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
10723 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
10724 to OPTION_MASK_ISA_AVX512VL - builtins that have both
10725 OPTION_MASK_ISA_MMX and some other bit set require both
10726 mmx and the ISAs without the mmx bit.
10727 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
10728 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
10729 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
10730 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
10731 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
10732 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
10733 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
10734 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
10735 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
10736 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
10737 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
10738 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
10739 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
10740 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
10741 Add OPTION_MASK_ISA_MMX.
10743 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
10745 * config.gcc (armv7*-*-freebsd*): New target.
10746 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
10748 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
10750 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
10751 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
10754 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
10756 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
10757 a template parameter.
10758 (WIDE_INT_REF_FOR): Update accordingly.
10759 * tree.h (wi::int_traits <const_tree>): Delete.
10760 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
10761 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
10762 (wi::tree_to_wide_ref): New typedef.
10763 (wi::to_wide): New function.
10764 * calls.c (get_size_range): Use wi::to_wide when operating on
10765 trees as wide_ints.
10766 * cgraph.c (cgraph_node::create_thunk): Likewise.
10767 * config/i386/i386.c (ix86_data_alignment): Likewise.
10768 (ix86_local_alignment): Likewise.
10769 * dbxout.c (stabstr_O): Likewise.
10770 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
10771 * expr.c (const_vector_from_tree): Likewise.
10772 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
10773 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
10774 (fold_negate_expr_1, int_const_binop_1, const_binop)
10775 (fold_convert_const_int_from_real, optimize_bit_field_compare)
10776 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
10777 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
10778 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
10779 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
10780 (fold_not_const, round_up_loc): Likewise.
10781 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
10782 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
10783 (alloca_call_type): Likewise.
10784 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
10785 * godump.c (go_output_typedef): Likewise.
10786 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
10787 * internal-fn.c (get_min_precision): Likewise.
10788 * ipa-cp.c (ipcp_store_vr_results): Likewise.
10789 * ipa-polymorphic-call.c
10790 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
10791 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
10792 (ipa_modify_call_arguments): Likewise.
10793 * match.pd: Likewise.
10794 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
10795 * print-tree.c (print_node_brief, print_node): Likewise.
10796 * stmt.c (expand_case): Likewise.
10797 * stor-layout.c (layout_type): Likewise.
10798 * tree-affine.c (tree_to_aff_combination): Likewise.
10799 * tree-cfg.c (group_case_labels_stmt): Likewise.
10800 * tree-data-ref.c (dr_analyze_indices): Likewise.
10801 (prune_runtime_alias_test_list): Likewise.
10802 * tree-dump.c (dequeue_and_dump): Likewise.
10803 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
10804 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
10805 * tree-pretty-print.c (dump_generic_node): Likewise.
10806 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
10807 (simple_iv_with_niters): Likewise.
10808 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
10809 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
10810 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
10811 * tree-ssa-loop-niter.c (split_to_var_and_offset)
10812 (refine_value_range_using_guard, number_of_iterations_ne_max)
10813 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
10814 (get_cst_init_from_scev, record_nonwrapping_iv)
10815 (scev_var_range_cant_overflow): Likewise.
10816 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
10817 * tree-ssa-pre.c (compute_avail): Likewise.
10818 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
10819 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
10820 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
10821 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
10822 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
10823 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
10824 (dump_case_nodes, try_switch_expansion): Likewise.
10825 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
10826 (vect_do_peeling): Likewise.
10827 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
10828 * tree-vect-stmts.c (vectorizable_load): Likewise.
10829 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
10830 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
10831 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
10832 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
10833 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
10834 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
10835 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
10836 (evrp_dom_walker::before_dom_children): Likewise.
10837 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
10838 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
10839 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
10840 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
10841 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
10842 (get_range_pos_neg): Likewise.
10843 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
10844 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
10845 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
10846 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
10847 * config/avr/avr.c (avr_fold_builtin): Likewise.
10848 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
10849 * config/msp430/msp430.c (msp430_attr): Likewise.
10850 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
10851 * config/powerpcspe/powerpcspe-c.c
10852 (altivec_resolve_overloaded_builtin): Likewise.
10853 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
10854 (rs6000_expand_ternop_builtin): Likewise.
10855 * config/rs6000/rs6000-c.c
10856 (altivec_resolve_overloaded_builtin): Likewise.
10857 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
10858 (rs6000_expand_ternop_builtin): Likewise.
10859 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
10861 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10863 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
10864 when copying loop nest with only one inner loop.
10866 2017-10-10 Richard Biener <rguenther@suse.de>
10868 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
10869 blocks if SCEV is active.
10870 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
10872 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
10873 (scev_initialize): Assert we are not yet initialized.
10875 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10877 * tree-loop-distribution.c (generate_loops_for_partition): Remove
10878 inner loop's exit stmt by making it always exit the loop, otherwise
10879 we would generate an infinite empty loop.
10881 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10883 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
10884 renaming variables in new preheader if it's deleted.
10886 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10888 * tree-loop-distribution.c (struct partition): Remove unused field
10889 loops of the structure.
10890 (partition_alloc, partition_free): Ditto.
10891 (build_rdg_partition_for_vertex): Ditto.
10893 2017-10-09 Jeff Law <law@redhat.com>
10895 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
10896 return type to match prototype and documentation.
10898 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10900 * config/rs6000/rs6000.c (processor_costs): Move to ...
10901 * config/rs6000/rs6000.h: ... here.
10902 (rs6000_cost): Declare.
10904 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
10906 * except.c (setjmp_fn): New global variable.
10907 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
10908 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
10909 if DONT_USE_BUILTIN_SETJMP is defined.
10911 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10913 * target.def (insn_cost): New hook.
10914 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
10915 * doc/tm.texi: Regenerate.
10916 * rtlanal.c (insn_cost): Use the new hook.
10918 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10920 * combine.c (combine_validate_cost): Compute the new insn_cost,
10921 not just pattern_cost.
10922 (try_combine): Adjust comment.
10924 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10926 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
10928 * combine.c (uid_insn_cost): Adjust comment.
10929 (combine_validate_cost): Adjust comment. Use pattern_cost instead
10931 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
10932 * dse.c (find_shift_sequence): Ditto.
10933 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
10934 (bb_valid_for_noce_process_p): Use pattern_cost.
10935 * rtl.h (insn_rtx_cost): Delete.
10936 (pattern_cost): New prototype.
10937 (insn_cost): New prototype.
10938 * rtlanal.c (insn_rtx_cost): Rename to...
10939 (pattern_cost): ... this.
10942 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
10944 * config/i386/i386.md (*jcc_2): Remove insn pattern.
10945 (*jcc<mode>_0_r_i387): Ditto.
10946 (*jccxf_r_i387): Ditto.
10947 (*jcc<mode>_r_i387): Ditto.
10948 (*jccu<mode>_r_i387): Ditto.
10949 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
10950 (*jcc): Rename from *jcc_1.
10952 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10954 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
10955 deferred rescans after the lvx/stvx recombination pre-pass.
10957 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
10959 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
10960 memory operation instruction support.
10961 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
10962 (rs6000-ibm-aix[789]*): Likewise.
10963 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
10964 Document new functions.
10966 2017-10-09 Richard Biener <rguenther@suse.de>
10968 PR tree-optimization/82397
10969 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
10970 equality only for semantically equal trees.
10972 2017-10-09 Richard Biener <rguenther@suse.de>
10974 PR tree-optimization/82449
10975 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
10976 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
10977 allow constant addresses.
10978 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
10981 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10983 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
10986 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10989 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
10992 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10995 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
10997 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11000 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11001 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11003 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11005 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11006 (WI_BINARY_PREDICATE_RESULT): Likewise.
11007 (wi::binary_traits::operator_result): New type.
11008 (wi::binary_traits::predicate_result): Likewise.
11009 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11010 (generic_wide_int::operator==, generic_wide_int::operator!=)
11011 (generic_wide_int::operator&, generic_wide_int::and_not)
11012 (generic_wide_int::operator|, generic_wide_int::or_not)
11013 (generic_wide_int::operator^, generic_wide_int::operator+
11014 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11016 (operator~, unary operator-, operator==, operator!=, operator&)
11017 (operator|, operator^, operator+, binary operator-, operator*): New
11019 * expr.c (get_inner_reference): Use wi::bit_and_not.
11020 * fold-const.c (fold_binary_loc): Likewise.
11021 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11022 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11023 (bit_value_binop): Likewise.
11024 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11025 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11026 (extract_range_from_binary_expr_1): Likewise.
11027 (masked_increment): Likewise.
11028 (simplify_bit_ops_using_ranges): Likewise.
11030 2017-10-09 Martin Jambor <mjambor@suse.cz>
11033 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11034 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11035 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11037 (hsa_fixup_mov_insn_type): New function.
11038 (hsa_op_with_type::get_in_type): Use it.
11039 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11040 immediates in an assert.
11041 (hsa_op_with_type::extend_int_to_32bit): New method.
11042 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
11043 types. Convert to dest type if necessary.
11044 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
11045 (reg_for_gimple_ssa): Pass false as min32int to
11046 hsa_type_for_scalar_tree_type.
11047 (gen_hsa_addr): Fixup type when creating addresable temporary.
11048 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
11049 (gen_hsa_unary_operation): Extend operands and convert to dest type if
11050 necessary. Call hsa_fixup_mov_insn_type.
11051 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
11052 extend operands and convert to dest type if necessary.
11053 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
11054 to dest type if necessary.
11055 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
11056 if conversion nt necessary and size matches.
11057 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
11058 to dest type if necessary.
11059 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
11060 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
11062 (gen_hsa_clrsb): Likewise.
11063 (gen_hsa_ffs): Likewise.
11064 (gen_hsa_divmod): Extend operands and convert to dest type if
11066 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
11068 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11070 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
11071 Remove empty default arguments. Use a brace block as output
11073 (conditional return): Ditto.
11075 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
11076 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
11078 (group_ending_nop): Ditto.
11079 (doloop_end): Ditto.
11080 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
11081 (splitters for those): Ditto.
11083 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11085 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
11086 a conditional jump (and the compare for it) so that pc_rtx is the
11088 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
11089 for the deleted and renamed ctr<mode>_internal[234] patterns.
11090 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
11091 Delete second conditional return pattern.
11092 (ctr<mode>_internal2): Delete this second bdnz pattern.
11093 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
11094 (ctr<mode>_internal4): Delete this second bdz pattern.
11096 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11098 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
11099 (always_initialized_rtx_for_ssa_name_p): New predicate.
11100 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
11101 (finish_out_of_ssa): Free new field of SA.
11102 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
11103 * tree-ssa-coalesce.c: Include tree-ssa.h.
11104 (get_parm_default_def_partitions): Remove extern keyword.
11105 (get_undefined_value_partitions): New function.
11106 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
11107 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
11108 uninitialized bits.
11109 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
11111 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11113 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
11115 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11117 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
11118 for targets that preffer 128bit.
11120 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11122 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
11124 2017-10-08 Olivier Hainque <hainque@adacore.com>
11126 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
11127 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
11129 (thumb_set_return_address): Likewise.
11131 2017-10-08 Olivier Hainque <hainque@adacore.com>
11133 * common/config/arm/arm-common.c (arm_except_unwind_info):
11134 Handle DWARF2_UNWIND_INFO.
11136 2017-10-07 Michael Collison <michael.collison@arm.com>
11138 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
11141 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
11143 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
11144 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
11145 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
11146 defined, force the creation of a new block for a dispatch label.
11148 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
11150 * invoke.texi (Wsuggest-attribute=cold): Document.
11151 * common.opt (Wsuggest-attribute=cold): New
11152 * ipa-pure-const.c (warn_function_cold): New function.
11153 * predict.c (compute_function_frequency): Use it.
11154 * predict.h (warn_function_cold): Declare.
11156 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
11158 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
11161 2017-10-06 Martin Liska <mliska@suse.cz>
11163 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
11164 keyword for member functions.
11165 (struct sanopt_tree_couple): New struct.
11166 (struct sanopt_tree_couple_hash): New function.
11167 (struct sanopt_ctx): Add new hash_map.
11168 (has_dominating_ubsan_ptr_check): New function.
11169 (record_ubsan_ptr_check_stmt): Likewise.
11170 (maybe_optimize_ubsan_ptr_ifn): Likewise.
11171 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
11172 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
11174 2017-10-06 Sudakshina Das <sudi.das@arm.com>
11177 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
11178 aarch64_simd_valid_immediate on CONST_VECTORs.
11179 (aarch64_reg_or_bic_imm): Likewise.
11181 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
11183 PR rtl-optimization/82396
11184 * haifa-sched.c (ready_sort_real): Disable qsort checking.
11186 2017-10-06 Sebastian Pop <sebpop@gmail.com>
11188 * graphite-dependences.c (scop_get_reads): Move code to...
11189 (scop_get_must_writes): Move code to...
11190 (scop_get_may_writes): Move code to...
11191 (scop_get_reads_and_writes): ... here.
11192 (scop_get_dependences): Call scop_get_reads_and_writes.
11194 2017-10-06 Jakub Jelinek <jakub@redhat.com>
11196 PR tree-optimization/82434
11197 * fold-const.h (can_native_encode_type_p,
11198 can_native_encode_string_p): Remove.
11199 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
11200 don't encode anything, just return what would be otherwise returned.
11201 (native_encode_fixed, native_encode_complex, native_encode_vector):
11203 (native_encode_string): Likewise. Inline by hand
11204 can_native_encode_string_p.
11205 (can_native_encode_type_p): Remove.
11206 (can_native_encode_string_p): Remove.
11207 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
11208 STRING_CSTs using can_native_encode_string_p, test all
11209 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
11210 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
11211 argument from native_encode_expr.
11212 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
11213 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
11216 2017-10-06 Richard Biener <rguenther@suse.de>
11218 PR tree-optimization/82397
11219 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
11220 operand_equal_p but rely on data_ref_compare_tree for detecting
11222 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
11223 to match up with dr_group_sort_cmp.
11225 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11228 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
11230 * config/s390/s390-builtin-types.def: Regenerate.
11232 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11235 * config/s390/s390-builtin-types.def: Regenerate.
11236 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
11237 Change flag from B_VXE to B_VX.
11238 (s390_vec_min_dbl): Remove B_VXE flag.
11240 2017-10-06 Richard Biener <rguenther@suse.de>
11242 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
11243 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
11244 translate_isl_ast_to_gimple::is_valid_rename,
11245 translate_isl_ast_to_gimple::get_rename,
11246 translate_isl_ast_to_gimple::get_def_bb_for_const,
11247 translate_isl_ast_to_gimple::get_new_name,
11248 translate_isl_ast_to_gimple::collect_all_ssa_names,
11249 translate_isl_ast_to_gimple::copy_loop_phi_args,
11250 translate_isl_ast_to_gimple::collect_all_ssa_names,
11251 translate_isl_ast_to_gimple::copy_loop_phi_args,
11252 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
11253 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
11254 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
11255 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
11256 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
11257 translate_isl_ast_to_gimple::copy_cond_phi_args,
11258 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
11259 translate_isl_ast_to_gimple::edge_for_new_close_phis,
11260 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
11261 translate_isl_ast_to_gimple::rename_uses,
11262 translate_isl_ast_to_gimple::rename_all_uses): Remove.
11263 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
11264 (set_rename_for_each_def): Likewise.
11265 (graphite_copy_stmts_from_block): Handle debug stmt resetting
11266 here. Handle rewriting SCEV analyzable uses here.
11267 (copy_bb_and_scalar_dependences): Generate code for PHI
11269 (graphite_regenerate_ast_isl): Adjust.
11270 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
11271 (add_write, add_read): New functions.
11272 (build_cross_bb_scalars_def): Use it and simplify.
11273 (build_cross_bb_scalars_use): Likewise.
11274 (graphite_find_cross_bb_scalar_vars): Inline into...
11275 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
11276 simulating out-of-SSA. Compute liveout and add dependencies.
11277 (build_scops): Force an empty entry block.
11278 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
11280 (sese_build_liveouts): Declare.
11281 (sese_trivially_empty_bb_p): Likewise.
11282 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
11283 compute liveout and debug_liveout.
11284 (sese_bad_liveouts_use): Remove.
11285 (sese_reset_debug_liveouts_bb): Likewise.
11286 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
11287 (sese_build_liveouts): Build liveout and debug_liveout and store
11289 (new_sese_info): Adjust.
11290 (free_sese_info): Likewise.
11291 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
11292 do not build liveout here.
11293 (move_sese_in_condition): Adjust region entry.
11294 (scev_analyzable_p): Match up with chrec_apply requirements.
11295 (sese_trivially_empty_bb_p): New.
11296 * tree-into-ssa.c (get_reaching_def): Properly support generating
11297 default-defs for incremental rewrite of anonymous names.
11299 2017-10-06 Richard Biener <rguenther@suse.de>
11301 * graphite-sese-to-poly.c (extract_affine): For casts increasing
11302 precision do not perform modulo reduction.
11304 2017-10-06 Richard Biener <rguenther@suse.de>
11306 PR tree-optimization/82436
11307 * tree-vect-slp.c (vect_supported_load_permutation_p): More
11308 conservatively choose the vectorization factor when checking
11309 whether we can perform the required load permutation.
11310 (vect_transform_slp_perm_load): Assert when we may not fail.
11312 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
11314 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
11315 message for incompatible -msdata=* and -mcall-* options.
11317 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11319 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
11320 rate for post-reload scheduling.
11322 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11324 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
11326 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11328 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
11329 to improve monte carlo in scimark.
11331 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11333 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
11334 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
11335 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
11336 pentium4_cost, nocona_cost): Set reassociation width to 1.
11337 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
11338 width to 2 for fp operations and 1 otherwise.
11339 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
11341 (atom_cost): Set reassociation width to 2.
11342 (slm_cost, generic_cost): Set fp reassociation width
11343 to 2 and 1 otherwise.
11344 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
11345 (core_cost): Set fp reassociation width to 4 and vector to 2.
11346 (ix86_reassociation_width): Rewrite using cost table; special case
11347 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
11348 and TARGET_AVX128_OPTIMAL.
11349 * config/i386/i386.h (processor_costs): Add
11350 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
11351 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
11352 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
11353 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
11354 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
11355 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
11357 2017-10-05 Nathan Sidwell <nathan@acm.org>
11359 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
11361 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11363 * config/arm/arm.c (arm_test_fpu_data): New.
11364 (arm_run_selftests): Call arm_test_fpu_data.
11366 2017-10-04 Nathan Sidwell <nathan@acm.org>
11368 * toplev.c (toplev::main): Remove excess parens on pretty_printer
11370 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
11372 2017-10-04 Sudakshina Das <sudi.das@arm.com>
11374 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
11375 check type for aarch64_simd_valid_immediate.
11376 (aarch64_output_simd_mov_immediate): Update prototype.
11377 (aarch64_simd_valid_immediate): Update prototype.
11378 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
11379 support for ORR-immediate.
11380 (and<mode>3): modified pattern to add support for BIC-immediate.
11381 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
11382 now checks for valid immediate for BIC and ORR based on new enum
11384 (aarch64_output_simd_mov_immediate): Function now used to output
11385 BIC/ORR imm as well based on new enum argument.
11386 * config/aarch64/constraints.md (Do): New vector immediate constraint.
11388 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
11389 (aarch64_reg_or_bic_imm): Likewise.
11391 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11393 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
11394 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
11396 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11400 PR rtl-optimization/82396
11401 * haifa-sched.c (autopref_multipass_init): Simplify
11403 (autopref_rank_data): Simplify sort order.
11404 * sched-int.h (autopref_multipass_data_): Remove
11405 multi_mem_insn_p, min_offset and max_offset.
11407 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11409 * doc/sourcebuild.texi: Document vect_peeling_profitable.
11411 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11413 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
11414 vect_doubleint_cvt.
11416 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11418 * doc/sourcebuild.texi: Document vect_long_mult.
11420 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
11422 PR tree-optimization/82413
11423 * fold-const.c (build_range_check): Use widest_int when comparing
11424 the maximum ETYPE value with HIGH.
11426 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11428 PR rtl-optimization/82396
11429 * haifa-sched.c (autopref_multipass_init): Simplify
11431 (autopref_rank_data): Simplify sort order.
11432 * sched-int.h (autopref_multipass_data_): Remove
11433 multi_mem_insn_p, min_offset and max_offset.
11435 2017-10-04 Jakub Jelinek <jakub@redhat.com>
11437 PR tree-optimization/82381
11438 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
11439 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
11442 PR tree-optimization/82374
11443 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
11444 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
11445 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
11446 current_function_decl to the new decl.
11448 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
11450 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
11451 helper macro for IEEE float128 hardware built-in functions.
11452 (SQRTF128_ODD): Add built-in functions with the round-to-odd
11454 (TRUNCF128_ODD): Likewise.
11455 (ADDF128_ODD): Likewise.
11456 (SUBF128_ODD): Likewise.
11457 (MULF128_ODD): Likewise.
11458 (DIVF128_ODD): Likewise.
11459 (FMAF128_ODD): Likewise.
11460 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
11461 UNSPEC_TRUNC_ROUND_TO_ODD.
11462 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
11463 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
11464 floating point round to odd instructions.
11465 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
11466 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
11467 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
11468 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
11469 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
11470 (trunc<mode>sf2_hw): Change the truncate with round to odd
11471 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
11472 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
11473 to odd hardware instructions.
11474 (sub<mode>3_odd): Likewise.
11475 (mul<mode>3_odd): Likewise.
11476 (div<mode>3_odd): Likewise.
11477 (sqrt<mode>2_odd): Likewise.
11478 (fma<mode>4_odd): Likewise.
11479 (fms<mode>4_odd): Likewise.
11480 (nfma<mode>4_odd): Likewise.
11481 (nfms<mode>4_odd): Likewise.
11482 (trunc<mode>df2_odd): Change the truncate with round to odd
11483 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
11485 * doc/extend.texi (PowerPC built-in functions): Update documentation
11486 for existing IEEE float128-bit built-in functions. Add built-in
11487 functions that generate the IEEE 128-bit floating point round to
11490 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
11492 PR rtl-optimization/77729
11493 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
11494 to (X&(C1&~C2))|C2 transformations.
11496 2017-10-03 Martin Jambor <mjambor@suse.cz>
11498 PR tree-optimization/82363
11499 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
11500 mismatch, mark lacc written regardless of racc.
11502 2017-10-03 Jakub Jelinek <jakub@redhat.com>
11504 PR tree-optimization/82381
11505 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
11506 stmt_to_insert nor wheather SSA_NAMEs are default defs.
11507 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
11508 fallthrough into reassoc_stmt_dominates_stmt_p.
11511 * combine.c (combine_instructions): Don't combine in unreachable
11514 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
11517 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
11518 function to not use the have_cpu variable. Do not set cpu_index,
11519 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
11520 or the default cpu.
11521 (rs6000_valid_attribute_p): Remove duplicate initializations of
11522 old_optimize and func_optimize.
11523 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
11524 (rs6000_activate_target_options): Make global.
11525 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
11528 2017-10-02 Jakub Jelinek <jakub@redhat.com>
11530 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
11531 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
11532 Set *poffset to 0 and *psize and *pmax_size to -1 if
11533 *poffset + *psize overflows in HOST_WIDE_INT.
11535 PR tree-optimization/82387
11536 PR tree-optimization/82388
11537 PR tree-optimization/82389
11538 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
11539 instead of live_bytes non-NULL.
11541 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
11544 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
11547 2017-10-02 Richard Biener <rguenther@suse.de>
11549 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
11550 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
11551 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
11553 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11555 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
11556 requested precision matches the type's.
11557 * calls.c (alloc_max_size): Calculate the new candidate size as
11558 a widest_int and use wi::to_widest when comparing it with the
11559 current candidate size.
11560 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
11561 zero rather than integer_zero_node.
11562 * match.pd: Check for a no-op conversion before using wi::add
11563 rather than after. Use tree_to_uhwi when summing small shift
11564 counts into an unsigned int.
11566 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11567 Alan Hayward <alan.hayward@arm.com>
11568 David Sherwood <david.sherwood@arm.com>
11571 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
11572 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
11573 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
11574 POINTER_AND_FP_REGS.
11576 2017-10-02 Richard Biener <rguenther@suse.de>
11578 PR tree-optimization/82355
11579 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11580 a mapping for the enclosing loop but avoid generating one for
11581 the loop tree root.
11582 (copy_bb_and_scalar_dependences): Remove premature codegen
11583 error on PHIs in blocks duplicated into multiple places.
11584 * graphite-scop-detection.c
11585 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11586 in the region use it as loop and nest to analyze the DR in.
11587 (try_generate_gimple_bb): Likewise.
11588 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11589 (add_loop_constraints): For blocks in a loop not in the region
11590 create a dimension with a single iteration.
11591 * sese.h (gbb_loop_at_index): Remove assert.
11593 2017-10-01 Kevin Buettner <kevinb@redhat.com>
11595 * omp-expand.c (adjust_context_scope): New function.
11596 (expand_parallel_call): Call adjust_context_scope.
11598 2017-10-01 Jeff Law <law@redhat.com>
11600 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
11601 dom_opt_dom_walker class with direct access to private members.
11602 Add comments. Call test_for_singularity.
11603 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
11604 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
11605 m_dummy_cond anymore.
11606 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
11608 (pass_dominator:execute): Build the dummy_cond here and pass it
11609 to the dom_opt_dom_walker ctor.
11610 (test_for_singularity): New function.
11612 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
11613 Maya Rashish <coypu@sdf.org>
11615 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
11616 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
11617 (alpha*-*-netbsd*) Use nbsd_tm_file.
11618 (arm*-*-netbsdelf*) Likewise.
11619 (i[34567]86-*-netbsdelf*) Likewise.
11620 (x86_64-*-netbsd*) Likewise.
11621 (mips*-*-netbsd*) Likewise.
11622 (powerpc-*-netbsd*) Likewise.
11623 (sh*-*-netbsd*) Likewise.
11624 (sparc-*-netbsdelf*) Likewise.
11625 (sparc64-*-netbsd*) Likewise.
11626 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
11628 (vax-*-netbsdelf*) Likewise.
11629 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
11630 (UINT_FAST8_TYPE) Likewise.
11631 (INT_FAST16_TYPE) Check CHAR_FAST16.
11632 (UINT_FAST16_TYPE) Likewise.
11634 2017-09-30 Jakub Jelinek <jakub@redhat.com>
11637 * config/i386/i386.md
11638 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
11639 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
11640 *divmodsi4_zext_2): New define_insn_and_split.
11641 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
11642 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
11643 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
11644 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
11645 New define_insn_and_split.
11646 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
11647 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
11648 operands[1] having DImode when mode is SImode.
11650 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
11651 always SImode for DIV and MOD in REG_EQUAL notes.
11653 2017-09-29 Yury Gribov <tetra2005@gmail.com>
11655 PR middle-end/82319
11656 * match.pd: Fix handling of NaNs in pattern.
11658 2017-09-29 Jeff Law <law@redhat.com>
11660 * sbitmap.c (bitmap_bit_in_range_p): New function.
11661 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
11662 * tree-ssa-dse.c (live_bytes_read): New function.
11663 (dse_classify_store): Ignore reads of dead bytes.
11665 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
11666 typos and whitespace errors.
11667 * config/i386/predicates.md (address_no_seg_operand): Likewise.
11668 * config/s390/s390.c (s390_emit_prologue): Likewise.
11670 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11673 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
11674 with a symbol for LRA.
11676 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11678 PR rtl-optimization/82338
11679 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
11681 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
11683 * genmodes.c (calc_wider_mode): Suppress qsort macro.
11684 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
11685 (qsort_chk): Declare.
11686 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
11687 (qsort_chk): New function.
11689 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11691 PR tree-optimization/82337
11692 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
11693 phi definition if the PHI result appears in an abnormal PHI.
11694 (find_basis_for_base_expr): Don't record a basis if the LHS of the
11695 basis appears in an abnormal PHI.
11697 2017-09-29 Richard Biener <rguenther@suse.de>
11699 * graphite-isl-ast-to-gimple.c
11700 (translate_isl_ast_to_gimple::set_codegen_error): New function.
11701 (binary_op_to_tree): Use it.
11702 (get_rename_from_scev): Likewise.
11703 (copy_loop_phi_nodes): Likewise.
11704 (copy_bb_and_scalar_dependences): Likewise.
11705 (translate_pending_phi_nodes): Likewise.
11707 2017-09-29 Jakub Jelinek <jakub@redhat.com>
11710 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
11711 for movabsq $(i32 << shift), r64.
11713 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
11715 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
11716 index when encoding %esp as %rsp to avoid 0x67 prefix.
11718 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
11720 * config/i386/i386.md (*movsf_internal, *movdf_internal):
11721 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
11723 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
11725 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
11726 Extensions with more than 16 double VFP registers.
11727 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
11728 to_clear_mask and all code related to it. Replace the remaining
11729 entry by a sbitmap and adapt code accordingly.
11731 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
11733 * brig-builtins.def: Change pure attributes to const.
11735 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11737 * config.gcc (default_gnu_indirect_function): Default to yes for
11738 sparc*-*-linux* with glibc.
11740 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11742 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
11743 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
11744 when creating .init_array and .fini_array sections with priority
11747 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
11750 * config/aarch64/aarch64.c
11751 (aarch64_builtin_support_vector_misalignment): Always return false
11752 when misalignment is unknown.
11754 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11756 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
11757 this function to return false if the definition used by the swap
11758 instruction is artificial, or if the memory address from which the
11759 constant value is loaded is not represented by a base address held
11760 in a register or if the base address register is a frame or stack
11761 pointer. Additionally, return false if the base address of the
11762 loaded constant is a SYMBOL_REF but is not considered to be a
11764 (replace_swapped_load_constant): New function.
11765 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
11766 loaded constant vector with a load of a swapped constant vector.
11768 2017-09-27 Carl Love <cel@us.ibm.com>
11770 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
11771 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
11772 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
11775 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
11777 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
11778 first, always call autopref_rank_data otherwise.
11780 2017-09-27 Richard Biener <rguenther@suse.de>
11782 * graphite-scop-detection.c (find_scop_parameters): Move
11783 loop bound handling ...
11784 (gather_bbs::before_dom_children): ... here, avoiding the need
11785 to build scop_info->loop_nest.
11786 (record_loop_in_sese): Remove.
11787 * sese.h (sese_info_t::loop_nest): Remove.
11788 * sese.c (new_sese_info): Do not allocate loop_nest.
11789 (free_sese_info): Do not free loop_nest.
11791 2017-09-27 Jakub Jelinek <jakub@redhat.com>
11794 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
11795 lhs from calls if the lhs has addressable type.
11797 2017-09-27 Richard Biener <rguenther@suse.de>
11799 * graphite.h (scop::max_alias_set): New member.
11800 * graphite-scop-detection.c: Remove references to non-existing
11801 --param in comments.
11802 (build_alias_sets): Record the maximum alias set used for drs.
11803 (build_scops): Support zero as unlimited for
11804 --param graphite-max-arrays-per-scop.
11805 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
11806 and inline into ...
11807 (build_poly_sr_1): ... here. Compute alias set based on the
11808 maximum alias set used for drs rather than
11809 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
11811 2017-09-27 Richard Biener <rguenther@suse.de>
11813 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
11814 --param loop-block-tile-size=0 to disable tiling.
11816 2017-09-27 Richard Biener <rguenther@suse.de>
11818 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
11819 (graphite-max-nb-scop-params): Document special value zero.
11820 * domwalk.h (dom_walker::STOP): New symbolical constant.
11821 (dom_walker::dom_walker): Add optional parameter for bb to
11823 (dom_walker::~dom_walker): Declare.
11824 (dom_walker::before_dom_children): Document STOP return value.
11825 (dom_walker::m_user_bb_to_rpo): New member.
11826 (dom_walker::m_bb_to_rpo): Likewise.
11827 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
11828 mapping here if not provided by the user.
11829 (dom_walker::~dom_walker): Free bb to RPO mapping if not
11830 provided by the user.
11831 (dom_walker::STOP): Define.
11832 (dom_walker::walk): Do not compute bb to RPO mapping here.
11833 Support STOP return value from before_dom_children to stop
11835 * graphite-optimize-isl.c (optimize_isl): If the schedule
11836 is the same still generate code if -fgraphite-identity
11837 or -floop-parallelize-all are given.
11838 * graphite-scop-detection.c: Include cfganal.h.
11839 (gather_bbs::gather_bbs): Get and pass through bb to RPO
11841 (gather_bbs::before_dom_children): Return STOP for BBs
11843 (build_scops): Compute bb to RPO mapping and pass it to
11844 the domwalk. Treat --param graphite-max-nb-scop-params=0
11845 as not limiting the number of params.
11846 * graphite.c (graphite_initialize): Remove limit on the
11847 number of basic-blocks in a function.
11848 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
11849 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
11850 default value of 10.
11852 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11854 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
11855 Adjust code to eliminate needing to do the shift right 32-bits
11856 operation after XSCVDPSPN.
11858 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11860 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
11861 ((X / Y) != 0 -> X >= Y): Likewise.
11863 2017-09-26 Carl Love <cel@us.ibm.com>
11865 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
11866 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
11867 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
11868 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
11869 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
11870 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
11871 definitions and overloading.
11872 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
11873 statement for P9V_BUILTIN_XST_LEN_R.
11874 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
11875 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
11876 define_expand and define_insn for the instructions and builtins.
11877 * doc/extend.texi: Update the built-in documentation file for the new
11878 built-in functions.
11879 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
11880 define_insn for the instructions
11882 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
11885 * gcc/config/netbsd-protos.h: New file.
11886 * gcc/config/netbsd.c: New file.
11887 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
11888 * gcc/config/t-netbsd: New file.
11889 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
11890 (tmake_file) Add t-netbsd.
11891 (extra_objs) Add netbsd.o.
11893 2017-09-26 Janus Weil <janus@gcc.gnu.org>
11897 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
11899 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11901 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
11902 sign extension from a vector register to a GPR by doing a 32-bit
11903 direct move and then an EXTSW.
11904 (extendsi<mode>2 splitter): Likewise.
11905 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
11906 right or vector extract after doing XSCVDPSPN. Use
11907 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
11909 (movdi_from_sf_zero_ext): Likewise.
11910 (reload_gpr_from_vsxsf): Likewise.
11911 (p8_mfvsrd_4_disf): Delete, no longer used.
11912 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
11913 then needing to move the SFmode to a GPR to use the XSCVDPSP
11914 instruction instead of FRSP and XSCVDPSPN.
11915 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
11916 it is adjacent to the other XSCVSPDP insns.
11917 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
11918 SFmode to be in traditional Altivec registers.
11919 (vsx_xscvdpspn): Eliminate useless alternative constraint.
11920 (vsx_xscvspdpn): Likewise.
11921 (vsx_xscvspdpn_scalar): Likewise.
11923 2017-09-26 Martin Jambor <mjambor@suse.cz>
11925 * tree-sra.c (compare_access_positions): Put integral types first,
11926 stabilize sorting of integral types, remove conditions putting
11927 non-full-precision integers last.
11928 (sort_and_splice_var_accesses): Disable scalarization if a
11929 non-integert would be represented by a non-full-precision integer.
11931 2017-09-26 Joseph Myers <joseph@codesourcery.com>
11933 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
11934 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
11935 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
11936 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
11937 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
11938 conditionals inside the function instead of around it. Call
11939 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
11940 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
11942 2017-09-26 Richard Biener <rguenther@suse.de>
11944 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
11946 (scop_detection::build_scop_breadth): ... this. Removed.
11947 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
11948 (scop_detection::harmful_stmt_in_bb): Likewise.
11949 (scop_detection::graphite_can_represent_stmt): Likewise.
11950 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
11951 (scop_detection::can_represent_loop): Remove recursion, fold in ...
11952 (scop_detection::can_represent_loop_1): ... this. Removed.
11953 (scop_detection::harmful_loop_in_region): Simplify after inlining
11954 the above and remove more quadraticness.
11955 (build_scops): Adjust.
11956 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
11959 2017-09-26 Jakub Jelinek <jakub@redhat.com>
11962 * config/i386/i386.c (ix86_print_operand_address_as): Only test
11963 REGNO (base) == SP_REG if base is a REG.
11965 PR middle-end/35691
11966 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
11967 if it is different SSA_NAME.
11968 (optimize_range_tests_cmp_bitwise): New function.
11969 (optimize_range_tests): Call it.
11971 2017-09-26 Richard Biener <rguenther@suse.de>
11973 PR tree-optimization/82321
11974 * graphite.c (canonicalize_loop_closed_ssa): Properly check
11975 for the def being inside the loop.
11977 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11979 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
11981 * config/s390/s390-builtins.def: Fix constraint on op4.
11983 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11985 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
11986 independent expanders.
11987 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
11988 ("vec_ordered", "vec_unordered"): New expanders.
11990 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11992 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
11995 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11997 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
11998 vec_unpacks_lo_v16qi.
11999 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12001 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12003 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12004 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12005 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12007 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12009 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12011 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12013 ("*vec_slb<mode>"): New insn pattern.
12014 ("vec_shr_<mode>"): New expander.
12015 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12016 and force the shift count operand to V16QImode.
12017 ("vec_srb<mode>"): Set shift count mode to V16QI.
12019 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12021 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12022 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12023 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12025 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12028 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12030 2017-09-26 Richard Biener <rguenther@suse.de>
12032 PR tree-optimization/82320
12033 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12036 2017-09-25 Jeff Law <law@redhat.com>
12038 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12039 prototype for new argument.
12040 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12041 mostly extracted from rs6000_emit_allocate_stack.
12042 (rs6000_emit_probe_stack_range_stack_clash): New function.
12043 (rs6000_emit_allocate_stack): Call
12044 rs6000_emit_probe_stack_range_stack_clash as needed.
12045 (rs6000_emit_probe_stack_range): Add additional argument
12046 to call to gen_probe_stack_range{si,di}.
12047 (output_probe_stack_range): New.
12048 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
12049 (output_probe_stack_range_stack_clash): New.
12050 (rs6000_emit_prologue): Emit notes into dump file as requested.
12051 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
12052 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
12053 Add additional operand and pass it to output_probe_stack_range.
12055 2017-09-25 Bin Cheng <bin.cheng@arm.com>
12057 PR tree-optimization/82163
12058 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
12059 (checking_verify_loop_closed_ssa): New parameter.
12060 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
12061 (check_loop_closed_ssa_stmt): Delete.
12062 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
12063 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
12064 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
12067 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
12069 * brig-builtins.def: Treat HSAIL barrier builtins as
12070 setjmp/longjump style functions.
12072 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12074 * target.def (constant_alignment): New hook.
12075 * defaults.h (CONSTANT_ALIGNMENT): Delete.
12076 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
12077 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
12078 * doc/tm.texi: Regenerate.
12079 * targhooks.h (default_constant_alignment): Declare.
12080 (constant_alignment_word_strings): Likewise.
12081 * targhooks.c (default_constant_alignment): New function.
12082 (constant_alignment_word_strings): Likewise.
12083 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
12084 instead of CONSTANT_ALIGNMENT.
12085 * varasm.c (align_variable, get_variable_align, build_constant_desc)
12086 (force_const_mem): Likewise.
12087 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
12088 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
12089 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
12090 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12091 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
12093 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
12094 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12095 constant_alignment_word_strings.
12096 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
12097 (CONSTANT_ALIGNMENT): Likewise.
12098 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12099 (arm_constant_alignment): New function.
12100 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
12101 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12102 constant_alignment_word_strings.
12103 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
12104 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12105 constant_alignment_word_strings.
12106 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
12107 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12108 (cris_constant_alignment): New function.
12109 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
12110 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12111 (epiphany_constant_alignment): New function.
12112 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
12113 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12114 constant_alignment_word_strings.
12115 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
12116 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12117 constant_alignment_word_strings.
12118 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
12119 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12120 constant_alignment_word_strings.
12121 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
12122 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
12123 * config/i386/i386.c (ix86_constant_alignment): Make static.
12124 Use the same interface as the target hook.
12125 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12126 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
12127 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12128 constant_alignment_word_strings.
12129 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
12130 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
12131 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12132 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
12133 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12134 constant_alignment_word_strings.
12135 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
12136 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12137 constant_alignment_word_strings.
12138 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
12139 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12140 constant_alignment_word_strings.
12141 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
12142 * config/microblaze/microblaze.c (microblaze_constant_alignment):
12144 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12145 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
12146 * config/mips/mips.c (mips_constant_alignment): New function.
12147 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12148 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
12149 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
12150 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12151 (mmix_constant_alignment): Make static. Use the same interface
12152 as the target hook.
12153 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
12154 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12155 constant_alignment_word_strings.
12156 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
12157 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12158 constant_alignment_word_strings.
12159 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
12160 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12161 constant_alignment_word_strings.
12162 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
12163 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12164 (rs6000_constant_alignment): New function.
12165 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
12166 * config/riscv/riscv.c (riscv_constant_alignment): New function.
12167 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12168 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
12169 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12170 (rs6000_constant_alignment): New function.
12171 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
12172 * config/s390/s390.c (s390_constant_alignment): New function.
12173 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12174 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
12175 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12176 constant_alignment_word_strings.
12177 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
12178 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12179 (sparc_constant_alignment): New function.
12180 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
12181 * config/spu/spu.c (spu_constant_alignment): New function.
12182 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12183 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
12184 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12185 constant_alignment_word_strings.
12186 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
12187 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12188 constant_alignment_word_strings.
12189 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
12190 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12191 constant_alignment_word_strings.
12192 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
12193 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12194 (visium_constant_alignment): New function.
12195 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
12196 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12197 (xtensa_constant_alignment): New function.
12198 * system.h (CONSTANT_ALIGNMENT): Poison.
12200 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
12202 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12203 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
12204 (rs6000_builtin_valid_without_lhs): New helper function.
12205 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12206 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
12208 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12210 * target.h (vec_perm_indices): Use unsigned short rather than
12212 (auto_vec_perm_indices): Likewise.
12213 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
12214 Use unsigned int rather than unsigned char.
12215 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
12217 2017-09-25 Richard Biener <rguenther@suse.de>
12219 * cfgloop.h (sort_sibling_loops): Declare.
12220 * cfgloop.c (sort_sibling_loops_cmp): New helper.
12221 (sort_sibling_loops): New function sorting the sibling loop list
12223 * graphite.c (graphite_transform_loops): Sort sibling loops.
12225 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
12227 * target.def (vec_perm_const_ok): Change sel parameter to
12229 * optabs-query.c (can_vec_perm_p): Update accordingly.
12230 * doc/tm.texi: Regenerate.
12231 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
12232 auto_vec_perm_indices and remove separate nelt field.
12233 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
12234 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
12235 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
12236 (aarch64_expand_vec_perm_const): Update accordingly.
12237 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
12238 to vec_perm_indices.
12239 * config/arm/arm.c (expand_vec_perm_d): Change perm to
12240 auto_vec_perm_indices and remove separate nelt field.
12241 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
12242 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
12243 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
12245 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
12246 to vec_perm_indices.
12247 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
12248 sel to vec_perm_indices.
12249 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
12250 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
12251 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
12253 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
12256 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
12259 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
12260 on the FUNCTION_DECL function context if it has a DIE that is a
12263 2017-09-25 Richard Biener <rguenther@suse.de>
12265 PR tree-optimization/82285
12266 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
12269 2017-09-25 Tom de Vries <tom@codesourcery.com>
12273 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
12276 2017-09-25 Richard Biener <rguenther@suse.de>
12278 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
12279 ISL errors other than isl_error_quota happen. Dump if the
12280 schedule is the same.
12281 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
12282 errors instead of aborting inside ISL.
12284 2017-09-25 Iain Sandoe <iain@codesourcery.com>
12287 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
12288 of libgcc_eh for m64.
12289 * config/i386/darwin64.h: Likewise.
12291 2017-09-25 Richard Biener <rguenther@suse.de>
12293 PR middle-end/82144
12294 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
12295 attribute for incomplete types nor twice for complete ones.
12297 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
12300 * config/i386/i386.c (ix86_print_operand_address_as): Encode
12301 %esp as %rsp to avoid 0x67 prefix if there is no index or base
12304 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
12307 * config/i386/i386.opt (mprefer-avx256): Use
12308 ix86_target_flags variable.
12309 * config/i386/i386.c (ix86_target_string): Move
12310 -mprefer-avx256 to flag2_opts.
12312 2017-09-22 Jakub Jelinek <jakub@redhat.com>
12314 PR middle-end/35691
12315 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
12316 and x != -1 | y != -1 into (x & y) != -1.
12318 2017-09-22 Steve Ellcey <sellcey@cavium.com>
12320 * config.gcc: Add new case statement to set
12321 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
12322 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
12323 s390x-*-linux* case statements. Added aarch64 to the list of
12324 supported architectures.
12326 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12328 PR tree-optimization/82289
12329 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
12330 STMT_VINFO_RELEVANT_P.
12332 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12333 Alan Hayward <alan.hayward@arm.com>
12334 David Sherwood <david.sherwood@arm.com>
12336 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
12337 for VR_RANGE only; don't allow VR_ANTI_RANGE.
12338 (extract_range_from_binary_expr_1): Don't call
12339 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
12341 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12342 Alan Hayward <alan.hayward@arm.com>
12343 David Sherwood <david.sherwood@arm.com>
12345 * target.def (preferred_vector_alignment): New hook.
12346 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
12348 * doc/tm.texi: Regenerate.
12349 * targhooks.h (default_preferred_vector_alignment): Declare.
12350 * targhooks.c (default_preferred_vector_alignment): New function.
12351 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
12353 (DR_TARGET_ALIGNMENT): New macro.
12354 (aligned_access_p): Update commentary.
12355 (vect_known_alignment_in_bytes): New function.
12356 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
12358 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
12359 Calculate the misalignment based on the target alignment rather than
12361 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
12362 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
12363 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
12364 the target alignment, rather than masking the element misalignment
12365 with the number of elements in a vector. Also use the target
12366 alignment when calculating the maximum number of peels.
12367 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
12368 instead of TYPE_ALIGN_UNIT.
12369 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
12370 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
12371 (vect_create_addr_base_for_vector_ref): Update call accordingly.
12372 (vect_create_data_ref_ptr): Likewise.
12373 (vect_setup_realignment): Realign by ANDing with
12374 -DR_TARGET_MISALIGNMENT.
12375 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
12376 the number of peels based on DR_TARGET_ALIGNMENT.
12377 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
12378 with the guaranteed alignment boundary when deciding whether
12380 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
12381 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
12382 (ensure_base_align): Remove stmt_info parameter. Get the
12383 target base alignment from DR_TARGET_ALIGNMENT.
12384 (vectorizable_store): Update call accordingly. Interpret
12385 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
12387 (vectorizable_load): Likewise.
12389 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12390 Alan Hayward <alan.hayward@arm.com>
12391 David Sherwood <david.sherwood@arm.com>
12393 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
12394 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
12395 (vect_enhance_data_refs_alignment): Likewise.
12397 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12399 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12400 error. Only quit immediately if parsing is complete.
12401 (BEGIN): Initialize fatal_err and parse_done.
12402 (begin fpu, end fpu): Check number of arguments.
12403 (begin arch, end arch): Likewise.
12404 (begin cpu, end cpu): Likewise.
12405 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12406 (optalias): Likewise.
12408 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12410 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12411 * config/arm/arm-isa.h: Delete. Move definitions to ...
12412 * arm-cpus.in: ... here. Use new feature and fgroup values.
12413 * config/arm/arm.c (arm_option_override): Use lower case for feature
12415 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12416 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12417 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12419 (print_isa_bits_for): New function.
12420 (gen_isa): New function.
12421 (gen_comm_data): Use print_isa_bits_for.
12422 (define feature): New keyword.
12423 (define fgroup): New keyword.
12424 * config/arm/t-arm (TM_H): Remove.
12425 (GTM_H): Add arm-isa.h.
12426 (arm-isa.h): Add rule to generate file.
12427 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12428 case for feature bit names.
12430 2017-09-22 Richard Biener <rguenther@suse.de>
12432 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
12434 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
12435 print of no dependency loops ...
12436 * graphite.c (graphite_transform_loops): ... here.
12437 (canonicalize_loop_closed_ssa_form): Work from inner to outer
12439 (same_close_phi_node, remove_duplicate_close_phi,
12440 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
12441 (canonicalize_loop_closed_ssa): ... here and simplify.
12442 * graphite-optimize-isl.c: Include tree-vectorizer.h.
12443 (optimize_isl): Use dump_printf_loc to tell when we stopped
12444 optimizing because of an ISL timeout.
12446 2017-09-22 Richard Biener <rguenther@suse.de>
12448 PR tree-optimization/82291
12449 * tree-if-conv.c (predicate_mem_writes): Make sure to
12450 remove writes in blocks predicated with false.
12452 2017-09-22 Richard Biener <rguenther@suse.de>
12454 * sese.c: Include cfganal.h.
12455 (if_region_set_false_region): Remove.
12456 (create_if_region_on_edge): Likewise.
12457 (move_sese_in_condition): Re-implement without destroying
12460 2017-09-22 Richard Biener <rguenther@suse.de>
12462 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
12463 Verify both BBs contain loop PHI nodes before dispatching to
12464 copy_loop_phi_args.
12465 (graphite_regenerate_ast_isl): Do not recompute dominators,
12466 do not verify three times. Restructure for clarity.
12467 * graphite-scop-detection.c (same_close_phi_node,
12468 remove_duplicate_close_phi, make_close_phi_nodes_unique,
12469 defined_in_loop_p, canonicalize_loop_closed_ssa,
12470 canonicalize_loop_closed_ssa_form): Simplify, remove excess
12471 checking and SSA rewrite, move to ...
12472 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
12473 (graphite_initialize): Do not pass in ctx, do not reset the
12474 SCEV cache, compute only dominators.
12475 (graphite_transform_loops): Allocate ISL ctx after
12476 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
12477 Maintain post-dominators only around build_scops.
12478 * sese.c (if_region_set_false_region): Make static. Free
12479 and recompute dominators.
12480 (move_sese_in_condition): Assert we don't get called with
12481 post-dominators computed.
12482 * sese.h (if_region_set_false_region): Remove.
12484 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
12486 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
12487 mode attribute for TARGET_AVX512VL.
12489 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
12491 * config/i386/i386.opt (mprefer-avx256): New option.
12492 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
12494 (ix86_preferred_simd_mode): Return 256-bit AVX modes
12495 for TARGET_PREFER_AVX256.
12496 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
12498 2017-09-21 Jeff Law <law@redhat.com>
12500 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
12501 Fix dump output if the only stack space is for pushed registers.
12503 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12505 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
12508 2017-09-21 Martin Sebor <msebor@redhat.com>
12511 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
12512 code (in C++) or code that triggers warnings.
12514 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
12516 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
12518 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
12520 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
12521 * sched-rgn.c: Ditto.
12522 * sel-sched-ir.c: Ditto.
12524 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
12526 * toplev.h (set_random_seed): Adjust return type.
12527 * toplev.c (init_local_tick): Move eager initialization of random_seed
12528 to get_random_seed. Adjust comment.
12529 (init_random_seed): Inline to get_random_seed, delete.
12530 (get_random_seed): Initialize random_seed lazily.
12531 (set_random_seed): Do not return previous value.
12532 (print_switch_value): Do not call get_random_seed.
12534 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
12536 * cgraph.c (delete_function_version): New, broken out from...
12537 (cgraph_node::delete_function_version): ...here. Rename to
12538 cgraph_node::delete_function_version_by_decl. Update all uses.
12539 (cgraph_node::remove): Call delete_function_version.
12541 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12544 * tree-inline.c (expand_call_inline): Emit clobber stmts for
12545 VAR_DECLs to which addressable non-volatile parameters are mapped
12546 and for id->retvar after the return value assignment. Clear
12547 id->retval and id->retbnd after inlining.
12549 2017-09-21 Richard Biener <rguenther@suse.de>
12551 PR tree-optimization/82276
12552 PR tree-optimization/82244
12553 * tree-vrp.c (build_assert_expr_for): Set
12554 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
12556 (remove_range_assertions): Revert earlier change.
12558 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
12561 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
12563 2017-09-21 Richard Biener <rguenther@suse.de>
12565 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12566 Restore valid IL after code generation errors.
12567 * graphite.c (graphite_transform_loops): Diagnose code
12568 generation issues as MSG_MISSED_OPTIMIZATION and continue
12569 with processing SCOPs.
12571 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12572 Alan Hayward <alan.hayward@arm.com>
12573 David Sherwood <david.sherwood@arm.com>
12575 * calls.c (compute_argument_addresses): Use simplify_gen_binary
12576 rather than choosing between plus_constant and gen_rtx_<CODE>.
12577 * expr.c (emit_push_insn): Likewise.
12578 (expand_expr_real_2): Likewise.
12580 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12581 Alan Hayward <alan.hayward@arm.com>
12582 David Sherwood <david.sherwood@arm.com>
12584 * loop-unroll.c (split_iv): Call copy_rtx on the step.
12586 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12587 Alan Hayward <alan.hayward@arm.com>
12588 David Sherwood <david.sherwood@arm.com>
12590 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
12591 calling tree_to_uhwi.
12593 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12594 Alan Hayward <alan.hayward@arm.com>
12595 David Sherwood <david.sherwood@arm.com>
12597 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
12598 INTEGER_CST rather than a negative test for ADDR_EXPR.
12600 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12601 Alan Hayward <alan.hayward@arm.com>
12602 David Sherwood <david.sherwood@arm.com>
12604 * tree-vrp.c (extract_range_from_binary_expr_1): Check
12605 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
12607 2017-09-21 Richard Biener <rguenther@suse.de>
12609 PR tree-optimization/71351
12610 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
12611 graphite_create_new_loop_guard): Remove, fold remaining parts
12613 (translate_isl_ast_node_for): ... here and simplify.
12615 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12618 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
12619 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
12620 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
12621 alternative always use QI mode, for -Os imov (=R,R) alternative
12622 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
12625 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12626 Jeff Law <law@redhat.com>
12628 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
12629 (allocate_stack_space): New function, partially extracted from
12630 s390_emit_prologue.
12631 (s390_emit_prologue): Track offset to most recent stack probe.
12632 Code to allocate space moved into allocate_stack_space.
12633 Dump actions when no stack is allocated.
12634 (s390_prologue_plus_offset): New function.
12635 (s390_emit_stack_probe): Likewise.
12637 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
12639 * common.opt (Wa, Wl, Wp, g, gz=): Add
12641 (gno-column-info): Remove.
12642 (gcolumn-info): Drop RejectNegative.
12643 (gno-): New prefix.
12644 (gno-record-gcc-switches): Remove.
12645 (grecord-gcc-switches): Drop RejectNegative.
12646 (gno-split-dwarf): Remove.
12647 (gsplit-dwarf): Drop RejectNegative.
12648 (gno-strict-dwarf): Remove.
12649 (gstrict-dwarf): Drop RejectNegative.
12650 * config/darwin.opt (gfull, gused): Add RejectNegative.
12651 * dwarf2out.c (gen_producer_string): Drop
12652 gno-record-gcc-switches handler.
12653 * optc-gen.awk: Add g to prefixes with negative forms.
12654 * opts-common.c (remapping_prefix_p): New.
12655 (find_opt): Check it.
12656 (generate_canonical_option): Test g prefix.
12657 (option_map): Add -gno- mapping.
12658 (add_misspelling_candidates): Check remapping_prefix_p.
12660 2017-09-20 Jeff Law <law@redhat.com>
12662 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
12663 thinko in stack clash protection support.
12665 * explow.c (compute_stack_clash_protection_loop_data): Use
12666 CONST_INT_P instead of explicit test. Verify object is a
12667 CONST_INT_P before looking at INTVAL.
12668 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
12669 instead of explicit test.
12671 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
12674 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
12675 address instead of to r1 and r11.
12677 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
12679 * config.gcc: Support "knm".
12680 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
12681 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12683 * config/i386/i386.c (m_KNM): Define.
12684 (processor_target_table): Add "knm".
12686 (ix86_option_override_internal): Add "knm".
12687 (ix86_issue_rate): Add PROCESSOR_KNM.
12688 (ix86_adjust_cost): Ditto.
12689 (ia32_multipass_dfa_lookahead): Ditto.
12690 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
12691 (fold_builtin_cpu): Add M_INTEL_KNM.
12692 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
12693 (processor_type): Add PROCESSOR_KNM.
12694 * config/i386/x86-tune.def: Add m_KNM.
12695 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
12697 2017-09-20 Richard Biener <rguenther@suse.de>
12699 PR tree-optimization/80213
12700 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
12701 are allowed in empty BBs as well.
12702 (canonicalize_loop_closed_ssa): Also look for other complex
12704 (scop_detection::get_sese): Include the loop-closed PHI block
12706 (scop_detection::merge_sese): Remove code adding extra blocks.
12707 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
12708 (build_scops): Assert the final returned scop is invalid.
12710 2017-09-20 Richard Biener <rguenther@suse.de>
12712 PR tree-optimization/82264
12713 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
12715 (vn_phi_lookup): Likewise.
12716 (vn_phi_insert): Likewise.
12718 2017-09-20 Jakub Jelinek <jakub@redhat.com>
12720 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
12721 that fits into uhwi or shwi, add DW_AT_const_value regardless
12722 of early_dwarf without going through RTL, using add_AT_unsigned
12725 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
12726 (DEBUG_LTO_ABBREV_SECTION): Likewise.
12727 (DEBUG_LTO_MACINFO_SECTION): Likewise.
12728 (DEBUG_MACRO_SECTION): Likewise.
12729 (DEBUG_LTO_MACRO_SECTION): Likewise.
12730 (DEBUG_STR_DWO_SECTION): Likewise.
12731 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
12732 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
12733 (DEBUG_LTO_DWO_LINE_SECTION): Define.
12734 (DEBUG_LTO_LINE_STR_SECTION): Define.
12735 (init_sections_and_labels): Initialize debug_line_str_section
12736 variable. Initialize debug_loc_section for -gdwarf-5 to
12737 DEBUG_LOCLISTS_SECTION. Formatting fixes.
12739 2017-09-20 Richard Biener <rguenther@suse.de>
12741 * graphite-sese-to-poly.c (extract_affine): Properly handle
12742 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
12744 2017-09-20 Richard Biener <rguenther@suse.de>
12746 PR tree-optimization/81373
12747 * graphite-scop-detection.c (build_cross_bb_scalars_def):
12748 Force SESE live-out defs to be handled even if they are
12751 2017-09-19 Jeff Law <law@redhat.com>
12753 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
12754 nothing for stack adjustments with REG_STACK_CHECK.
12755 * sched-deps.c (parse_add_or_inc): Reject insns with
12756 REG_STACK_CHECK from dependency breaking.
12757 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
12758 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
12759 * reg-notes.def (STACK_CHECK): New note.
12761 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
12762 (ix86_expand_prologue): Dump stack clash info as needed.
12763 Call ix86_adjust_stack_and_probe_stack_clash as needed.
12765 * function.c (dump_stack_clash_frame_info): New function.
12766 * function.h (dump_stack_clash_frame_info): Prototype.
12767 (enum stack_clash_probes): New enum.
12769 * config/alpha/alpha.c (alpha_expand_prologue): Also check
12770 flag_stack_clash_protection.
12771 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
12772 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
12773 (arm_frame_pointer_required): Likewise.
12774 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
12775 (ia64_expand_prologue): Likewise.
12776 * config/mips/mips.c (mips_expand_prologue): Likewise.
12777 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
12778 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12779 (sparc_flat_expand_prologue): Likewise.
12780 * config/spu/spu.c (spu_expand_prologue): Likewise.
12782 * explow.c: Include "params.h".
12783 (anti_adjust_stack_and_probe_stack_clash): New function.
12784 (get_stack_check_protect): Likewise.
12785 (compute_stack_clash_protection_loop_data): Likewise.
12786 (emit_stack_clash_protection_loop_start): Likewise.
12787 (emit_stack_clash_protection_loop_end): Likewise.
12788 (allocate_dynamic_stack_space): Use get_stack_check_protect.
12789 Use anti_adjust_stack_and_probe_stack_clash.
12790 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
12791 (emit_stack_clash_protection_loop_start): Likewise.
12792 (emit_stack_clash_protection_loop_end): Likewise.
12793 * rtl.h (get_stack_check_protect): Prototype.
12794 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
12795 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
12796 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
12798 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
12800 * doc/tm.texi: Rebuilt.
12801 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
12802 get_stack_check_protect.
12803 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
12804 * config/arm/arm.c (arm_expand_prologue): Likewise.
12805 (arm_frame_pointer_required): Likewise.
12806 * config/i386/i386.c (ix86_expand_prologue): Likewise.
12807 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
12808 * config/mips/mips.c (mips_expand_prologue): Likewise.
12809 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
12810 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
12811 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12812 (sparc_flat_expand_prologue): Likewise.
12814 * common.opt (-fstack-clash-protection): New option.
12815 * flag-types.h (enum stack_check_type): Note difference between
12816 -fstack-check= and -fstack-clash-protection.
12817 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
12818 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
12819 * toplev.c (process_options): Issue warnings/errors for cases
12820 not handled with -fstack-clash-protection.
12821 * doc/invoke.texi (-fstack-clash-protection): Document new option.
12822 (-fstack-check): Note additional problem with -fstack-check=generic.
12823 Note that -fstack-check is primarily for Ada and refer users
12824 to -fstack-clash-protection for stack-clash-protection.
12825 Document new params for stack clash protection.
12827 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
12829 * config/i386/i386.c (ix86_split_long_move): Do not handle
12830 address used for LEA in a special way.
12832 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
12834 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
12836 2017-09-19 Martin Sebor <msebor@redhat.com>
12839 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
12840 of incompatible types.
12842 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
12844 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12845 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
12846 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12847 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
12849 2017-09-19 Richard Biener <rguenther@suse.de>
12851 PR tree-optimization/82244
12852 * tree-vrp.c (remove_range_assertions): Do not propagate
12853 a constant to abnormals but replace the assert with a copy.
12855 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
12857 PR rtl-optimization/57878
12858 PR rtl-optimization/68988
12859 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
12860 avoidance test involving non_reload_pseudos. Move frequency test
12861 below the general fragmentation avoidance test.
12863 2017-09-19 Richard Biener <rguenther@suse.de>
12865 PR tree-optimization/69728
12866 * graphite-sese-to-poly.c (schedule_error): New global.
12867 (add_loop_schedule): Handle empty domain by failing the
12869 (build_original_schedule): Handle schedule_error.
12871 2017-09-19 Richard Biener <rguenther@suse.de>
12873 * graphite-scop-detection.c (scop_detection::can_represent_loop):
12874 Do not iterate to sibling loops but only to siblings of inner
12877 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
12880 * config/m68k/m68k.md (moveq feeding equality comparison): Check
12881 that the registers are different.
12883 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
12885 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
12886 to processor_model and "amdfam17h" to arch_names_table.
12887 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
12889 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12892 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
12894 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12895 Alan Hayward <alan.hayward@arm.com>
12896 David Sherwood <david.sherwood@arm.com>
12898 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
12900 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
12902 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
12903 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
12904 vect_schedule_slp_instance.
12905 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
12906 Update call to vect_slp_analyze_node_operations. Simplify return
12908 (vect_slp_analyze_bb_1): Update call accordingly.
12909 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
12910 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
12911 (vect_schedule_slp): Update call accordingly.
12913 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12914 Alan Hayward <alan.hayward@arm.com>
12915 David Sherwood <david.sherwood@arm.com>
12917 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
12918 with types that aren't in fact scalar.
12920 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12922 * tree-vect-slp.c (vect_record_max_nunits): New function,
12924 (vect_build_slp_tree_1): ...here.
12925 (vect_build_slp_tree_2): Call it for phis too.
12927 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12929 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
12930 to vect_get_vec_def_for_operand when getting the mask operand.
12932 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12933 Alan Hayward <alan.hayward@arm.com>
12934 David Sherwood <david.sherwood@arm.com>
12936 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
12939 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12940 Alan Hayward <alan.hayward@arm.com>
12941 David Sherwood <david.sherwood@arm.com>
12943 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
12944 calculation for vector booleans.
12946 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12947 Alan Hayward <alan.hayward@arm.com>
12948 David Sherwood <david.sherwood@arm.com>
12950 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
12952 (vect_transform_stmt): ...here.
12953 (vect_analyze_stmt): Use it instead of calling
12954 vectorizable_live_operation directly.
12956 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
12958 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
12959 non-SIMT targets in acc vector loops.
12961 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
12963 * configure.ac: Add arc and check if assembler supports gdwarf2.
12964 * configure: Regenerate.
12966 2017-09-18 Richard Biener <rguenther@suse.de>
12968 PR tree-optimization/82220
12969 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
12970 epilogue niters from the min_profitable_iters compute.
12972 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12975 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
12976 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
12977 (ix86_init_pic_reg): Revert 2017-09-01 changes.
12979 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
12982 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
12983 switching to a new text section.
12985 2017-09-18 Richard Biener <rguenther@suse.de>
12987 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
12989 (build_alias_set): Reject aliases with no access function.
12991 2017-09-18 Richard Biener <rguenther@suse.de>
12993 PR tree-optimization/79622
12994 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
12996 (build_cross_bb_scalars_use): Likewise.
12998 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13000 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13002 2017-09-18 Alan Modra <amodra@gmail.com>
13005 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13006 stack_pointer_rtx for count 0. Update comments. Break up
13007 large rtl expression.
13009 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13011 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13012 Increase to 20 bytes.
13013 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13014 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13015 or avx version of the stub.
13017 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13020 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13021 compute the minimum stack alignment. Also update preferred stack
13022 boundary for leaf functions.
13024 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13026 PR tree-optimization/82228
13027 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13030 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13032 * common/config/nds32/nds32-common.c
13033 (nds32_option_optimization_table): Refine formatting.
13034 (nds32_option_optimization_table): Use -fsched-pressure and
13035 -fomit-frame-pointer for specific optimization level.
13037 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13039 * config/nds32/nds32.c: Refine formatting and comments.
13040 * config/nds32/nds32.h: Likewise.
13041 * config/nds32/nds32.md: Likewise.
13042 * config/nds32/nds32-cost.c: Likewise.
13043 * config/nds32/nds32-isr.c: Likewise.
13044 * config/nds32/nds32-md-auxiliary.c: Likewise.
13045 * config/nds32/nds32-multiple.md: Likewise.
13046 * config/nds32/nds32-predicates.c: Likewise.
13048 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
13049 Jakub Jelinek <jakub@redhat.com>
13051 Add support for -std=c++2a.
13052 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
13054 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
13056 2017-09-15 Steve Ellcey <sellcey@cavium.com>
13059 * doc/extend.texi (Common Function Attributes): Add
13060 references to ARM, AArch64, and S/390 specific attributes.
13061 (Function Specific Option Pragmas): Add AArch64 and S/390
13062 to list of back ends that support the target pragma.
13064 2017-09-15 Nathan Sidwell <nathan@acm.org>
13066 * doc/standards.texi: Fix C++17 description. Update URLs for
13069 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13071 * common.opt (Wcast-align=strict): New warning option.
13072 * doc/invoke.texi: Document -Wcast-align=strict.
13074 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
13076 * cgraph.h (cgraph_thunk_info): Add comments.
13077 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
13078 assert for VIRTUAL_* arguments stricter.
13080 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
13082 PR tree-optimization/71026
13083 * match.pd: Move RDIV patterns from fold-const.c
13084 * fold-const.c (distribute_real_division): Removed.
13085 (fold_binary_loc): Remove calls to distribute_real_divison.
13087 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13089 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
13090 c++1z and gnu++1z as deprecated. Change other references to
13091 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13092 Change -Wc++1z-compat to -Wc++17-compat.
13093 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
13094 * dwarf2out.c (highest_c_language): Handle C++17.
13095 (gen_compile_unit_die): Likewise.
13097 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13099 PR rtl-optimization/82192
13100 * combine.c (make_extraction): Don't look through non-paradoxical
13101 SUBREGs or TRUNCATE if pos + len is or might be bigger than
13104 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13105 Alan Hayward <alan.hayward@arm.com>
13106 David Sherwood <david.sherwood@arm.com>
13108 * target.def (function_arg_offset): New hook.
13109 * targhooks.h (default_function_arg_offset): Declare.
13110 * targhooks.c (default_function_arg_offset): New function.
13111 * function.c (locate_and_pad_parm): Use
13112 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
13113 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
13114 (TARGET_FUNCTION_ARG_OFFSET): ...this.
13115 * doc/tm.texi: Regenerate.
13116 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
13117 * config/spu/spu.c (spu_function_arg_offset): New function.
13118 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
13119 * system.h (FUNCTION_ARG_OFFSET): Poison.
13121 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13122 Alan Hayard <alan.hayward@arm.com>
13123 David Sherwood <david.sherwood@arm.com>
13125 * target.def (truly_noop_truncation): New hook.
13126 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
13127 than TRULY_NOOP_TRUNCATION.
13128 * hooks.h (hook_bool_uint_uint_true): Declare.
13129 * hooks.c (hook_bool_uint_uint_true): New function.
13130 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
13131 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
13132 * doc/tm.texi: Regenerate.
13133 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
13134 rather than TRULY_NOOP_TRUNCATION in comments.
13135 (simplify_comparison): Likewise.
13136 (record_truncated_value): Likewise.
13137 * expmed.c (extract_bit_field_1): Likewise.
13138 (extract_split_bit_field): Likewise.
13139 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
13140 instead of TRULY_NOOP_TRUNCATION.
13141 * function.c (assign_parm_setup_block): Likewise.
13142 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
13143 * rtlhooks.c: Include target.h.
13144 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
13145 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
13146 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
13147 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
13148 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
13149 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
13150 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
13151 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
13152 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
13153 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
13154 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
13155 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
13156 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
13157 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
13158 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
13159 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
13160 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
13161 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
13162 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
13163 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
13164 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
13165 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
13166 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
13167 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
13168 * config/mips/mips.c (mips_truly_noop_truncation): New function.
13169 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13170 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13171 rather than TRULY_NOOP_TRUNCATION in comments.
13172 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
13173 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
13174 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
13175 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
13176 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
13177 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
13178 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
13179 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
13180 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
13181 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
13182 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
13183 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13184 rather than TRULY_NOOP_TRUNCATION in comments.
13185 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
13186 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
13187 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
13188 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
13189 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
13190 TRULY_NOOP_TRUNCATION condition.
13191 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
13192 (TRULY_NOOP_TRUNCATION): Delete.
13193 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
13194 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
13195 * config/spu/spu.c (spu_truly_noop_truncation): New function.
13196 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13197 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
13198 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
13199 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
13200 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13201 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13202 rather than TRULY_NOOP_TRUNCATION in comments.
13203 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
13204 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
13205 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
13206 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
13207 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
13208 * system.h (TRULY_NOOP_TRUNCATION): Poison.
13210 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
13213 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
13214 (*cmp_ior): Likewise.
13215 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
13216 (*ior_scc_scc_cmp): Likewise.
13217 (*and_scc_scc): Likewise.
13218 (*and_scc_scc_cmp): Likewise.
13220 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13221 Alan Hayard <alan.hayward@arm.com>
13222 David Sherwood <david.sherwood@arm.com>
13224 * target.def (can_change_mode_class): New hook.
13225 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
13226 (hard_regno_nregs): Likewise.
13227 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
13228 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
13229 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
13230 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
13231 (LOAD_EXTEND_OP): Update accordingly.
13232 * doc/tm.texi: Regenerate.
13233 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
13234 CANNOT_CHANGE_MODE_CLASS.
13235 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
13236 (REG_CAN_CHANGE_MODE_P): ...this new macro.
13237 * combine.c (simplify_set): Update accordingly.
13238 * emit-rtl.c (validate_subreg): Likewise.
13239 * recog.c (general_operand): Likewise.
13240 * regcprop.c (mode_change_ok): Likewise.
13241 * reload1.c (choose_reload_regs): Likewise.
13242 (inherit_piecemeal_p): Likewise.
13243 * rtlanal.c (simplify_subreg_regno): Likewise.
13244 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
13245 instead of CANNOT_CHANGE_MODE_CLASS.
13246 (reload_cse_simplify_operands): Likewise.
13247 * reload.c (push_reload): Use targetm.can_change_mode_class
13248 instead of CANNOT_CHANGE_MODE_CLASS.
13249 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
13250 REG_CANNOT_CHANGE_MODE_P.
13251 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13252 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
13253 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13254 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13255 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13256 (arm_can_change_mode_class): New function.
13257 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13258 than CANNOT_CHANGE_MODE_CLASS in comments.
13259 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13260 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
13261 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
13262 (ix86_can_change_mode_class): ...this new function, inverting the
13263 sense of the return value.
13264 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13265 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13266 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13267 (ia64_can_change_mode_class): New function.
13268 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13269 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
13270 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
13271 (m32c_can_change_mode_class): ...this new function, inverting the
13272 sense of the return value.
13273 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13274 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13275 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
13276 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
13277 (mips_can_change_mode_class): ...this new function, inverting the
13278 sense of the return value.
13279 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13280 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13281 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13282 (msp430_can_change_mode_class): New function.
13283 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13284 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
13285 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13286 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13287 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13288 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
13289 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13290 (pa_cannot_change_mode_class): Replace with...
13291 (pa_can_change_mode_class): ...this new function, inverting the
13292 sense of the return value.
13293 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13294 than CANNOT_CHANGE_MODE_CLASS in comments.
13295 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13296 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
13297 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13298 (pdp11_cannot_change_mode_class): Replace with...
13299 (pdp11_can_change_mode_class): ...this new function, inverting the
13300 sense of the return value.
13301 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13302 * config/powerpcspe/powerpcspe-protos.h
13303 (rs6000_cannot_change_mode_class_ptr): Delete.
13304 * config/powerpcspe/powerpcspe.c
13305 (rs6000_cannot_change_mode_class_ptr): Delete.
13306 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13307 (rs6000_option_override_internal): Assign to
13308 targetm.can_change_mode_class instead of
13309 rs6000_cannot_change_mode_class_ptr.
13310 (rs6000_cannot_change_mode_class): Replace with...
13311 (rs6000_can_change_mode_class): ...this new function, inverting the
13312 sense of the return value.
13313 (rs6000_debug_cannot_change_mode_class): Replace with...
13314 (rs6000_debug_can_change_mode_class): ...this new function.
13315 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13316 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
13317 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13318 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13319 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
13321 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
13322 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13323 (rs6000_option_override_internal): Assign to
13324 targetm.can_change_mode_class instead of
13325 rs6000_cannot_change_mode_class_ptr.
13326 (rs6000_cannot_change_mode_class): Replace with...
13327 (rs6000_can_change_mode_class): ...this new function, inverting the
13328 sense of the return value.
13329 (rs6000_debug_cannot_change_mode_class): Replace with...
13330 (rs6000_debug_can_change_mode_class): ...this new function.
13331 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13332 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
13333 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
13334 (s390_can_change_mode_class): ...this new function, inverting the
13335 sense of the return value.
13336 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13337 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13338 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
13339 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13340 (sh_cannot_change_mode_class): Replace with...
13341 (sh_can_change_mode_class): ...this new function, inverting the
13342 sense of the return value.
13343 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13344 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13345 (sparc_can_change_mode_class): New function.
13346 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13347 * config/spu/spu.c (spu_can_change_mode_class): New function.
13348 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13349 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13350 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13351 (visium_can_change_mode_class): New function.
13352 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
13354 2017-09-15 Richard Biener <rguenther@suse.de>
13356 PR tree-optimization/82217
13357 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
13358 but not undefined case.
13360 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13363 * postreload.c (reload_cse_simplify_operands): Skip
13364 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
13366 2017-09-15 Richard Biener <rguenther@suse.de>
13368 PR tree-optimization/68823
13369 * graphite-scop-detection.c (build_alias_set): If we have a
13370 possible dependence check whether we can handle them by just
13371 looking at the DRs DR_ACCESS_FNs.
13372 (build_scops): If build_alias_set fails, fail the SCOP.
13374 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13376 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
13377 to support float128 built-in functions that require the ISA 3.0
13379 (BU_FLOAT128_3_HW): Likewise.
13380 (SQRTF128): Add support for the IEEE 128-bit square root and fma
13381 built-in functions.
13382 (FMAF128): Likewise.
13384 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13385 support for built-in functions that need the ISA 3.0 IEEE 128-bit
13386 floating point instructions.
13387 (rs6000_invalid_builtin): Likewise.
13388 (rs6000_builtin_mask_names): Likewise.
13389 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
13390 (RS6000_BTM_FLOAT128_HW): Likewise.
13391 (RS6000_BTM_COMMON): Likewise.
13392 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
13394 * doc/extend.texi (RS/6000 built-in functions): Document the
13395 IEEE 128-bit floating point square root and fused multiply-add
13396 built-in functions.
13398 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
13400 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
13401 reg (r2) isn't in the set of registers defined in the prologue.
13403 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13404 Alan Hayward <alan.hayward@arm.com>
13405 David Sherwood <david.sherwood@arm.com>
13407 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
13408 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
13409 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
13410 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
13411 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
13413 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13414 max_vectorization_factor.
13415 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
13417 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13418 Alan Hayward <alan.hayward@arm.com>
13419 David Sherwood <david.sherwood@arm.com>
13421 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
13422 (vect_worthwhile_without_simd_p): Declare.
13423 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
13424 (vectorizable_reduction): Use it.
13425 * tree-vect-stmts.c (vectorizable_shift): Likewise.
13426 (vectorizable_operation): Likewise.
13428 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13429 Alan Hayward <alan.hayward@arm.com>
13430 David Sherwood <david.sherwood@arm.com>
13432 * tree-vectorizer.h (vect_get_num_copies): New function.
13433 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
13434 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13435 (vectorizable_induction): Likewise.
13436 (vectorizable_live_operation): Likewise.
13437 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13438 (vectorizable_bswap): Likewise.
13439 (vectorizable_call): Likewise.
13440 (vectorizable_conversion): Likewise.
13441 (vectorizable_assignment): Likewise.
13442 (vectorizable_shift): Likewise.
13443 (vectorizable_operation): Likewise.
13444 (vectorizable_store): Likewise.
13445 (vectorizable_load): Likewise.
13446 (vectorizable_condition): Likewise.
13447 (vectorizable_comparison): Likewise.
13448 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
13450 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13451 Alan Hayward <alan.hayward@arm.com>
13452 David Sherwood <david.sherwood@arm.com>
13454 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
13455 of vect_init_vector.
13457 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13458 Alan Hayward <alan.hayward@arm.com>
13459 David Sherwood <david.sherwood@arm.com>
13461 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
13462 an inline wrapper that provides a location.
13463 (gimple_build_vector): Likewise.
13464 * gimple-fold.c (gimple_build_vector_from_val): New function.
13465 (gimple_build_vector): Likewise.
13466 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
13467 functions to build the initial value. Always return a gimple value.
13468 (get_initial_defs_for_reduction): Likewise. Only compute
13470 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
13471 vect_init_vector on the results from get_initial_def(s)_for_reduction.
13472 (vectorizable_induction): Use gimple_build_vector rather than
13475 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13476 Alan Hayward <alan.hayward@arm.com>
13477 David Sherwood <david.sherwood@arm.com>
13479 * target.h (vec_perm_indices): New typedef.
13480 (auto_vec_perm_indices): Likewise.
13481 * optabs-query.h: Include target.h
13482 (can_vec_perm_p): Take a vec_perm_indices *.
13483 * optabs-query.c (can_vec_perm_p): Likewise.
13484 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
13485 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13486 * tree-vect-generic.c (lower_vec_perm): Likewise.
13487 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
13488 (vect_grouped_load_supported): Likewise.
13489 (vect_shift_permute_load_chain): Likewise.
13490 (vect_permute_store_chain): Use auto_vec_perm_indices.
13491 (vect_permute_load_chain): Likewise.
13492 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
13493 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
13494 Update uses of can_vec_perm_p.
13495 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
13496 mode with a number of elements. Take a vec_perm_indices *.
13497 (vect_create_epilog_for_reduction): Update accordingly.
13498 Use auto_vec_perm_indices.
13499 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
13500 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
13501 (vect_transform_slp_perm_load): Likewise.
13502 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
13503 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
13504 (vect_gen_perm_mask_checked): Likewise.
13505 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
13506 (vect_gen_perm_mask_checked): Likewise.
13507 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
13508 (vectorizable_store): Likewise.
13509 (vectorizable_load): Likewise.
13510 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
13511 (vectorizable_bswap): Likewise.
13513 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13514 Alan Hayward <alan.hayward@arm.com>
13515 David Sherwood <david.sherwood@arm.com>
13517 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
13518 * tree.c (build_vector): Likewise.
13519 (build_vector_from_ctor): Update accordingly.
13520 (build_vector_from_val): Likewise.
13521 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13522 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13523 * tree-vect-generic.c (add_rshift): Likewise.
13524 (expand_vector_divmod): Likewise.
13525 (optimize_vector_constructor): Likewise.
13526 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
13527 (vect_transform_slp_perm_load): Likewise.
13528 (vect_schedule_slp_instance): Likewise.
13529 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
13530 (vectorizable_call): Likewise.
13531 (vect_gen_perm_mask_any): Likewise. Add elements in order.
13532 * expmed.c (make_tree): Likewise.
13533 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
13534 a vector passed to build_vector.
13535 (fold_convert_const): Likewise.
13536 (exact_inverse): Likewise.
13537 (fold_ternary_loc): Likewise.
13538 (fold_relational_const): Likewise.
13539 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
13540 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
13541 (const_unop): Likewise. Store the reduction accumulator in a
13542 variable rather than an array.
13543 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
13544 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
13545 the new vector, rather than constructing it after the input arrays.
13546 (native_interpret_vector): Use auto_vec<tree> when building
13547 a vector passed to build_vector. Add elements in order.
13548 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
13549 auto_vec<tree> when building a vector passed to build_vector.
13550 (vect_create_epilog_for_reduction): Likewise.
13551 (vectorizable_induction): Likewise.
13552 (get_initial_def_for_reduction): Likewise. Fix indentation of
13554 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
13556 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
13557 passed to build_vector.
13559 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13560 Alan Hayward <alan.hayward@arm.com>
13561 David Sherwood <david.sherwood@arm.com>
13563 * tree-core.h (tree_base::u): Add an "nelts" field.
13564 (tree_vector): Use VECTOR_CST_NELTS as the length.
13565 * tree.c (tree_size): Likewise.
13566 (make_vector): Initialize VECTOR_CST_NELTS.
13567 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
13568 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
13569 TYPE_VECTOR_SUBPARTS.
13570 * expr.c (const_vector_mask_from_tree): Consistently use "units"
13571 as the number of units, setting it from VECTOR_CST_NELTS.
13572 (const_vector_from_tree): Likewise.
13573 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
13574 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
13575 (fold_negate_expr_1): Likewise.
13576 (fold_convert_const): Likewise.
13577 (const_binop): Likewise. Differentiate the number of output and
13579 (const_unop): Likewise.
13580 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
13581 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
13582 in cases that did the opposite.
13584 2017-09-14 Richard Biener <rguenther@suse.de>
13586 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
13589 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
13591 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
13593 2017-09-14 Jakub Jelinek <jakub@redhat.com>
13596 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
13597 if and where to split a bb, except for splitting before debug insn
13598 sequences followed by non-label real insn. Delete debug insns
13599 in between basic blocks.
13601 * combine.c (make_compound_operation_int): Formatting fixes.
13603 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
13604 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13605 * config/netbsd.h (LINK_EH_SPEC): Likewise.
13606 * config/sol2.h (LINK_EH_SPEC): Likewise.
13607 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13608 * config/s390/linux.h (LINK_SPEC): Likewise.
13609 * config/freebsd.h (LINK_EH_SPEC): Likewise.
13610 * config/openbsd.h (LINK_EH_SPEC): Likewise.
13611 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13612 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
13613 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
13614 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13615 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
13616 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
13618 2017-09-13 Jakub Jelinek <jakub@redhat.com>
13620 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
13622 (ENDFILE_LINUX_SPEC): Likewise.
13623 (LINK_EH_SPEC): Likewise.
13624 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
13625 (LINK_OS_LINUX_SPEC32): Likewise.
13626 (LINK_OS_LINUX_SPEC64): Likewise.
13627 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
13628 (LINK_OS_LINUX_SPEC): Likewise.
13630 2017-09-13 Martin Liska <mliska@suse.cz>
13632 PR middle-end/82154
13633 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
13634 CASE_HIGH is NULL_TREE.
13636 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13637 Alan Hayward <alan.hayward@arm.com>
13638 David Sherwood <david.sherwood@arm.com>
13640 * target.def (secondary_memory_needed): New hook.
13641 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
13642 instead of SECONDARY_MEMORY_NEEDED.
13643 (secondary_memory_needed_mode): Likewise.
13644 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
13645 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
13646 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
13647 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
13648 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
13649 * doc/tm.texi: Regenerate.
13650 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
13651 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
13652 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13653 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
13654 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
13655 * config/i386/i386.c (inline_secondary_memory_needed): Put the
13656 mode argument first and change the reg_class arguments to reg_class_t.
13657 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
13658 Make static. Update the call to inline_secondary_memory_needed.
13659 (ix86_register_move_cost): Update the call to
13660 inline_secondary_memory_needed.
13661 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13662 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
13664 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
13665 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13667 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
13668 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
13669 * config/mips/mips.c (mips_secondary_memory_needed): Make static
13670 and match hook interface. Add comment from mips.h.
13671 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13672 * config/mmix/mmix.md (truncdfsf2): Refer to
13673 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13675 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
13676 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
13677 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13678 (pa_secondary_memory_needed): New function.
13679 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
13680 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
13681 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13682 (pdp11_secondary_memory_needed): Make static and match hook interface.
13683 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
13684 * config/powerpcspe/powerpcspe-protos.h
13685 (rs6000_secondary_memory_needed_ptr): Delete.
13686 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
13688 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13689 (rs6000_option_override_internal): Assign to
13690 targetm.secondary_memory_needed rather than
13691 rs6000_secondary_memory_needed_ptr.
13692 (rs6000_secondary_memory_needed): Match hook interface.
13693 (rs6000_debug_secondary_memory_needed): Likewise.
13694 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
13695 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
13696 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
13697 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13698 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
13699 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
13701 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
13702 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13703 (rs6000_option_override_internal): Assign to
13704 targetm.secondary_memory_needed rather than
13705 rs6000_secondary_memory_needed_ptr.
13706 (rs6000_secondary_memory_needed): Match hook interface.
13707 (rs6000_debug_secondary_memory_needed): Likewise.
13708 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
13709 * config/s390/s390.c (s390_secondary_memory_needed): New function.
13710 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13711 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
13712 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13713 (sparc_secondary_memory_needed): New function.
13714 * lra-constraints.c (check_and_process_move): Refer to
13715 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13717 (curr_insn_transform): Likewise.
13718 (process_alt_operands): Use targetm.secondary_memory_needed
13719 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13720 (check_secondary_memory_needed_p): Likewise.
13721 (choose_split_class): Likewise.
13722 * reload.c: Unconditionally include code that was previously
13723 conditional on SECONDARY_MEMORY_NEEDED.
13724 (push_secondary_reload): Use targetm.secondary_memory_needed
13725 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13726 (push_reload): Likewise.
13727 * reload1.c: Unconditionally include code that was previously
13728 conditional on SECONDARY_MEMORY_NEEDED.
13729 (choose_reload_regs): Use targetm.secondary_memory_needed
13730 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13731 (gen_reload): Likewise.
13732 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
13734 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13735 Alan Hayward <alan.hayward@arm.com>
13736 David Sherwood <david.sherwood@arm.com>
13738 * target.def (secondary_memory_needed_mode): New hook:
13739 * targhooks.c (default_secondary_memory_needed_mode): Declare.
13740 * targhooks.h (default_secondary_memory_needed_mode): New function.
13741 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
13742 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
13743 * doc/tm.texi: Regenerate.
13744 * lra-constraints.c (check_and_process_move): Use
13745 targetm.secondary_memory_needed_mode instead of
13746 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
13747 (curr_insn_transform): Likewise.
13748 * reload.c (get_secondary_mem): Likewise.
13749 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13750 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
13752 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13753 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13754 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
13755 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13756 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
13758 * config/powerpcspe/powerpcspe-protos.h
13759 (rs6000_secondary_memory_needed_mode): Delete.
13760 * config/powerpcspe/powerpcspe.c
13761 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13762 (rs6000_secondary_memory_needed_mode): Make static.
13763 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13764 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
13766 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13768 (rs6000_secondary_memory_needed_mode): Make static.
13769 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13770 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
13771 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13772 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13773 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13775 (sparc_secondary_memory_needed_mode): New function.
13776 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
13778 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
13780 * config/aarch64/constraints.md (Umq): New constraint.
13781 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
13783 (mov<mode>): Update condition.
13785 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13787 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
13788 when bitposition is the same.
13790 2017-09-13 Richard Biener <rguenther@suse.de>
13792 * dwarf2out.c (output_die_symbol): Remove.
13793 (output_die): Do not output a DIEs symbol.
13795 2017-09-13 Richard Biener <rguenther@suse.de>
13797 PR middle-end/82128
13798 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
13799 default-def to avoid breaking iterator update with the weird
13800 interaction with cgraph_update_edges_for_call_stmt_node.
13802 2017-09-13 Richard Biener <rguenther@suse.de>
13804 * tree-cfg.c (verify_gimple_assign_binary): Add verification
13805 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
13806 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
13807 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
13809 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13811 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
13812 Disable pc relative literal load irrespective of
13813 TARGET_FIX_ERR_A53_84341 for default.
13815 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
13817 * config/sparc/sparc.c (output_return): Output the source location of
13818 the insn in the delay slot, if any.
13819 (output_sibcall): Likewise.
13821 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
13824 * common.opt (-static-pie): New alias.
13825 (shared): Negate static-pie.
13826 (-no-pie): Update help text.
13828 (static-pie): New option.
13829 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
13830 -static-pie support.
13831 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
13832 (LINK_EH_SPEC): Likewise.
13833 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13834 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13835 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13836 * gcc.c (LINK_COMMAND_SPEC): Likewise.
13837 (init_gcc_specs): Likewise.
13838 (init_spec): Likewise.
13839 (display_help): Update help message for -pie.
13840 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
13843 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
13845 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
13846 (movdi_aarch64): Likewise.
13847 (movti_aarch64): Likewise.
13849 2017-09-12 Simon Wright <simon@pushface.org>
13852 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
13853 calculation of the minor version, always output as 0.
13855 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13858 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13859 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
13860 on it early, rather than manual conversion late. For
13861 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
13862 instead of performing manual conversion.
13864 2017-09-12 Carl Love <cel@us.ibm.com>
13866 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13867 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
13869 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13870 VMULOSW): Add definitions.
13871 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13872 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13873 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13874 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13875 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13877 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13879 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
13881 (movti_aarch64): Likewise.
13882 (movdf_aarch64): Likewise.
13883 (movtf_aarch64): Likewise.
13884 (load_pairdi): Likewise.
13885 (store_pairdi): Likewise.
13886 (load_pairdf): Likewise.
13887 (store_pairdf): Likewise.
13888 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
13889 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
13890 (ldr_got_small_<mode>): Likewise.
13891 (ldr_got_small_28k_<mode>): Likewise.
13892 (ldr_got_tiny): Likewise.
13893 * config/aarch64/iterators.md (ldst_sz): New.
13894 (ldpstp_sz): Likewise.
13895 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
13897 (thunderx_load): Split load_8 to load_16.
13898 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
13900 (thunderx2t99_storepair_basic): Split store_8 to store_16.
13901 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
13902 (xgene1_store_pair): Split store_8 to store_16.
13903 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
13904 (falkor_st_0_st_sd): Split store_8 to store_16.
13906 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13908 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
13909 and store1/2/3/4 to store_4/8/12/16.
13910 * config/aarch64/aarch64.md: Update for rename.
13911 * config/arm/arm.md: Likewise.: Likewise.
13912 * config/arm/arm.c: Likewise.
13913 * config/arm/thumb1.md: Likewise.
13914 * config/arm/thumb2.md: Likewise.
13915 * config/arm/vfp.md: Likewise.
13916 * config/arm/arm-generic.md: Likewise.
13917 * config/arm/arm1020e.md: Likewise.
13918 * config/arm/arm1026ejs.md: Likewise.
13919 * config/arm/arm1136jfs.md: Likewise.
13920 * config/arm/arm926ejs.md: Likewise.
13921 * config/arm/cortex-a15.md: Likewise.
13922 * config/arm/cortex-a17.md: Likewise.
13923 * config/arm/cortex-a5.md: Likewise.
13924 * config/arm/cortex-a53.md: Likewise.
13925 * config/arm/cortex-a57.md: Likewise.
13926 * config/arm/cortex-a7.md: Likewise.
13927 * config/arm/cortex-a8.md: Likewise.
13928 * config/arm/cortex-a9.md: Likewise.
13929 * config/arm/cortex-m4.md: Likewise.
13930 * config/arm/cortex-m7.md: Likewise.
13931 * config/arm/cortex-r4.md: Likewise.
13932 * config/arm/exynos-m1.md: Likewise.
13933 * config/arm/fa526.md: Likewise.
13934 * config/arm/fa606te.md: Likewise.
13935 * config/arm/fa626te.md: Likewise.
13936 * config/arm/fa726te.md: Likewise.
13937 * config/arm/fmp626.md: Likewise.
13938 * config/arm/iwmmxt.md: Likewise.
13939 * config/arm/ldmstm.md: Likewise.
13940 * config/arm/marvell-pj4.md: Likewise.
13941 * config/arm/xgene1.md: Likewise.
13942 * config/aarch64/thunderx.md: Likewise.
13943 * config/aarch64/thunderx2t99.md: Likewise.
13944 * config/aarch64/falkor.md: Likewise.
13946 2017-09-12 Martin Liska <mliska@suse.cz>
13948 * attribs.c (private_lookup_attribute): New function.
13949 * attribs.h (private_lookup_attribute): Declared here.
13950 (lookup_attribute): Called from this place.
13952 2017-09-12 Richard Biener <rguenther@suse.de>
13954 PR tree-optimization/82157
13955 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
13956 stmts with side-effects.
13958 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13959 Alan Hayward <alan.hayward@arm.com>
13960 David Sherwood <david.sherwood@arm.com>
13962 * target.def (hard_regno_nregs): New hook.
13963 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
13964 * targhooks.h (default_hard_regno_nregs): Declare.
13965 * targhooks.c (default_hard_regno_nregs): New function.
13966 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
13967 (TARGET_HARD_REGNO_NREGS): ...this hook.
13968 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
13969 (CLASS_MAX_NREGS): Likewise.
13970 * doc/tm.texi: Regenerate.
13971 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
13972 instead of HARD_REGNO_NREGS.
13973 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
13974 HARD_REGNO_NREGS in the comment.
13975 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
13976 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
13977 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
13978 Return an unsigned int.
13979 (TARGET_HARD_REGNO_NREGS): Redefine.
13980 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
13981 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
13982 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13983 (arc_hard_regno_nregs): New function.
13984 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
13985 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
13986 (arm_hard_regno_nregs): New function.
13987 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
13988 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
13989 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
13990 (TARGET_HARD_REGNO_NREGS): Redefine.
13991 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
13992 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
13993 (HARD_REGNO_NREGS): Delete.
13994 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
13995 (cr16_hard_regno_nregs): New function.
13996 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
13997 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
13998 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
13999 (cris_hard_regno_nregs): New function.
14000 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14001 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14002 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14003 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14004 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14005 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14006 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14007 (frv_hard_regno_nregs): Make static. Take and return an
14009 (frv_class_max_nregs): Remove outdated copy of documentation.
14010 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14011 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14012 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14013 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14014 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14015 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14016 (TARGET_HARD_REGNO_NREGS): Redefine.
14017 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14018 (CLASS_MAX_NREGS): Update comment.
14019 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14020 (ia64_hard_regno_nregs): New function.
14021 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14022 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14023 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14024 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14025 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14027 (m32c_hard_regno_nregs): Likewise. Make static.
14028 (TARGET_HARD_REGNO_NREGS): Redefine.
14029 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14030 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14031 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14032 (m68k_hard_regno_nregs): New function.
14033 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14034 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14035 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14036 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14037 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14038 Take and return an unsigned int.
14039 (TARGET_HARD_REGNO_NREGS): Redefine.
14040 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14041 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14042 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
14043 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
14044 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
14045 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
14046 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
14047 (msp430_hard_regno_nregs): Make static. Take and return an
14049 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
14050 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
14051 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
14052 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
14053 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
14054 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
14055 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
14056 (TARGET_HARD_REGNO_NREGS): Redefine.
14057 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
14058 (PA_HARD_REGNO_NREGS): ...this.
14059 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
14060 (PA_HARD_REGNO_NREGS): ...this.
14061 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14062 (pa_hard_regno_nregs): New function.
14063 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
14064 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
14065 (pdp11_hard_regno_nregs): New function.
14066 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
14067 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
14068 (rs6000_hard_regno_nregs_hook): New function.
14069 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
14070 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
14071 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
14072 Take and return an unsigned int. Move earlier in file.
14073 (TARGET_HARD_REGNO_NREGS): Redefine.
14074 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
14075 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
14076 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
14077 (rl78_hard_regno_nregs): Make static. Take and return an
14079 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
14080 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
14081 (rs6000_hard_regno_nregs_hook): New function.
14082 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
14083 * config/rx/rx.c (rx_hard_regno_nregs): New function.
14084 (TARGET_HARD_REGNO_NREGS): Redefine.
14085 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
14086 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
14087 instead of HARD_REGNO_NREGS.
14088 (s390_hard_regno_nregs): New function.
14089 (s390_hard_regno_mode_ok): Add comment from s390.h.
14090 (TARGET_HARD_REGNO_NREGS): Redefine.
14091 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
14092 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
14093 (sh_hard_regno_nregs): New function.
14094 (sh_pass_in_reg_p): Use it.
14095 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
14096 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14097 (sparc_hard_regno_nregs): New function.
14098 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
14099 * config/spu/spu.c (spu_hard_regno_nregs): New function.
14100 (spu_function_arg_advance): Use it, supplying a valid register number.
14101 (TARGET_HARD_REGNO_NREGS): Redefine.
14102 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
14103 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
14104 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
14105 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
14106 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
14107 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
14108 (CLASS_MAX_NREGS): Remove copy of old documentation.
14109 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
14110 (visium_hard_regno_nregs): New function.
14111 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
14112 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
14113 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14114 xtensa_hard_regno_nregs): New function.
14115 * system.h (HARD_REGNO_NREGS): Poison.
14117 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14119 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
14120 hard_regno_nregs instead of HARD_REGNO_NREGS.
14121 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
14122 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
14123 (c6x_expand_epilogue): Likewise.
14124 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
14125 (frv_read_iacc_argument): Likewise.
14126 * config/sh/sh.c: Include regs.h.
14127 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14128 (regs_used): Likewise.
14129 (output_stack_adjust): Likewise.
14130 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
14131 * expmed.c: Include regs.h.
14132 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14133 * ree.c: Include regs.h.
14134 (combine_reaching_defs): Use hard_regno_nregs instead of
14136 (add_removable_extension): Likewise.
14138 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14140 * regs.h (hard_regno_nregs): Turn into a function.
14141 (end_hard_regno): Update accordingly.
14142 * caller-save.c (setup_save_areas): Likewise.
14143 (save_call_clobbered_regs): Likewise.
14144 (replace_reg_with_saved_mem): Likewise.
14145 (insert_restore): Likewise.
14146 (insert_save): Likewise.
14147 * combine.c (can_change_dest_mode): Likewise.
14148 (move_deaths): Likewise.
14149 (distribute_notes): Likewise.
14150 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
14151 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
14152 (rs6000_split_multireg_move): Likewise.
14153 (rs6000_register_move_cost): Likewise.
14154 (rs6000_memory_move_cost): Likewise.
14155 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
14156 (rs6000_split_multireg_move): Likewise.
14157 (rs6000_register_move_cost): Likewise.
14158 (rs6000_memory_move_cost): Likewise.
14159 * cselib.c (cselib_reset_table): Likewise.
14160 (cselib_lookup_1): Likewise.
14161 * emit-rtl.c (set_mode_and_regno): Likewise.
14162 * function.c (aggregate_value_p): Likewise.
14163 * ira-color.c (setup_profitable_hard_regs): Likewise.
14164 (check_hard_reg_p): Likewise.
14165 (calculate_saved_nregs): Likewise.
14166 (assign_hard_reg): Likewise.
14167 (improve_allocation): Likewise.
14168 (calculate_spill_cost): Likewise.
14169 * ira-emit.c (modify_move_list): Likewise.
14170 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
14171 (ira_hard_reg_in_set_p): Likewise.
14172 * ira.c (setup_reg_mode_hard_regset): Likewise.
14173 (clarify_prohibited_class_mode_regs): Likewise.
14174 (check_allocation): Likewise.
14175 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14176 (lra_setup_reg_renumber): Likewise.
14177 (setup_try_hard_regno_pseudos): Likewise.
14178 (spill_for): Likewise.
14179 (assign_hard_regno): Likewise.
14180 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
14181 * lra-constraints.c (in_class_p): Likewise.
14182 (lra_constraint_offset): Likewise.
14183 (simplify_operand_subreg): Likewise.
14184 (lra_constraints): Likewise.
14185 (split_reg): Likewise.
14186 (split_if_necessary): Likewise.
14187 (invariant_p): Likewise.
14188 (inherit_in_ebb): Likewise.
14189 * lra-lives.c (process_bb_lives): Likewise.
14190 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
14191 (get_hard_regs): Likewise.
14192 (do_remat): Likewise.
14193 * lra-spills.c (assign_spill_hard_regs): Likewise.
14194 * mode-switching.c (create_pre_exit): Likewise.
14195 * postreload.c (reload_combine_recognize_pattern): Likewise.
14196 * recog.c (peep2_find_free_register): Likewise.
14197 * regcprop.c (kill_value_regno): Likewise.
14198 (set_value_regno): Likewise.
14199 (copy_value): Likewise.
14200 (maybe_mode_change): Likewise.
14201 (find_oldest_value_reg): Likewise.
14202 (copyprop_hardreg_forward_1): Likewise.
14203 * regrename.c (check_new_reg_p): Likewise.
14204 (regrename_do_replace): Likewise.
14205 * reload.c (push_reload): Likewise.
14206 (combine_reloads): Likewise.
14207 (find_dummy_reload): Likewise.
14208 (operands_match_p): Likewise.
14209 (find_reloads): Likewise.
14210 (find_equiv_reg): Likewise.
14211 (reload_adjust_reg_for_mode): Likewise.
14212 * reload1.c (count_pseudo): Likewise.
14213 (count_spilled_pseudo): Likewise.
14214 (find_reg): Likewise.
14215 (clear_reload_reg_in_use): Likewise.
14216 (free_for_value_p): Likewise.
14217 (allocate_reload_reg): Likewise.
14218 (choose_reload_regs): Likewise.
14219 (reload_adjust_reg_for_temp): Likewise.
14220 (emit_reload_insns): Likewise.
14221 (delete_output_reload): Likewise.
14222 * rtlanal.c (subreg_get_info): Likewise.
14223 * sched-deps.c (sched_analyze_reg): Likewise.
14224 * sel-sched.c (init_regs_for_mode): Likewise.
14225 (mark_unavailable_hard_regs): Likewise.
14226 (choose_best_reg_1): Likewise.
14227 (verify_target_availability): Likewise.
14228 * valtrack.c (dead_debug_insert_temp): Likewise.
14229 * var-tracking.c (track_loc_p): Likewise.
14230 (emit_note_insn_var_location): Likewise.
14231 * varasm.c (make_decl_rtl): Likewise.
14232 * reginfo.c (choose_hard_reg_mode): Likewise.
14233 (init_reg_modes_target): Refer directly to
14234 this_target_regs->x_hard_regno_nregs.
14236 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14238 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
14239 instead of hard_regno_nregs.
14241 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14243 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
14244 end_hard_regno instead of hard_regno_nregs.
14245 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
14246 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
14247 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
14248 * ira-color.c (improve_allocation): Likewise.
14249 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14250 * lra-lives.c (mark_regno_live): Likewise.
14251 (mark_regno_dead): Likewise.
14252 * lra-remat.c (operand_to_remat): Likewise.
14253 * lra.c (collect_non_operand_hard_regs): Likewise.
14254 * postreload.c (reload_combine_note_store): Likewise.
14255 (move2add_valid_value_p): Likewise.
14256 * reload.c (regno_clobbered_p): Likewise.
14258 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14260 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
14262 * config/v850/v850.c (v850_reorg): Likewise.
14263 * reload.c (refers_to_regno_for_reload_p): Likewise.
14264 (find_equiv_reg): Likewise.
14265 * reload1.c (reload_reg_reaches_end_p): Likewise.
14267 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14269 * caller-save.c (add_used_regs): Use REG_NREGS instead of
14271 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
14272 * config/arm/arm.c (output_move_neon): Likewise.
14273 (arm_attr_length_move_neon): Likewise.
14274 (neon_split_vcombine): Likewise.
14275 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
14276 (c6x_mark_reg_written): Likewise.
14277 (c6x_dwarf_register_span): Likewise.
14278 * config/i386/i386.c (ix86_save_reg): Likewise.
14279 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
14280 (rws_access_reg): Likewise.
14281 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
14282 * mode-switching.c (create_pre_exit): Likewise.
14283 * ree.c (combine_reaching_defs): Likewise.
14284 (add_removable_extension): Likewise.
14285 * regcprop.c (find_oldest_value_reg): Likewise.
14286 (copyprop_hardreg_forward_1): Likewise.
14287 * reload.c (reload_inner_reg_of_subreg): Likewise.
14288 (push_reload): Likewise.
14289 (combine_reloads): Likewise.
14290 (find_dummy_reload): Likewise.
14291 (reload_adjust_reg_for_mode): Likewise.
14292 * reload1.c (find_reload_regs): Likewise.
14293 (forget_old_reloads_1): Likewise.
14294 (reload_reg_free_for_value_p): Likewise.
14295 (reload_adjust_reg_for_temp): Likewise.
14296 (emit_reload_insns): Likewise.
14297 (delete_output_reload): Likewise.
14298 * sel-sched.c (choose_best_reg_1): Likewise.
14299 (choose_best_pseudo_reg): Likewise.
14301 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14302 Alan Hayward <alan.hayward@arm.com>
14303 David Sherwood <david.sherwood@arm.com>
14305 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
14306 * target.def (slow_unaligned_access): New hook.
14307 * targhooks.h (default_slow_unaligned_access): Declare.
14308 * targhooks.c (default_slow_unaligned_access): New function.
14309 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
14310 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
14311 * doc/tm.texi: Regenerate.
14312 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
14313 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
14314 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
14316 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
14317 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
14319 (rs6000_slow_unaligned_access): New function.
14320 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14321 (expand_block_compare): Likewise.
14322 (expand_strn_compare): Likewise.
14323 (rs6000_rtx_costs): Likewise.
14324 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
14325 (riscv_slow_unaligned_access): Likewise.
14326 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
14327 (riscv_slow_unaligned_access_p): ...this and make static.
14328 (riscv_option_override): Update accordingly.
14329 (riscv_slow_unaligned_access): New function.
14330 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14331 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
14332 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14333 (rs6000_slow_unaligned_access): New function.
14334 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14335 (rs6000_rtx_costs): Likewise.
14336 * config/rs6000/rs6000-string.c (expand_block_compare)
14337 (expand_strn_compare): Use targetm.slow_unaligned_access instead
14338 of SLOW_UNALIGNED_ACCESS.
14339 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
14340 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
14341 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
14342 of SLOW_UNALIGNED_ACCESS.
14343 * expmed.c (simple_mem_bitfield_p): Likewise.
14344 * expr.c (alignment_for_piecewise_move): Likewise.
14345 (emit_group_load_1): Likewise.
14346 (emit_group_store): Likewise.
14347 (copy_blkmode_from_reg): Likewise.
14348 (emit_push_insn): Likewise.
14349 (expand_assignment): Likewise.
14350 (store_field): Likewise.
14351 (expand_expr_real_1): Likewise.
14352 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14353 * lra-constraints.c (simplify_operand_subreg): Likewise.
14354 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14355 * gimple-ssa-store-merging.c: Likewise in block comment at start
14357 * tree-ssa-strlen.c: Include target.h.
14358 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
14359 of SLOW_UNALIGNED_ACCESS.
14360 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
14362 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14364 PR rtl-optimization/82185
14365 * expmed.c (emit_store_flag_int): Only test tem if it has been
14368 2017-09-12 Richard Biener <rguenther@suse.de>
14370 PR middle-end/82149
14371 * match.pd ((FTYPE) N CMP CST): Fix typo.
14373 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
14375 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
14377 (mips_near_type_p): Add 'short_call' attribute as a synonym
14379 * doc/extend.texi (short_call): Document new function attribute.
14381 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14384 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
14385 assertion check that in the condition.
14386 (get_atomic_generic_size): Likewise. Before testing if parameter
14387 has pointer type, if it has array type, call for C++
14388 default_conversion to perform array-to-pointer conversion.
14390 2017-09-12 Richard Biener <rguenther@suse.de>
14392 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
14393 for operations we cannot scalarize.
14395 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
14397 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
14398 vectors heap vectors. Clean up comments.
14399 Make visited_bbs a reference.
14400 (profitable_jump_thread_path): Make GC
14401 vectors heap vectors. Clean up comments.
14403 (convert_and_register_jump_thread_path): Make GC vectors heap
14405 (check_subpath_and_update_thread_path): Same. Clean up comments.
14406 Make visited_bbs a reference.
14407 (handle_phi): Abstract common code to to
14408 register_jump_thread_path_if_profitable.
14409 Rename VAR_BB to DEF_BB.
14411 Make GC vectors heap vectors.
14412 Make visited_bbs a reference.
14413 (handle_assignment): Same.
14414 (register_jump_thread_path_if_profitable): New.
14415 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
14417 Make GC vectors heap vectors. Clean up comments.
14418 Make visited_bbs a reference.
14419 (find_jump_threads_backwards): Make visited_bbs live in the stack.
14420 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
14423 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
14426 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
14427 words of E_DImode object are reachable by xtensa_uimm8x4 access.
14429 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
14431 Revert r251800 and r251799.
14433 2017-09-11 Martin Jambor <mjambor@suse.cz>
14436 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
14437 arguments in advance.
14438 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
14439 use it to find predecessor edges.
14440 (naive_outof_ssa): Collect vector of predecessors.
14442 2017-09-08 Jason Merrill <jason@redhat.com>
14444 PR c++/70029 - ICE with ref-qualifier and -flto
14445 * langhooks.h (struct lang_hooks_for_types): Add
14446 copy_lang_qualifiers.
14447 * attribs.c (build_type_attribute_qual_variant): Use it.
14448 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
14450 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
14451 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
14453 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
14456 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
14457 (*mulsi3_sp64): New instruction.
14458 (mulsi3): New expander.
14460 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
14462 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
14464 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14466 * sancov.c: Include memmodel.h.
14468 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
14471 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
14474 2017-09-07 Carl Love <cel@us.ibm.com>
14476 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
14477 the sldi instruction.
14479 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
14481 * sancov.c: Include tm_p.h.
14483 2017-09-07 Jakub Jelinek <jakub@redhat.com>
14486 * output.h (switch_to_other_text_partition): New declaration.
14487 * varasm.c (switch_to_other_text_partition): New function.
14488 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
14489 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
14490 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
14491 to the other text partition before emitting LCL label and switch back
14492 after emitting the word after it.
14494 2017-09-07 Richard Biener <rguenther@suse.de>
14496 * passes.def (pass_split_crit_edges): Remove instance before PRE.
14497 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
14498 critical edges here, after loop init.
14499 (pass_data_pre): Remove PROP_no_crit_edges flags.
14500 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
14501 for valueization of call args to avoid leaking VN_TOP.
14502 (visit_use): Assert we do not visit default defs.
14503 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
14504 Use error_mark_node to more easily detect leaking VN_TOP.
14505 All default-defs are varying, not VN_TOP. Mark them visited.
14506 (run_scc_vn): Make code match comment.
14508 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14510 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
14511 OPTION_MASK_FLOAT128_KEYWORD.
14512 (POWERPC_MASKS): Likewise.
14513 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
14514 support for the -mfloat128-type option, and make -mfloat128
14515 default on PowerPC Linux systems. Define or undefine
14516 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
14517 Define __float128 to be __ieee128 if IEEE 128-bit support is
14518 enabled, or undefine it.
14519 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
14520 Delete defining __FLOAT128_TYPE__.
14521 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
14522 -mfloat128-type option and make -mfloat128 default on PowerPC
14524 (TARGET_FLOAT128_TYPE): Likewise.
14525 (-mfloat128-type): Likewise.
14526 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14527 Delete the -mfloat128-type option and make -mfloat128 default on
14528 PowerPC Linux systems. Always use __ieee128 to be the keyword for
14529 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
14530 128-bit floating point is enabled. Change tests from using
14531 -mfloat128-type to -mfloat128.
14532 (rs6000_mangle_type): Use the correct mangling for the __float128
14533 type even if normal long double is restricted to 64-bits.
14534 (floatn_mode): Enable the _Float128 type by default on VSX Linux
14536 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
14537 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
14538 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
14540 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
14541 documentation for -mfloat128.
14543 2017-09-06 Olivier Hainque <hainque@adacore.com>
14545 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
14547 2017-09-06 Wish Wu <wishwu007@gmail.com>
14548 Jakub Jelinek <jakub@redhat.com>
14550 * asan.c (initialize_sanitizer_builtins): Add
14551 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
14552 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
14553 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
14554 BT_FN_VOID_UINT64_PTR variables.
14555 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
14556 (BT_FN_VOID_UINT16_UINT16): Likewise.
14557 (BT_FN_VOID_UINT32_UINT32): Likewise.
14558 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
14559 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
14560 (BT_FN_VOID_UINT64_PTR): Likewise.
14561 * common.opt (flag_sanitize_coverage): New variable.
14562 (fsanitize-coverage=trace-pc): Remove.
14563 (fsanitize-coverage=): Add.
14564 * flag-types.h (enum sanitize_coverage_code): New enum.
14565 * fold-const.c (fold_range_test): Disable non-short-circuit
14566 optimization if flag_sanitize_coverage.
14567 (fold_truth_andor): Likewise.
14568 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14569 * opts.c (COVERAGE_SANITIZER_OPT): Define.
14570 (coverage_sanitizer_opts): New array.
14571 (get_closest_sanitizer_option): Add OPTS argument, handle also
14572 OPT_fsanitize_coverage_.
14573 (parse_sanitizer_options): Adjusted to also handle
14574 OPT_fsanitize_coverage_.
14575 (common_handle_option): Add OPT_fsanitize_coverage_.
14576 * sancov.c (instrument_comparison, instrument_switch): New function.
14577 (sancov_pass): Add trace-cmp support.
14578 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
14579 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
14580 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
14581 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
14582 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
14583 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
14584 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
14585 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
14586 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
14587 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
14589 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14591 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14592 error. Only quit immediately if parsing is complete.
14593 (BEGIN): Initialize fatal_err and parse_done.
14594 (begin fpu, end fpu): Check number of arguments.
14595 (begin arch, end arch): Likewise.
14596 (begin cpu, end cpu): Likewise.
14597 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14598 (optalias): Likewise.
14600 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14602 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14603 * config/arm/arm-isa.h: Delete. Move definitions to ...
14604 * arm-cpus.in: ... here. Use new feature and fgroup values.
14605 * config/arm/arm.c (arm_option_override): Use lower case for feature
14607 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14608 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14609 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14611 (print_isa_bits_for): New function.
14612 (gen_isa): New function.
14613 (gen_comm_data): Use print_isa_bits_for.
14614 (define feature): New keyword.
14615 (define fgroup): New keyword.
14616 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
14617 (arm-isa.h): Add rule to generate file.
14618 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14619 case for feature bit names.
14621 2017-09-06 Richard Biener <rguenther@suse.de>
14623 * tree-ssa-pre.c (NECESSARY): Remove.
14624 (create_expression_by_pieces): Do not touch pass-local flags.
14625 (insert_into_preds_of_block): Likewise.
14626 (do_pre_regular_insertion): Likewise.
14627 (eliminate_insert): Likewise.
14628 (eliminate_dom_walker::before_dom_children): Likewise.
14629 (fini_eliminate): Do not look at inserted_exprs.
14630 (mark_operand_necessary): Remove.
14631 (remove_dead_inserted_code): Replace with simple work-list
14632 algorithm based on inserted_exprs and SSA uses.
14633 (pass_pre::execute): Re-order fini_eliminate and
14634 remove_dead_inserted_code.
14636 2017-09-06 Olivier Hainque <hainque@adacore.com>
14638 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
14639 for VxWorks 7. Adjust surrounding comments.
14641 2017-09-06 Richard Biener <rguenther@suse.de>
14643 * gimple-ssa-strength-reduction.c
14644 (find_candidates_dom_walker::before_dom_children): Also allow
14647 2017-09-06 Richard Biener <rguenther@suse.de>
14649 PR tree-optimization/82108
14650 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
14651 for gap in the non-permutation SLP case.
14653 2017-09-06 Martin Jambor <mjambor@suse.cz>
14655 PR tree-optimization/82078
14656 * tree-sra.c (sort_and_splice_var_accesses): Move call to
14657 add_access_to_work_queue...
14658 (build_accesses_from_assign): ...here.
14659 (propagate_all_subaccesses): Make sure racc is the group
14660 representative, if there is one.
14662 2017-09-06 Jakub Jelinek <jakub@redhat.com>
14664 PR middle-end/82095
14665 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
14668 2017-09-06 Richard Biener <rguenther@suse.de>
14670 * gimple-ssa-strength-reduction.c
14671 (find_candidates_dom_walker::before_doom_children): Use a
14672 type and not a mode check.
14674 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
14677 * config/arm/predicates.md (arm_general_adddi_operand): Create new
14679 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
14681 2017-09-05 Jeff Law <law@redhat.com>
14683 PR tree-optimization/64910
14684 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
14685 cases where we have 3 or more operands.
14687 2017-09-05 Jakub Jelinek <jakub@redhat.com>
14689 PR middle-end/81768
14690 * omp-low.c (lower_omp_for): Recompute tree invariant if
14691 gimple_omp_for_initial/final is ADDR_EXPR.
14693 PR middle-end/81768
14694 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
14695 into gimple val before gimplification fo the COND_EXPR.
14697 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
14699 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
14700 REGION_COPY argument.
14701 (thread_through_all_blocks): Remove unused argument to
14702 duplicate_thread_path.
14704 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14705 Alan Hayward <alan.hayward@arm.com>
14706 David Sherwood <david.sherwood@arm.com>
14708 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
14709 Take a scalar_mode rather than a machine_mode.
14710 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14711 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
14712 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14713 (aarch64_gen_adjusted_ldpstp): Likewise.
14714 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
14716 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14717 Alan Hayward <alan.hayward@arm.com>
14718 David Sherwood <david.sherwood@arm.com>
14720 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
14721 Take a scalar_int_mode instead of a machine_mode.
14722 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14723 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14724 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14725 (aarch64_simd_attr_length_rglist): Delete.
14726 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
14727 a scalar_int_mode instead of a machine_mode.
14728 (aarch64_add_offset): Likewise.
14729 (aarch64_internal_mov_immediate): Likewise
14730 (aarch64_add_constant_internal): Likewise.
14731 (aarch64_add_constant): Likewise.
14732 (aarch64_movw_imm): Likewise.
14733 (aarch64_rtx_arith_op_extract_p): Likewise.
14734 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14735 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14736 Remove assert that the mode isn't a vector.
14737 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14738 (aarch64_expand_mov_immediate): Update calls after above changes.
14739 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
14740 (aarch64_and_bitmask_imm): Check for scalar integer modes.
14741 (aarch64_move_imm): Likewise.
14742 (aarch64_can_const_movi_rtx_p): Likewise.
14743 (aarch64_strip_extend): Likewise.
14744 (aarch64_extr_rtx_p): Likewise.
14745 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
14746 a CONST_INT when the mode parameter is VOIDmode.
14747 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
14749 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14751 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
14752 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14753 (bitwise_type_for_mode): Update accordingly.
14755 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14757 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
14758 * stor-layout.c (mode_for_size_tree): Likewise.
14759 (mode_for_array): Update accordingly.
14760 (layout_decl): Likewise.
14761 (compute_record_mode): Likewise. Only set the mode once.
14763 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14765 * target.def (get_mask_mode): Change return type to opt_mode.
14767 * doc/tm.texi: Regenerate.
14768 * targhooks.h (default_get_mask_mode): Return an opt_mode.
14769 * targhooks.c (default_get_mask_mode): Likewise.
14770 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14771 * optabs-query.c (can_vec_mask_load_store_p): Update use of
14772 targetm.get_mask_mode.
14773 * tree.c (build_truth_vector_type): Likewise.
14775 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14777 * machmode.h (mode_for_vector): Return an opt_mode.
14778 * stor-layout.c (mode_for_vector): Likewise.
14779 (mode_for_int_vector): Update accordingly.
14780 (layout_type): Likewise.
14781 * config/i386/i386.c (emit_memmov): Likewise.
14782 (ix86_expand_set_or_movmem): Likewise.
14783 (ix86_expand_vector_init): Likewise.
14784 (ix86_get_mask_mode): Likewise.
14785 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
14787 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
14788 * expmed.c (extract_bit_field_1): Likewise.
14789 * expr.c (expand_expr_real_2): Likewise.
14790 * optabs-query.c (can_vec_perm_p): Likewise.
14791 (can_vec_mask_load_store_p): Likewise.
14792 * optabs.c (expand_vec_perm): Likewise.
14793 * targhooks.c (default_get_mask_mode): Likewise.
14794 * tree-vect-stmts.c (vectorizable_store): Likewise.
14795 (vectorizable_load): Likewise.
14796 (get_vectype_for_scalar_type_and_size): Likewise.
14798 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14800 * machmode.h (mode_for_int_vector): New function.
14801 * stor-layout.c (mode_for_int_vector): Likewise.
14802 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
14803 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
14804 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14805 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14806 (s390_expand_vcond): Likewise.
14808 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14810 * machmode.h (opt_machine_mode): New type.
14811 (opt_mode<T>): Allow construction from anything that can be
14813 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
14814 (mode_for_size): Return an opt_machine_mode.
14815 * stor-layout.c (mode_for_size): Likewise.
14816 (mode_for_size_tree): Update call accordingly.
14817 (bitwise_mode_for_mode): Likewise.
14818 (make_fract_type): Likewise.
14819 (make_accum_type): Likewise.
14820 * caller-save.c (replace_reg_with_saved_mem): Update call
14822 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14823 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14824 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14825 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14826 * expmed.c (extract_bit_field_1): Likewise.
14827 * reload.c (get_secondary_mem): Likewise.
14828 * varasm.c (assemble_integer): Likewise.
14829 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
14832 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14834 * machmode.h (decimal_float_mode_for_size): New function.
14835 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
14836 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
14837 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
14838 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
14839 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
14840 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
14842 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14844 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14845 (get_builtin_sync_mode): Likewise.
14846 (expand_ifn_atomic_compare_exchange): Likewise.
14847 (expand_builtin_atomic_clear): Likewise.
14848 (expand_builtin_atomic_test_and_set): Likewise.
14849 (fold_builtin_atomic_always_lock_free): Likewise.
14850 * calls.c (compute_argument_addresses): Likewise.
14851 (emit_library_call_value_1): Likewise.
14852 (store_one_arg): Likewise.
14853 * combine.c (combine_instructions): Likewise.
14854 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
14855 * config/arm/arm.c (arm_function_value): Likewise.
14856 (aapcs_allocate_return_reg): Likewise.
14857 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
14858 * config/i386/i386.c (construct_container): Likewise.
14859 (ix86_gimplify_va_arg): Likewise.
14860 (ix86_expand_sse_cmp): Likewise.
14861 (emit_memmov): Likewise.
14862 (emit_memset): Likewise.
14863 (expand_small_movmem_or_setmem): Likewise.
14864 (ix86_expand_pextr): Likewise.
14865 (ix86_expand_pinsr): Likewise.
14866 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
14867 * config/microblaze/microblaze.c (microblaze_block_move_straight):
14869 * config/mips/mips.c (mips_function_value_1) Likewise.
14870 (mips_block_move_straight): Likewise.
14871 (mips_expand_ins_as_unaligned_store): Likewise.
14872 * config/powerpcspe/powerpcspe.c
14873 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14874 (rs6000_darwin64_record_arg_flush): Likewise.
14875 * config/rs6000/rs6000.c
14876 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14877 (rs6000_darwin64_record_arg_flush): Likewise.
14878 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
14879 (sparc_function_value_1): Likewise.
14880 * config/spu/spu.c (adjust_operand): Likewise.
14881 (spu_emit_branch_or_set): Likewise.
14882 (arith_immediate_p): Likewise.
14883 * emit-rtl.c (gen_lowpart_common): Likewise.
14884 * expr.c (expand_expr_real_1): Likewise.
14885 * function.c (assign_parm_setup_block): Likewise.
14886 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
14887 * reload1.c (alter_reg): Likewise.
14888 * stor-layout.c (mode_for_vector): Likewise.
14889 (layout_type): Likewise.
14891 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14893 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
14894 (spu_convert_move): Likewise.
14895 * lower-subreg.c (resolve_simple_move): Likewise.
14897 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14900 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
14901 define_insn to a define_expand.
14902 (altivec_vsum2sws_direct): New define_insn.
14903 (altivec_vsumsws): Convert from a define_insn to a define_expand.
14905 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14907 * config/arm/arm.c (arm_option_params_internal): Improve setting of
14910 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
14915 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
14916 realign stack if stack alignment needed is less than incoming
14919 2017-09-05 Marek Polacek <polacek@redhat.com>
14922 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
14925 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14927 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
14929 2017-09-05 Richard Biener <rguenther@suse.de>
14931 PR tree-optimization/82084
14932 * fold-const.c (can_native_encode_string_p): Handle wide characters.
14934 2017-09-05 Richard Biener <rguenther@suse.de>
14936 PR tree-optimization/82102
14937 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
14939 2017-09-05 Martin Liska <mliska@suse.cz>
14941 PR tree-optimization/82032
14942 * tree-cfg.c (generate_range_test): New function.
14943 * tree-cfg.h (generate_range_test): Declared here.
14944 * tree-cfgcleanup.c (convert_single_case_switch): New function.
14945 (cleanup_control_expr_graph): Use it.
14946 * tree-switch-conversion.c (try_switch_expansion): Remove
14948 (emit_case_nodes): Use generate_range_test.
14950 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14953 * config/i386/i386.md (*<btsc><mode>_mask): Add
14954 TARGET_USE_BT to insn constraint.
14955 (*btr<mode>_mask): Ditto.
14957 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
14959 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
14960 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
14962 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
14965 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
14966 TARGET_NEON and TARGET_IWMMXT.
14967 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
14968 TARGET_NEON and TARGET_IWMMXT.
14969 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
14971 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14973 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
14974 (ix86_rewrite_tls_address): Ditto.
14975 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
14976 (ix86_rewrite_tls_address_1): Ditto.
14977 (ix86_rewrite_tls_address): Ditto.
14978 * config/i386/predicates.md (tls_address_pattern): New predicate.
14979 * config/i386/i386.md (TLS address splitter): New splitter.
14981 2017-09-04 Richard Biener <rguenther@suse.de>
14983 PR tree-optimization/82084
14984 * fold-const.h (can_native_encode_string_p): Declare.
14985 * fold-const.c (can_native_encode_string_p): Factor out from ...
14986 (native_encode_string): ... here.
14987 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
14988 vectorizing stores from constants we later cannot handle.
14990 2017-09-04 Marek Polacek <polacek@redhat.com>
14993 * doc/invoke.texi: Update -Wtautological-compare documentation.
14995 2017-09-04 Jeff Law <law@redhat.com>
14997 PR tree-optimization/64910
14998 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
14999 swap the first and last operand if the last is a constant.
15001 2017-09-04 Marek Polacek <polacek@redhat.com>
15004 * convert.c (do_narrow): When sanitizing signed integer overflows,
15005 bail out for signed types.
15006 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15008 2017-09-04 Richard Biener <rguenther@suse.de>
15010 PR tree-optimization/82060
15011 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15012 Move devirtualization after stmt folding and before EH/AB/noreturn
15013 cleanup to get the stmt refs canonicalized. Use a bool instead
15014 of gimple_modified_p since that doesn't work for NOPs. Schedule
15015 NOPs generated by folding for removal.
15017 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15018 Alan Hayward <alan.hayward@arm.com>
15019 David Sherwood <david.sherwood@arm.com>
15021 * coretypes.h (pad_direction): New enum.
15022 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15023 (FUNCTION_ARG_PADDING): Likewise.
15024 * target.def (function_arg_padding): New hook.
15025 * targhooks.h (default_function_arg_padding): Declare.
15026 * targhooks.c (default_function_arg_padding): New function.
15027 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15028 (TARGET_FUNCTION_ARG_PADDING): ...this.
15029 * doc/tm.texi: Regenerate.
15030 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15031 instead of direction.
15032 (compute_argument_addresses): Likewise.
15033 (load_register_parameters): Likewise.
15034 (emit_library_call_value_1): Likewise.
15035 (store_one_arg): Use targetm.calls.function_arg_padding instead
15036 of FUNCTION_ARG_PADDING.
15037 (must_pass_in_stack_var_size_or_pad): Likewise.
15038 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15039 (emit_group_store): Likewise.
15040 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15041 instead of FUNCTION_ARG_PADDING.
15042 (emit_push_insn): Likewise, and propagate enum change throughout
15044 * function.h (direction): Delete.
15045 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
15047 * function.c (assign_parm_find_stack_rtl): Likewise.
15048 (assign_parm_setup_block_p): Likewise.
15049 (assign_parm_setup_block): Likewise.
15050 (gimplify_parameters): Likewise.
15051 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
15052 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
15054 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
15055 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15056 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
15057 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
15058 (aarch64_function_arg_padding): ...this new function.
15059 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
15060 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15061 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
15062 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15063 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
15064 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15065 (arm_pad_arg_upward): Replace with...
15066 (arm_function_arg_padding): ...this new function.
15067 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
15069 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
15070 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
15071 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15072 (ia64_hpux_function_arg_padding): Replace with...
15073 (ia64_function_arg_padding): ...this new function. Use pad_direction
15074 instead of direction. Check for TARGET_HPUX.
15075 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
15076 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15077 (iq2000_function_arg_padding): New function.
15078 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
15079 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
15080 (mips_function_arg_padding): ...this new function.
15081 (mips_pad_reg_upward): Update accordingly.
15082 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15083 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
15084 targetm.calls.function_arg_padding.
15085 (FUNCTION_ARG_PADDING): Delete.
15086 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15087 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
15088 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15089 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
15090 (nios2_block_reg_padding): Return pad_direction instead of direction.
15091 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
15092 instead of direction.
15093 (nios2_function_arg_padding): Likewise. Make static.
15094 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15095 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
15096 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
15097 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
15098 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15099 (pa_function_arg_padding): Make static. Return pad_direction instead
15101 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
15102 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15103 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
15104 instead of direction. Use targetm.calls.function_arg_padding.
15105 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
15106 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15107 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
15108 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
15109 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
15111 (function_arg_padding): Rename to...
15112 (rs6000_function_arg_padding): ...this. Make static. Return
15113 pad_direction instead of direction.
15114 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15115 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
15116 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15117 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
15118 instead of direction. Use targetm.calls.function_arg_padding.
15119 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
15120 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15121 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
15122 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
15123 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15124 (function_arg_padding): Rename to...
15125 (rs6000_function_arg_padding): ...this. Make static. Return
15126 pad_direction instead of direction.
15127 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15128 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
15129 * config/s390/s390.c (s390_function_arg_padding): New function.
15130 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15131 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
15132 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
15133 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15134 (function_arg_padding): Rename to...
15135 (sparc_function_arg_padding): ...this. Make static. Return
15136 pad_direction instead of direction.
15137 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
15138 * config/spu/spu.c (spu_function_arg_padding): New function.
15139 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15140 * system.h (FUNCTION_ARG_PADDING): Poison.
15142 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15143 Alan Hayward <alan.hayward@arm.com>
15144 David Sherwood <david.sherwood@arm.com>
15146 * target.def (modes_tieable_p): New hook.
15147 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
15148 (TARGET_MODES_TIEABLE_P): ...this.
15149 * doc/tm.texi.in: Regenerate.
15150 * hooks.h (hook_bool_mode_mode_true): Declare.
15151 * hooks.c (hook_bool_mode_mode_true): New function.
15152 * combine.c (subst): Use targetm.modes_tieable_p instead of
15154 * dse.c (find_shift_sequence): Likewise.
15155 * expmed.c (extract_low_bits): Likewise.
15156 * lower-subreg.c: Include target.h.
15157 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
15159 * rtlanal.c (rtx_cost): Likewise.
15160 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
15161 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
15162 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
15163 (TARGET_MODES_TIEABLE_P): Redefine.
15164 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
15165 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
15166 (TARGET_MODES_TIEABLE_P): Redefine.
15167 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
15168 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
15169 (arc_modes_tieable_p): New function.
15170 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
15171 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
15172 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
15173 (arm_modes_tieable_p): Make static.
15174 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
15175 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
15176 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
15177 (TARGET_MODES_TIEABLE_P): Redefine.
15178 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
15179 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
15180 (TARGET_MODES_TIEABLE_P): Redefine.
15181 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
15182 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
15183 (cr16_modes_tieable_p): New function.
15184 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
15185 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
15186 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
15187 (TRULY_NOOP_TRUNCATION): Update comment.
15188 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
15189 (TRULY_NOOP_TRUNCATION): Update comment.
15190 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
15191 (frv_modes_tieable_p): New function.
15192 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
15193 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
15194 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
15195 (TARGET_MODES_TIEABLE_P): Redefine.
15196 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
15197 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
15198 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
15199 (TARGET_MODES_TIEABLE_P): Redefine.
15200 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
15201 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
15202 (ia64_modes_tieable_p): New function.
15203 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
15204 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
15205 (iq2000_modes_tieable_p): New function.
15206 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
15207 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
15208 (lm32_modes_tieable_p): New function.
15209 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
15210 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
15211 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
15212 (TARGET_MODES_TIEABLE_P): Redefine.
15213 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
15214 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
15215 (m32r_modes_tieable_p): New function.
15216 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
15217 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
15218 (m68k_modes_tieable_p): New function.
15219 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
15220 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
15221 (mcore_modes_tieable_p): New function.
15222 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
15223 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
15225 (TARGET_MODES_TIEABLE_P): Redefine.
15226 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
15227 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
15228 * config/mips/mips.c (mips_modes_tieable_p): Make static.
15229 (TARGET_MODES_TIEABLE_P): Redefine.
15230 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
15231 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
15232 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
15233 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
15234 (mn10300_modes_tieable_p): ...this and make static.
15235 (TARGET_MODES_TIEABLE_P): Redefine.
15236 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
15237 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
15238 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
15239 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
15240 (msp430_modes_tieable_p): Make static.
15241 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
15242 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
15243 (TARGET_MODES_TIEABLE_P): Redefine.
15244 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
15245 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
15246 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
15247 (TARGET_MODES_TIEABLE_P): Redefine.
15248 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
15249 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
15250 * config/pa/pa.c (pa_modes_tieable_p): Make static.
15251 (TARGET_MODES_TIEABLE_P): Redefine.
15252 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
15253 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
15254 (pdp11_modes_tieable_p): New function.
15255 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
15256 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
15257 (rs6000_modes_tieable_p): New function.
15258 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15259 * config/powerpcspe/powerpcspe.md: Update comment.
15260 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
15261 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
15262 (TARGET_MODES_TIEABLE_P): Redefine.
15263 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
15264 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
15265 (rl78_modes_tieable_p): New function.
15266 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
15267 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
15268 (rs6000_modes_tieable_p): New function.
15269 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15270 * config/rs6000/rs6000.md: Update comment.
15271 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
15272 * config/rx/rx.c (rx_modes_tieable_p): New function.
15273 (TARGET_MODES_TIEABLE_P): Redefine.
15274 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
15275 * config/s390/s390.c (s390_modes_tieable_p): New function.
15276 (TARGET_MODES_TIEABLE_P): Redefine.
15277 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
15278 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
15279 (sh_modes_tieable_p): New function.
15280 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
15281 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
15282 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
15283 (sparc_modes_tieable_p): Make static.
15284 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
15285 * config/spu/spu.c (spu_modes_tieable_p): New function.
15286 (TARGET_MODES_TIEABLE_P): Redefine.
15287 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
15288 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
15289 (TARGET_MODES_TIEABLE_P): Redefine.
15290 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
15291 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
15292 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
15293 * config/v850/v850.c (v850_modes_tieable_p): New function.
15294 (TARGET_MODES_TIEABLE_P): Redefine.
15295 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
15296 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
15297 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
15298 (visium_modes_tieable_p): New function.
15299 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
15300 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
15301 (xtensa_modes_tieable_p): New function.
15302 * system.h (MODES_TIEABLE_P): Poison.
15304 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15305 Alan Hayward <alan.hayward@arm.com>
15306 David Sherwood <david.sherwood@arm.com>
15308 * target.def (hard_regno_mode_ok): New hook.
15309 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
15310 (TARGET_HARD_REGNO_MODE_OK): ...this.
15311 * doc/tm.texi.in: Regenerate.
15312 * hooks.h (hook_bool_uint_mode_true): Declare.
15313 * hooks.c (hook_bool_uint_mode_true): New function.
15314 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
15315 HARD_REGNO_MODE_OK.
15316 * genpreds.c (write_insn_preds_c): Add an include of target.h.
15317 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
15318 instead of HARD_REGNO_MODE_OK.
15319 * caller-save.c: Include target.h.
15320 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
15321 HARD_REGNO_MODE_OK.
15322 * combine.c (can_combine_p): Likewise.
15323 (combinable_i3pat): Likewise.
15324 (can_change_dest_mode): Likewise.
15325 * expr.c (init_expr_target): Likewise.
15326 (convert_move): Likewise.
15327 (convert_modes): Likewise.
15328 * ira.c (setup_prohibited_class_mode_regs): Likewise.
15329 (setup_prohibited_mode_move_regs): Likewise.
15330 * ira.h (target_ira): Likewise.
15331 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15332 * lra-constraints.c (process_alt_operands): Likewise.
15333 (split_reg): Likewise.
15334 * recog.c (peep2_find_free_register): Likewise.
15335 * ree.c (combine_reaching_defs): Likewise.
15336 * regcprop.c (maybe_mode_change): Likewise.
15337 * reginfo.c (init_reg_sets_1): Likewise.
15338 (choose_hard_reg_mode): Likewise.
15339 (simplifiable_subregs): Likewise.
15340 * regrename.c (check_new_reg_p): Likewise.
15341 * reload.c (find_valid_class): Likewise.
15342 (find_valid_class_1): Likewise.
15343 (reload_inner_reg_of_subreg): Likewise.
15344 (push_reload): Likewise.
15345 (combine_reloads): Likewise.
15346 (find_dummy_reload): Likewise.
15347 (find_reloads): Likewise.
15348 * reload1.c (find_reg): Likewise.
15349 (set_reload_reg): Likewise.
15350 (allocate_reload_reg): Likewise.
15351 (choose_reload_regs): Likewise.
15352 (reload_adjust_reg_for_temp): Likewise.
15353 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
15354 (simplify_subreg_regno): Likewise.
15355 * sel-sched.c (init_regs_for_mode): Likewise.
15356 * varasm.c (make_decl_rtl): Likewise.
15357 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
15358 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
15359 HARD_REGNO_MODE_OK.
15360 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
15361 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
15362 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15363 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
15364 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
15365 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15366 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
15367 (arc_mode_class): Delete.
15368 (HARD_REGNO_MODE_OK): Delete.
15369 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15370 (arc_hard_regno_mode_ok): Rename old array to...
15371 (arc_hard_regno_mode_ok_modes): ...this.
15372 (arc_conditional_register_usage): Update accordingly.
15373 (arc_mode_class): Make static.
15374 (arc_hard_regno_mode_ok): New function.
15375 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
15376 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
15377 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15378 (arm_hard_regno_mode_ok): Make static.
15379 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
15380 HARD_REGNO_MODE_OK.
15381 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
15382 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
15383 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
15385 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15386 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
15387 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
15388 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
15389 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
15390 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15391 * config/bfin/predicates.md (valid_reg_operand): Use
15392 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
15393 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
15394 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
15395 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15396 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
15397 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
15398 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15399 (cr16_hard_regno_mode_ok): Make static and return a bool.
15400 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
15401 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15402 (cris_hard_regno_mode_ok): New function.
15403 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
15404 (epiphany_mode_class): Delete.
15405 (HARD_REGNO_MODE_OK): Delete.
15406 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
15407 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15408 (hard_regno_mode_ok): Rename to...
15409 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
15410 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
15411 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
15412 HARD_REGNO_MODE_OK.
15413 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
15414 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
15415 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15416 (frv_hard_regno_mode_ok): Make static and return a bool.
15417 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
15418 HARD_REGNO_MODE_OK.
15419 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
15420 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
15421 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
15422 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
15424 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15425 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
15426 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
15427 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
15429 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15430 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
15431 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15432 (ia64_hard_regno_mode_ok): New function.
15433 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
15434 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15435 (iq2000_hard_regno_mode_ok): New function.
15436 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
15437 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15438 (lm32_hard_regno_mode_ok): New function.
15439 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
15440 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
15441 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
15442 instead of HARD_REGNO_MODE_OK.
15443 (m32c_hard_regno_ok): Rename to...
15444 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
15445 (m32c_cannot_change_mode_class): Update accordingly.
15446 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15447 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
15448 (m32r_mode_class): Delete.
15449 (HARD_REGNO_MODE_OK): Delete.
15450 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15451 (m32r_hard_regno_mode_ok): Rename to...
15452 (m32r_hard_regno_modes): ...this.
15453 (m32r_mode_class): Make static.
15454 (m32r_hard_regno_mode_ok): New function.
15455 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
15456 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
15457 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15458 (m68k_hard_regno_mode_ok): Make static.
15459 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
15460 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15461 (mcore_hard_regno_mode_ok): New function.
15462 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
15463 (HARD_REGNO_MODE_OK): Delete.
15464 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
15466 (microblaze_hard_regno_mode_ok_p): ...this and make static.
15467 (microblaze_hard_regno_mode_ok): New function.
15468 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15469 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
15470 (mips_hard_regno_mode_ok): Delete.
15471 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
15472 (mips_hard_regno_mode_ok_p): ...this and make static.
15473 (mips_hard_regno_mode_ok_p): Rename to...
15474 (mips_hard_regno_mode_ok_uncached): ...this.
15475 (mips_hard_regno_mode_ok): New function.
15476 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
15477 of HARD_REGNO_MODE_OK.
15478 (mips_option_override): Update after above name changes.
15479 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15480 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
15481 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
15482 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
15483 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
15484 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15485 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
15486 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
15487 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
15488 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15489 (msp430_hard_regno_mode_ok): Make static and return a bool.
15490 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
15491 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
15492 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
15494 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15495 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
15496 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
15497 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
15498 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
15499 (PA_HARD_REGNO_MODE_OK): ...this
15500 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
15501 (PA_HARD_REGNO_MODE_OK): ...this.
15502 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15503 (pa_hard_regno_mode_ok): New function.
15504 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
15505 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15506 (pdp11_hard_regno_mode_ok): New function.
15507 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
15508 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
15510 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
15512 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15513 (rs6000_hard_regno_mode_ok): Rename to...
15514 (rs6000_hard_regno_mode_ok_uncached): ...this.
15515 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15516 (rs6000_hard_regno_mode_ok): New function.
15517 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
15518 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
15519 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
15520 (riscv_hard_regno_mode_ok): ...this and make static.
15521 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15522 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
15523 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
15524 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15525 (rl78_hard_regno_mode_ok): Make static and return bool.
15526 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
15527 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
15529 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
15530 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15531 (rs6000_hard_regno_mode_ok): Rename to...
15532 (rs6000_hard_regno_mode_ok_uncached): ...this.
15533 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15534 (rs6000_hard_regno_mode_ok): New function.
15535 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
15536 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
15537 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15538 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
15539 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
15540 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
15541 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15542 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
15543 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
15544 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15545 (sh_hard_regno_mode_ok): Make static.
15546 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
15547 instead of HARD_REGNO_MODE_OK.
15548 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
15549 (sparc_mode_class): Delete.
15550 (HARD_REGNO_MODE_OK): Delete.
15551 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15552 (hard_regno_mode_classes): Make static.
15553 (sparc_mode_class): Likewise.
15554 (sparc_hard_regno_mode_ok): New function.
15555 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
15556 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
15557 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
15559 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15560 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
15561 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
15562 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
15563 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
15564 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15565 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
15566 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
15567 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15568 (visium_hard_regno_mode_ok): New function.
15569 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
15570 instead of HARD_REGNO_MODE_OK.
15571 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
15572 (HARD_REGNO_MODE_OK): Delete.
15573 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
15574 (xtensa_hard_regno_mode_ok_p): ...this and make static.
15575 (xtensa_option_override): Update accordingly.
15576 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15577 (xtensa_hard_regno_mode_ok): New function.
15578 * system.h (HARD_REGNO_MODE_OK): Poison.
15580 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15581 Alan Hayward <alan.hayward@arm.com>
15582 David Sherwood <david.sherwood@arm.com>
15584 * target.def (hard_regno_call_part_clobbered): New hook.
15585 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
15586 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
15587 * doc/tm.texi: Regenerate.
15588 * hooks.h (hook_bool_uint_mode_false): Declare.
15589 * hooks.c (hook_bool_uint_mode_false): New function.
15590 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15591 * cselib.c (cselib_process_insn): Use
15592 targetm.hard_regno_call_part_clobbered instead of
15593 HARD_REGNO_CALL_PART_CLOBBERED.
15594 * ira-conflicts.c (ira_build_conflicts): Likewise.
15595 * ira-costs.c (ira_tune_allocno_costs): Likewise.
15596 * lra-constraints.c (need_for_call_save_p): Likewise.
15597 * lra-lives.c: Include target.h.
15598 (check_pseudos_live_through_calls): Use
15599 targetm.hard_regno_call_part_clobbered instead of
15600 HARD_REGNO_CALL_PART_CLOBBERED.
15601 * regcprop.c: Include target.h.
15602 (copyprop_hardreg_forward_1): Use
15603 targetm.hard_regno_call_part_clobbered instead of
15604 HARD_REGNO_CALL_PART_CLOBBERED.
15605 * reginfo.c (choose_hard_reg_mode): Likewise.
15606 * regrename.c (check_new_reg_p): Likewise.
15607 * reload.c (find_equiv_reg): Likewise.
15608 * reload1.c (emit_reload_insns): Likewise.
15609 * sched-deps.c (deps_analyze_insn): Likewise.
15610 * sel-sched.c (init_regs_for_mode): Likewise.
15611 (mark_unavailable_hard_regs): Likewise.
15612 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
15613 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15614 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
15616 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15617 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15618 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
15620 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
15622 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15623 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15624 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
15626 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15627 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15628 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
15630 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15631 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
15633 * config/powerpcspe/powerpcspe.c
15634 (rs6000_hard_regno_call_part_clobbered): New function.
15635 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15636 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15637 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
15639 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15640 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15641 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
15643 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15644 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15645 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
15647 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15648 Alan Hayward <alan.hayward@arm.com>
15649 David Sherwood <david.sherwood@arm.com>
15651 * rtl.h (subreg_memory_offset): Declare.
15652 * emit-rtl.c (subreg_memory_offset): New function.
15653 * expmed.c (store_bit_field_1): Use it.
15654 * expr.c (undefined_operand_subword_p): Likewise.
15655 * simplify-rtx.c (simplify_subreg): Likewise.
15657 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
15659 PR rtl-optimization/57448
15662 * optabs.c (expand_atomic_load): Place compiler memory barriers if
15663 using atomic_load pattern.
15664 (expand_atomic_store): Likewise.
15666 2017-09-04 Jakub Jelinek <jakub@redhat.com>
15669 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
15670 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
15671 handling. Use replace_call_with_value with NULL instead of
15672 gsi_replace, unlink_stmt_vdef and release_defs.
15674 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
15677 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
15679 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15682 * rtl.h (emit_library_call_value_1): Declare.
15683 (emit_library_call): Replace declaration with a series of overloads.
15684 Remove the parameter count argument.
15685 (emit_library_call_value): Likewise.
15686 * calls.c (emit_library_call_value_1): Make global. Replace varargs
15687 with an "rtx_mode_t *".
15688 (emit_library_call_value): Delete.
15689 (emit_library_call): Likewise.
15690 * asan.c (asan_emit_stack_protection): Update calls accordingly.
15691 (asan_emit_allocas_unpoison): Likewise.
15692 * builtins.c (expand_builtin_powi): Likewise.
15693 (expand_asan_emit_allocas_unpoison): Likewise.
15694 * cfgexpand.c (expand_main_function): Likewise.
15695 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
15696 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
15697 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
15698 * config/arm/arm.c (arm_trampoline_init): Likewise.
15699 (arm_call_tls_get_addr): Likewise.
15700 (arm_expand_divmod_libfunc): Likewise.
15701 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
15702 (smulsi3_highpart): Likewise.
15703 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
15704 (c6x_expand_compare): Likewise.
15705 (c6x_expand_movmem): Likewise.
15706 * config/frv/frv.c (frv_trampoline_init): Likewise.
15707 * config/i386/i386.c (ix86_trampoline_init): Likewise.
15708 (ix86_expand_divmod_libfunc): Likewise.
15709 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
15710 (ia64_expand_compare): Likewise.
15711 (ia64_profile_hook): Likewise.
15712 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
15713 (nonlocal_goto): Likewise.
15714 (restore_stack_nonlocal): Likewise.
15715 * config/m32r/m32r.c (block_move_call): Likewise.
15716 (m32r_trampoline_init): Likewise.
15717 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
15718 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
15719 (m68k_call_m68k_read_tp): Likewise.
15720 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
15721 (microblaze_expand_divide): Likewise.
15722 * config/mips/mips.h (mips_args): Likewise.
15723 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
15724 (MIPS_ICACHE_SYNC): Likewise.
15725 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
15726 (nios2_trampoline_init): Likewise.
15727 * config/pa/pa.c (hppa_tls_call): Likewise.
15728 (pa_trampoline_init): Likewise.
15729 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
15730 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
15731 (expand_strn_compare): Likewise.
15732 (rs6000_generate_compare): Likewise.
15733 (rs6000_expand_float128_convert): Likewise.
15734 (output_profile_hook): Likewise.
15735 (rs6000_trampoline_init): Likewise.
15736 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
15737 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
15738 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
15739 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
15740 (rs6000_generate_compare): Likewise.
15741 (rs6000_expand_float128_convert): Likewise.
15742 (output_profile_hook): Likewise.
15743 (rs6000_trampoline_init): Likewise.
15744 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
15745 * config/sh/sh.c (sh_trampoline_init): Likewise.
15746 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
15747 (sparc_emit_float_lib_cmp): Likewise.
15748 (sparc32_initialize_trampoline): Likewise.
15749 (sparc64_initialize_trampoline): Likewise.
15750 (sparc_profile_hook): Likewise.
15751 * config/spu/spu.c (ea_load_store): Likewise.
15752 * config/spu/spu.md (floatunssidf2): Likewise.
15753 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
15754 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
15755 * config/visium/visium.c (expand_block_move_4): Likewise.
15756 (expand_block_move_2): Likewise.
15757 (expand_block_move_1): Likewise.
15758 (expand_block_set_4): Likewise.
15759 (expand_block_set_2): Likewise.
15760 (expand_block_set_1): Likewise.
15761 (visium_trampoline_init): Likewise.
15762 (visium_profile_hook): Likewise.
15763 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
15764 (xtensa_setup_frame_addresses): Likewise.
15765 (xtensa_trampoline_init): Likewise.
15766 * except.c (sjlj_emit_function_enter): Likewise.
15767 (sjlj_emit_function_exit): Likewise.
15768 * explow.c (allocate_dynamic_stack_space): Likewise.
15769 (probe_stack_range): Likewise.
15770 * expr.c (convert_mode_scalar): Likewise.
15771 * optabs.c (expand_binop): Likewise.
15772 (expand_twoval_binop_libfunc): Likewise.
15773 (expand_unop): Likewise.
15774 (prepare_cmp_insn): Likewise.
15775 (prepare_float_lib_cmp): Likewise.
15776 (expand_float): Likewise.
15777 (expand_fix): Likewise.
15778 (expand_fixed_convert): Likewise.
15779 (maybe_emit_sync_lock_test_and_set): Likewise.
15780 (expand_atomic_compare_and_swap): Likewise.
15781 (expand_mem_thread_fence): Likewise.
15782 (expand_atomic_fetch_op): Likewise.
15784 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
15786 * doc/generic.texi (OpenACC): Adjust URL.
15787 * doc/invoke.texi (C Dialect Options): Ditto.
15789 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
15791 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
15792 predicate for operand 1. Add (m,<S>) constraint.
15793 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
15794 Prevent memory operand 1 with register operand 2.
15796 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
15798 PR rtl-optimization/82024
15799 * combine.c (try_combine): If the combination result is a PARALLEL,
15800 and we only need to retain the SET in there that would be placed
15801 at I2, check that we can place that at I3 instead, before doing so.
15803 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15806 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
15808 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
15809 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
15812 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
15813 Jeff Law <law@redhat.com>
15815 * varasm.c (bss_initializer_p): Do not put constants into .bss
15816 (categorize_decl_for_section): Handle bss_initializer_p returning
15817 false when DECL_INITIAL is NULL.
15819 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15822 * config/s390/s390.c (s390_can_inline_p): New function.
15824 2017-09-01 Jeff Law <law@redhat.com>
15826 PR tree-optimization/82052
15827 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15828 Always initialize the returned slot after a hash table miss
15829 when INSERT is true.
15831 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
15833 * config/s390/s390.md (mem_signal_fence): Remove.
15834 * doc/md.texi (mem_signal_fence): Remove.
15835 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
15837 * target-insns.def (mem_signal_fence): Remove.
15839 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15842 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
15845 * asan.c (create_odr_indicator): Strip name encoding from assembler
15846 name before appending it after __odr_asan_.
15848 2017-09-01 Martin Liska <mliska@suse.cz>
15850 PR tree-optimization/82059
15851 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
15852 frequency only when an edge is redirected.
15854 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15856 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
15857 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
15858 (arc_conditional_register_usage): Remove ARC600 lp_count
15860 (arc_file_start): Emit Tag_ARC_CPU_variation.
15861 (arc_can_use_doloop_p): New conditions to use ZOLs.
15862 (hwloop_fail): New function.
15863 (hwloop_optimize): Likewise.
15864 (hwloop_pattern_reg): Likewise.
15865 (arc_doloop_hooks): New struct, to be used with reorg_loops.
15866 (arc_reorg_loops): New function, calls reorg_loops.
15867 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
15868 (arc600_corereg_hazard): Remove ZOL checking, case handled by
15870 (arc_loop_hazard): Remove function, functionality moved into
15872 (arc_hazard): Remove arc_loop_hazard call.
15873 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
15874 into hwloop_optimize.
15875 (arc_label_align): Remove ZOL handling.
15876 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
15877 * config/arc/arc.md (doloop_begin): Remove pattern.
15878 (doloop_begin_i): Likewise.
15879 (doloop_end_i): Likewise.
15880 (doloop_fallback): Likewise.
15881 (doloop_fallback_m): Likewise.
15882 (doloop_end): Reimplement expand.
15883 (arc_lp): New pattern for LP instruction.
15884 (loop_end): New pattern.
15885 (loop_fail): Likewise.
15886 (decrement_and_branch_until_zero): Likewise.
15887 * config/arc/arc.opt (mlpc-width): New option.
15888 * doc/invoke.texi (mlpc-width): Document option.
15890 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15892 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
15893 (arc_ccfsm_advance): Fix checking for delay slots.
15894 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
15896 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15898 * config/arc/arc.md (movqi_insn): Add stores to save constant long
15900 (movhi_insn): Update store instruction constraint which are saving
15901 6-bit short immediates.
15902 (movsi_insn): Consider also short scaled load operations.
15903 (zero_extendhisi2_i): Use Usd constraint instead of T.
15904 (extendhisi2_i): Add q constraint.
15905 (arc_clzsi2): Add type and length attributes.
15906 (arc_ctzsi2): Likewise.
15907 * config/arc/constraints.md (Usc): Update constraint, the
15908 assembler can parse two relocations for a single instruction.
15910 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15912 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
15913 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
15915 2017-08-31 Olivier Hainque <hainque@adacore.com>
15917 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
15918 match as powerpc-wrs-vxworks*.
15920 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
15922 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
15923 register constraint for by-element operand.
15924 (aarch64_mls_elt_merge<mode>): Likewise.
15926 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15928 * config/arc/arc.c (arc_can_follow_jump): Check for short
15931 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15933 * config.gcc: Use g.opt for arc.
15934 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
15935 functionality moved to ...
15936 (legitimate_scaled_address_p): New function, ...here.
15937 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
15938 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
15939 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
15941 (arc_override_options): Handle G option.
15942 (arc_output_pic_addr_const): Correct function definition.
15943 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
15944 (arc_decl_anon_ns_mem_p): Delete.
15945 (arc_in_small_data_p): Overhaul this function to take into
15946 consideration the value given via G option.
15947 (arc_rewrite_small_data_1): Renamed and corrected old
15948 arc_rewrite_small_data function.
15949 (arc_rewrite_small_data): New function.
15950 (small_data_pattern): Don't use pic_offset_table_rtx.
15951 * config/arc/arc.h (CC1_SPEC): Recognize G option.
15952 * config/arc/simdext.md (movmisalignv2hi): Use
15953 prepare_move_operands function.
15955 (movmisalign*): Likewise.
15956 * doc/invoke.texi (ARC options): Document -G option.
15958 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15960 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
15962 * config/arc/arc.c (arc_print_operand): Output scalled address for
15963 sdata whenever is possible.
15964 (arc_in_small_data_p): Allow sdata for 64bit datum when double
15965 load/stores are available.
15966 (compact_sda_memory_operand): Check for the alignment required by
15967 code density instructions.
15968 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
15970 * config/arc/constraints.md (Usd): Update constraint.
15971 (Us0): New constraint.
15972 (Usc): Update constraint.
15974 2017-08-31 Richard Biener <rguenther@suse.de>
15976 PR middle-end/82054
15977 * dwarf2out.c (dwarf2out_early_global_decl): Process each
15978 function only once.
15980 2017-08-31 Tamar Christina <tamar.christina@arm.com>
15982 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
15983 Resize type_signature.
15985 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15986 Alan Hayward <alan.hayward@arm.com>
15987 David Sherwood <david.sherwood@arm.com>
15989 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
15990 subregs whose inner modes can be stored in GPRs.
15991 (aarch64_classify_index): Likewise.
15993 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15994 Alan Hayward <alan.hayward@arm.com>
15995 David Sherwood <david.sherwood@arm.com>
15997 * config/aarch64/iterators.md (V_cmp_result): Rename to...
15998 (V_INT_EQUIV): ...this.
15999 (v_cmp_result): Rename to...
16000 (v_int_equiv): ...this.
16001 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16002 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16003 (copysign<mode>3): Likewise.
16004 (aarch64_simd_bsl<mode>_internal): Likewise.
16005 (aarch64_simd_bsl<mode>): Likewise.
16006 (vec_cmp<mode><mode>): Likewise.
16007 (vcond<mode><mode>): Likewise.
16008 (vcond<v_cmp_mixed><mode>): Likewise.
16009 (vcondu<mode><v_cmp_mixed>): Likewise.
16010 (aarch64_cm<optab><mode>): Likewise.
16011 (aarch64_cmtst<mode>): Likewise.
16012 (aarch64_fac<optab><mode>): Likewise.
16013 (vec_perm_const<mode>): Likewise.
16014 (vcond_mask_<mode><v_cmp_result>): Rename to...
16015 (vcond_mask_<mode><v_int_equiv>): ...this.
16016 (vec_cmp<mode><v_cmp_result>): Rename to...
16017 (vec_cmp<mode><v_int_equiv>): ...this.
16019 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16020 Alan Hayward <alan.hayward@arm.com>
16021 David Sherwood <david.sherwood@arm.com>
16023 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16025 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16026 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16027 (UNSPEC_LD4_DREG): New unspecs.
16028 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16029 (aarch64_ld2<mode>_dreg_be): Replace with...
16030 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16032 (aarch64_ld3<mode>_dreg_le)
16033 (aarch64_ld3<mode>_dreg_be): Replace with...
16034 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16036 (aarch64_ld4<mode>_dreg_le)
16037 (aarch64_ld4<mode>_dreg_be): Replace with...
16038 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16041 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16043 PR tree-optimization/81987
16044 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
16045 insert an initializer in a location not dominated by the stride
16048 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16050 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
16051 on the entire header of the finally block in the fallthru case.
16053 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16055 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
16057 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
16059 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
16060 rs6000_emit_move_from_cr and call renamed function.
16061 (rs6000_emit_prologue): Call renamed functions.
16062 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16063 movesi_from_cr, remove volatile CRs.
16065 2017-08-30 Jon Beniston <jon@beniston.com>
16066 Richard Biener <rguenther@suse.de>
16068 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
16069 of VECTOR_MODE_P check.
16070 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
16071 element vector types.
16073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16075 * df.h (df_read_modify_subreg_p): Remove in favor of...
16076 * rtl.h (read_modify_subreg_p): ...this new function. Take a
16077 const_rtx instead of an rtx.
16078 * cprop.c (local_cprop_find_used_regs): Update accordingly.
16079 * df-problems.c (df_word_lr_mark_ref): Likewise.
16080 * ira-lives.c (mark_pseudo_reg_live): Likewise.
16081 (mark_pseudo_reg_dead): Likewise.
16082 (mark_ref_dead): Likewise.
16083 * reginfo.c (init_subregs_of_mode): Likewise.
16084 * sched-deps.c (sched_analyze_1): Likewise.
16085 * df-scan.c (df_def_record_1): Likewise.
16086 (df_uses_record): Likewise.
16087 (df_read_modify_subreg_p): Remove in favor of...
16088 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
16089 const_rtx instead of an rtx.
16091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16092 Alan Hayward <alan.hayward@arm.com>
16093 David Sherwood <david.sherwood@arm.com>
16095 * rtl.h (partial_subreg_p): New function.
16096 * caller-save.c (save_call_clobbered_regs): Use it.
16097 * calls.c (expand_call): Likewise.
16098 * combine.c (combinable_i3pat): Likewise.
16099 (simplify_set): Likewise.
16100 (make_extraction): Likewise.
16101 (make_compound_operation_int): Likewise.
16102 (gen_lowpart_or_truncate): Likewise.
16103 (force_to_mode): Likewise.
16104 (make_field_assignment): Likewise.
16105 (reg_truncated_to_mode): Likewise.
16106 (record_truncated_value): Likewise.
16107 (move_deaths): Likewise.
16108 * cse.c (record_jump_cond): Likewise.
16109 (cse_insn): Likewise.
16110 * cselib.c (cselib_lookup_1): Likewise.
16111 * expmed.c (extract_bit_field_using_extv): Likewise.
16112 * function.c (assign_parm_setup_reg): Likewise.
16113 * ifcvt.c (noce_convert_multiple_sets): Likewise.
16114 * ira-build.c (create_insn_allocnos): Likewise.
16115 * lra-coalesce.c (merge_pseudos): Likewise.
16116 * lra-constraints.c (match_reload): Likewise.
16117 (simplify_operand_subreg): Likewise.
16118 (curr_insn_transform): Likewise.
16119 * lra-lives.c (process_bb_lives): Likewise.
16120 * lra.c (new_insn_reg): Likewise.
16121 (lra_substitute_pseudo): Likewise.
16122 * regcprop.c (mode_change_ok): Likewise.
16123 (maybe_mode_change): Likewise.
16124 (copyprop_hardreg_forward_1): Likewise.
16125 * reload.c (push_reload): Likewise.
16126 (find_reloads): Likewise.
16127 (find_reloads_subreg_address): Likewise.
16128 * reload1.c (alter_reg): Likewise.
16129 (eliminate_regs_1): Likewise.
16130 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16132 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
16134 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
16135 back to if statements, including unpack.
16137 2017-08-30 Martin Liska <mliska@suse.cz>
16139 PR inline-asm/82001
16140 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
16142 (func_checker::compare_asm_inputs_outputs): ... this function.
16143 (func_checker::compare_gimple_asm): Use the function to compare
16144 also ASM constrains.
16145 * ipa-icf-gimple.h: Rename the function.
16147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16148 Alan Hayward <alan.hayward@arm.com>
16149 David Sherwood <david.sherwood@arm.com>
16151 * coretypes.h (complex_mode): New type.
16152 * gdbhooks.py (build_pretty_printer): Handle it.
16153 * machmode.h (complex_mode): New class.
16154 (complex_mode::includes_p): New function.
16155 (is_complex_int_mode): Likewise.
16156 (is_complex_float_mode): Likewise.
16157 * genmodes.c (get_mode_class): Handle complex mode classes.
16158 * function.c (expand_function_end): Use is_complex_int_mode.
16160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16161 Alan Hayward <alan.hayward@arm.com>
16162 David Sherwood <david.sherwood@arm.com>
16164 * coretypes.h (scalar_mode_pod): New typedef.
16165 * gdbhooks.py (build_pretty_printer): Handle it.
16166 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
16167 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
16168 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
16169 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
16170 as_a <scalar_mode>.
16172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16173 Alan Hayward <alan.hayward@arm.com>
16174 David Sherwood <david.sherwood@arm.com>
16176 * machmode.h (mode_for_vector): Take a scalar_mode instead
16178 * stor-layout.c (mode_for_vector): Likewise.
16179 * explow.c (promote_mode): Use as_a <scalar_mode>.
16180 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
16182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16183 Alan Hayward <alan.hayward@arm.com>
16184 David Sherwood <david.sherwood@arm.com>
16186 * target.def (preferred_simd_mode): Take a scalar_mode
16187 instead of a machine_mode.
16188 * targhooks.h (default_preferred_simd_mode): Likewise.
16189 * targhooks.c (default_preferred_simd_mode): Likewise.
16190 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
16191 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
16192 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
16193 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
16194 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
16195 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
16196 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
16197 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
16199 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
16200 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
16201 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
16202 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
16203 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
16204 * doc/tm.texi: Regenerate.
16205 * optabs-query.c (can_vec_mask_load_store_p): Return false for
16208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16209 Alan Hayward <alan.hayward@arm.com>
16210 David Sherwood <david.sherwood@arm.com>
16212 * target.def (scalar_mode_supported_p): Take a scalar_mode
16213 instead of a machine_mode.
16214 * targhooks.h (default_scalar_mode_supported_p): Likewise.
16215 * targhooks.c (default_scalar_mode_supported_p): Likewise.
16216 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
16217 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16218 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
16219 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
16220 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
16221 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
16222 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
16223 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
16224 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
16225 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
16226 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
16227 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
16229 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
16230 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
16231 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16232 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16233 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
16235 * doc/tm.texi: Regenerate.
16237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16238 Alan Hayward <alan.hayward@arm.com>
16239 David Sherwood <david.sherwood@arm.com>
16241 * coretypes.h (opt_scalar_mode): New typedef.
16242 * gdbhooks.py (build_pretty_printers): Handle it.
16243 * machmode.h (mode_iterator::get_2xwider): Add overload for
16245 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
16247 * expr.c (convert_mode_scalar): Likewise.
16248 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16249 * optabs.c (expand_float): Likewise.
16250 (expand_fix): Likewise.
16251 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16254 Alan Hayward <alan.hayward@arm.com>
16255 David Sherwood <david.sherwood@arm.com>
16257 * optabs.c (expand_float): Explicitly check for scalars before
16258 using a branching expansion.
16259 (expand_fix): Likewise.
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 * expr.c (convert_mode): Split scalar handling out into...
16266 (convert_mode_scalar): ...this new function. Treat the modes
16269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16270 Alan Hayward <alan.hayward@arm.com>
16271 David Sherwood <david.sherwood@arm.com>
16273 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
16275 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
16277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16278 Alan Hayward <alan.hayward@arm.com>
16279 David Sherwood <david.sherwood@arm.com>
16281 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
16282 rather than a machine_mode.
16283 (fixed_from_string): Likewise.
16284 (fixed_convert): Likewise.
16285 (fixed_convert_from_int): Likewise.
16286 (fixed_convert_from_real): Likewise.
16287 (real_convert_from_fixed): Likewise.
16288 * fixed-value.c (fixed_from_double_int): Likewise.
16289 (fixed_from_string): Likewise.
16290 (fixed_convert): Likewise.
16291 (fixed_convert_from_int): Likewise.
16292 (fixed_convert_from_real): Likewise.
16293 (real_convert_from_fixed): Likewise.
16294 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
16296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16297 Alan Hayward <alan.hayward@arm.com>
16298 David Sherwood <david.sherwood@arm.com>
16300 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
16301 of separate mode class checks. Do not allow vector modes here.
16302 (immed_wide_int_const): Use as_a <scalar_mode>.
16303 * explow.c (trunc_int_for_mode): Likewise.
16304 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
16305 (wi::shwi): Likewise.
16306 (wi::min_value): Likewise.
16307 (wi::max_value): Likewise.
16308 * dwarf2out.c (loc_descriptor): Likewise.
16309 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
16310 for CONST_WIDE_INT.
16312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16313 Alan Hayward <alan.hayward@arm.com>
16314 David Sherwood <david.sherwood@arm.com>
16316 * tree.h (SCALAR_TYPE_MODE): New macro.
16317 * expr.c (expand_expr_addr_expr_1): Use it.
16318 (expand_expr_real_2): Likewise.
16319 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
16320 (fold_convert_const_fixed_from_int): Likewise.
16321 (fold_convert_const_fixed_from_real): Likewise.
16322 (native_encode_fixed): Likewise
16323 (native_encode_complex): Likewise
16324 (native_encode_vector): Likewise.
16325 (native_interpret_fixed): Likewise.
16326 (native_interpret_real): Likewise.
16327 (native_interpret_complex): Likewise.
16328 (native_interpret_vector): Likewise.
16329 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
16330 (simd_clone_adjust_argument_types): Likewise.
16331 (simd_clone_init_simd_arrays): Likewise.
16332 (simd_clone_adjust): Likewise.
16333 * stor-layout.c (layout_type): Likewise.
16334 * tree.c (build_minus_one_cst): Likewise.
16335 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16336 * tree-inline.c (estimate_move_cost): Likewise.
16337 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
16338 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
16339 (vectorizable_reduction): Likewise.
16340 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
16341 (vect_recog_mixed_size_cond_pattern): Likewise.
16342 (check_bool_pattern): Likewise.
16343 (adjust_bool_pattern): Likewise.
16344 (search_type_for_mask_1): Likewise.
16345 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
16346 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16347 (vectorizable_load): Likewise.
16348 (vectorizable_store): Likewise.
16349 * ubsan.c (ubsan_encode_value): Likewise.
16350 * varasm.c (output_constant): Likewise.
16352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16353 Alan Hayward <alan.hayward@arm.com>
16354 David Sherwood <david.sherwood@arm.com>
16356 * coretypes.h (scalar_mode): New class.
16357 * machmode.h (scalar_mode): Likewise.
16358 (scalar_mode::includes_p): New function.
16359 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
16360 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
16361 * genmodes.c (get_mode_class): Handle remaining scalar modes.
16362 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
16363 * expmed.c (store_bit_field_1): Likewise.
16364 (extract_bit_field_1): Likewise.
16365 * expr.c (write_complex_part): Likewise.
16366 (read_complex_part): Likewise.
16367 (emit_move_complex_push): Likewise.
16368 (expand_expr_real_2): Likewise.
16369 * function.c (assign_parm_setup_reg): Likewise.
16370 (assign_parms_unsplit_complex): Likewise.
16371 * optabs.c (expand_binop): Likewise.
16372 * rtlanal.c (subreg_get_info): Likewise.
16373 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16374 * varasm.c (output_constant_pool_2): Likewise.
16376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16377 Alan Hayward <alan.hayward@arm.com>
16378 David Sherwood <david.sherwood@arm.com>
16380 * expmed.c (extract_high_half): Use scalar_int_mode and remove
16382 (expmed_mult_highpart_optab): Likewise.
16383 (expmed_mult_highpart): Likewise.
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 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
16390 instead of a machine_mode.
16391 (builtin_memset_read_str): Likewise.
16392 * builtins.c (c_readstr): Likewise.
16393 (builtin_memcpy_read_str): Likewise.
16394 (builtin_strncpy_read_str): Likewise.
16395 (builtin_memset_read_str): Likewise.
16396 (builtin_memset_gen_str): Likewise.
16397 (expand_builtin_signbit): Use scalar_int_mode for local variables.
16398 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
16399 instead of a machine_mode.
16400 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
16402 (make_extraction): Likewise.
16403 (try_widen_shift_mode): Take and return scalar_int_modes instead
16405 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
16406 a scalar_int_mode instead of a machine_mode.
16407 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
16408 (avr_addr_space_pointer_mode): Likewise.
16409 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
16410 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
16411 (msp430_unwind_word_mode): Likewise.
16412 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
16413 (spu_addr_space_pointer_mode): Likewise.
16414 (spu_addr_space_address_mode): Likewise.
16415 (spu_libgcc_cmp_return_mode): Likewise.
16416 (spu_libgcc_shift_count_mode): Likewise.
16417 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
16418 (rl78_addr_space_pointer_mode): Likewise.
16419 (fl78_unwind_word_mode): Likewise.
16420 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
16422 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
16423 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
16424 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
16425 (mips_valid_pointer_mode): Likewise.
16426 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
16427 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
16428 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
16430 (ft32_addr_space_address_mode): Likewise.
16431 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
16432 scalar_int_mode instead of a machine_mode.
16433 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
16435 (m32c_addr_space_address_mode): Likewise.
16436 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
16437 (rs6000_eh_return_filter_mode): Likewise.
16438 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
16439 (rs6000_eh_return_filter_mode): Likewise.
16440 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
16441 (s390_libgcc_shift_count_mode): Likewise.
16442 (s390_unwind_word_mode): Likewise.
16443 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
16445 * target.def (mode_rep_extended): Likewise.
16446 (valid_pointer_mode): Likewise.
16447 (addr_space.valid_pointer_mode): Likewise.
16448 (eh_return_filter_mode): Return a scalar_int_mode rather than
16450 (libgcc_cmp_return_mode): Likewise.
16451 (libgcc_shift_count_mode): Likewise.
16452 (unwind_word_mode): Likewise.
16453 (addr_space.pointer_mode): Likewise.
16454 (addr_space.address_mode): Likewise.
16455 * doc/tm.texi: Regenerate.
16456 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
16458 (do_jump): Use scalar_int_mode for local variables.
16459 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
16460 rather than a machine_mode.
16461 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
16462 (scompare_loc_descriptor_wide): Likewise.
16463 (scompare_loc_descriptor_narrow): Likewise.
16464 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
16466 * except.c (sjlj_emit_dispatch_table): Likewise.
16467 (expand_builtin_eh_copy_values): Likewise.
16468 * explow.c (convert_memory_address_addr_space_1): Likewise.
16469 Take a scalar_int_mode rather than a machine_mode.
16470 (convert_memory_address_addr_space): Take a scalar_int_mode rather
16471 than a machine_mode.
16472 (memory_address_addr_space): Use scalar_int_mode for local variables.
16473 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
16474 rather than a machine_mode.
16475 * expmed.c (mask_rtx): Likewise.
16476 (init_expmed_one_conv): Likewise.
16477 (expand_mult_highpart_adjust): Likewise.
16478 (extract_high_half): Likewise.
16479 (expmed_mult_highpart_optab): Likewise.
16480 (expmed_mult_highpart): Likewise.
16481 (expand_smod_pow2): Likewise.
16482 (expand_sdiv_pow2): Likewise.
16483 (emit_store_flag_int): Likewise.
16484 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
16486 (extract_low_bits): Likewise.
16487 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
16489 * expr.c (pieces_addr::adjust): Likewise.
16490 (can_store_by_pieces): Likewise.
16491 (store_by_pieces): Likewise.
16492 (clear_by_pieces_1): Likewise.
16493 (expand_expr_addr_expr_1): Likewise.
16494 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
16495 (expand_expr_real_1): Likewise.
16496 (try_casesi): Likewise.
16497 * final.c (shorten_branches): Likewise.
16498 * fold-const.c (fold_convert_const_int_from_fixed): Change the
16499 type of "mode" to machine_mode.
16500 * internal-fn.c (expand_arith_overflow_result_store): Take a
16501 scalar_int_mode rather than a machine_mode.
16502 (expand_mul_overflow): Use scalar_int_mode for local variables.
16503 * loop-doloop.c (doloop_modify): Likewise.
16504 (doloop_optimize): Likewise.
16505 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
16506 than a machine_mode.
16507 (expand_doubleword_shift_condmove): Likewise.
16508 (expand_doubleword_shift): Likewise.
16509 (expand_doubleword_clz): Likewise.
16510 (expand_doubleword_popcount): Likewise.
16511 (expand_doubleword_parity): Likewise.
16512 (expand_absneg_bit): Use scalar_int_mode for local variables.
16513 (prepare_float_lib_cmp): Likewise.
16514 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
16515 rather than a machine_mode.
16516 (convert_memory_address_addr_space): Likewise.
16517 (get_mode_bounds): Likewise.
16518 (get_address_mode): Return a scalar_int_mode rather than a
16520 * rtlanal.c (get_address_mode): Likewise.
16521 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
16522 than a machine_mode.
16523 * targhooks.c (default_mode_rep_extended): Likewise.
16524 (default_valid_pointer_mode): Likewise.
16525 (default_addr_space_valid_pointer_mode): Likewise.
16526 (default_eh_return_filter_mode): Return a scalar_int_mode rather
16527 than a machine_mode.
16528 (default_libgcc_cmp_return_mode): Likewise.
16529 (default_libgcc_shift_count_mode): Likewise.
16530 (default_unwind_word_mode): Likewise.
16531 (default_addr_space_pointer_mode): Likewise.
16532 (default_addr_space_address_mode): Likewise.
16533 * targhooks.h (default_eh_return_filter_mode): Likewise.
16534 (default_libgcc_cmp_return_mode): Likewise.
16535 (default_libgcc_shift_count_mode): Likewise.
16536 (default_unwind_word_mode): Likewise.
16537 (default_addr_space_pointer_mode): Likewise.
16538 (default_addr_space_address_mode): Likewise.
16539 (default_mode_rep_extended): Take a scalar_int_mode rather than
16541 (default_valid_pointer_mode): Likewise.
16542 (default_addr_space_valid_pointer_mode): Likewise.
16543 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
16545 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
16546 rather than a machine_mode.
16547 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
16548 for local variables.
16549 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16550 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
16551 than a machine_mode.
16553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16554 Alan Hayward <alan.hayward@arm.com>
16555 David Sherwood <david.sherwood@arm.com>
16557 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
16558 the mode argument to scalar_int_mode.
16559 (do_jump_by_parts_zero_rtx): Likewise.
16560 (do_jump_by_parts_equality_rtx): Likewise.
16561 (do_jump_by_parts_greater): Take a mode argument.
16562 (do_jump_by_parts_equality): Likewise.
16563 (do_jump_1): Update calls accordingly.
16565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16566 Alan Hayward <alan.hayward@arm.com>
16567 David Sherwood <david.sherwood@arm.com>
16569 * is-a.h (safe_dyn_cast): New function.
16570 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
16571 (jump_table_for_label): Likewise.
16572 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
16573 instead of an rtx_insn *.
16574 (shorten_branches): Use dyn_cast instead of LABEL_P and
16575 JUMP_TABLE_DATA_P. Use jump_table_for_label and
16576 rtx_jump_table_data::get_data_mode.
16577 (final_scan_insn): Likewise.
16579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16580 Alan Hayward <alan.hayward@arm.com>
16581 David Sherwood <david.sherwood@arm.com>
16583 * combine.c (try_combine): Use is_a <scalar_int_mode> when
16584 trying to combine a full-register integer set with a subreg
16587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16588 Alan Hayward <alan.hayward@arm.com>
16589 David Sherwood <david.sherwood@arm.com>
16591 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
16592 that is always either address_mode or pointer_mode.
16594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16595 Alan Hayward <alan.hayward@arm.com>
16596 David Sherwood <david.sherwood@arm.com>
16598 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
16599 when the two are known to be equal.
16601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16602 Alan Hayward <alan.hayward@arm.com>
16603 David Sherwood <david.sherwood@arm.com>
16605 * simplify-rtx.c (simplify_const_unary_operation): Use
16606 is_a <scalar_int_mode> instead of checking for a nonzero
16607 precision. Forcibly convert op_mode to a scalar_int_mode
16608 in that case. More clearly differentiate the operand and
16609 result modes and use the former when deciding what the value
16610 of a count-bits operation should be. Use is_int_mode instead
16611 of checking for a MODE_INT. Remove redundant check for whether
16612 this mode has a zero precision.
16614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16615 Alan Hayward <alan.hayward@arm.com>
16616 David Sherwood <david.sherwood@arm.com>
16618 * optabs.c (widen_leading): Change the type of the mode argument
16619 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
16620 (widen_bswap): Likewise.
16621 (expand_parity): Likewise.
16622 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
16623 (expand_ffs): Likewise.
16624 (epand_unop): Check for scalar integer modes before calling the
16627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16628 Alan Hayward <alan.hayward@arm.com>
16629 David Sherwood <david.sherwood@arm.com>
16631 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
16633 (expand_expr_real_1): Update call accordingly.
16635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16636 Alan Hayward <alan.hayward@arm.com>
16637 David Sherwood <david.sherwood@arm.com>
16639 * expmed.c (store_bit_field_using_insv): Add op0_mode and
16640 value_mode arguments. Use scalar_int_mode internally.
16641 (store_bit_field_1): Rename the new integer mode from imode
16642 to op0_mode and use it instead of GET_MODE (op0). Update calls
16643 to store_split_bit_field, store_bit_field_using_insv and
16644 store_fixed_bit_field.
16645 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
16646 Use scalar_int_mode internally. Use a bit count rather than a mode
16647 when calculating the largest bit size for get_best_mode.
16648 Update calls to store_split_bit_field and store_fixed_bit_field_1.
16649 (store_fixed_bit_field_1): Add mode and value_mode arguments.
16650 Remove assertion that OP0 has a scalar integer mode.
16651 (store_split_bit_field): Add op0_mode and value_mode arguments.
16652 Update calls to extract_fixed_bit_field.
16653 (extract_bit_field_using_extv): Add an op0_mode argument.
16654 Use scalar_int_mode internally.
16655 (extract_bit_field_1): Rename the new integer mode from imode to
16656 op0_mode and use it instead of GET_MODE (op0). Update calls to
16657 extract_split_bit_field, extract_bit_field_using_extv and
16658 extract_fixed_bit_field.
16659 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
16660 to extract_split_bit_field and extract_fixed_bit_field_1.
16661 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
16662 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
16663 on the target mode.
16664 (extract_split_bit_field): Add an op0_mode argument. Update call
16665 to extract_fixed_bit_field.
16667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16668 Alan Hayward <alan.hayward@arm.com>
16669 David Sherwood <david.sherwood@arm.com>
16671 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
16672 * explow.c (hard_function_value): Likewise.
16673 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
16674 convert_to_mode call outside the loop.
16675 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
16676 for the mode iterator. Require the mode specified by max_pieces
16678 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
16680 (copy_blkmode_to_reg): Likewise.
16681 (set_storage_via_setmem): Likewise.
16682 * optabs.c (prepare_cmp_insn): Likewise.
16683 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16684 * stor-layout.c (finish_bitfield_representative): Likewise.
16686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16687 Alan Hayward <alan.hayward@arm.com>
16688 David Sherwood <david.sherwood@arm.com>
16690 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
16691 * expr.c (convert_move): Use them.
16692 (convert_modes): Likewise.
16693 (store_expr_with_bounds): Likewise.
16695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16696 Alan Hayward <alan.hayward@arm.com>
16697 David Sherwood <david.sherwood@arm.com>
16699 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
16700 parameter for the mode of "x". Remove the "known_x", "known_mode"
16701 and "known_ret" arguments. Change the type of the mode argument
16702 to scalar_int_mode.
16703 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
16704 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
16705 (reg_num_sign_bit_copies_for_combine): Likewise.
16706 * rtlanal.c (nonzero_bits1): Likewise.
16707 (num_sign_bit_copies1): Likewise.
16708 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
16709 (reg_num_sign_bit_copies_general): Likewise.
16710 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
16711 (reg_nonzero_bits_general): Likewise.
16713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16714 Alan Hayward <alan.hayward@arm.com>
16715 David Sherwood <david.sherwood@arm.com>
16717 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
16718 than in subroutines. Return 1 for non-integer modes.
16719 (cached_num_sign_bit_copies): Change the type of the mode parameter
16720 to scalar_int_mode.
16721 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
16722 classes. Handle CONST_INT_P first and then check whether X also
16723 has a scalar integer mode. Check the same thing for inner registers
16724 of a SUBREG and for values that are being extended or truncated.
16726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16727 Alan Hayward <alan.hayward@arm.com>
16728 David Sherwood <david.sherwood@arm.com>
16730 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
16731 in subroutines. Return the mode mask for non-integer modes.
16732 (cached_nonzero_bits): Change the type of the mode parameter
16733 to scalar_int_mode.
16734 (nonzero_bits1): Likewise. Remove early exit for other mode
16735 classes. Handle CONST_INT_P first and then check whether X
16736 also has a scalar integer mode.
16738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16739 Alan Hayward <alan.hayward@arm.com>
16740 David Sherwood <david.sherwood@arm.com>
16742 * expr.c (widest_int_mode_for_size): Make the comment match the code.
16743 Return a scalar_int_mode and assert that the size is greater than
16745 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
16746 (op_by_pieces_d::op_by_pieces_d): Likewise.
16747 (op_by_pieces_d::run): Likewise.
16748 (can_store_by_pieces): Likewise.
16750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16751 Alan Hayward <alan.hayward@arm.com>
16752 David Sherwood <david.sherwood@arm.com>
16754 * combine.c (extract_left_shift): Add a mode argument and update
16756 (make_compound_operation_int): Change the type of the mode parameter
16757 to scalar_int_mode and update the call to extract_left_shift.
16759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16760 Alan Hayward <alan.hayward@arm.com>
16761 David Sherwood <david.sherwood@arm.com>
16763 * combine.c (simplify_and_const_int): Change the type of the mode
16764 parameter to scalar_int_mode.
16765 (simplify_and_const_int_1): Likewise. Update recursive call.
16767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16768 Alan Hayward <alan.hayward@arm.com>
16769 David Sherwood <david.sherwood@arm.com>
16771 * combine.c (simplify_compare_const): Check that the mode is a
16772 scalar_int_mode (rather than VOIDmode) before testing its
16774 (simplify_comparison): Move COMPARISON_P handling out of the
16775 loop and restrict the latter part of the loop to scalar_int_modes.
16776 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
16777 and when considering SUBREG_REGs. Use is_int_mode instead of
16778 checking GET_MODE_CLASS against MODE_INT.
16780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16781 Alan Hayward <alan.hayward@arm.com>
16782 David Sherwood <david.sherwood@arm.com>
16784 * combine.c (try_widen_shift_mode): Move check for equal modes to...
16785 (simplify_shift_const_1): ...here. Use scalar_int_mode for
16786 shift_unit_mode and for modes involved in scalar shifts.
16788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16789 Alan Hayward <alan.hayward@arm.com>
16790 David Sherwood <david.sherwood@arm.com>
16792 * combine.c (force_int_to_mode): New function, split out from...
16793 (force_to_mode): ...here. Keep xmode up-to-date and use it
16794 instead of GET_MODE (x).
16796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16797 Alan Hayward <alan.hayward@arm.com>
16798 David Sherwood <david.sherwood@arm.com>
16800 * optabs-query.h (extraction_insn::struct_mode): Change type to
16801 opt_scalar_int_mode and update comment.
16802 (extraction_insn::field_mode): Change type to scalar_int_mode.
16803 (extraction_insn::pos_mode): Likewise.
16804 * combine.c (make_extraction): Update accordingly.
16805 * optabs-query.c (get_traditional_extraction_insn): Likewise.
16806 (get_optab_extraction_insn): Likewise.
16807 * recog.c (simplify_while_replacing): Likewise.
16808 * expmed.c (narrow_bit_field_mem): Change the type of the mode
16809 parameter to opt_scalar_int_mode.
16811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16812 Alan Hayward <alan.hayward@arm.com>
16813 David Sherwood <david.sherwood@arm.com>
16815 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
16816 to a scalar_int_mode instead of a machine_mode.
16817 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
16818 (get_best_mode): Return a boolean and use a pointer argument to store
16819 the selected mode. Replace the limit mode parameter with a bit limit.
16820 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
16821 for the values returned by bit_field_mode_iterator::next_mode.
16822 (store_bit_field): Update call to get_best_mode.
16823 (store_fixed_bit_field): Likewise.
16824 (extract_fixed_bit_field): Likewise.
16825 * expr.c (optimize_bitfield_assignment_op): Likewise.
16826 * fold-const.c (optimize_bit_field_compare): Likewise.
16827 (fold_truth_andor_1): Likewise.
16828 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
16829 Update for new type of m_mode.
16830 (get_best_mode): As above.
16832 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16833 Alan Hayward <alan.hayward@arm.com>
16834 David Sherwood <david.sherwood@arm.com>
16836 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
16837 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
16838 (store_bit_field): Check is_a <scalar_int_mode> before calling
16839 strict_volatile_bitfield_p.
16840 (extract_bit_field): Likewise.
16842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16843 Alan Hayward <alan.hayward@arm.com>
16844 David Sherwood <david.sherwood@arm.com>
16846 * target.def (cstore_mode): Return a scalar_int_mode.
16847 * doc/tm.texi: Regenerate.
16848 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
16849 * targhooks.h (default_cstore_mode): Likewise.
16850 * targhooks.c (default_cstore_mode): Likewise, using a forced
16852 * expmed.c (emit_cstore): Expect the target of the cstore to be
16855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16856 Alan Hayward <alan.hayward@arm.com>
16857 David Sherwood <david.sherwood@arm.com>
16859 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
16861 (niter_desc): Likewise mode.
16862 (iv_analyze): Add a mode parameter.
16864 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
16865 and change its type to scalar_int_mode.
16866 * loop-iv.c: Update commentary at head of file.
16867 (iv_constant): Pass the mode paraeter before the rtx it describes
16868 and change its type to scalar_int_mode. Remove VOIDmode handling.
16869 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
16870 (iv_extend): Likewise.
16871 (shorten_into_mode): Likewise.
16872 (iv_add): Use scalar_int_mode.
16873 (iv_mult): Likewise.
16874 (iv_shift): Likewise.
16875 (canonicalize_iv_subregs): Likewise.
16876 (get_biv_step_1): Pass the outer_mode parameter before the rtx
16877 it describes and change its mode to scalar_int_mode. Also change
16878 the type of the returned inner_mode to scalar_int_mode.
16879 (get_biv_step): Likewise, turning outer_mode from a pointer
16880 into a direct parameter. Update call to get_biv_step_1.
16881 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
16882 iv_constant and get_biv_step.
16883 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
16884 and change its type to scalar_int_mode. Don't initialise iv->mode
16885 to VOIDmode and remove later checks for its still being VOIDmode.
16886 Update calls to iv_analyze_op and iv_analyze_expr. Check
16887 is_a <scalar_int_mode> when changing the mode under consideration.
16888 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
16889 Update call to iv_analyze_expr.
16890 (iv_analyze_op): Add a mode parameter. Reject subregs whose
16891 inner register is not also a scalar_int_mode. Update call to
16893 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
16894 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
16895 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
16896 separate mode class checks. Update calls to iv_analyze. Remove
16897 fix-up of VOIDmodes after iv_analyze_biv.
16898 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
16899 don't have a scalar_int_mode. Update call to biv_p.
16901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16902 Alan Hayward <alan.hayward@arm.com>
16903 David Sherwood <david.sherwood@arm.com>
16905 * cfgexpand.c (convert_debug_memory_address): Use
16906 as_a <scalar_int_mode>.
16907 * combine.c (expand_compound_operation): Likewise.
16908 (make_extraction): Likewise.
16909 (change_zero_ext): Likewise.
16910 (simplify_comparison): Likewise.
16911 * cse.c (cse_insn): Likewise.
16912 * dwarf2out.c (minmax_loc_descriptor): Likewise.
16913 (mem_loc_descriptor): Likewise.
16914 (loc_descriptor): Likewise.
16915 * expmed.c (init_expmed_one_mode): Likewise.
16916 (synth_mult): Likewise.
16917 (emit_store_flag_1): Likewise.
16918 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
16919 of a comparison with size.
16920 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
16921 (reduce_to_bit_field_precision): Likewise.
16922 * function.c (expand_function_end): Likewise.
16923 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
16924 * loop-doloop.c (doloop_modify): Likewise.
16925 * optabs.c (expand_binop): Likewise.
16926 (expand_unop): Likewise.
16927 (expand_copysign_absneg): Likewise.
16928 (prepare_cmp_insn): Likewise.
16929 (maybe_legitimize_operand): Likewise.
16930 * recog.c (const_scalar_int_operand): Likewise.
16931 * rtlanal.c (get_address_mode): Likewise.
16932 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16933 (simplify_cond_clz_ctz): Likewise.
16934 * tree-nested.c (get_nl_goto_field): Likewise.
16935 * tree.c (build_vector_type_for_mode): Likewise.
16936 * var-tracking.c (use_narrower_mode): Likewise.
16938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16939 Alan Hayward <alan.hayward@arm.com>
16940 David Sherwood <david.sherwood@arm.com>
16942 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
16943 * builtins.c (expand_builtin_signbit): Use it.
16944 * cfgexpand.c (expand_debug_expr): Likewise.
16945 * dojump.c (do_jump): Likewise.
16946 (do_compare_and_jump): Likewise.
16947 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
16948 * expmed.c (make_tree): Likewise.
16949 * expr.c (expand_expr_real_2): Likewise.
16950 (expand_expr_real_1): Likewise.
16951 (try_casesi): Likewise.
16952 * fold-const-call.c (fold_const_call_ss): Likewise.
16953 * fold-const.c (unextend): Likewise.
16954 (extract_muldiv_1): Likewise.
16955 (fold_single_bit_test): Likewise.
16956 (native_encode_int): Likewise.
16957 (native_encode_string): Likewise.
16958 (native_interpret_int): Likewise.
16959 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
16960 * internal-fn.c (expand_addsub_overflow): Likewise.
16961 (expand_neg_overflow): Likewise.
16962 (expand_mul_overflow): Likewise.
16963 (expand_arith_overflow): Likewise.
16964 * match.pd: Likewise.
16965 * stor-layout.c (layout_type): Likewise.
16966 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16967 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
16968 * tree-ssanames.c (get_range_info): Likewise.
16969 * tree-switch-conversion.c (array_value_type) Likewise.
16970 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
16971 (vect_recog_divmod_pattern): Likewise.
16972 (vect_recog_mixed_size_cond_pattern): Likewise.
16973 * tree-vrp.c (extract_range_basic): Likewise.
16974 (simplify_float_conversion_using_ranges): Likewise.
16975 * tree.c (int_fits_type_p): Likewise.
16976 * ubsan.c (instrument_bool_enum_load): Likewise.
16977 * varasm.c (mergeable_string_section): Likewise.
16978 (narrowing_initializer_constant_valid_p): Likewise.
16979 (output_constant): Likewise.
16981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16982 Alan Hayward <alan.hayward@arm.com>
16983 David Sherwood <david.sherwood@arm.com>
16985 * machmode.h (NARROWEST_INT_MODE): New macro.
16986 * expr.c (alignment_for_piecewise_move): Use it instead of
16987 GET_CLASS_NARROWEST_MODE (MODE_INT).
16988 (push_block): Likewise.
16989 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
16991 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16994 Alan Hayward <alan.hayward@arm.com>
16995 David Sherwood <david.sherwood@arm.com>
16997 * postreload.c (move2add_valid_value_p): Change the type of the
16998 mode parameter to scalar_int_mode.
16999 (move2add_use_add2_insn): Add a mode parameter and use it instead
17001 (move2add_use_add3_insn): Likewise.
17002 (reload_cse_move2add): Update accordingly.
17004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17005 Alan Hayward <alan.hayward@arm.com>
17006 David Sherwood <david.sherwood@arm.com>
17008 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17010 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17011 * optabs.c (expand_unop): Likewise.
17013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17014 Alan Hayward <alan.hayward@arm.com>
17015 David Sherwood <david.sherwood@arm.com>
17017 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17018 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17019 (popcount_loc_descriptor): Likewise.
17020 (bswap_loc_descriptor): Likewise.
17021 (rotate_loc_descriptor): Likewise.
17022 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17023 calling the functions above.
17025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17026 Alan Hayward <alan.hayward@arm.com>
17027 David Sherwood <david.sherwood@arm.com>
17029 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17031 (try_combine): Likewise.
17032 (simplify_if_then_else): Likewise.
17033 * cse.c (cse_insn): Likewise.
17034 * dwarf2out.c (mem_loc_descriptor): Likewise.
17035 * emit-rtl.c (gen_lowpart_common): Likewise.
17036 * simplify-rtx.c (simplify_truncation): Likewise.
17037 (simplify_binary_operation_1): Likewise.
17038 (simplify_const_relational_operation): Likewise.
17039 (simplify_ternary_operation): Likewise.
17040 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
17042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17043 Alan Hayward <alan.hayward@arm.com>
17044 David Sherwood <david.sherwood@arm.com>
17046 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
17047 * reload.c (push_reload): Likewise.
17048 (find_reloads): Likewise.
17050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17051 Alan Hayward <alan.hayward@arm.com>
17052 David Sherwood <david.sherwood@arm.com>
17054 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
17055 (make_compound_operation_int): Likewise.
17056 (change_zero_ext): Likewise.
17057 * expr.c (convert_move): Likewise.
17058 (convert_modes): Likewise.
17059 * fwprop.c (forward_propagate_subreg): Likewise.
17060 * loop-iv.c (get_biv_step_1): Likewise.
17061 * optabs.c (widen_operand): Likewise.
17062 * postreload.c (move2add_valid_value_p): Likewise.
17063 * recog.c (simplify_while_replacing): Likewise.
17064 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17065 (simplify_binary_operation_1): Likewise. Remove redundant
17066 mode equality check.
17068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17069 Alan Hayward <alan.hayward@arm.com>
17070 David Sherwood <david.sherwood@arm.com>
17072 * combine.c (combine_simplify_rtx): Add checks for
17073 is_a <scalar_int_mode>.
17074 (simplify_if_then_else): Likewise.
17075 (make_field_assignment): Likewise.
17076 (simplify_comparison): Likewise.
17077 * ifcvt.c (noce_try_bitop): Likewise.
17078 * loop-invariant.c (canonicalize_address_mult): Likewise.
17079 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17082 Alan Hayward <alan.hayward@arm.com>
17083 David Sherwood <david.sherwood@arm.com>
17085 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
17086 is_a <scalar_int_mode> instead of != BLKmode.
17088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17089 Alan Hayward <alan.hayward@arm.com>
17090 David Sherwood <david.sherwood@arm.com>
17092 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
17093 instead of != VOIDmode.
17094 * combine.c (if_then_else_cond): Likewise.
17095 (change_zero_ext): Likewise.
17096 * dwarf2out.c (mem_loc_descriptor): Likewise.
17097 (loc_descriptor): Likewise.
17098 * rtlanal.c (canonicalize_condition): Likewise.
17099 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
17101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17102 Alan Hayward <alan.hayward@arm.com>
17103 David Sherwood <david.sherwood@arm.com>
17105 * simplify-rtx.c (simplify_binary_operation_1): Use
17106 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
17108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17109 Alan Hayward <alan.hayward@arm.com>
17110 David Sherwood <david.sherwood@arm.com>
17112 * wide-int.h (int_traits<unsigned char>) New class.
17113 (int_traits<unsigned short>) Likewise.
17114 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
17115 Use GET_MODE_UNIT_PRECISION and remove redundant test for
17117 * combine.c (set_nonzero_bits_and_sign_copies): Use
17118 is_a <scalar_int_mode>.
17119 (find_split_point): Likewise.
17120 (combine_simplify_rtx): Likewise.
17121 (simplify_logical): Likewise.
17122 (expand_compound_operation): Likewise.
17123 (expand_field_assignment): Likewise.
17124 (make_compound_operation): Likewise.
17125 (extended_count): Likewise.
17126 (change_zero_ext): Likewise.
17127 (simplify_comparison): Likewise.
17128 * dwarf2out.c (scompare_loc_descriptor): Likewise.
17129 (ucompare_loc_descriptor): Likewise.
17130 (minmax_loc_descriptor): Likewise.
17131 (mem_loc_descriptor): Likewise.
17132 (loc_descriptor): Likewise.
17133 * expmed.c (init_expmed_one_mode): Likewise.
17134 * lra-constraints.c (lra_constraint_offset): Likewise.
17135 * optabs.c (prepare_libcall_arg): Likewise.
17136 * postreload.c (move2add_note_store): Likewise.
17137 * reload.c (operands_match_p): Likewise.
17138 * rtl.h (load_extend_op): Likewise.
17139 * rtlhooks.c (gen_lowpart_general): Likewise.
17140 * simplify-rtx.c (simplify_truncation): Likewise.
17141 (simplify_unary_operation_1): Likewise.
17142 (simplify_binary_operation_1): Likewise.
17143 (simplify_const_binary_operation): Likewise.
17144 (simplify_const_relational_operation): Likewise.
17145 (simplify_subreg): Likewise.
17146 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17147 * var-tracking.c (adjust_mems): Likewise.
17148 (prepare_call_arguments): Likewise.
17150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17151 Alan Hayward <alan.hayward@arm.com>
17152 David Sherwood <david.sherwood@arm.com>
17154 * machmode.h (is_int_mode): New fuction.
17155 * combine.c (find_split_point): Use it.
17156 (combine_simplify_rtx): Likewise.
17157 (simplify_if_then_else): Likewise.
17158 (simplify_set): Likewise.
17159 (simplify_shift_const_1): Likewise.
17160 (simplify_comparison): Likewise.
17161 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
17162 * cse.c (notreg_cost): Likewise.
17163 (cse_insn): Likewise.
17164 * cselib.c (cselib_lookup_1): Likewise.
17165 * dojump.c (do_jump_1): Likewise.
17166 (do_compare_rtx_and_jump): Likewise.
17167 * dse.c (get_call_args): Likewise.
17168 * dwarf2out.c (rtl_for_decl_init): Likewise.
17169 (native_encode_initializer): Likewise.
17170 * expmed.c (emit_store_flag_1): Likewise.
17171 (emit_store_flag): Likewise.
17172 * expr.c (convert_modes): Likewise.
17173 (store_field): Likewise.
17174 (expand_expr_real_1): Likewise.
17175 * fold-const.c (fold_read_from_constant_string): Likewise.
17176 * gimple-ssa-sprintf.c (get_format_string): Likewise.
17177 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
17178 * optabs.c (expand_binop): Likewise.
17179 (expand_unop): Likewise.
17180 (expand_abs_nojump): Likewise.
17181 (expand_one_cmpl_abs_nojump): Likewise.
17182 * simplify-rtx.c (mode_signbit_p): Likewise.
17183 (val_signbit_p): Likewise.
17184 (val_signbit_known_set_p): Likewise.
17185 (val_signbit_known_clear_p): Likewise.
17186 (simplify_relational_operation_1): Likewise.
17187 * tree.c (vector_type_mode): Likewise.
17189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17190 Alan Hayward <alan.hayward@arm.com>
17191 David Sherwood <david.sherwood@arm.com>
17193 * machmode.h (smallest_mode_for_size): Fix formatting.
17194 (smallest_int_mode_for_size): New function.
17195 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
17196 instead of smallest_mode_for_size.
17197 * combine.c (make_extraction): Likewise.
17198 * config/arc/arc.c (arc_expand_movmem): Likewise.
17199 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
17200 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17201 * config/s390/s390.c (s390_expand_insv): Likewise.
17202 * config/sparc/sparc.c (assign_int_registers): Likewise.
17203 * config/spu/spu.c (spu_function_value): Likewise.
17204 (spu_function_arg): Likewise.
17205 * coverage.c (get_gcov_type): Likewise.
17206 (get_gcov_unsigned_t): Likewise.
17207 * dse.c (find_shift_sequence): Likewise.
17208 * expmed.c (store_bit_field_1): Likewise.
17209 * expr.c (convert_move): Likewise.
17210 (store_field): Likewise.
17211 * internal-fn.c (expand_arith_overflow): Likewise.
17212 * optabs-query.c (get_best_extraction_insn): Likewise.
17213 * optabs.c (expand_twoval_binop_libfunc): Likewise.
17214 * stor-layout.c (layout_type): Likewise.
17215 (initialize_sizetypes): Likewise.
17216 * targhooks.c (default_get_mask_mode): Likewise.
17217 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
17219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17220 Alan Hayward <alan.hayward@arm.com>
17221 David Sherwood <david.sherwood@arm.com>
17223 * machmode.h (opt_mode::else_blk): New function.
17224 (int_mode_for_mode): Declare.
17225 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
17226 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
17228 * cfgexpand.c (expand_debug_expr): Likewise.
17229 * combine.c (gen_lowpart_or_truncate): Likewise.
17230 (gen_lowpart_for_combine): Likewise.
17231 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
17232 * config/avr/avr.c (avr_to_int_mode): Likewise.
17233 (avr_out_plus_1): Likewise.
17234 (avr_out_plus): Likewise.
17235 (avr_out_round): Likewise.
17236 * config/i386/i386.c (ix86_split_to_parts): Likewise.
17237 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17238 (s390_expand_vcond): Likewise.
17239 * config/spu/spu.c (spu_split_immediate): Likewise.
17240 (spu_expand_mov): Likewise.
17241 * dse.c (get_stored_val): Likewise.
17242 * expmed.c (store_bit_field_1): Likewise.
17243 (convert_extracted_bit_field): Use int_mode_for_mode instead of
17245 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
17246 (extract_low_bits): Likewise.
17247 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
17248 handling rather than repeating the check.
17249 (emit_group_store): Likewise.
17250 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
17251 * optabs.c (expand_absneg_bit): Likewise.
17252 (expand_copysign_absneg): Likewise.
17253 (expand_copysign_bit): Likewise.
17254 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
17255 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
17256 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
17257 * var-tracking.c (prepare_call_arguments): Likewise.
17258 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
17259 int_mode_for_mode instead of mode_for_size.
17260 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17263 Alan Hayward <alan.hayward@arm.com>
17264 David Sherwood <david.sherwood@arm.com>
17266 * machmode.h (int_mode_for_size): New function.
17267 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
17268 instead of mode_for_size.
17269 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
17271 * combine.c (expand_field_assignment): Use int_mode_for_size
17272 instead of mode_for_size.
17273 (make_extraction): Likewise.
17274 (simplify_shift_const_1): Likewise.
17275 (simplify_comparison): Likewise.
17276 * dojump.c (do_jump): Likewise.
17277 * dwarf2out.c (mem_loc_descriptor): Likewise.
17278 * emit-rtl.c (init_derived_machine_modes): Likewise.
17279 * expmed.c (flip_storage_order): Likewise.
17280 (convert_extracted_bit_field): Likewise.
17281 * expr.c (copy_blkmode_from_reg): Likewise.
17282 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
17283 * internal-fn.c (expand_mul_overflow): Likewise.
17284 * lower-subreg.c (simple_move): Likewise.
17285 * optabs-libfuncs.c (init_optabs): Likewise.
17286 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17287 * tree.c (vector_type_mode): Likewise.
17288 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
17289 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
17290 * tree-vect-generic.c (expand_vector_parallel): Likewise.
17291 * tree-vect-stmts.c (vectorizable_load): Likewise.
17292 (vectorizable_store): Likewise.
17294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17295 Alan Hayward <alan.hayward@arm.com>
17296 David Sherwood <david.sherwood@arm.com>
17298 * coretypes.h (pod_mode): New type.
17299 (scalar_int_mode_pod): New typedef.
17300 * machmode.h (pod_mode): New class.
17301 (int_n_data_t::m): Change type to scalar_int_mode_pod.
17302 * genmodes.c (emit_mode_int_n): Update accordingly.
17303 * lower-subreg.h (target_lower_subreg): Change type to
17304 scalar_int_mode_pod.
17305 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
17306 scalar_int_mode_pod.
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 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
17313 machine_mode to scalar_int_mode.
17314 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
17315 (rs6000_option_override_internal): Remove cast to int.
17316 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
17317 machine_mode to scalar_int_mode.
17318 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
17319 (rs6000_option_override_internal): Remove cast to int.
17320 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
17321 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
17323 * config/s390/s390.c (s390_expand_builtin): Likewise.
17324 * coretypes.h (scalar_int_mode): New type.
17325 (opt_scalar_int_mode): New typedef.
17326 * machmode.h (scalar_int_mode): New class.
17327 (scalar_int_mode::includes_p): New function.
17328 (byte_mode): Change type to scalar_int_mode.
17329 (word_mode): Likewise.
17330 (ptr_mode): Likewise.
17331 * emit-rtl.c (byte_mode): Likewise.
17332 (word_mode): Likewise.
17333 (ptr_mode): Likewise.
17334 (init_derived_machine_modes): Update accordingly.
17335 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
17336 and MODE_PARTIAL_INT.
17337 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
17338 opt_scalar_int_mode.
17340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17341 Alan Hayward <alan.hayward@arm.com>
17342 David Sherwood <david.sherwood@arm.com>
17344 * target.def (libgcc_floating_mode_supported_p): Take a
17346 * doc/tm.texi: Regenerate.
17347 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
17349 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17350 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
17353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17354 Alan Hayward <alan.hayward@arm.com>
17355 David Sherwood <david.sherwood@arm.com>
17357 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
17358 * doc/tm.texi: Regenerate.
17359 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
17360 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
17361 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
17362 * targhooks.h (default_floatn_mode): Likewise.
17363 * targhooks.c (default_floatn_mode): Likewise.
17364 * tree.c (build_common_tree_nodes): Update accordingly.
17366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17367 Alan Hayward <alan.hayward@arm.com>
17368 David Sherwood <david.sherwood@arm.com>
17370 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
17371 (mode_iterator::iterate_p): Likewise.
17372 (mode_iterator::get_wider): Likewise.
17373 * expr.c (init_expr_target): Use opt_scalar_float_mode.
17375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17376 Alan Hayward <alan.hayward@arm.com>
17377 David Sherwood <david.sherwood@arm.com>
17379 * coretypes.h (opt_scalar_float_mode): New typedef.
17380 * machmode.h (float_mode_for_size): New function.
17381 * emit-rtl.c (double_mode): Delete.
17382 (init_emit_once): Use float_mode_for_size.
17383 * stor-layout.c (layout_type): Likewise.
17384 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
17386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17387 Alan Hayward <alan.hayward@arm.com>
17388 David Sherwood <david.sherwood@arm.com>
17390 * output.h (assemble_real): Take a scalar_float_mode.
17391 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
17392 * config/arm/arm.md (consttable_4): Likewise.
17393 (consttable_8): Likewise.
17394 (consttable_16): Likewise.
17395 * config/mips/mips.md (consttable_float): Likewise.
17396 * config/s390/s390.c (s390_output_pool_entry): Likewise.
17397 * varasm.c (assemble_real): Take a scalar_float_mode.
17398 (output_constant_pool_2): Update accordingly.
17399 (output_constant): Likewise.
17401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17402 Alan Hayward <alan.hayward@arm.com>
17403 David Sherwood <david.sherwood@arm.com>
17405 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
17406 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
17407 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
17408 (native_encode_real): Likewise.
17409 (native_interpret_real): Likewise.
17410 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
17411 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17414 Alan Hayward <alan.hayward@arm.com>
17415 David Sherwood <david.sherwood@arm.com>
17417 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
17418 <scalar_float_mode>. Simplify.
17419 (gen_extend_conv_libfunc): Likewise.
17421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17422 Alan Hayward <alan.hayward@arm.com>
17423 David Sherwood <david.sherwood@arm.com>
17425 * coretypes.h (scalar_float_mode): New type.
17426 * machmode.h (mode_traits::from_int): Use machine_mode if
17427 USE_ENUM_MODES is defined.
17428 (is_a): New function.
17430 (dyn_cast): Likewise.
17431 (scalar_float_mode): New class.
17432 (scalar_float_mode::includes_p): New function.
17433 (is_float_mode): Likewise.
17434 * gdbhooks.py (MachineModePrinter): New class.
17435 (build_pretty_printer): Use it for scalar_float_mode.
17436 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
17437 (format_helper::format_helper): Turn into a template.
17438 * genmodes.c (get_mode_class): New function.
17439 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
17440 or machine_mode if none.
17441 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
17442 as_a <scalar_float_mode>.
17443 * dwarf2out.c (mem_loc_descriptor): Likewise.
17444 (insert_float): Likewise.
17445 (add_const_value_attribute): Likewise.
17446 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17447 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
17448 (expand_unop): Update accordingly.
17449 (expand_abs_nojump): Likewise.
17450 (expand_copysign_absneg): Take a scalar_float_mode.
17451 (expand_copysign_bit): Likewise.
17452 (expand_copysign): Update accordingly.
17454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17455 Alan Hayward <alan.hayward@arm.com>
17456 David Sherwood <david.sherwood@arm.com>
17458 * coretypes.h (opt_mode): New class.
17459 * machmode.h (opt_mode): Likewise.
17460 (opt_mode::else_void): New function.
17461 (opt_mode::require): Likewise.
17462 (opt_mode::exists): Likewise.
17463 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
17464 (GET_MODE_2XWIDER_MODE): Likewise.
17465 (mode_iterator::get_wider): Update accordingly.
17466 (mode_iterator::get_2xwider): Likewise.
17467 (mode_iterator::get_known_wider): Likewise, turning into a template.
17468 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
17469 forcing a wider mode to exist.
17470 * config/cr16/cr16.h (LONG_REG_P): Likewise.
17471 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17472 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
17473 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
17474 * lower-subreg.c (init_lower_subreg): Likewise.
17475 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
17476 on the final iteration.
17477 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
17478 a wider mode exists before asking for a move pattern.
17479 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
17480 forcing a wider mode to exist.
17481 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
17482 returning false if no such mode exists.
17483 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
17484 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
17485 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
17486 Avoid checking for a MODE_INT if we already know the mode is not a
17488 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
17489 forcing a wider mode to exist.
17490 (expmed_mult_highpart_optab): Likewise.
17491 (expmed_mult_highpart): Likewise.
17492 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
17494 * lto-streamer-in.c (lto_input_mode_table): Likewise.
17495 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
17496 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
17497 * internal-fn.c (expand_mul_overflow): Update use of
17498 GET_MODE_2XWIDER_MODE.
17499 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17500 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
17501 GET_MODE_WIDER_MODE.
17502 (convert_plusminus_to_widen): Likewise.
17503 * tree-switch-conversion.c (array_value_type): Likewise.
17504 * var-tracking.c (emit_note_insn_var_location): Likewise.
17505 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17506 Return false inside rather than outside the loop if no wider mode
17508 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
17509 and GET_MODE_2XWIDER_MODE
17510 (can_compare_p): Use else_void.
17511 * gdbhooks.py (OptMachineModePrinter): New class.
17512 (build_pretty_printer): Use it for opt_mode.
17514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17515 Alan Hayward <alan.hayward@arm.com>
17516 David Sherwood <david.sherwood@arm.com>
17518 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
17519 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
17521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17522 Alan Hayward <alan.hayward@arm.com>
17523 David Sherwood <david.sherwood@arm.com>
17525 * machmode.h (mode_traits): New structure.
17526 (get_narrowest_mode): New function.
17527 (mode_iterator::start): Likewise.
17528 (mode_iterator::iterate_p): Likewise.
17529 (mode_iterator::get_wider): Likewise.
17530 (mode_iterator::get_known_wider): Likewise.
17531 (mode_iterator::get_2xwider): Likewise.
17532 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
17533 (FOR_EACH_MODE): Likewise.
17534 (FOR_EACH_MODE_FROM): Likewise.
17535 (FOR_EACH_MODE_UNTIL): Likewise.
17536 (FOR_EACH_WIDER_MODE): Likewise.
17537 (FOR_EACH_2XWIDER_MODE): Likewise.
17538 * builtins.c (expand_builtin_strlen): Use new mode iterators.
17539 * combine.c (simplify_comparison): Likewise
17540 * config/i386/i386.c (type_natural_mode): Likewise.
17541 * cse.c (cse_insn): Likewise.
17542 * dse.c (find_shift_sequence): Likewise.
17543 * emit-rtl.c (init_derived_machine_modes): Likewise.
17544 (init_emit_once): Likewise.
17545 * explow.c (hard_function_value): Likewise.
17546 * expmed.c (extract_fixed_bit_field_1): Likewise.
17547 (extract_bit_field_1): Likewise.
17548 (expand_divmod): Likewise.
17549 (emit_store_flag_1): Likewise.
17550 * expr.c (init_expr_target): Likewise.
17551 (convert_move): Likewise.
17552 (alignment_for_piecewise_move): Likewise.
17553 (widest_int_mode_for_size): Likewise.
17554 (emit_block_move_via_movmem): Likewise.
17555 (copy_blkmode_to_reg): Likewise.
17556 (set_storage_via_setmem): Likewise.
17557 (compress_float_constant): Likewise.
17558 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17559 * optabs-query.c (get_best_extraction_insn): Likewise.
17560 * optabs.c (expand_binop): Likewise.
17561 (expand_twoval_unop): Likewise.
17562 (expand_twoval_binop): Likewise.
17563 (widen_leading): Likewise.
17564 (widen_bswap): Likewise.
17565 (expand_parity): Likewise.
17566 (expand_unop): Likewise.
17567 (prepare_cmp_insn): Likewise.
17568 (prepare_float_lib_cmp): Likewise.
17569 (expand_float): Likewise.
17570 (expand_fix): Likewise.
17571 (expand_sfix_optab): Likewise.
17572 * postreload.c (move2add_use_add2_insn): Likewise.
17573 * reg-stack.c (reg_to_stack): Likewise.
17574 * reginfo.c (choose_hard_reg_mode): Likewise.
17575 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17576 * stor-layout.c (mode_for_size): Likewise.
17577 (smallest_mode_for_size): Likewise.
17578 (mode_for_vector): Likewise.
17579 (finish_bitfield_representative): Likewise.
17580 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
17581 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
17582 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17583 * var-tracking.c (prepare_call_arguments): Likewise.
17585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17586 Alan Hayward <alan.hayward@arm.com>
17587 David Sherwood <david.sherwood@arm.com>
17589 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
17590 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
17591 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
17592 * machmode.h (mode_size): Move earlier in file.
17593 (mode_precision): Likewise.
17594 (mode_inner): Likewise.
17595 (mode_nunits): Likewise.
17596 (mode_unit_size): Likewise.
17597 (unit_unit_precision): Likewise.
17598 (mode_wider): Likewise.
17599 (mode_2xwider): Likewise.
17600 (machine_mode): New class.
17601 (mode_to_bytes): New function.
17602 (mode_to_bits): Likewise.
17603 (mode_to_precision): Likewise.
17604 (mode_to_inner): Likewise.
17605 (mode_to_unit_size): Likewise.
17606 (mode_to_unit_precision): Likewise.
17607 (mode_to_nunits): Likewise.
17608 (GET_MODE_SIZE): Use mode_to_bytes.
17609 (GET_MODE_BITSIZE): Use mode_to_bits.
17610 (GET_MODE_PRECISION): Use mode_to_precision.
17611 (GET_MODE_INNER): Use mode_to_inner.
17612 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
17613 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
17614 (GET_MODE_NUNITS): Use mode_to_nunits.
17615 * system.h (ALWAYS_INLINE): New macro.
17616 * config/powerpcspe/powerpcspe-c.c
17617 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
17618 int for arg1_mode and arg2_mode.
17620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17621 Alan Hayward <alan.hayward@arm.com>
17622 David Sherwood <david.sherwood@arm.com>
17624 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17625 Prefix mode names with E_ in case statements.
17626 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17627 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
17628 (aarch64_split_simd_move): Likewise.
17629 (aarch64_gen_storewb_pair): Likewise.
17630 (aarch64_gen_loadwb_pair): Likewise.
17631 (aarch64_gen_store_pair): Likewise.
17632 (aarch64_gen_load_pair): Likewise.
17633 (aarch64_get_condition_code_1): Likewise.
17634 (aarch64_constant_pool_reload_icode): Likewise.
17635 (get_rsqrte_type): Likewise.
17636 (get_rsqrts_type): Likewise.
17637 (get_recpe_type): Likewise.
17638 (get_recps_type): Likewise.
17639 (aarch64_gimplify_va_arg_expr): Likewise.
17640 (aarch64_simd_container_mode): Likewise.
17641 (aarch64_emit_load_exclusive): Likewise.
17642 (aarch64_emit_store_exclusive): Likewise.
17643 (aarch64_expand_compare_and_swap): Likewise.
17644 (aarch64_gen_atomic_cas): Likewise.
17645 (aarch64_emit_bic): Likewise.
17646 (aarch64_emit_atomic_swap): Likewise.
17647 (aarch64_emit_atomic_load_op): Likewise.
17648 (aarch64_evpc_trn): Likewise.
17649 (aarch64_evpc_uzp): Likewise.
17650 (aarch64_evpc_zip): Likewise.
17651 (aarch64_evpc_ext): Likewise.
17652 (aarch64_evpc_rev): Likewise.
17653 (aarch64_evpc_dup): Likewise.
17654 (aarch64_gen_ccmp_first): Likewise.
17655 (aarch64_gen_ccmp_next): Likewise.
17656 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17657 (alpha_emit_xfloating_libcall): Likewise.
17658 (emit_insxl): Likewise.
17659 (alpha_arg_type): Likewise.
17660 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
17661 (arc_preferred_simd_mode): Likewise.
17662 (arc_secondary_reload): Likewise.
17663 (get_arc_condition_code): Likewise.
17664 (arc_print_operand): Likewise.
17665 (arc_legitimate_constant_p): Likewise.
17666 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17667 * config/arc/arc.md (casesi_load): Likewise.
17668 (casesi_compact_jump): Likewise.
17669 * config/arc/predicates.md (proper_comparison_operator): Likewise.
17670 (cc_use_register): Likewise.
17671 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17672 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17673 (arm_init_iwmmxt_builtins): Likewise.
17674 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
17675 (neon_expand_vector_init): Likewise.
17676 (arm_attr_length_move_neon): Likewise.
17677 (maybe_get_arm_condition_code): Likewise.
17678 (arm_emit_vector_const): Likewise.
17679 (arm_preferred_simd_mode): Likewise.
17680 (arm_output_iwmmxt_tinsr): Likewise.
17681 (thumb1_output_casesi): Likewise.
17682 (thumb2_output_casesi): Likewise.
17683 (arm_emit_load_exclusive): Likewise.
17684 (arm_emit_store_exclusive): Likewise.
17685 (arm_expand_compare_and_swap): Likewise.
17686 (arm_evpc_neon_vuzp): Likewise.
17687 (arm_evpc_neon_vzip): Likewise.
17688 (arm_evpc_neon_vrev): Likewise.
17689 (arm_evpc_neon_vtrn): Likewise.
17690 (arm_evpc_neon_vext): Likewise.
17691 (arm_validize_comparison): Likewise.
17692 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
17693 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
17694 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
17695 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
17696 (c6x_preferred_simd_mode): Likewise.
17697 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
17698 (epiphany_rtx_costs): Likewise.
17699 * config/epiphany/predicates.md (proper_comparison_operator):
17701 * config/frv/frv.c (condexec_memory_operand): Likewise.
17702 (frv_emit_move): Likewise.
17703 (output_move_single): Likewise.
17704 (output_condmove_single): Likewise.
17705 (frv_hard_regno_mode_ok): Likewise.
17706 (frv_matching_accg_mode): Likewise.
17707 * config/h8300/h8300.c (split_adds_subs): Likewise.
17708 (h8300_rtx_costs): Likewise.
17709 (h8300_print_operand): Likewise.
17710 (compute_mov_length): Likewise.
17711 (output_logical_op): Likewise.
17712 (compute_logical_op_length): Likewise.
17713 (compute_logical_op_cc): Likewise.
17714 (h8300_shift_needs_scratch_p): Likewise.
17715 (output_a_shift): Likewise.
17716 (compute_a_shift_length): Likewise.
17717 (compute_a_shift_cc): Likewise.
17718 (expand_a_rotate): Likewise.
17719 (output_a_rotate): Likewise.
17720 * config/i386/i386.c (classify_argument): Likewise.
17721 (function_arg_advance_32): Likewise.
17722 (function_arg_32): Likewise.
17723 (function_arg_64): Likewise.
17724 (function_value_64): Likewise.
17725 (ix86_gimplify_va_arg): Likewise.
17726 (ix86_legitimate_constant_p): Likewise.
17727 (put_condition_code): Likewise.
17728 (split_double_mode): Likewise.
17729 (ix86_avx256_split_vector_move_misalign): Likewise.
17730 (ix86_expand_vector_logical_operator): Likewise.
17731 (ix86_split_idivmod): Likewise.
17732 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
17733 (ix86_build_const_vector): Likewise.
17734 (ix86_build_signbit_mask): Likewise.
17735 (ix86_match_ccmode): Likewise.
17736 (ix86_cc_modes_compatible): Likewise.
17737 (ix86_expand_branch): Likewise.
17738 (ix86_expand_sse_cmp): Likewise.
17739 (ix86_expand_sse_movcc): Likewise.
17740 (ix86_expand_int_sse_cmp): Likewise.
17741 (ix86_expand_vec_perm_vpermi2): Likewise.
17742 (ix86_expand_vec_perm): Likewise.
17743 (ix86_expand_sse_unpack): Likewise.
17744 (ix86_expand_int_addcc): Likewise.
17745 (ix86_split_to_parts): Likewise.
17746 (ix86_vectorize_builtin_gather): Likewise.
17747 (ix86_vectorize_builtin_scatter): Likewise.
17748 (avx_vpermilp_parallel): Likewise.
17749 (inline_memory_move_cost): Likewise.
17750 (ix86_tieable_integer_mode_p): Likewise.
17751 (x86_maybe_negate_const_int): Likewise.
17752 (ix86_expand_vector_init_duplicate): Likewise.
17753 (ix86_expand_vector_init_one_nonzero): Likewise.
17754 (ix86_expand_vector_init_one_var): Likewise.
17755 (ix86_expand_vector_init_concat): Likewise.
17756 (ix86_expand_vector_init_interleave): Likewise.
17757 (ix86_expand_vector_init_general): Likewise.
17758 (ix86_expand_vector_set): Likewise.
17759 (ix86_expand_vector_extract): Likewise.
17760 (emit_reduc_half): Likewise.
17761 (ix86_emit_i387_round): Likewise.
17762 (ix86_mangle_type): Likewise.
17763 (ix86_expand_round_sse4): Likewise.
17764 (expand_vec_perm_blend): Likewise.
17765 (canonicalize_vector_int_perm): Likewise.
17766 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
17767 (expand_vec_perm_1): Likewise.
17768 (expand_vec_perm_interleave3): Likewise.
17769 (expand_vec_perm_even_odd_pack): Likewise.
17770 (expand_vec_perm_even_odd_1): Likewise.
17771 (expand_vec_perm_broadcast_1): Likewise.
17772 (ix86_vectorize_vec_perm_const_ok): Likewise.
17773 (ix86_expand_vecop_qihi): Likewise.
17774 (ix86_expand_mul_widen_hilo): Likewise.
17775 (ix86_expand_sse2_abs): Likewise.
17776 (ix86_expand_pextr): Likewise.
17777 (ix86_expand_pinsr): Likewise.
17778 (ix86_preferred_simd_mode): Likewise.
17779 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
17780 * config/i386/sse.md (*andnot<mode>3): Likewise.
17781 (<mask_codefor><code><mode>3<mask_name>): Likewise.
17782 (*<code><mode>3): Likewise.
17783 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
17784 (ia64_expand_atomic_op): Likewise.
17785 (ia64_arg_type): Likewise.
17786 (ia64_mode_to_int): Likewise.
17787 (ia64_scalar_mode_supported_p): Likewise.
17788 (ia64_vector_mode_supported_p): Likewise.
17789 (expand_vec_perm_broadcast): Likewise.
17790 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
17791 (iq2000_function_arg_advance): Likewise.
17792 (iq2000_function_arg): Likewise.
17793 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
17794 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
17795 (m68k_libcall_value): Likewise.
17796 (m68k_function_value): Likewise.
17797 (sched_attr_op_type): Likewise.
17798 * config/mcore/mcore.c (mcore_output_move): Likewise.
17799 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
17801 (microblaze_function_arg): Likewise.
17802 * config/mips/mips.c (mips16_build_call_stub): Likewise.
17803 (mips_print_operand): Likewise.
17804 (mips_mode_ok_for_mov_fmt_p): Likewise.
17805 (mips_vector_mode_supported_p): Likewise.
17806 (mips_preferred_simd_mode): Likewise.
17807 (mips_expand_vpc_loongson_even_odd): Likewise.
17808 (mips_expand_vec_unpack): Likewise.
17809 (mips_expand_vi_broadcast): Likewise.
17810 (mips_expand_vector_init): Likewise.
17811 (mips_expand_vec_reduc): Likewise.
17812 (mips_expand_msa_cmp): Likewise.
17813 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
17814 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
17815 (cc_flags_for_mode): Likewise.
17816 * config/msp430/msp430.c (msp430_print_operand): Likewise.
17817 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
17818 (nds32_output_casesi_pc_relative): Likewise.
17819 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17820 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
17821 (nvptx_gen_unpack): Likewise.
17822 (nvptx_gen_pack): Likewise.
17823 (nvptx_gen_shuffle): Likewise.
17824 (nvptx_gen_wcast): Likewise.
17825 (nvptx_preferred_simd_mode): Likewise.
17826 * config/pa/pa.c (pa_secondary_reload): Likewise.
17827 * config/pa/predicates.md (base14_operand): Likewise.
17828 * config/powerpcspe/powerpcspe-c.c
17829 (altivec_resolve_overloaded_builtin): Likewise.
17830 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
17832 (rs6000_preferred_simd_mode): Likewise.
17833 (output_vec_const_move): Likewise.
17834 (rs6000_expand_vector_extract): Likewise.
17835 (rs6000_split_vec_extract_var): Likewise.
17836 (reg_offset_addressing_ok_p): Likewise.
17837 (rs6000_legitimate_offset_address_p): Likewise.
17838 (rs6000_legitimize_address): Likewise.
17839 (rs6000_emit_set_const): Likewise.
17840 (rs6000_const_vec): Likewise.
17841 (rs6000_emit_move): Likewise.
17842 (spe_build_register_parallel): Likewise.
17843 (rs6000_darwin64_record_arg_recurse): Likewise.
17844 (swap_selector_for_mode): Likewise.
17845 (spe_init_builtins): Likewise.
17846 (paired_init_builtins): Likewise.
17847 (altivec_init_builtins): Likewise.
17848 (do_load_for_compare): Likewise.
17849 (rs6000_generate_compare): Likewise.
17850 (rs6000_expand_float128_convert): Likewise.
17851 (emit_load_locked): Likewise.
17852 (emit_store_conditional): Likewise.
17853 (rs6000_output_function_epilogue): Likewise.
17854 (rs6000_handle_altivec_attribute): Likewise.
17855 (rs6000_function_value): Likewise.
17856 (emit_fusion_gpr_load): Likewise.
17857 (emit_fusion_p9_load): Likewise.
17858 (emit_fusion_p9_store): Likewise.
17859 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
17860 (fusion_gpr_mem_load): Likewise.
17861 (fusion_addis_mem_combo_load): Likewise.
17862 (fusion_addis_mem_combo_store): Likewise.
17863 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
17864 (fusion_gpr_mem_load): Likewise.
17865 (fusion_addis_mem_combo_load): Likewise.
17866 (fusion_addis_mem_combo_store): Likewise.
17867 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17869 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
17870 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17871 (rs6000_preferred_simd_mode): Likewise.
17872 (output_vec_const_move): Likewise.
17873 (rs6000_expand_vector_extract): Likewise.
17874 (rs6000_split_vec_extract_var): Likewise.
17875 (reg_offset_addressing_ok_p): Likewise.
17876 (rs6000_legitimate_offset_address_p): Likewise.
17877 (rs6000_legitimize_address): Likewise.
17878 (rs6000_emit_set_const): Likewise.
17879 (rs6000_const_vec): Likewise.
17880 (rs6000_emit_move): Likewise.
17881 (rs6000_darwin64_record_arg_recurse): Likewise.
17882 (swap_selector_for_mode): Likewise.
17883 (paired_init_builtins): Likewise.
17884 (altivec_init_builtins): Likewise.
17885 (rs6000_expand_float128_convert): Likewise.
17886 (emit_load_locked): Likewise.
17887 (emit_store_conditional): Likewise.
17888 (rs6000_output_function_epilogue): Likewise.
17889 (rs6000_handle_altivec_attribute): Likewise.
17890 (rs6000_function_value): Likewise.
17891 (emit_fusion_gpr_load): Likewise.
17892 (emit_fusion_p9_load): Likewise.
17893 (emit_fusion_p9_store): Likewise.
17894 * config/rx/rx.c (rx_gen_move_template): Likewise.
17895 (flags_from_mode): Likewise.
17896 * config/s390/predicates.md (s390_alc_comparison): Likewise.
17897 (s390_slb_comparison): Likewise.
17898 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
17899 (s390_vector_mode_supported_p): Likewise.
17900 (s390_cc_modes_compatible): Likewise.
17901 (s390_match_ccmode_set): Likewise.
17902 (s390_canonicalize_comparison): Likewise.
17903 (s390_emit_compare_and_swap): Likewise.
17904 (s390_branch_condition_mask): Likewise.
17905 (s390_rtx_costs): Likewise.
17906 (s390_secondary_reload): Likewise.
17907 (__SECONDARY_RELOAD_CASE): Likewise.
17908 (s390_expand_cs): Likewise.
17909 (s390_preferred_simd_mode): Likewise.
17910 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
17911 * config/sh/sh.c (sh_print_operand): Likewise.
17912 (dump_table): Likewise.
17913 (sh_secondary_reload): Likewise.
17914 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17915 * config/sh/sh.md (casesi_worker_1): Likewise.
17916 (casesi_worker_2): Likewise.
17917 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
17918 (fcc_comparison_operator): Likewise.
17919 * config/sparc/sparc.c (sparc_expand_move): Likewise.
17920 (emit_soft_tfmode_cvt): Likewise.
17921 (sparc_preferred_simd_mode): Likewise.
17922 (output_cbranch): Likewise.
17923 (sparc_print_operand): Likewise.
17924 (sparc_expand_vec_perm_bmask): Likewise.
17925 (vector_init_bshuffle): Likewise.
17926 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17927 (spu_vector_mode_supported_p): Likewise.
17928 (spu_expand_insv): Likewise.
17929 (spu_emit_branch_or_set): Likewise.
17930 (spu_handle_vector_attribute): Likewise.
17931 (spu_builtin_splats): Likewise.
17932 (spu_builtin_extract): Likewise.
17933 (spu_builtin_promote): Likewise.
17934 (spu_expand_sign_extend): Likewise.
17935 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17936 (tilegx_simd_int): Likewise.
17937 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
17938 (tilepro_simd_int): Likewise.
17939 * config/v850/v850.c (const_double_split): Likewise.
17940 (v850_print_operand): Likewise.
17941 (ep_memory_offset): Likewise.
17942 * config/vax/vax.c (vax_rtx_costs): Likewise.
17943 (vax_output_int_move): Likewise.
17944 (vax_output_int_add): Likewise.
17945 (vax_output_int_subtract): Likewise.
17946 * config/visium/predicates.md (visium_branch_operator): Likewise.
17947 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
17948 (visium_print_operand_address): Likewise.
17949 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17950 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
17951 (xtensa_expand_conditional_branch): Likewise.
17952 (xtensa_copy_incoming_a7): Likewise.
17953 (xtensa_output_literal): Likewise.
17954 * dfp.c (decimal_real_maxval): Likewise.
17955 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17958 Alan Hayward <alan.hayward@arm.com>
17959 David Sherwood <david.sherwood@arm.com>
17961 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
17962 (mode_nunits_inline): Likewise.
17963 (mode_inner_inline): Likewise.
17964 (mode_unit_size_inline): Likewise.
17965 (mode_unit_precision_inline): Likewise.
17966 (emit_insn_modes_h): Likewise. Also emit a #define of the
17968 (emit_mode_wider): Add an E_ prefix to mode names.
17969 (emit_mode_complex): Likewise.
17970 (emit_mode_inner): Likewise.
17971 (emit_mode_adjustments): Likewise.
17972 (emit_mode_int_n): Likewise.
17973 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
17974 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
17975 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
17976 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
17977 (CRC32_BUILTIN, ENTRY): Likewise.
17978 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
17979 (aarch64_pop_regs): Likewise.
17980 (aarch64_process_components): Likewise.
17981 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
17982 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
17983 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
17984 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
17985 * config/arm/arm.c (arm_init_libfuncs): Likewise.
17986 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
17988 * config/i386/i386-builtin.def (pcmpestr): Likewise.
17989 (pcmpistr): Likewise.
17990 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
17991 * config/mmix/mmix.c (mmix_output_condition): Likewise.
17992 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17994 * config/rl78/rl78.c (mduc_regs): Likewise.
17995 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17996 (htm_expand_builtin): Likewise.
17997 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
17998 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
17999 * config/xtensa/xtensa.c (print_operand): Likewise.
18000 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18001 (NUM_MODE_VECTOR_INT): Likewise.
18002 * genoutput.c (null_operand): Likewise.
18003 (output_operand_data): Likewise.
18004 * genrecog.c (print_parameter_value): Likewise.
18005 * lra.c (debug_operand_data): Likewise.
18007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18009 * dwarf2out.c (scompare_loc_descriptor_wide)
18010 (scompare_loc_descriptor_narrow): New functions, split out from...
18011 (scompare_loc_descriptor): ...here.
18012 * expmed.c (emit_store_flag_int): New function, split out from...
18013 (emit_store_flag): ...here.
18015 2017-08-30 Richard Biener <rguenther@suse.de>
18017 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18018 (dwarf2out_early_finish): Move setting of AT_pubnames from
18019 early debug output to early finish.
18021 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18023 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18024 and -mdata-region to the assembler.
18026 2017-08-30 Richard Biener <rguenther@suse.de>
18028 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18030 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18031 (dwarf2out_early_global_decl): Only generate a DIE for the
18032 abstract origin if it doesn't already exist or is a declaration DIE.
18033 (resolve_addr): Do not add the linkage name twice when
18034 generating a stub DIE for the DW_TAG_GNU_call_site target.
18036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18038 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18039 Use machine_mode rather than int for arg1_mode.
18041 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18044 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
18045 that the second argument of the built-in functions to unpack
18046 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
18047 switch statement instead a lot of if statements.
18048 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
18049 Allow 64-bit values to be in Altivec registers as well as
18050 traditional floating point registers.
18051 (pack<mode>, FMOVE128_VSX iterator): Likewise.
18053 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
18055 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
18056 MAX_REGS_PER_ADDRESS == 1.
18058 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
18060 * config/i386/i386.opt (flag_fentry): Do not init to -1.
18061 * config/i386/i386.c (ix86_option_override_internal): Simplify
18062 setting of opts->x_flag_entry.
18064 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18065 Jakub Jelinek <jakub@redhat.com>
18066 Richard Biener <rguenther@suse.de>
18068 PR tree-optimization/81503
18069 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
18070 folded constant fits in the target type; reorder tests for clarity.
18072 2017-08-29 Martin Liska <mliska@suse.cz>
18074 * passes.def: Include pass_lower_switch.
18075 * stmt.c (dump_case_nodes): Remove and move to
18076 tree-switch-conversion.
18077 (case_values_threshold): Likewise.
18078 (expand_switch_as_decision_tree_p): Likewise.
18079 (emit_case_decision_tree): Likewise.
18080 (expand_case): Likewise.
18081 (balance_case_nodes): Likewise.
18082 (node_has_low_bound): Likewise.
18083 (node_has_high_bound): Likewise.
18084 (node_is_bounded): Likewise.
18085 (emit_case_nodes): Likewise.
18086 (struct simple_case_node): New struct.
18087 (add_case_node): Remove.
18088 (emit_case_dispatch_table): Use vector instead of case_list.
18089 (reset_out_edges_aux): Remove.
18090 (compute_cases_per_edge): Likewise.
18091 (expand_case): Build list of simple_case_node.
18092 (expand_sjlj_dispatch_table): Use it.
18093 * tree-switch-conversion.c (struct case_node): Moved from
18094 stmt.c and adjusted.
18095 (emit_case_nodes): Likewise.
18096 (node_has_low_bound): Likewise.
18097 (node_has_high_bound): Likewise.
18098 (node_is_bounded): Likewise.
18099 (case_values_threshold): Likewise.
18100 (reset_out_edges_aux): Likewise.
18101 (compute_cases_per_edge): Likewise.
18102 (add_case_node): Likewise.
18103 (dump_case_nodes): Likewise.
18104 (balance_case_nodes): Likewise.
18105 (expand_switch_as_decision_tree_p): Likewise.
18106 (emit_jump): Likewise.
18107 (emit_case_decision_tree): Likewise.
18108 (try_switch_expansion): Likewise.
18109 (do_jump_if_equal): Likewise.
18110 (emit_cmp_and_jump_insns): Likewise.
18111 (fix_phi_operands_for_edge): New function.
18112 (record_phi_operand_mapping): Likewise.
18113 (class pass_lower_switch): New pass.
18114 (pass_lower_switch::execute): New function.
18115 (make_pass_lower_switch): Likewise.
18116 (conditional_probability):
18117 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
18118 * tree-pass.h: Add make_pass_lower_switch.
18120 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18123 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
18126 2017-08-29 Richard Biener <rguenther@suse.de>
18128 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
18129 we do not add a DW_AT_inline attribute twice.
18130 (gen_subprogram_die): Remove code setting DW_AT_inline on
18131 DECL_ABSTRACT_P nodes.
18133 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
18135 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
18136 calls to internal functions.
18137 (gimplify_modify_expr): Likewise.
18138 * tree-call-cdce.c (use_internal_fn): Likewise.
18139 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
18140 (convert_to_divmod): Set the nothrow flag.
18141 * tree-if-conv.c (predicate_mem_writes): Likewise.
18142 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18143 (vectorizable_call): Likewise.
18144 (vectorizable_store): Likewise.
18145 (vectorizable_load): Likewise.
18146 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
18147 (vect_recog_mask_conversion_pattern): Likewise.
18149 2017-08-29 Martin Liska <mliska@suse.cz>
18152 * gcc.c (driver_handle_option): Add new argument.
18153 * opts-common.c (handle_option): Pass
18154 target_option_override_hook.
18155 * opts-global.c (lang_handle_option): Add new option.
18156 (set_default_handlers): Add new argument.
18157 (decode_options): Likewise.
18158 * opts.c (target_handle_option): Likewise.
18159 (common_handle_option): Call target_option_override_hook.
18160 * opts.h (struct cl_option_handler_func): Add hook for
18161 target option override.
18162 (struct cl_option_handlers): Likewise.
18163 (set_default_handlers): Add new argument.
18164 (decode_options): Likewise.
18165 (common_handle_option): Likewise.
18166 (target_handle_option): Likewise.
18167 * toplev.c (toplev::main): Pass targetm.target_option.override
18170 2017-08-29 Richard Biener <rguenther@suse.de>
18171 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
18173 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
18174 life to the active subtree.
18176 2017-08-28 Jeff Law <law@redhat.com>
18178 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
18179 derive_equivalences.
18180 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
18181 Code moved into....
18182 (edge_info::derive_equivalences): New private member function
18184 * tree-ssa-dom.c (class edge_info): Changed from a struct
18185 to a class. Add ctor/dtor, methods and data members.
18186 (edge_info::edge_info): Renamed from allocate_edge_info.
18187 Initialize additional members.
18188 (edge_info::~edge_info): New.
18189 (free_dom_edge_info): Delete the edge info.
18190 (record_edge_info): Use new class & associated member functions.
18191 Tighten forms for testing for edge equivalences.
18192 (record_temporary_equivalences): Iterate over the simple
18193 equivalences rather than assuming there's only one per edge.
18194 (cprop_into_successor_phis): Iterate over the simple
18195 equivalences rather than assuming there's only one per edge.
18196 (optimize_stmt): Use operand_equal_p rather than pointer
18197 equality for mini-DSE code.
18199 2017-08-28 Nathan Sidwell <nathan@acm.org>
18201 * gcc.c (execute): Fold SIGPIPE handling into switch
18202 statement. Adjust internal error message.
18204 2017-08-28 Richard Biener <rguenther@suse.de>
18207 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
18208 Do nothing for removed DIEs.
18210 2017-08-28 Richard Biener <rguenther@suse.de>
18212 PR tree-optimization/81977
18213 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
18216 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
18219 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
18220 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
18221 using targetm.gen_mem_thread_fence.
18223 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
18226 * config/i386/i386.md (*<btsc><mode>): Change operand 2
18227 predicate to register_operand. Reorder operands.
18228 (*btr<mode>): Ditto.
18229 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
18230 (*btr<mode>_mask): Ditto.
18232 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
18234 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
18235 * config/rs6000/xmmintrin.h: New file.
18236 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
18238 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18241 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
18242 parameter and_insn and return it.
18243 (recombine_lvx_pattern): Insert a copy to ensure availability of
18244 the base register of the copied masking operation at the point of
18245 the instruction replacement.
18246 (recombine_stvx_pattern): Likewise.
18248 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
18250 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
18251 undocumented switches.
18252 (-mpower9-dform-vector): Likewise.
18253 (-mpower9-dform): Likewise.
18254 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18255 comments to delete references to -mpower9-dform* switches.
18256 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
18257 Delete reference to -mpower9-dform* switches, test for
18258 -mpower9-vector instead.
18259 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
18260 (OTHER_P9_VECTOR_MASKS): Likewise.
18261 (POWERPC_MASKS): Likewise.
18262 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
18263 tests against -mpower9-dform* to -mpower9-vector. Delete code
18264 that checked for -mpower9-dform* consistancy with other options.
18265 Add test for -mpower9-misc to enable other power9 switches.
18266 (rs6000_init_hard_regno_mode_ok): Likewise.
18267 (rs6000_option_override_internal): Likewise.
18268 (rs6000_emit_prologue): Likewise.
18269 (rs6000_emit_epilogue): Likewise.
18270 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
18271 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
18272 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
18274 (emit_fusion_p9_store): Likewise.
18275 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
18276 resetting these macros if the assembler does not support ISA 3.0
18278 (TARGET_P9_DFORM_VECTOR): Likewise.
18279 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
18280 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
18282 2017-08-25 Alan Modra <amodra@gmail.com>
18284 PR rtl-optimization/81747
18285 * cse.c (cse_extended_basic_block): Don't attempt to record
18286 equivalences for degenerate conditional branches that branch
18287 to their fall-through.
18289 2017-08-24 Martin Sebor <msebor@redhat.com>
18291 PR middle-end/81908
18292 * gimple-fold.c (size_must_be_zero_p): New function.
18293 (gimple_fold_builtin_memory_op): Call it.
18295 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
18297 * config/rs6000/mm_malloc.h: New file.
18299 2017-08-24 Bin Cheng <bin.cheng@arm.com>
18301 PR tree-optimization/81913
18302 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
18303 analysis when either IVs in condition can wrap.
18305 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
18307 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
18308 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
18310 2017-08-24 Richard Biener <rguenther@suse.de>
18313 * targhooks.c (default_target_can_inline_p): Properly
18314 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
18315 is present and always compare.
18316 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
18317 infer -mfpmath=sse from TARGET_SSE_P.
18318 (ix86_can_inline_p): Properly use target_option_default_node when
18319 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
18321 2017-08-24 Richard Biener <rguenther@suse.de>
18324 * dwarf2out.c (output_die): Handle flag_generate_offload like
18326 (output_comp_unit): Likewise.
18327 (gen_array_type_die): Likewise.
18328 (dwarf2out_early_finish): Likewise.
18329 (note_variable_value_in_expr): Likewise.
18330 (dwarf2out_finish): Likewise. Adjust assert.
18331 * cgraphunit.c (symbol_table::compile): Move setting of
18332 flag_generate_offload earlier ...
18333 (symbol_table::finalize_compilation_unit): ... here, before
18334 early debug finalization.
18336 2017-08-24 Richard Biener <rguenther@suse.de>
18338 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
18339 and ipa-fnsummary.h.
18340 (ix86_can_inline_p): When ix86_fpmath flags do not match
18341 check whether the callee uses FP math at all.
18343 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
18345 PR middle-end/81931
18346 * tree-ssanames.c (get_nonzero_bits): Use element_precision
18347 instead of TYPE_PRECISION.
18349 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
18350 Alan Hayward <alan.hayward@arm.com>
18351 David Sherwood <david.sherwood@arm.com>
18353 * combine.c (make_extraction): Use subreg_offset_from_lsb.
18355 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
18357 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
18359 (ix86_frame::stack_realign_allocate): New field.
18360 (struct machine_frame_state): Modify comments.
18361 (machine_frame_state::sp_realigned_fp_end): New field.
18362 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
18363 layout calculation.
18364 (sp_valid_at): Add assertion to assure no attempt to access invalid
18365 offset of a realigned stack.
18366 (fp_valid_at): Likewise.
18367 (choose_baseaddr): Modify comments.
18368 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
18369 ix86_expand_prologue.
18370 (ix86_expand_prologue): Modify stack realignment and allocation.
18371 (ix86_expand_epilogue): Modify comments.
18372 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
18373 avx2_runtime, avx512f, and avx512f_runtime.
18375 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
18377 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
18378 (mstackrealign): Do not init to -1.
18379 * config/i386/i386.c (ix86_option_override_internal):
18380 Check opts_set, not opts when setting default value of
18381 opts->x_ix86_force_align_arg_pointer.
18383 2017-08-23 Richard Biener <rguenther@suse.de>
18385 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
18386 lang_hooks.decl_printable_name.
18387 * print-rtl-function.c (print_rtx_function): Likewise.
18388 * tree-pretty-print.c (dump_function_header): Likewise.
18390 2017-08-23 Richard Biener <rguenther@suse.de>
18393 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
18394 -g0 at compile-time.
18396 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18398 PR middle-end/19706
18399 * doc/sourcebuild.texi (Other hardware attributes):
18402 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18404 PR middle-end/19706
18405 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
18406 Fix single-use check.
18408 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
18410 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
18412 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18414 * doc/install.texi: Modify to add more details on running selected
18417 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18419 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
18420 is combined with -mabi=ms.
18421 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
18424 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18426 PR tree-optimization/81488
18427 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
18428 and cached_basis fields.
18429 (MAX_SPREAD): New constant.
18430 (alloc_cand_and_find_basis): Initialize new fields.
18431 (clear_visited): New function.
18432 (create_phi_basis_1): Rename from create_phi_basis, set visited
18433 and cached_basis fields.
18434 (create_phi_basis): New wrapper function.
18435 (phi_add_costs_1): Rename from phi_add_costs, add spread
18436 parameter, set visited field, short-circuit when limits reached.
18437 (phi_add_costs): New wrapper function.
18438 (record_phi_increments_1): Rename from record_phi_increments, set
18440 (record_phi_increments): New wrapper function.
18441 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
18442 (phi_incr_cost): New wrapper function.
18443 (all_phi_incrs_profitable_1): Rename from
18444 all_phi_incrs_profitable, set visited field.
18445 (all_phi_incrs_profitable): New wrapper function.
18447 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18448 Alan Hayward <alan.hayward@arm.com>
18449 David Sherwood <david.sherwood@arm.com>
18451 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
18452 that takes the outer and inner modes.
18453 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
18454 comparison as the canonical test for a paradoxical subreg.
18455 * combine.c (simplify_set): Use paradoxical_subreg_p.
18456 (make_extraction): Likewise.
18457 (force_to_mode): Likewise.
18458 (rtx_equal_for_field_assignment_p): Likewise.
18459 (gen_lowpart_for_combine): Likewise.
18460 (simplify_comparison): Likewise.
18461 * cse.c (equiv_constant): Likewise.
18462 * expmed.c (store_bit_field_1): Likewise.
18463 * final.c (alter_subreg): Likewise.
18464 * fwprop.c (propagate_rtx): Likewise.
18465 (forward_propagate_subreg): Likewise.
18466 * ira-conflicts.c (ira_build_conflicts): Likewise.
18467 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
18468 * lra-constraints.c (curr_insn_transform): Likewise.
18469 (split_reg): Likewise.
18470 * lra-eliminations.c (move_plus_up): Likewise.
18471 (lra_eliminate_regs_1): Likewise.
18472 * recog.c (general_operand): Likewise.
18473 * ree.c (combine_reaching_defs): Likewise.
18474 * reload.c (push_reload): Likewise.
18475 (find_reloads): Likewise.
18476 * reload1.c (elimination_effects): Likewise.
18477 (compute_reload_subreg_offset): Likewise.
18478 (choose_reload_regs): Likewise.
18479 * rtlanal.c (subreg_lsb_1): Likewise.
18480 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18481 (simplify_subreg): Likewise.
18482 * var-tracking.c (track_loc_p): Likewise.
18483 * emit-rtl.c (byte_lowpart_offset): Likewise.
18484 (paradoxical_subreg_p): Delete out-of-line definition.
18486 2017-08-22 Jeff Law <law@redhat.com>
18488 PR tree-optimization/81741
18489 PR tree-optimization/71947
18490 * tree-ssa-dom.c: Include tree-inline.h.
18491 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
18492 equivalences if one is more expensive to compute than the other.
18493 * tree-ssa-scopedtables.h (class const_or_copies): Make
18494 record_const_or_copy_raw method private.
18495 (class avail_exprs_stack): New method simplify_binary_operation.
18496 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
18497 avail_exprs_stack::simplify_binary_operation as needed.
18498 (avail_exprs_stack::simplify_binary_operation): New function.
18500 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
18502 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
18503 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
18504 (DOT_SYMBOLS): Likewise.
18505 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
18506 (RELOCATABLE_NEEDS_FIXUP): Likewise.
18507 (RS6000_ABI_NAME): Likewise.
18508 (TARGET_CMODEL): Likewise.
18509 (TOC_SECTION_ASM_OP): Likewise.
18510 (SET_CMODEL): New macro.
18511 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
18513 2017-08-22 Richard Biener <rguenther@suse.de>
18515 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
18516 to allow for free-lang-data replacements similar to verify_type_variant.
18518 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18519 Alan Hayward <alan.hayward@arm.com>
18520 David Sherwood <david.sherwood@arm.com>
18522 * config/aarch64/aarch64.md (casesi): Use DImode rather than
18523 VOIDmode for the LABEL_REF.
18525 2017-08-22 Richard Biener <rguenther@suse.de>
18527 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
18529 2017-08-22 Richard Biener <rguenther@suse.de>
18531 * common.opt (feliminate-dwarf2-dups): Ignore.
18532 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
18533 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
18534 same_die_p_wrap, compute_section_prefix,
18535 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
18536 (comdat_symbol_id, comdat_symbol_number): Likewise.
18537 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
18539 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
18540 (output_die): Mark unreachable path unreachable.
18541 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
18542 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
18543 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
18544 (dwarf2out_early_finish): Likewise.
18546 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
18548 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
18550 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
18553 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
18554 not VAR_P. Filter attribute warnings with OPT_Wattributes.
18555 (avr_attribute_table) <io, io_low, address>: Initialize
18556 .decl_required with true.
18558 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
18560 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
18561 undocumented debugging options.
18562 (-mvsx-scalar-double): Likewise.
18563 (-mallow-df-permute): Likewise.
18564 (-mvectorize-builtins): Likewise.
18565 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18566 (rs6000_builtin_vectorized_function): Likewise.
18567 (rs6000_builtin_md_vectorized_function): Likewise.
18568 (rs6000_opt_vars): Likewise.
18570 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
18573 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
18574 (*btrq_imm): Rename from *btrq.
18575 (*btcq_imm): Rename from *btcq.
18576 (btsc): New code attribute.
18577 (*<btsc><mode>): New insn pattern.
18578 (*btr<mode>): Ditto.
18579 (*<btsc><mode>_mask): New insn_and_split pattern.
18580 (*btr<mode>_mask): Ditto.
18582 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18583 Alan Hayward <alan.hayward@arm.com>
18584 David Sherwood <david.sherwood@arm.com>
18586 * function.c (pad_below): Simplify padding calculation.
18588 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18589 Alan Hayward <alan.hayward@arm.com>
18590 David Sherwood <david.sherwood@arm.com>
18592 * target.def (function_prologue): Remove frame size argument.
18593 (function_epilogue): Likewise.
18594 * doc/tm.texi: Regenerate.
18595 * final.c (final_start_function): Update call to function_prologue.
18596 (final_end_function): Update call to function_epilogue.
18597 (default_function_pro_epilogue): Remove frame size argument.
18598 * output.h (default_function_pro_epilogue): Likewise.
18599 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
18600 (arm_output_function_prologue): Likewise.
18601 * config/frv/frv.c (frv_function_prologue): Likewise.
18602 (frv_function_epilogue): Likewise.
18603 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
18604 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
18605 (ia64_output_function_epilogue): Likewise.
18606 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
18607 (m32r_output_function_epilogue): Likewise.
18608 * config/microblaze/microblaze.c (microblaze_function_prologue)
18609 (microblaze_function_epilogue): Likewise.
18610 * config/mips/mips.c (mips_output_function_prologue): Likewise.
18611 (mips_output_function_epilogue): Likewise.
18612 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
18613 (mmix_target_asm_function_epilogue): Likewise.
18614 * config/msp430/msp430.c (msp430_start_function): Likewise.
18615 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
18616 (nds32_asm_function_epilogue): Likewise.
18617 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
18618 * config/pa/pa.c (pa_output_function_prologue): Likewise.
18619 (pa_output_function_epilogue): Likewise.
18620 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
18621 (rs6000_output_function_epilogue): Likewise.
18622 * config/rl78/rl78.c (rl78_start_function): Likewise.
18623 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
18624 (rs6000_output_function_epilogue): Likewise.
18625 * config/rx/rx.c (rx_output_function_prologue): Likewise.
18626 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
18627 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
18628 (sparc_asm_function_epilogue): Likewise.
18630 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18632 * tree.h (type_has_mode_precision_p): New function.
18633 * convert.c (convert_to_integer_1): Use it.
18634 * expr.c (expand_expr_real_2): Likewise.
18635 (expand_expr_real_1): Likewise.
18636 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
18637 * match.pd: Likewise.
18638 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
18639 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
18640 * tree-tailcall.c (process_assignment): Likewise.
18641 * tree-vect-loop.c (vectorizable_reduction): Likewise.
18642 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
18643 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
18644 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18645 (vectorizable_assignment): Likewise.
18646 (vectorizable_shift): Likewise.
18647 (vectorizable_operation): Likewise.
18648 * tree-vrp.c (register_edge_assert_for_2): Likewise.
18650 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
18652 * match.pd: Add pow (C, x) simplification.
18654 2017-08-21 Richard Biener <rguenther@suse.de>
18656 PR tree-optimization/81900
18657 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
18658 for blocks with abnormal predecessors.
18659 (compute_antic): Do not set visited flag prematurely.
18661 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
18664 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
18666 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18668 * stor-layout.h (vector_type_mode): Move to...
18669 * tree.h (vector_type_mode): ...here.
18670 * stor-layout.c (vector_type_mode): Move to...
18671 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
18673 2017-08-21 Richard Biener <rguenther@suse.de>
18675 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
18676 register_external_die hooks.
18677 (debug_false_tree_charstarstar_uhwistar): Declare.
18678 (debug_nothing_tree_charstar_uhwi): Likewise.
18679 * debug.c (do_nothing_debug_hooks): Adjust.
18680 (debug_false_tree_charstarstar_uhwistar): New do nothing.
18681 (debug_nothing_tree_charstar_uhwi): Likewise.
18682 * dbxout.c (dbx_debug_hooks): Adjust.
18683 (xcoff_debug_hooks): Likewise.
18684 * sdbout.c (sdb_debug_hooks): Likewise.
18685 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18686 * dwarf2out.c (macinfo_label_base): New global.
18687 (dwarf2out_register_external_die): New function for the
18688 register_external_die hook.
18689 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
18690 (dwarf2_debug_hooks): Use them.
18691 (dwarf2_lineno_debug_hooks): Adjust.
18692 (struct die_struct): Add with_offset flag.
18693 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
18694 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
18695 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
18696 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
18697 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
18698 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
18699 defining section names for the early LTO debug variants.
18700 (reset_indirect_string): New helper.
18701 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
18702 (print_dw_val): Add support for offsetted symbol references.
18703 (get_ultimate_context): Split out from is_cxx.
18704 (is_cxx): Use get_ultimate_context.
18705 (is_fortran): Add decl overload.
18706 (compute_comp_unit_symbol): Split out worker from
18707 compute_section_prefix.
18708 (compute_section_prefix): Call compute_comp_unit_symbol and
18709 set comdat_type_p here.
18710 (output_die): Skip DIE symbol output for the LTO added one.
18711 Handle DIE symbol references with offset.
18712 (output_comp_unit): Guard section name mangling properly.
18713 For LTO debug sections emit a symbol at the section beginning
18714 which we use to refer to its DIEs.
18715 (add_abstract_origin_attribute): For DIEs registered via
18716 dwarf2out_register_external_die directly refer to the early
18717 DIE rather than indirectly through the shadow one we created.
18718 Remove obsolete call to dwarf2out_abstract_function for
18719 non-function/block origins.
18720 (gen_array_type_die): When generating early LTO debug do
18721 not emit DW_AT_string_length.
18722 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
18723 late when in LTO. As suggested place a gcc_unreachable for
18724 the DECL_ABSTRACT_P case.
18725 (gen_subprogram_die): Avoid another specification DIE
18726 for early built declarations/definitions for the late LTO case.
18727 (gen_variable_die): Add type references for late duplicated VLA dies
18729 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
18730 we have the abstract instance already.
18731 (process_scope_var): Adjust decl DIE contexts in LTO which
18732 first puts them in limbo.
18733 (gen_decl_die): Do not generate type DIEs late apart from
18734 types for VLAs or for decls we do not yet have a DIE. Do not
18735 call dwarf2out_abstract_function late.
18736 (dwarf2out_early_global_decl): Make sure to create DIEs
18737 for abstract instances of a decl first.
18738 (dwarf2out_late_global_decl): Adjust comment.
18739 (output_macinfo_op): With multiple macro sections use
18740 macinfo_label_base to distinguish labels.
18741 (output_macinfo): Likewise. Update macinfo_label_base.
18742 Pass in the line info label.
18743 (note_variable_value_in_expr): When generating LTO resolve
18744 all variable values here by generating DIEs as needed.
18745 (init_sections_and_labels): Add early LTO debug flag parameter
18746 and generate different sections and names if set. Add generation
18747 counter for the labels so we can have multiple of them.
18748 (reset_dies): Helper to allow DIEs to be output multiple times.
18749 (dwarf2out_finish): When outputting DIEs to the fat part of an
18750 LTO object first reset DIEs.
18751 (dwarf2out_early_finish): Output early DIEs when generating LTO.
18752 (modified_type_die): Check for decl_ultimate_origin being self
18754 (gen_type_die_with_usage): Likewise.
18755 (gen_typedef_die): Allow decl_ultimate_origin being self.
18756 (set_decl_abstract_flags): Remove.
18757 (set_block_abstract_flags): Likewise.
18758 (dwarf2out_abstract_function): Treat the early generated DIEs
18759 as the abstract copy and only add DW_AT_inline and
18760 DW_AT_artificial here and call set_decl_origin_self.
18761 If the DIE has an abstract origin don't do anything.
18762 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
18763 if we have none yet (Go fails to build one, PR78628).
18764 (variably_modified_type_p): Prevent endless recursion for Ada
18765 cyclic pointer types.
18766 * lto-streamer-in.c: Include debug.h.
18767 (dref_queue): New global.
18768 (lto_read_tree_1): Stream in DIE references.
18769 (lto_input_tree): Register DIE references.
18770 (input_function): Stream DECL_DEBUG_ARGS.
18771 * lto-streamer-out.c: Include debug.h.
18772 (lto_write_tree_1): Output DIE references.
18773 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
18774 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
18775 (output_function): Stream DECL_DEBUG_ARGS.
18776 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
18777 Stream DECL_ABSTRACT_ORIGIN.
18778 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
18779 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
18780 DECL_CONTEXT for file-scope decls.
18781 * lto-streamer.h (struct dref_entry): Declare.
18782 (dref_queue): Likewise.
18783 * cfgexpand.c (pass_expand::execute): Do not call the
18784 outlining_inline_function hook here.
18785 * lto-wrapper.c (debug_obj): New global.
18786 (tool_cleanup): Unlink it if required.
18787 (debug_objcopy): New function.
18788 (run_gcc): Handle early debug sections in the IL files by
18789 extracting them to separate files, partially linkin them and
18790 feeding the result back as result to the linker.
18791 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
18792 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
18793 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
18794 sections into a separate segment.
18795 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
18797 (darwin_asm_dwarf_section): Likewise.
18798 (darwin_asm_output_dwarf_offset): Likewise.
18799 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
18801 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18802 Alan Hayward <alan.hayward@arm.com>
18803 David Sherwood <david.sherwood@arm.com>
18805 * read-md.h (md_reader::record_potential_iterator_use): Replace
18806 pointer argument with an rtx and an index.
18807 * read-rtl.c (iterator_group::apply_iterator): Likewise.
18808 (apply_mode_iterator): Likewise.
18809 (apply_code_iterator): Likewise.
18810 (apply_int_iterator): Likewise.
18811 (apply_subst_iterator): Likewise.
18812 (record_iterator_use): Likewise.
18813 (record_attribute_use): Likewise.
18814 (md_reader::record_potential_iterator_use): Likewise. Update calls
18815 to record_iterator_use and apply_iterator.
18816 (iterator_use): Replace ptr with x and index.
18817 (attribute_use): Likewise.
18818 (apply_attribute_uses): Update calls to apply_iterator.
18819 (apply_iterators): Likewise. Update initialization of iterator_use.
18820 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
18821 and record_potential_iterator_use.
18822 (rtx_reader::read_rtx_operand): Likewise.
18824 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18825 Alan Hayward <alan.hayward@arm.com>
18826 David Sherwood <david.sherwood@arm.com>
18828 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
18831 2017-08-21 Richard Biener <rguenther@suse.de>
18833 PR middle-end/81884
18834 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
18835 at struct end conservatively when comparing common bases.
18837 2017-08-21 Richard Biener <rguenther@suse.de>
18839 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
18840 (mem_ref_in_stmt): Remove.
18841 (determine_max_movement): Use ref index to get at the reference.
18842 (invariantness_dom_walker::before_dom_children): Deal with
18843 lim data already initialized.
18844 (gather_mem_refs_stmt): Initialize lim data and record ref index.
18846 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18848 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
18849 (TARGET_ISA_ROUND): Ditto.
18850 (TARGET_ROUND): Ditto.
18851 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
18852 * config/i386/i386.md: Ditto.
18853 * config/i386/sse.md: Ditto.
18854 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
18855 with OPTION_MASK_ISA_SSE4_1.
18857 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18860 * doc/extend.texi (x86 Built-in Functions): Correct the name of
18861 __builtin_ia32_lzcnt_u16.
18863 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
18866 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
18867 (rs6000_set_current_function): Rewrite function to use it.
18869 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
18872 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
18873 and TYPE_WARN_IF_NOT_ALIGN.
18874 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
18875 (handle_warn_if_not_align): New.
18876 (place_union_field): Call handle_warn_if_not_align.
18877 (place_field): Call handle_warn_if_not_align.
18878 Copy TYPE_WARN_IF_NOT_ALIGN.
18879 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
18880 (layout_type): Likewise.
18881 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
18883 (tree_decl_common): Add warn_if_not_align.
18884 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
18885 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
18886 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
18887 (DECL_WARN_IF_NOT_ALIGN): Likewise.
18888 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
18889 * doc/extend.texi: Document warn_if_not_aligned attribute.
18890 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
18892 2017-08-17 Martin Liska <mliska@suse.cz>
18895 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
18896 (get_data_dependence): Use it as pointer type.
18897 (distribute_loop): Likewise.
18899 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18901 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
18902 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
18903 (p8_vmrgow_<mode>_direct): New define_insn.
18904 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
18905 handle endianness for vmrgew and vmrgow permute patterns.
18907 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18909 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
18910 * config/rs6000/rs6000-cpus.def: Remove comment.
18911 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
18912 (POWERPC_MASKS): Likewise.
18913 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
18914 use of TARGET_VSX_TIMODE.
18915 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
18916 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
18917 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
18918 (rs6000_option_override_internal): Remove dead code.
18919 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
18920 (rs6000_legitimize_reload_address): Likewise.
18921 (rs6000_legitimate_address_p): Likewise.
18922 (rs6000_opt_masks): Delete "vsx-timode".
18923 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
18924 from function comment.
18925 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
18926 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
18927 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
18929 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
18930 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
18931 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
18932 (VSX_TI): Likewise.
18934 (define_peephole2): Likewise.
18936 2017-08-17 Martin Sebor <msebor@redhat.com>
18939 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
18940 past the end of an array.
18941 (test_pp_format): Add test cases.
18943 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
18945 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
18946 missing ECF_NOTHROW flags.
18948 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18951 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
18952 operands residing in integer registers.
18953 (*vsx_le_perm_load_<mode>): Likewise.
18954 (*vsx_le_perm_store_<mode>): Likewise.
18955 (define_peephole2): Add peepholes to optimize the above.
18957 2017-08-17 Marek Polacek <polacek@redhat.com>
18959 PR middle-end/81814
18960 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
18961 to mimic what shorten_compare did. Change the return type to bool.
18962 (fold_cond_expr_with_comparison): Update call to
18963 operand_equal_for_comparison_p.
18964 (fold_ternary_loc): Likewise.
18966 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
18968 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
18970 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
18972 2017-08-17 Richard Biener <rguenther@suse.de>
18974 * tree-ssa-structalias.c (solve_graph): When propagating
18975 to successors update the graphs succ edges and avoid duplicate work.
18977 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
18980 * config/i386/i386.c (ix86_option_override_internal): Save target
18981 specific options after ix86_stack_protector_guard_reg was changed.
18983 2017-08-17 Richard Biener <rguenther@suse.de>
18985 PR tree-optimization/81827
18986 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
18987 (new_var_info): Initialize it conservatively.
18988 (get_call_vi): Mark register vars.
18989 (new_scalar_tmp_constraint_exp): Likewise.
18990 (handle_rhs_call): Likewise.
18991 (handle_const_call): Likewise.
18992 (create_function_info_for): Likewise.
18993 (solve_constraints): Sort varinfos to separate register from
18994 non-register vars to pack points-to solution bitmaps during
18997 2017-08-17 Marek Polacek <polacek@redhat.com>
18999 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19001 2017-08-17 Richard Biener <rguenther@suse.de>
19003 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19004 to true when overflow is undefined and we saturated the result.
19006 2017-08-17 Alan Modra <amodra@gmail.com>
19009 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19010 Don't use store multiple if only one reg needs saving.
19011 (interesting_frame_related_regno): New function.
19012 (rs6000_frame_related): Don't emit frame info for regs that
19014 (rs6000_emit_epilogue): Likewise.
19016 2017-08-16 Nathan Sidwell <nathan@acm.org>
19018 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19019 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19020 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19021 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19022 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19023 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19024 (verify_type): Adjust for TYPE_BINFO move.
19025 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19026 process TYPE_BINFO directly.
19027 (hash_tree): Likewise.
19028 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19030 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19033 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19035 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19037 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19040 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19041 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
19042 Add ix86_binary_operator_ok to insn constraint.
19043 (*iordi_1_bts): Ditto.
19044 (*xordi_1_btc): Ditto.
19045 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
19046 Update corresponding peephole2 pattern.
19050 2017-08-16 Bin Cheng <bin.cheng@arm.com>
19052 PR tree-optimization/81832
19053 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
19054 copy loop header which has IFN_LOOP_DIST_ALIAS call.
19056 2017-08-16 Marek Polacek <polacek@redhat.com>
19059 * fold-const.c (fold_indirect_ref_1): Restore original behavior
19060 regarding size_zero_node.
19062 2017-08-16 Martin Liska <mliska@suse.cz>
19065 * config.gcc: Respect previously set extra_objs in case
19068 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
19070 PR tree-optimization/81835
19071 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
19072 the phi SSA_NAME. Check that the condition in a COND_EXPR does
19073 not depend on the phi.
19075 2017-08-16 Alan Modra <amodra@gmail.com>
19077 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
19080 2017-08-16 Alan Modra <amodra@gmail.com>
19082 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
19083 (save_reg_p): ..into this. Update all callers.
19084 (first_reg_to_save): Simplify.
19086 2017-08-16 Alan Modra <amodra@gmail.com>
19088 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
19091 2017-08-15 Joseph Myers <joseph@codesourcery.com>
19095 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
19096 constant count if that count is less than 32.
19098 2017-08-15 Nathan Sidwell <nathan@acm.org>
19100 * gcc.c (execute): Emit friendlier message if inferior is killed
19101 by an external cause.
19103 2017-08-15 Richard Biener <rguenther@suse.de>
19105 PR tree-optimization/81790
19106 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
19107 CONSTRUCTORs from simplifying and VN.
19109 2017-08-14 Martin Sebor <msebor@redhat.com>
19111 * builtin-attrs.def: Add comments.
19113 2017-08-14 Martin Sebor <msebor@redhat.com>
19116 * doc/extend.texi (attribute nonstring): Document new attribute.
19118 2017-08-14 Martin Sebor <msebor@redhat.com>
19121 * tree-diagnostic.c (default_tree_printer): Handle %G.
19122 * gimple-pretty-print.h (percent_G_format): Declare new function.
19123 * gimple-pretty-print.c (percent_G_format): Define.
19124 * tree-pretty-print.c (percent_K_format): Add argument.
19126 2017-08-14 Martin Sebor <msebor@redhat.com>
19128 PR translation/79998
19129 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
19130 Remove a stray space.
19132 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
19135 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
19136 (*iordi_1_bts): Ditto.
19137 (*xordi_1_btc): Ditto.
19139 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19142 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
19143 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19145 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
19146 quoted strings, and make more translator-friendly.
19147 (darwin_rs6000_override_options): Likewise.
19148 (rs6000_option_override_internal): Likewise.
19149 (rs6000_return_in_memory): Fix overlong line.
19150 (init_cmulative_args): Use quoted strings, and make more
19151 translator-friendly.
19152 (rs6000_pass_by_reference): Fix overlong line.
19153 (def_builtin): Use quoted strings.
19154 (altivec_expand_predicate_builtin): Use quoted strings, and make
19155 more translator-friendly.
19156 (htm_expand_builtin): Use quoted strings.
19157 (cpu_expand_builtin): Use quoted strings, and make more
19158 translator-friendly.
19159 (altivec_expand_builtin): Likewise.
19160 (paired_expand_predicate_builtin): Likewise.
19161 (rs6000_invalid_builtin): Likewise.
19162 (builtin_function_type): Use quoted strings.
19163 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
19164 more translator-friendly.
19165 (rs6000_trampoline_init): Likewise.
19166 (rs6000_handle_altivec_attribute): Likewise.
19167 (rs6000_inner_target_options): Use quoted strings.
19168 (rs6000_disable_incompatible_switches): Likewise.
19169 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
19170 strings, and make more translator-friendly.
19171 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
19173 2017-08-14 Bin Cheng <bin.cheng@arm.com>
19175 PR tree-optimization/81799
19176 * tree-loop-distribution.c (version_loop_by_alias_check): Force
19177 cond_expr to simple gimple operand.
19179 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
19181 PR middle-end/46932
19182 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
19184 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
19188 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
19189 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
19190 TARGET_GASISR_PROLOGUES.
19191 * config/avr/avr.c (avr_option_override): Same.
19192 (avr_pass_pre_proep::execute): Same.
19194 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
19197 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
19198 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
19200 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
19202 * config/i386/i386.md (*load_tp_<mode>): Redefine as
19203 define_insn_and_split. Split to a memory load from 0 in
19204 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
19205 using PTR mode iterator.
19206 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
19207 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
19208 (*add_tp_<mode>): Redefine as define_insn_and_split.
19209 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
19210 address space. Merge with *add_tp_x32 using PTR mode iterator.
19211 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
19212 Split to an add with a memory load from 0 in
19213 DEFAULT_TLS_SEG_REG address space.
19215 2017-08-12 Andrew Pinski <apinski@cavium.com>
19217 * config/aarch64/aarch64-option-extensions.def (rdma):
19218 Fix feature string to what Linux prints out in /proc/cpuinfo.
19220 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
19223 * dwarf2out.c (modified_type_die): For C typedef types that have
19224 an ultimate origin, process the ultimate origin instead of the
19226 (gen_typedef_die): Assert that input DECLs have no ultimate
19228 (gen_type_die_with_usage): For typedef variants that have an
19229 ultimate origin, just call gen_decl_die on the original DECL.
19230 (process_scope_var): Avoid creating DIEs for local typedefs and
19231 concrete static variables.
19233 2017-08-12 Alan Modra <amodra@gmail.com>
19237 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
19238 match gnu-user.h startfile.
19239 (ENDFILE_LINUX_SPEC): Similarly.
19241 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
19244 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
19246 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
19248 2017-08-11 Tamar Christina <tamar.christina@arm.com>
19249 * config/aarch64/aarch64.md (mov<mode>): Change.
19250 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19251 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
19252 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
19254 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
19256 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
19257 for storage order barriers.
19259 2017-08-11 Martin Liska <mliska@suse.cz>
19262 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
19263 variables of void type.
19265 2017-08-11 Martin Liska <mliska@suse.cz>
19267 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
19268 TARGET_SUPPORTS_ALIASES.
19269 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
19270 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
19271 (optimize_weakref): Likewise.
19272 * symtab.c (symtab_node::noninterposable_alias): Likewise.
19273 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
19274 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
19276 2017-08-11 Martin Liska <mliska@suse.cz>
19279 * config/i386/i386.c (make_resolver_func): Do complete
19280 refactoring of the function.
19282 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19285 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
19286 * config/i386/i386.c (ix86_stack_protect_guard): Use
19287 ix86_stack_protect_guard_symbol_str to generate varible declaration.
19288 * doc/invoke.texi (x86 Options): Document
19289 -mstack-protector-guard-symbol= option.
19291 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19293 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
19294 * config/i386/i386.c (ix86_split_stack_guard): New function.
19295 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
19296 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
19297 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
19298 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
19299 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
19300 (split_stack_space_check): Call ix86_split_stack_guard.
19302 2017-08-10 Martin Sebor <msebor@redhat.com>
19304 * print-tree.c (print_node): Print location using the established
19306 Replace spaces with colons.
19307 (debug_raw, debug): Ditto.
19309 2017-08-10 Martin Sebor <msebor@redhat.com>
19312 * pretty-print.c (pp_format): Correct the handling of %s precision.
19314 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
19317 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19319 (ix86_finalize_stack_frame_flags): This. Also clear
19320 frame_pointer_needed if -fno-omit-frame-pointer is used without
19322 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19323 with ix86_finalize_stack_frame_flags.
19324 (ix86_expand_epilogue): Likewise.
19325 (ix86_expand_split_stack_prologue): Likewise.
19326 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
19328 2017-08-10 Martin Liska <mliska@suse.cz>
19331 * c-attribs.c (handle_target_attribute):
19332 Report warning for an empty string argument of target attribute.
19334 2017-08-09 Jakub Jelinek <jakub@redhat.com>
19337 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
19339 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
19340 or DECL_NONLOCAL labels.
19341 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
19342 or DECL_NONLOCAL labels here.
19344 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
19346 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
19347 to indicate when early gimple folding has been disabled.
19348 (rs6000_gimple_fold_builtin): Add debug content.
19349 (rs6000_invalid_builtin): Fix whitespace.
19350 (rs6000_expand_builtin): Fix whitespace.
19351 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
19353 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
19356 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
19357 SAVE_MULTIPLE if not all the registers that saves, should be saved.
19359 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
19361 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
19362 (qdf24xx): Likewise.
19363 * config/aarch64/aarch64.md: Include falkor.md.
19364 * config/aarch64/falkor.md: New.
19366 2017-08-09 Marek Polacek <polacek@redhat.com>
19369 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
19370 * diagnostic.c (emit_diagnostic): Add a comment.
19371 (emit_diagnostic_valist): New function.
19373 2017-08-09 Marek Polacek <polacek@redhat.com>
19376 * input.c (make_location): New overload.
19377 * input.h (make_location): Declare.
19379 2017-08-08 Alan Modra <amodra@gmail.com>
19380 H.J. Lu <hongjiu.lu@intel.com>
19383 * gcc.c (NO_PIE_SPEC): Delete.
19384 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
19386 (LINK_PIE_SPEC): ..to here.
19387 (LINK_COMMAND_SPEC): Support -no-pie.
19388 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
19389 chain of crtbegin*.o selection, update for PIE_SPEC changes and
19391 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
19392 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
19393 (ENDFILE_CRTEND_SPEC): Similarly.
19395 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
19398 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
19399 (mstack-protector-guard-offset=): Ditto.
19400 * config/i386/i386.c (ix86_option_override): Handle
19401 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
19403 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
19404 ix86_stack_protect_guard_offset variables.
19405 (TARGET_STACK_PROTECT_GUARD): Always define.
19406 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
19407 and -mstack-protector-guard-offset= options.
19409 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19411 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
19412 boundary case for the last candidate.
19414 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19416 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
19417 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
19419 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19421 PR middle-end/19706
19422 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
19423 * config/aarch64/aarch64-builtins.c
19424 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
19425 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
19426 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
19428 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19429 Andrew Pinski <pinskia@gmail.com>
19431 PR middle-end/19706
19432 * internal-fn.def (XORSIGN): New.
19433 * optabs.def (xorsign_optab): New.
19434 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
19435 (convert_expand_mult_copysign): New.
19436 (pass_optimize_widening_mul::execute): Call
19437 convert_expand_mult_copysign.
19439 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19441 PR tree-optimization/81354
19442 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
19443 Insert on edges rather than explicitly creating landing pads.
19444 (analyze_candidates_and_replace): Commit edge inserts.
19446 2017-08-08 Richard Biener <rguenther@suse.de>
19448 PR middle-end/81719
19449 * tree-ssa-loop-niter.c: Include tree-dfa.h.
19450 (expand_simple_operations): Also look through ADDR_EXPRs with
19451 MEM_REF bases treating them as POINTER_PLUS_EXPR.
19453 2017-08-08 Richard Biener <rguenther@suse.de>
19455 PR tree-optimization/81723
19456 * tree-vect-slp.c (struct bst_traits): New hash traits.
19457 (bst_fail): New global.
19458 (vect_build_slp_tree_2): New worker, split out from ...
19459 (vect_build_slp_tree): ... this now wrapping it with using
19460 bst_fail set to cache SLP tree build fails. Properly handle
19462 (vect_analyze_slp_instance): Allocate and free bst_fail.
19464 2017-08-08 Martin Liska <mliska@suse.cz>
19467 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
19468 LABEL_DECLs that can be from a different function.
19470 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19472 PR tree-optimization/81744
19473 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
19474 loop's number of iterations.
19476 2017-08-08 Martin Liska <mliska@suse.cz>
19478 * asan.c: Include header files.
19479 * attribs.c (build_decl_attribute_variant): New function moved
19481 (build_type_attribute_qual_variant): Likewise.
19482 (cmp_attrib_identifiers): Likewise.
19483 (simple_cst_list_equal): Likewise.
19484 (omp_declare_simd_clauses_equal): Likewise.
19485 (attribute_value_equal): Likewise.
19486 (comp_type_attributes): Likewise.
19487 (build_type_attribute_variant): Likewise.
19488 (lookup_ident_attribute): Likewise.
19489 (remove_attribute): Likewise.
19490 (merge_attributes): Likewise.
19491 (merge_type_attributes): Likewise.
19492 (merge_decl_attributes): Likewise.
19493 (merge_dllimport_decl_attributes): Likewise.
19494 (handle_dll_attribute): Likewise.
19495 (attribute_list_equal): Likewise.
19496 (attribute_list_contained): Likewise.
19497 * attribs.h (lookup_attribute): New function moved from tree.[ch].
19498 (lookup_attribute_by_prefix): Likewise.
19499 * bb-reorder.c: Include header files.
19500 * builtins.c: Likewise.
19501 * calls.c: Likewise.
19502 * cfgexpand.c: Likewise.
19503 * cgraph.c: Likewise.
19504 * cgraphunit.c: Likewise.
19505 * convert.c: Likewise.
19506 * dwarf2out.c: Likewise.
19507 * final.c: Likewise.
19508 * fold-const.c: Likewise.
19509 * function.c: Likewise.
19510 * gimple-expr.c: Likewise.
19511 * gimple-fold.c: Likewise.
19512 * gimple-pretty-print.c: Likewise.
19513 * gimple.c: Likewise.
19514 * gimplify.c: Likewise.
19515 * hsa-common.c: Likewise.
19516 * hsa-gen.c: Likewise.
19517 * internal-fn.c: Likewise.
19518 * ipa-chkp.c: Likewise.
19519 * ipa-cp.c: Likewise.
19520 * ipa-devirt.c: Likewise.
19521 * ipa-fnsummary.c: Likewise.
19522 * ipa-inline.c: Likewise.
19523 * ipa-visibility.c: Likewise.
19525 * lto-cgraph.c: Likewise.
19526 * omp-expand.c: Likewise.
19527 * omp-general.c: Likewise.
19528 * omp-low.c: Likewise.
19529 * omp-offload.c: Likewise.
19530 * omp-simd-clone.c: Likewise.
19531 * opts-global.c: Likewise.
19532 * passes.c: Likewise.
19533 * predict.c: Likewise.
19534 * sancov.c: Likewise.
19535 * sanopt.c: Likewise.
19536 * symtab.c: Likewise.
19537 * toplev.c: Likewise.
19538 * trans-mem.c: Likewise.
19539 * tree-chkp.c: Likewise.
19540 * tree-eh.c: Likewise.
19541 * tree-into-ssa.c: Likewise.
19542 * tree-object-size.c: Likewise.
19543 * tree-parloops.c: Likewise.
19544 * tree-profile.c: Likewise.
19545 * tree-ssa-ccp.c: Likewise.
19546 * tree-ssa-live.c: Likewise.
19547 * tree-ssa-loop.c: Likewise.
19548 * tree-ssa-sccvn.c: Likewise.
19549 * tree-ssa-structalias.c: Likewise.
19550 * tree-ssa.c: Likewise.
19551 * tree-streamer-in.c: Likewise.
19552 * tree-vectorizer.c: Likewise.
19553 * tree-vrp.c: Likewise.
19554 * tsan.c: Likewise.
19555 * ubsan.c: Likewise.
19556 * varasm.c: Likewise.
19557 * varpool.c: Likewise.
19558 * tree.c: Remove functions moved to attribs.[ch].
19559 * tree.h: Likewise.
19560 * config/aarch64/aarch64.c: Add attrs.h header file.
19561 * config/alpha/alpha.c: Likewise.
19562 * config/arc/arc.c: Likewise.
19563 * config/arm/arm.c: Likewise.
19564 * config/avr/avr.c: Likewise.
19565 * config/bfin/bfin.c: Likewise.
19566 * config/c6x/c6x.c: Likewise.
19567 * config/cr16/cr16.c: Likewise.
19568 * config/cris/cris.c: Likewise.
19569 * config/darwin.c: Likewise.
19570 * config/epiphany/epiphany.c: Likewise.
19571 * config/fr30/fr30.c: Likewise.
19572 * config/frv/frv.c: Likewise.
19573 * config/ft32/ft32.c: Likewise.
19574 * config/h8300/h8300.c: Likewise.
19575 * config/i386/winnt.c: Likewise.
19576 * config/ia64/ia64.c: Likewise.
19577 * config/iq2000/iq2000.c: Likewise.
19578 * config/lm32/lm32.c: Likewise.
19579 * config/m32c/m32c.c: Likewise.
19580 * config/m32r/m32r.c: Likewise.
19581 * config/m68k/m68k.c: Likewise.
19582 * config/mcore/mcore.c: Likewise.
19583 * config/microblaze/microblaze.c: Likewise.
19584 * config/mips/mips.c: Likewise.
19585 * config/mmix/mmix.c: Likewise.
19586 * config/mn10300/mn10300.c: Likewise.
19587 * config/moxie/moxie.c: Likewise.
19588 * config/msp430/msp430.c: Likewise.
19589 * config/nds32/nds32-isr.c: Likewise.
19590 * config/nds32/nds32.c: Likewise.
19591 * config/nios2/nios2.c: Likewise.
19592 * config/nvptx/nvptx.c: Likewise.
19593 * config/pa/pa.c: Likewise.
19594 * config/pdp11/pdp11.c: Likewise.
19595 * config/powerpcspe/powerpcspe.c: Likewise.
19596 * config/riscv/riscv.c: Likewise.
19597 * config/rl78/rl78.c: Likewise.
19598 * config/rx/rx.c: Likewise.
19599 * config/s390/s390.c: Likewise.
19600 * config/sh/sh.c: Likewise.
19601 * config/sol2.c: Likewise.
19602 * config/sparc/sparc.c: Likewise.
19603 * config/spu/spu.c: Likewise.
19604 * config/stormy16/stormy16.c: Likewise.
19605 * config/tilegx/tilegx.c: Likewise.
19606 * config/tilepro/tilepro.c: Likewise.
19607 * config/v850/v850.c: Likewise.
19608 * config/vax/vax.c: Likewise.
19609 * config/visium/visium.c: Likewise.
19610 * config/xtensa/xtensa.c: Likewise.
19612 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19615 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
19616 constraints since the -mupper-regs-* switches have been
19618 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
19619 into a vector from a double word element that was extracted from
19620 another vector, and eliminate extra XXPERMDI instructions.
19621 (vsx_concat_<mode>_2): Likewise.
19622 (vsx_concat_<mode>_3): Likewise.
19623 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
19624 concat to allow optimizing inserts from previous extracts.
19626 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
19628 * config/i386/i386.c (ix86_stack_protect_guard): Generate
19629 memory reference to a SSP offset in TLS address space.
19630 (ix86_print_operand) <case '@'>: Remove.
19631 (ix86_print_operand_punct_valid_p): Remove '@' code.
19632 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
19633 UNSPEC_SP_TLS_TEST.
19634 (stack_tls_protect_set_<mode>): Remove.
19635 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
19636 (stack_tls_protect_test_<mode>): Remove.
19637 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
19639 2017-08-07 Olivier Hainque <hainque@adacore.com>
19642 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
19644 2017-08-07 Douglas Rupp <rupp@adacore.com>
19646 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
19647 variable was referenced as multidir in command.
19649 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19652 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
19655 2017-08-07 Martin Liska <mliska@suse.cz>
19657 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
19658 * config/rl78/rl78.c: Add include of attribs.h.
19659 * config/sh/sh.c: Likewise.
19660 * config/v850/v850.c: Likewise.
19662 2017-08-07 Tom de Vries <tom@codesourcery.com>
19664 PR middle-end/78266
19665 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
19667 2017-08-07 Martin Liska <mliska@suse.cz>
19669 * config/mips/mips.c: Include attribs.h.
19671 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19674 * doc/invoke.texi: Document change in behvaior for -Ofast for
19677 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
19679 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
19681 (aarch64_pop_regs): Likewise.
19682 (aarch64_gen_load_pair): Likewise.
19683 (aarch64_save_callee_saves): Likewise.
19684 (aarch64_restore_callee_saves): Likewise.
19686 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19688 * config/i386/i386.c: Revert the last change.
19690 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19693 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19695 (ix86_finalize_stack_frame_flags): This. Also clear
19696 frame_pointer_needed if -fno-omit-frame-pointer is used without
19698 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19699 with ix86_finalize_stack_frame_flags.
19700 (ix86_expand_epilogue): Likewise.
19701 (ix86_expand_split_stack_prologue): Likewise.
19703 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19706 * config/i386/i386.c (get_builtin_code_for_version): Set priority
19707 to P_AES for Westmere.
19709 2017-08-07 Jonathan Yong <10walls@gmail.com>
19711 * config/i386/mingw.opt (fset-stack-executable): Removed.
19712 * config/i386/cygming.opt (fset-stack-executable): Moved
19714 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
19716 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
19718 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
19720 2017-08-07 Marek Polacek <polacek@redhat.com>
19722 PR middle-end/81737
19723 * fold-const.c (fold_indirect_ref_1): Check type_domain.
19725 2017-08-07 Martin Liska <mliska@suse.cz>
19727 * attribs.h (canonicalize_attr_name): New function.
19728 (cmp_attribs): Move from c-format.c and adjusted.
19729 (is_attribute_p): Moved from tree.h.
19730 * tree-inline.c: Add new includes.
19731 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
19732 (private_is_attribute_p): Remove.
19733 (private_lookup_attribute): Likewise.
19734 (private_lookup_attribute_by_prefix): Simplify.
19735 (remove_attribute): Use is_attribute_p.
19736 * tree.h: Remove removed declarations.
19738 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19740 PR middle-end/81698
19741 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
19742 instead of computing it in the function. Formatting fix.
19743 (expand_case): Don't rely on default_edge being the first edge,
19744 clear it if removing it, pass default_edge to
19745 emit_case_dispatch_table.
19746 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
19749 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
19751 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
19752 insn in the function, emit NOP after the insn.
19754 2017-08-06 Tom de Vries <tom@codesourcery.com>
19756 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
19759 2017-08-06 Tom de Vries <tom@codesourcery.com>
19761 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
19764 2017-08-04 Yury Gribov <tetra2005@gmail.com>
19766 PR tree-optimization/57371
19767 * match.pd: New pattern.
19769 2017-08-04 Marek Polacek <polacek@redhat.com>
19771 PR middle-end/81695
19772 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
19773 perform the computation in offset_int.
19775 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19777 PR tree-optimization/81136
19778 * tree-vectorizer.h: Include tree-hash-traits.h.
19779 (vec_base_alignments): New typedef.
19780 (vec_info): Add a base_alignments field.
19781 (vect_record_base_alignments): Declare.
19782 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
19784 (DR_IS_CONDITIONAL_IN_STMT): New macro.
19785 (create_data_ref): Add an is_conditional_in_stmt argument.
19786 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
19787 the is_conditional_in_stmt field.
19788 (data_ref_loc): Add an is_conditional_in_stmt field.
19789 (get_references_in_stmt): Set the is_conditional_in_stmt field.
19790 (find_data_references_in_stmt): Update call to create_data_ref.
19791 (graphite_find_data_references_in_stmt): Likewise.
19792 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
19793 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
19794 (vect_record_base_alignment): New function.
19795 (vect_record_base_alignments): Likewise.
19796 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
19797 for nested statements even if we fail to compute a misalignment.
19798 Use pooled base alignments for unconditional references.
19799 (vect_find_same_alignment_drs): Compare base addresses instead
19801 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
19802 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
19804 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19806 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
19807 Add an explicit name for the enum. Use auto_vec for slp_instances
19808 and grouped_stores.
19809 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
19811 (_bb_vec_info): Add a constructor and destructor.
19812 (vinfo_for_stmt): Return NULL for uids of -1 as well.
19813 (destroy_loop_vec_info): Delete.
19814 (vect_destroy_datarefs): Likewise.
19815 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
19816 (vec_info::vec_info): New function.
19817 (vec_info::~vec_info): Likewise.
19818 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
19819 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
19820 destroy_loop_vec_info.
19821 * tree-vect-loop.c (new_loop_vec_info): Replace with...
19822 (_loop_vec_info::_loop_vec_info): ...this.
19823 (destroy_loop_vec_info): Replace with...
19824 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
19825 the stmt_vec_infos. Leave handling of vec_info information to its
19826 destructor. Remove explicit vector releases.
19827 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
19828 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
19829 * tree-vect-slp.c (new_bb_vec_info): Replace with...
19830 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
19831 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
19832 (destroy_bb_vec_info): Replace with...
19833 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
19834 information to its destructor.
19835 (vect_slp_analyze_bb_1): Use new and delete instead of
19836 new_bb_vec_info and destroy_bb_vec_info.
19837 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
19840 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19842 * tree-data-ref.h (subscript): Add access_fn field.
19843 (data_dependence_relation): Add could_be_independent_p.
19844 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
19845 (same_access_functions): Move to tree-data-ref.c.
19846 * tree-data-ref.c (ref_contains_union_access_p): New function.
19847 (access_fn_component_p): Likewise.
19848 (access_fn_components_comparable_p): Likewise.
19849 (dr_analyze_indices): Add a reference to access_fn_component_p.
19850 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
19852 (constant_access_functions): Likewise.
19853 (add_other_self_distances): Likewise.
19854 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
19855 (initialize_data_dependence_relation): Use XCNEW and remove
19856 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
19857 of access functions that have the same type. Allow the
19858 subsequence to end with different bases in some circumstances.
19859 Record the chosen access functions in SUB_ACCESS_FN.
19860 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
19861 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
19862 (subscript_dependence_tester_1): Likewise dra and drb.
19863 (build_classic_dist_vector): Update calls accordingly.
19864 (subscript_dependence_tester): Likewise.
19865 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
19866 DDR_COULD_BE_INDEPENDENT_P.
19867 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
19868 comp_alias_ddrs instead of may_alias_ddrs.
19869 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
19871 (vect_analyze_data_ref_dependence): Use it if
19872 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
19873 distance vectors if that fails.
19874 (dependence_distance_ge_vf): New function.
19875 (vect_prune_runtime_alias_test_list): Use it. Don't clear
19876 LOOP_VINFO_MAY_ALIAS_DDRS.
19878 2017-08-04 Richard Biener <rguenther@suse.de>
19880 PR middle-end/81705
19881 * fold-const.c (fold_binary_loc): Properly restrict
19882 minus_var0 && minus_var1 case when associating undefined overflow
19885 2017-08-04 Tom de Vries <tom@codesourcery.com>
19887 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
19889 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19891 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19892 Don't start diagnostic messages with a capital letter.
19893 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19895 (rs6000_invalid_builtin): Likewise.
19896 (rs6000_trampoline_init): Likewise.
19898 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19901 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
19902 after setting changeable df flags.
19904 2017-08-03 Richard Biener <rguenther@suse.de>
19906 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
19907 up if the use is in USE - X.
19909 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
19911 * toplev.c (dumpfile.h): New include.
19912 (internal_error_reentered): New static function. Use it...
19913 (internal_error_function): ...here to handle reentered internal_error.
19915 2017-08-03 Richard Biener <rguenther@suse.de>
19917 PR middle-end/81148
19918 * fold-const.c (split_tree): Add minus_var and minus_con
19919 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
19920 here but always use minus_*.
19921 (associate_trees): Assert we never associate with MINUS_EXPR
19922 and NULL first operand. Do not recurse for PLUS_EXPR operands
19923 when associating as MINUS_EXPR either.
19924 (fold_binary_loc): Track minus_var and minus_con.
19926 2017-08-03 Tom de Vries <tom@codesourcery.com>
19929 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
19930 ACCEL_COMPILER, apply finish_options on
19931 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
19933 2017-08-03 Tom de Vries <tom@codesourcery.com>
19936 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
19937 function_entry_patch_area_size > 0.
19939 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19942 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
19943 instead of 10??LU, perform unit multiplication in wide_int,
19944 don't change alloc_object_size_limit if the limit is larger
19947 PR tree-optimization/81655
19948 PR tree-optimization/81588
19949 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
19950 the case when ranges[i].low and high are 1 for unsigned type with
19953 PR middle-end/81052
19954 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
19955 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
19957 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19959 * tree-vrp.h: Add include guard.
19961 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
19964 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
19965 (ud2): New insn pattern.
19966 * config/i386/i386.c (ix86_expand_epilogue):
19967 For naked functions, generate ud2 instead of trap insn.
19969 2017-08-02 Marek Polacek <polacek@redhat.com>
19972 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
19974 2017-08-02 Tom de Vries <tom@codesourcery.com>
19975 Cesar Philippidis <cesar@codesourcery.com>
19977 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
19978 Add missing edge probabilities.
19980 2017-08-02 Tamar Christina <tamar.christina@arm.com>
19982 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
19983 Correct endianness.
19985 2017-08-02 Jakub Jelinek <jakub@redhat.com>
19987 PR middle-end/79499
19988 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
19989 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
19990 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
19991 prologue_seq sequences - if any.
19993 2017-08-02 Richard Biener <rguenther@suse.de>
19995 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
19996 via vectors if supported, integer extracts via punning if supported
19997 or otherwise vector extracts.
19999 2017-08-02 Richard Biener <rguenther@suse.de>
20001 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20003 (bitmap_insert_into_set): ... this.
20005 2017-08-02 Richard Biener <rguenther@suse.de>
20007 PR tree-optimization/81633
20009 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20011 PR tree-optimization/71752
20012 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20014 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20016 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20017 (machine_function::call_ms2sysv_pad_out): Remove field.
20018 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20019 (ix86_compute_frame_layout): Likewise.
20021 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20024 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20025 attribute with interrupt attribute.
20027 2017-08-01 Andrew Pinski <apinski@cavium.com>
20029 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20030 BIT_INSERT_EXPR's operand 1
20031 to see if the types precision matches.
20033 2017-08-01 Martin Liska <mliska@suse.cz>
20035 PR middle-end/70140
20036 * builtins.c (expand_builtin_memcpy_args): Remove.
20037 (expand_builtin_memcpy): Call newly added function
20038 expand_builtin_memory_copy_args.
20039 (expand_builtin_memcpy_with_bounds): Likewise.
20040 (expand_builtin_mempcpy): Remove last argument.
20041 (expand_builtin_mempcpy_with_bounds): Likewise.
20042 (expand_builtin_memory_copy_args): New function created from
20043 expand_builtin_mempcpy_args with small modifications.
20044 (expand_builtin_mempcpy_args): Remove.
20045 (expand_builtin_stpcpy): Remove unused argument.
20046 (expand_builtin): Likewise.
20047 (expand_builtin_with_bounds): Likewise.
20049 2017-08-01 Martin Liska <mliska@suse.cz>
20052 Make mempcpy more optimal (PR middle-end/70140).
20054 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20057 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
20058 __builtin_vec_cmpne verify both arguments are compatible vectors
20059 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
20060 verify arg1_type is a pointer or array type. For __builtin_vec_st,
20061 move computation of aligned to after checking the argument types.
20065 * config/rs6000/vsx.md (vextract_fp_from_shorth,
20066 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
20069 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
20070 Doug Rupp <rupp@adacore.com>
20071 Olivier Hainque <hainque@adacore.com>
20073 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
20074 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
20076 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
20077 for TARGET_OS_CPP_BUILTIN.
20078 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
20079 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
20081 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
20082 passing required abi options to the assembler for EABI configurations.
20083 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
20084 of .text.hot and .text.unlikely sections for kernel modules when
20085 using ARM style exceptions.
20086 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
20087 options. Add EXTRA_CC1_SPEC.
20088 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
20090 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
20092 (ARM_TARGET2_DWARF_FORMAT): Define.
20093 * config/arm/t-vxworks: Adjust multilib control to removal of the
20094 Diab command line options.
20096 2017-08-01 Martin Liska <mliska@suse.cz>
20098 PR gcov-profile/81561
20099 * gcov.c (unblock): Make unblocking safe as we need to preserve
20100 index correspondence of blocks and block_lists.
20102 2017-08-01 Richard Biener <rguenther@suse.de>
20104 PR tree-optimization/81181
20105 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
20106 (compute_antic): ... end of iteration here.
20108 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
20110 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
20111 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
20112 (ftree-slp-vectorize): Likewise.
20113 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
20114 can no longer be set independent of flag_tree_loop_vectorize.
20115 * omp-general.c (emp_max_vf): Likewise.
20116 * opts.c (enable_fdo_optimizations): Remove references to
20117 flag_tree_vectorize, these are now implicit.
20118 (common_handle_option): Remove handling for OPT_ftree_vectorize,
20119 and leave it for the options machinery.
20121 2017-08-01 Martin Liska <mliska@suse.cz>
20123 PR middle-end/70140
20124 * builtins.c (expand_builtin_memcpy_args): Remove.
20125 (expand_builtin_memcpy): Call newly added function
20126 expand_builtin_memory_copy_args.
20127 (expand_builtin_memcpy_with_bounds): Likewise.
20128 (expand_builtin_mempcpy): Remove last argument.
20129 (expand_builtin_mempcpy_with_bounds): Likewise.
20130 (expand_builtin_memory_copy_args): New function created from
20131 expand_builtin_mempcpy_args with small modifications.
20132 (expand_builtin_mempcpy_args): Remove.
20133 (expand_builtin_stpcpy): Remove unused argument.
20134 (expand_builtin): Likewise.
20135 (expand_builtin_with_bounds): Likewise.
20137 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20140 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
20141 print "ds:" only for immediates in generic address space.
20143 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20146 * config/i386/i386.c (ix86_funciton_naked): New prototype.
20147 (ix86_function_ok_for_sibcall): Return false for naked functions.
20149 2017-08-01 Richard Biener <rguenther@suse.de>
20151 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
20152 (compute_antic): Seed worklist with exit block predecessors.
20153 * cfganal.c (dfs_find_deadend): For a cycle return the source
20154 of the edge closing it.
20156 2017-08-01 Tamar Christina <tamar.christina@arm.com>
20158 * config/aarch64/aarch64.c
20159 (aarch64_can_const_movi_rtx_p): Move 0 check.
20161 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20163 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
20164 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
20166 * match.pd: Ditto in address comparison pattern.
20168 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20170 PR tree-optimization/81627
20171 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
20172 closed ssa form for store-store chain.
20174 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20176 PR tree-optimization/81620
20177 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
20178 for store-store chain.
20180 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20182 PR tree-optimization/81588
20183 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
20184 ranges[i].in_p, invert comparison code ccode. For >/>=,
20185 swap rhs1 and rhs2 and comparison code unconditionally,
20186 for </<= don't do that. Don't swap rhs1/rhs2 again if
20187 ranges[i].in_p, instead invert comparison code ccode if
20188 opcode or oe->rank is BIT_IOR_EXPR.
20191 * optabs.def (vec_extract_optab, vec_init_optab): Change from
20192 a direct optab to conversion optab.
20193 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
20194 with GET_MODE_INNER as last argument instead of optab_handler.
20195 * expmed.c (extract_bit_field_1): Likewise. Use vector from
20196 vector extraction if possible and optab is available.
20197 * expr.c (store_constructor): Use convert_optab_handler instead
20198 of optab_handler. Use vector initialization from smaller
20199 vectors if possible and optab is available.
20200 * tree-vect-stmts.c (vectorizable_load): Likewise.
20201 * doc/md.texi (vec_extract, vec_init): Document that the optabs
20202 now have two modes.
20203 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
20204 of vec_init from half-sized vectors with the same element mode.
20205 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
20206 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
20207 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
20208 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
20209 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
20210 after mode in gen_vec_extract* calls.
20211 (vec_extract<mode>): Renamed to ...
20212 (vec_extract<mode><ssescalarmodelower>): ... this.
20213 (vec_extract<mode><ssehalfvecmodelower>): New expander.
20214 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
20215 element mode after mode in gen_vec_init* calls.
20216 (VEC_INIT_HALF_MODE): New mode iterator.
20217 (vec_init<mode>): Renamed to ...
20218 (vec_init<mode><ssescalarmodelower>): ... this.
20219 (vec_init<mode><ssehalfvecmodelower>): New expander.
20220 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
20221 (vec_extractv2sfsf): ... this.
20222 (vec_initv2sf): Renamed to ...
20223 (vec_initv2sfsf): ... this.
20224 (vec_extractv2si): Renamed to ...
20225 (vec_extractv2sisi): ... this.
20226 (vec_initv2si): Renamed to ...
20227 (vec_initv2sisi): ... this.
20228 (vec_extractv4hi): Renamed to ...
20229 (vec_extractv4hihi): ... this.
20230 (vec_initv4hi): Renamed to ...
20231 (vec_initv4hihi): ... this.
20232 (vec_extractv8qi): Renamed to ...
20233 (vec_extractv8qiqi): ... this.
20234 (vec_initv8qi): Renamed to ...
20235 (vec_initv8qiqi): ... this.
20236 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
20237 (vec_init<mode>): Renamed to ...
20238 (vec_init<mode><VEC_base_l>): ... this.
20239 (vec_extract<mode>): Renamed to ...
20240 (vec_extract<mode><VEC_base_l>): ... this.
20241 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
20242 (vec_initv2sfsf): ... this.
20243 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
20244 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20245 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
20246 element mode after mode in gen_vec_init* calls.
20247 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
20248 (vec_init<mode><Vel>): ... this.
20249 (vec_extract<mode>): Renamed to ...
20250 (vec_extract<mode><Vel>): ... this.
20251 * config/aarch64/iterators.md (Vel): New mode attribute.
20252 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
20253 Add element mode after mode in gen_vec_extract* calls.
20254 * config/s390/vector.md (non_vec_l): New mode attribute.
20255 (vec_extract<mode>): Renamed to ...
20256 (vec_extract<mode><non_vec_l>): ... this.
20257 (vec_init<mode>): Renamed to ...
20258 (vec_init<mode><non_vec_l>): ... this.
20259 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
20260 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
20262 * config/arm/iterators.md (V_elem_l): New mode attribute.
20263 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
20264 (vec_extract<mode><V_elem_l>): ... this.
20265 (vec_extractv2di): Renamed to ...
20266 (vec_extractv2didi): ... this.
20267 (vec_init<mode>): Renamed to ...
20268 (vec_init<mode><V_elem_l>): ... this.
20269 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
20270 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
20271 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
20272 Add element mode after gen_vec_extract* calls.
20273 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
20274 (vec_init<mode><unitmode>): ... this.
20275 (vec_extract<mode>): Renamed to ...
20276 (vec_extract<mode><unitmode>): ... this.
20277 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
20278 (vec_init<mode><unitmode>): ... this.
20279 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
20280 (vec_initv2sfsf): ... this.
20281 (vec_extractv2sf): Renamed to ...
20282 (vec_extractv2sfsf): ... this.
20283 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
20284 Add element mode after gen_vec_extract* calls.
20285 * config/mips/mips.md (unitmode): New mode iterator.
20286 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
20287 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
20288 * config/spu/spu.md (inner_l): New mode attribute.
20289 (vec_init<mode>): Renamed to ...
20290 (vec_init<mode><inner_l>): ... this.
20291 (vec_extract<mode>): Renamed to ...
20292 (vec_extract<mode><inner_l>): ... this.
20293 * config/sparc/sparc.md (veltmode): New mode iterator.
20294 (vec_init<VMALL:mode>): Renamed to ...
20295 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
20296 * config/ia64/vect.md (vec_initv2si): Renamed to ...
20297 (vec_initv2sisi): ... this.
20298 (vec_initv2sf): Renamed to ...
20299 (vec_initv2sfsf): ... this.
20300 (vec_extractv2sf): Renamed to ...
20301 (vec_extractv2sfsf): ... this.
20302 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
20303 (vec_init<mode>): Renamed to ...
20304 (vec_init<mode><VEC_base_l>): ... this.
20305 (vec_extract<mode>): Renamed to ...
20306 (vec_extract<mode><VEC_base_l>): ... this.
20307 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
20308 (vec_initv2sfsf): ... this.
20309 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
20310 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20311 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
20312 gen_vec_init* calls.
20314 2017-08-01 Richard Biener <rguenther@suse.de>
20316 PR tree-optimization/81297
20317 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
20318 TREE_OVERFLOW from INTEGER_CSTs.
20320 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
20322 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
20324 2017-07-31 Carl Love <cel@us.ibm.com>
20326 * config/rs6000/rs6000-c: Add support for built-in functions
20327 vector signed char vec_xl_be (signed long long, signed char *);
20328 vector unsigned char vec_xl_be (signed long long, unsigned char *);
20329 vector signed int vec_xl_be (signed long long, signed int *);
20330 vector unsigned int vec_xl_be (signed long long, unsigned int *);
20331 vector signed long long vec_xl_be (signed long long, signed long long *);
20332 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
20333 vector signed short vec_xl_be (signed long long, signed short *);
20334 vector unsigned short vec_xl_be (signed long long, unsigned short *);
20335 vector double vec_xl_be (signed long long, double *);
20336 vector float vec_xl_be (signed long long, float *);
20337 * config/rs6000/altivec.h (vec_xl_be): Add #define.
20338 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
20339 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
20341 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
20342 (altivec_expand_builtin): Add switch statement to call
20343 altivec_expand_xl_be for each builtin.
20344 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
20345 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
20346 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
20347 __builtin_vsx_le_be_v16qi.
20348 * doc/extend.texi: Update the built-in documentation file for the
20349 new built-in functions.
20351 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20354 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
20356 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20358 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20360 * config.gcc: Add z14.
20361 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
20362 CPU model numbers for z13s and z14.
20363 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
20365 * config/s390/s390-opts.h (enum processor_type): Rename
20366 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20367 * config/s390/s390.c (processor_table): Add field for CPU name to
20368 be passed to Binutils.
20369 (s390_asm_output_machine_for_arch): Use the new field in
20370 processor_table for Binutils.
20371 (s390_expand_builtin): Replace arch12 with z14.
20372 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20373 (s390_get_sched_attrmask): Likewise.
20374 (s390_get_unit_mask): Likewise.
20375 * config/s390/s390.opt: Add z14 to processor_type enum.
20377 2017-07-31 Martin Jambor <mjambor@suse.cz>
20380 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
20381 regardless of optimization level.
20383 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
20384 Martin Liska <mliska@suse.cz>
20386 * predict.def: Remove old comment and adjust probability.
20387 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
20388 PREDICT statements.
20390 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20393 * config/i386/i386.c (ix86_function_naked): New function.
20394 (ix86_can_use_return_insn_p): Return false for naked functions.
20395 (ix86_expand_prologue): Skip prologue for naked functions.
20396 (ix86_expand_epilogue): Skip epilogue for naked functions
20397 and emit trap instruction.
20398 (ix86_warn_func_return): New function.
20399 (ix86_attribute_table): Add "naked" attribute specification.
20400 (TARGET_WARN_FUNC_RETURN): Define.
20401 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
20403 2017-07-31 Martin Liska <mliska@suse.cz>
20405 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
20406 (dump_gimple_bb_header): Always dump BB info.
20407 (pp_cfg_jump): Do not append info about BB when dumping a jump.
20409 2017-07-31 Martin Liska <mliska@suse.cz>
20412 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
20413 also with current_function_decl non-null equality.
20415 2017-07-31 Jakub Jelinek <jakub@redhat.com>
20418 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
20419 change type to the element type, instead add eltype variable and
20420 use it where we are interested in the element type.
20422 PR tree-optimization/81603
20423 * ipa-polymorphic-call.c
20424 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
20425 offset arithmetic in offset_int, bail out if the resulting bit offset
20426 doesn't fit into shwi.
20428 2017-07-31 Martin Liska <mliska@suse.cz>
20430 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
20431 (gimplify_save_expr): Fix comment.
20433 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
20436 * config/i386/i386.c (ix86_function_arg): Update arguments for
20438 (ix86_compute_frame_layout): Set the initial stack offset to
20439 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
20440 INCOMING_FRAME_SP_OFFSET.
20441 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
20442 stack before exception handler returns.
20443 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
20444 the 'ERROR_CODE' for exception handler.
20446 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
20448 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
20449 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
20450 (ASM_OUTPUT_REG_POP): Ditto.
20451 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
20452 instead of asm_fprintf to output pure string.
20454 2017-07-29 Jakub Jelinek <jakub@redhat.com>
20456 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
20457 to imported_module_or_decl hook.
20458 (debug_nothing_tree_tree_tree_bool): Remove.
20459 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
20460 * debug.c (do_nothing_debug_hooks): Use
20461 debug_nothing_tree_tree_tree_bool_bool instead of
20462 debug_nothing_tree_tree_tree_bool.
20463 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20464 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
20465 * sdbout.c (sdb_debug_hooks): Likewise.
20466 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
20467 (gen_namespace_die): Add DW_AT_export_symbols attribute if
20469 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
20470 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
20471 attribute, don't add anything.
20473 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20475 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
20476 (fold_build2_stat_loc): Likewise.
20477 (fold_build3_stat_loc): Likewise.
20478 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
20479 (fold_build1_loc): Remove macro.
20480 (fold_build2_loc): Likewise.
20481 (fold_build3_loc): Likewise.
20483 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20485 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
20486 (gimple_build_debug_bind_source_stat): Likewise.
20487 * gimple.h (gimple_build_debug_bind): Remove macro.
20488 (gimple_build_debug_bind_source): Likewise.
20490 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20492 * bitmap.c (bitmap_alloc): Adjust.
20493 (bitmap_gc_alloc): Likewise.
20494 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
20496 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20498 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
20499 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
20500 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
20501 (bitmap_gc_alloc_stat): Likewise.
20502 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
20504 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20506 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
20507 * rtl.h (shallow_copy_rtx): Remove macro.
20509 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20511 * emit-rtl.c (gen_raw_REG): Adjust.
20512 * gengenrtl.c (gendef): Likewise.
20513 * rtl.c (rtx_alloc_stat): Remove _stat from name.
20514 * rtl.h (rtx_alloc): Remove macro.
20516 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20518 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
20519 (build_tree_list_stat): Likewise.
20520 * tree.h (build_tree_list): Remove macro.
20521 (build_tree_list_vec): Likewise.
20523 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20525 * tree.c (make_vector_stat): Remove _stat from name.
20526 (build_vector_stat): Likewise.
20527 * tree.h (make_vector_stat): Remove macro.
20528 (build_vector_stat): Likewise.
20530 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20532 * tree.h (build_var_debug_value): Remove prototype.
20534 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20536 * tree.c (tree_cons_stat): Remove _stat from name.
20537 * tree.h (tree_cons): Remove macro.
20539 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20541 * tree.c (build_vl_exp_stat): Remove _stat from name.
20542 * tree.h (build_vl_exp): Remove macro.
20544 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20546 * tree.c (build_decl_stat): Remove _stat from name.
20547 * tree.h (build_decl): Remove macro.
20549 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20551 * gimple.c (gimple_build_with_ops_stat): Adjust.
20552 (gimple_alloc_stat): Remove _stat from name.
20553 * gimple.h (gimple_alloc): Remove macro.
20555 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20557 * tree.c (make_tree_vec_stat): Remove _stat from name.
20558 (grow_tree_vec_stat): Likewise.
20559 * tree.h (make_tree_vec_stat): Adjust prototype.
20560 (grow_tree_vec_stat): Likewise.
20561 (make_tree_vec): Remove macro.
20562 (grow_tree_vec): Likewise.
20564 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20566 * fold-const.c (fold_build1_stat_loc): Adjust.
20567 (fold_build2_stat_loc): Likewise.
20568 (fold_build3_stat_loc): Likewise.
20569 * tree.c (build0_stat): Remove _stat from name.
20570 (build1_stat): Likewise.
20571 (build2_stat): Likewise.
20572 (build3_stat): Likewise.
20573 (build4_stat): Likewise.
20574 (build5_stat): Likewise.
20575 * tree.h (build1_loc): Remove macro, and rename _stat function
20577 (build2_loc): Likewise.
20578 (build3_loc): Likewise.
20579 (build4_loc): Likewise.
20580 (build5_loc): Likewise.
20582 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20584 * tree.c (make_int_cst_stat): Remove _stat from name.
20585 * tree.h (make_int_cst_stat): Adjust prototype.
20586 (make_int_cst): Remove macro.
20588 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20590 * tree.c (make_tre_binfo_stat): Remove _stat from name.
20591 * tree.h (make_tree_binfo_stat): Adjust prototype.
20592 (make_tree_binfo): Remove.
20594 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20596 * tree.c (copy_node_stat): Rename to copy_node.
20597 (build_distinct_type_copy): Adjust.
20598 * tree.h (copy_node_stat): Adjust prototype.
20599 (copy_node): Remove macro.
20601 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20603 * tree.c (make_node_stat): rename to make_node.
20604 (build_tree_list_stat): Adjust.
20605 (build0_stat): Likewise.
20606 (build2_stat): Likewise.
20607 (build3_stat): Likewise.
20608 (build4_stat): Likewise.
20609 (build5_stat): Likewise.
20610 (build_decl_stat): Likewise.
20611 * tree.h (make_node_stat): Adjust prototype.
20612 (make_node): remove macro.
20614 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
20616 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
20617 (PPC_FEATURE2_SCV): Likewise.
20618 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
20620 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20622 * config/aarch64/aarch64.c
20623 (aarch64_internal_mov_immediate): Add new special pattern.
20624 * config/aarch64/aarch64.md (*movdi_aarch64):
20625 Add reg/32bit const mov case.
20627 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20628 Richard Sandiford <richard.sandiford@linaro.org>
20630 * config/aarch64/aarch64.md (mov<mode>): Generalize.
20631 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20632 Add integer and movi cases.
20633 (movi-split-hf-df-sf split, fp16): New.
20634 (enabled): Added TARGET_FP_F16INST.
20635 * config/aarch64/iterators.md (GPF_HF): New.
20636 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
20638 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20640 * config/aarch64/aarch64.c
20641 (aarch64_simd_container_mode): Add prototype.
20642 (aarch64_expand_mov_immediate): Add HI support.
20643 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
20644 (aarch64_can_const_movi_rtx_p): New.
20645 (aarch64_preferred_reload_class):
20646 Remove restrictions of using FP registers for certain SIMD operations.
20647 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
20648 (aarch64_valid_floating_const): Add integer move validation.
20649 (aarch64_simd_imm_scalar_p): Remove.
20650 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
20651 (aarch64_legitimate_constant_p): Expand list of supported cases.
20652 * config/aarch64/aarch64-protos.h
20653 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
20654 (aarch64_reinterpret_float_as_int): New.
20655 (aarch64_simd_imm_scalar_p): Remove.
20656 * config/aarch64/constraints.md (Uvi): New.
20657 (Dd): Split into Ds and new Dd.
20658 * config/aarch64/aarch64.md (*movsi_aarch64):
20660 (*movdi_aarch64): Add SIMD mov case.
20662 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20664 * tree-predcom.c: (struct chain): Handle store-store chain in which
20665 stores for elimination only store loop invariant values.
20666 (execute_pred_commoning_chain): Ditto.
20667 (prepare_initializers_chain_store_elim): Ditto.
20668 (prepare_finalizers): Ditto.
20669 (is_inv_store_elimination_chain): New function.
20670 (initialize_root_vars_store_elim_1): New function.
20672 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20674 * tree-predcom.c: Revise general description of the pass.
20675 (enum chain_type): New enum type for store elimination.
20676 (struct chain): New field supporting store elimination.
20677 (struct component): Ditto.
20678 (dump_chain): Dump store-stores chain.
20679 (release_chain): Release resources.
20680 (split_data_refs_to_components): Compute and create component
20681 contains only stores for elimination.
20682 (get_chain_last_ref_at): New function.
20683 (make_invariant_chain): Initialization.
20684 (make_rooted_chain): Specify chain type in parameter and record it.
20685 (add_looparound_copies): Skip for store-stores chain.
20686 (determine_roots_comp): Compute type of chain and pass it to
20688 (initialize_root_vars_store_elim_2): New function.
20689 (finalize_eliminated_stores): New function.
20690 (remove_stmt): Handle store for elimination.
20691 (execute_pred_commoning_chain): Execute predictive commoning on
20692 store-store chains.
20693 (determine_unroll_factor): Skip unroll for store-stores chain.
20694 (prepare_initializers_chain_store_elim): New function.
20695 (prepare_initializers_chain): Hanlde store-store chain.
20696 (prepare_finalizers_chain, prepare_finalizers): New function.
20697 (tree_predictive_commoning_loop): Return integer value indicating
20698 if loop is unrolled or lcssa form is corrupted.
20699 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
20701 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20703 * tree-predcom.c (initialize_root): Delete.
20704 (execute_pred_commoning_chain): Initialize root vars and replace
20705 reference of non-combined chain directly, rather than call above
20708 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20710 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
20711 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
20713 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20715 * tree-predcom.c (struct chain): New field init_seq.
20716 (release_chain): Release init_seq.
20717 (prepare_initializers_chain): Record intialization stmts in above
20719 (insert_init_seqs): New function.
20720 (tree_predictive_commoning_loop): Call insert_init_seqs.
20722 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20724 * tree-predcom.c (determine_roots_comp): Skip trivial components.
20726 2017-07-28 Richard Biener <rguenther@suse.de>
20728 * match.pd: Remove superfluous :c.
20729 * genmatch.c (simplify::id): Add member.
20730 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
20732 (current_id): New global.
20733 (dt_node::parent): Move from ...
20734 (dt_operand::parent): ... here. Add for_id member.
20735 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
20736 (decision_tree::find_node): Relax order requirement when
20737 merging DT_TRUE nodes to ones inbetween the current simplify
20738 and the one we try to merge with. Add diagnostic whenever
20739 we need to enforce pattern order by not merging.
20740 (decision_tree::insert): Set current_id.
20741 (decision_tree::print_node): Dump parent node and for_id.
20742 (parser::last_id): Add member.
20743 (parser::push_simplify): Assign unique id.
20744 (parser::parser): Initialize last_id.
20746 2017-07-28 Martin Liska <mliska@suse.cz>
20749 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
20750 gimple_build_debug_bind.
20752 2017-07-28 Richard Biener <rguenther@suse.de>
20754 PR tree-optimization/81502
20755 * match.pd: Add pattern combining BIT_INSERT_EXPR with
20757 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
20759 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
20760 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
20761 for BIT_FIELD_REF args.
20762 * fold-const.c (make_bit_field_ref): Likewise.
20763 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
20765 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20768 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
20769 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
20770 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
20771 Or it into SANITIZER_UNDEFINED.
20772 * ubsan.c: Include gimple-fold.h and varasm.h.
20773 (ubsan_expand_ptr_ifn): New function.
20774 (instrument_pointer_overflow): New function.
20775 (maybe_instrument_pointer_overflow): New function.
20776 (instrument_object_size): Formatting fix.
20777 (pass_ubsan::execute): Call instrument_pointer_overflow
20778 and maybe_instrument_pointer_overflow.
20779 * internal-fn.c (expand_UBSAN_PTR): New function.
20780 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
20781 * sanitizer.def (__ubsan_handle_pointer_overflow,
20782 __ubsan_handle_pointer_overflow_abort): New builtins.
20783 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
20784 * internal-fn.def (UBSAN_PTR): New internal function.
20785 * opts.c (sanitizer_opts): Add pointer-overflow.
20786 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
20787 * fold-const.c (build_range_check): Compute pointer range check in
20788 integral type if pointer arithmetics would be needed. Formatting
20791 2017-07-28 Martin Liska <mliska@suse.cz>
20794 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
20795 parameters that are of a variable-length.
20797 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
20799 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
20801 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
20802 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
20803 (CRT_CALL_STATIC_FUNCTION): Likewise.
20804 (ASM_DEFAULT_SPEC): New define.
20805 (ASM_SPEC32): Likewise.
20806 (ASM_SPEC64): Likewise.
20807 (ASM_SPEC_COMMON): Likewise.
20808 (ASM_SPEC): Likewise.
20809 (INVALID_64BIT): Likewise.
20810 (LINK_OS_DEFAULT_SPEC): Likewise.
20811 (LINK_OS_SPEC32): Likewise.
20812 (LINK_OS_SPEC64): Likewise.
20813 (POWERPC_LINUX): Likewise.
20814 (PTRDIFF_TYPE): Likewise.
20815 (RESTORE_FP_PREFIX): Likewise.
20816 (RESTORE_FP_SUFFIX): Likewise.
20817 (SAVE_FP_PREFIX): Likewise.
20818 (SAVE_FP_SUFFIX): Likewise.
20819 (SIZE_TYPE): Likewise.
20820 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
20821 (TARGET_64BIT): Likewise.
20822 (TARGET_64BIT): Likewise.
20823 (TARGET_AIX): Likewise.
20824 (WCHAR_TYPE_SIZE): Likewise.
20825 (WCHAR_TYPE): Undefine.
20826 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
20827 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
20828 (CPP_OS_RTEMS_SPEC): Delete.
20829 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
20830 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
20832 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
20834 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20836 PR tree-optimization/81578
20837 * tree-parloops.c (build_new_reduction): Bail out if
20838 reduction_code isn't one of the standard OpenMP reductions.
20839 Move the details printing after that decision.
20841 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20843 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
20844 related to reload_in_progress.
20845 (splat_input_operand): Likewise.
20846 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
20848 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
20850 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
20851 (TARGET_INSTANTIATE_DECLS): Likewise.
20852 (legitimate_indexed_address_p): Delete reload_in_progress code.
20853 (rs6000_debug_legitimate_address_p): Likewise.
20854 (rs6000_eliminate_indexed_memrefs): Likewise.
20855 (rs6000_emit_le_vsx_store): Likewise.
20856 (rs6000_emit_move_si_sf_subreg): Likewise.
20857 (rs6000_emit_move): Likewise.
20858 (register_to_reg_type): Likewise.
20859 (rs6000_pre_atomic_barrier): Likewise.
20860 (rs6000_machopic_legitimize_pic_address): Likewise.
20861 (rs6000_allocate_stack_temp): Likewise.
20862 (rs6000_address_for_fpconvert): Likewise.
20863 (rs6000_address_for_altivec): Likewise.
20864 (rs6000_secondary_memory_needed_rtx): Delete function.
20865 (rs6000_check_sdmode): Likewise.
20866 (rs6000_alloc_sdmode_stack_slot): Likewise.
20867 (rs6000_instantiate_decls): Likewise.
20868 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
20869 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
20870 Delete reload_in_progress.
20871 (*vec_reload_and_plus_<mptrsize>): Likewise.
20872 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
20873 (vsx_div_v2di): Likewise.
20874 (vsx_udiv_v2di): Likewise.
20876 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20878 * config/rs6000/rs6000.opt (mlra): Replace with stub.
20879 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
20880 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
20881 (rs6000_debug_reg_global): Delete print of LRA status.
20882 (rs6000_option_override_internal): Delete dead LRA related code.
20883 (rs6000_lra_p): Delete function.
20884 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
20886 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
20888 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
20889 * config/riscv/rtems.h: New file.
20891 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20892 Sudakshina Das <sudi.das@arm.com>
20894 * config/aarch64/aarch64.md
20895 (define_split for and<mode>3nr_compare): Move
20896 non aarch64_logical_operand to a register.
20897 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
20898 register immediate operand to a register.
20899 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
20901 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20903 PR middle-end/81564
20904 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
20906 2017-07-27 Richard Biener <rguenther@suse.de>
20908 PR tree-optimization/81573
20909 PR tree-optimization/81494
20910 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
20911 multi defuse cycle case.
20913 2017-07-27 Richard Biener <rguenther@suse.de>
20915 PR tree-optimization/81571
20916 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
20919 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
20921 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
20922 earlier and only if MASK_FPU is set. Adjust formatting.
20924 2017-07-27 Martin Liska <mliska@suse.cz>
20926 * opt-functions.awk: Add validation of value of Init.
20927 * optc-gen.awk: Pass new argument.
20929 2017-07-27 Martin Liska <mliska@suse.cz>
20931 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
20932 Fix wrong condition.
20934 2017-07-27 Martin Liska <mliska@suse.cz>
20936 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
20937 BBs and edges seen by autoFDO.
20939 2017-07-27 Richard Biener <rguenther@suse.de>
20941 PR tree-optimization/81502
20942 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
20943 with incompatible but same sized type.
20944 (execute_update_addresses_taken): Likewise.
20946 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
20948 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
20949 flag_tree_loop_vectorize rather than flag_tree_vectorize.
20951 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20954 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
20955 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
20956 Change s_operand to memory_operand.
20958 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
20960 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
20961 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
20962 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
20963 Emit instructions rather than returning an expression. Handle TFmode
20964 and KFmode by casting to TImode.
20965 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
20966 (rs6000_emit_le_vsx_store): Likewise.
20967 * config/rs6000/vsx.md (VSX_TI): New iterator.
20968 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
20969 (*vsx_le_undo_permute_<mode>): Likewise.
20970 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
20971 emit the split sequence.
20972 (*vsx_le_perm_store_<mode>): Likewise.
20974 2017-07-27 Jakub Jelinek <jakub@redhat.com>
20976 PR tree-optimization/81555
20977 PR tree-optimization/81556
20978 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
20979 if true, force CHANGED for the recursive invocation.
20980 (reassociate_bb): Remember original length of ops array, pass
20981 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
20983 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
20984 attributes for noipa attribute. For naked attribute use
20985 lookup_attribute first before lookup_attribute_spec.
20986 * final.c (rest_of_handle_final): Disable IPA RA for functions with
20988 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
20989 for functions with noipa attribute.
20990 (cgraph_externally_visible_p): Return true for functions with noipa
20992 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
20993 for functions with noipa attribute.
20994 * doc/extend.texi: Document noipa function attribute.
20995 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
20996 also for functions with noipa attribute.
20997 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
20999 2017-07-26 Andrew Pinski <apinski@cavium.com>
21001 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21002 vec_unalign_load_cost and vec_unalign_store_cost.
21004 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21006 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21007 -mvsx-small-integer option.
21008 (ISA_3_0_MASKS_IEEE): Likewise.
21009 (OTHER_VSX_VECTOR_MASKS): Likewise.
21010 (POWERPC_MASKS): Likewise.
21011 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21012 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21013 code, only testing for DImode being allowed in non-VSX floating
21015 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21016 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21018 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21019 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21020 TARGET_P8_VECTOR test.
21021 (rs6000_secondary_reload_simple_move): Likewise.
21022 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21023 since TARGET_P9_VECTOR was already tested.
21024 (rs6000_opt_masks): Remove -mvsx-small-integer.
21025 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21026 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21028 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21029 test for TARGET_VEXTRACTUB was used, and that uses
21031 (p9 extract splitter): Likewise.
21032 (vsx_extract_<mode>_di_p9): Likewise.
21033 (vsx_extract_<mode>_store_p9): Likewise.
21034 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21035 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21036 the elimination of TARGET_VSX_SMALL_INTEGER.
21037 (vsx_extract_<mode>_p8): Likewise.
21038 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21039 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21040 (vsx_set_<mode>_p9): Likewise.
21041 (vsx_set_v4sf_p9): Likewise.
21042 (vsx_set_v4sf_p9_zero): Likewise.
21043 (vsx_insert_extract_v4sf_p9): Likewise.
21044 (vsx_insert_extract_v4sf_p9_2): Likewise.
21045 * config/rs6000/rs6000.md (sign extend splitter): Change
21046 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
21047 (floatsi<mode>2_lfiwax_mem): Likewise.
21048 (floatunssi<mode>2_lfiwzx_mem): Likewise.
21049 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
21050 since a test for TARGET_P9_VECTOR was used.
21051 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21052 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
21053 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21054 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21055 TARGET_P8_VECTOR test.
21056 (fix_trunc<mode>si2_stfiwx): Likewise.
21057 (fix_trunc<mode>si2_internal): Likewise.
21058 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
21059 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21061 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21062 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21063 TARGET_P8_VECTOR test.
21064 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21065 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
21066 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21068 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21069 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
21070 since a test for TARGET_P9_VECTOR was used.
21071 (splitter for loading small constants): Likewise.
21073 2017-07-26 Andrew Pinski <apinski@cavium.com>
21075 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
21078 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
21081 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
21082 (fp_valid_at): Likewise.
21084 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21086 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
21087 (qdf24xx_addrcost_table): Likewise.
21088 (cortexa57_tunings): Update to use generic_branch_cost.
21089 (cortexa72_tunings): Likewise.
21090 (cortexa73_tunings): Likewise.
21091 (qdf24xx_tunings): Likewise.
21093 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21095 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
21096 (thunderx2t99_branch_cost): Likewise.
21097 (cortexa35_tunings): Update to use generic_branch_cost.
21098 (cortexa53_tunings): Likewise.
21099 (cortexa57_tunings): Likewise.
21100 (cortexa72_tunings): Likewise.
21101 (cortexa73_tunings): Likewise.
21102 (thunderx2t99_tunings): Likewise.
21104 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21106 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
21107 (sparc_option_override): Honour MASK_FSMULD.
21108 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
21109 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
21110 * config/sparc/sparc.opt (mfsmuld): New option.
21111 * doc/invoke.texi (mfsmuld): Document option.
21113 2017-07-26 Marek Polacek <polacek@redhat.com>
21115 PR middle-end/70992
21116 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
21118 2017-07-26 Richard Biener <rguenther@suse.de>
21120 * gimple-match-head.c (do_valueize): Return OP if valueize
21122 (get_def): New helper to get at the def stmt of a SSA name
21123 if valueize allows.
21124 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
21125 do_valueize to get at the def stmt.
21126 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
21128 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
21130 PR middle-end/46932
21131 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
21133 2017-07-26 Martin Liska <mliska@suse.cz>
21136 * function.c (expand_function_start): Make expansion of
21137 nonlocal_goto_save_area after parm_birth_insn.
21139 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21141 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
21142 from all CPU target flags enable members.
21144 2017-07-26 Richard Biener <rguenther@suse.de>
21146 * genmatch.c (dt_simplify::gen): Make iterator vars const.
21147 (decision_tree::gen): Make 'type' const.
21148 (write_predicate): Likewise.
21150 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21152 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
21153 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
21154 (rs6000_option_override_internal): Likewise.
21155 (rs6000_expand_vector_set): Likewise.
21156 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
21157 (TARGET_UPPER_REGS_SF): Likewise.
21158 (TARGET_UPPER_REGS_DI): Likewise.
21159 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
21160 (TARGET_DIRECT_MOVE_64BIT): Likewise.
21161 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
21162 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21163 (Splitters for DI constants in Altivec registers): Likewise.
21164 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
21165 (vsx_set_v4sf_p9): Likewise.
21166 (vsx_set_v4sf_p9_zero): Likewise.
21167 (vsx_insert_extract_v4sf_p9): Likewise.
21168 (vsx_insert_extract_v4sf_p9_2): Likewise.
21170 2017-07-25 Carl Love <cel@us.ibm.com>
21172 * doc/extend.texi: Update the built-in documentation file for the
21173 existing built-in functions
21174 vector signed char vec_cnttz (vector signed char);
21175 vector unsigned char vec_cnttz (vector unsigned char);
21176 vector signed short vec_cnttz (vector signed short);
21177 vector unsigned short vec_cnttz (vector unsigned short);
21178 vector signed int vec_cnttz (vector signed int);
21179 vector unsigned int vec_cnttz (vector unsigned int);
21180 vector signed long long vec_cnttz (vector signed long long);
21181 vector unsigned long long vec_cnttz (vector unsigned long long);
21183 2017-07-25 Andrew Pinski <apinski@cavium.com>
21185 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
21186 accesses where the use is for the first operand of a BIT_INSERT.
21188 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21191 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21192 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21194 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21196 * config/i386/gstabs.h: Delete.
21197 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
21199 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
21201 * config/i386/i386.c (ix86_decompose_address): Do not check for
21202 register RTX when looking at index_reg or base_reg.
21203 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
21205 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
21207 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
21208 to update EH info here.
21210 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21212 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
21214 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21216 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
21218 2017-07-25 Torsten Duwe <duwe@suse.de>
21220 * common.opt: Introduce -fpatchable-function-entry
21221 command line option, and its variables function_entry_patch_area_size
21222 and function_entry_patch_area_start.
21223 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
21224 including a two-value parser.
21225 * target.def (print_patchable_function_entry): New target hook.
21226 * targhooks.h (default_print_patchable_function_entry): New function.
21227 * targhooks.c (default_print_patchable_function_entry): Likewise.
21228 * toplev.c (process_options): Switch off IPA-RA if
21229 patchable function entries are being generated.
21230 * varasm.c (assemble_start_function): Look at the
21231 patchable-function-entry command line switch and current
21232 function attributes and maybe generate NOP instructions by
21233 calling the print_patchable_function_entry hook.
21234 * doc/extend.texi: Document patchable_function_entry attribute.
21235 * doc/invoke.texi: Document -fpatchable_function_entry
21236 command line option.
21237 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
21239 * doc/tm.texi: Re-generate.
21241 2017-07-25 Jakub Jelinek <jakub@redhat.com>
21244 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
21245 TARGET_AVX512DQ rather than TARGET_AVX512BW.
21247 2017-07-25 Tamar Christina <tamar.christina@arm.com>
21249 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
21251 2017-07-25 Richard Biener <rguenther@suse.de>
21253 PR tree-optimization/81455
21254 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
21255 not walk in cycles when looking for guards.
21257 2017-07-25 Richard Biener <rguenther@suse.de>
21259 PR tree-optimization/81529
21260 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
21261 when optimizing backedge uses.
21263 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
21265 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
21267 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
21268 to dl_section_ref. On AIX, append an expression to subtract
21269 the size of the section length to dl_section_ref.
21271 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
21273 * configure.ac: If any of the config.* scripts fail, exit 1.
21274 * configure: Regenerate.
21276 2017-07-25 Richard Biener <rguenther@suse.de>
21278 PR middle-end/81546
21279 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
21280 of immediate uses, be more verbose on errors.
21282 2017-07-25 Richard Biener <rguenther@suse.de>
21284 PR tree-optimization/81510
21285 * tree-vect-loop.c (vect_is_simple_reduction): When the
21286 reduction stmt is not inside the loop bail out.
21288 2017-07-25 Richard Biener <rguenther@suse.de>
21290 PR tree-optimization/81303
21291 * tree-vect-loop-manip.c (vect_loop_versioning): Build
21292 profitability check against LOOP_VINFO_NITERSM1.
21294 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21296 * domwalk.c (cmp_bb_postorder): Simplify.
21297 (sort_bbs_postorder): New function. Use it...
21298 (dom_walker::walk): ...here to optimize common cases.
21300 2017-07-25 Martin Liska <mliska@suse.cz>
21303 * ipa-visibility.c (function_and_variable_visibility): Make the
21304 redirection just on target that supports aliasing.
21305 Fix GNU coding style.
21307 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
21310 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
21312 * config/arm/rtems.h: Likewise.
21313 * config/bfin/rtems.h: Likewise.
21314 * config/i386/rtemself.h: Likewise.
21315 * config/lm32/rtems.h: Likewise.
21316 * config/m32c/rtems.h: Likewise.
21317 * config/m68k/rtemself.h: Likewise.
21318 * config/microblaze/rtems.h: Likewise.
21319 * config/mips/rtems.h: Likewise.
21320 * config/moxie/rtems.h: Likewise.
21321 * config/nios2/rtems.h: Likewise.
21322 * config/powerpcspe/rtems.h: Likewise.
21323 * config/rs6000/rtems.h: Likewise.
21324 * config/rtems.h: Likewise.
21325 * config/sh/rtems.h: Likewise.
21326 * config/sh/rtemself.h: Likewise.
21327 * config/sparc/rtemself.h: Likewise.
21329 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21332 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
21333 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
21334 * tree-ssa-structalias.c (alias_get_name): Same.
21336 2017-07-25 Bin Cheng <bin.cheng@arm.com>
21339 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
21340 instructions if no du chain is found.
21342 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21344 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
21346 2017-07-25 Richard Biener <rguenther@suse.de>
21348 PR middle-end/81505
21349 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
21352 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21354 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
21355 upper-regs options.
21356 (ISA_2_7_MASKS_SERVER): Likewise.
21357 (ISA_3_0_MASKS_IEEE): Likewise.
21358 (OTHER_P8_VECTOR_MASKS): Likewise.
21359 (OTHER_VSX_VECTOR_MASKS): Likewise.
21360 (POWERPC_MASKS): Likewise.
21361 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
21362 duplicate list of options.
21363 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
21364 explicit -mupper-regs options.
21365 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
21366 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
21367 alias for -mupper-regs-df.
21368 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
21369 (rs6000_init_hard_regno_mode_ok): Likewise.
21370 (rs6000_option_override_internal): Likewise.
21371 (rs6000_opt_masks): Likewise.
21372 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
21373 options in terms of whether -mvsx or -mpower8-vector was used.
21374 (TARGET_UPPER_REGS_DI): Likewise.
21375 (TARGET_UPPER_REGS_SF): Likewise.
21376 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
21377 -mupper-regs-* options.
21379 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21381 * passes.c (emergency_dump_function): Print some empty lines and a
21382 header before the RTL dump.
21384 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21386 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
21388 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
21391 * config/aarch64/aarch64.c (aarch64_classify_symbol):
21392 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
21394 2017-07-24 Carl Love <cel@us.ibm.com>
21396 * config/rs6000/rs6000-c.c: Add support for built-in functions
21397 vector float vec_extract_fp32_from_shorth (vector unsigned short);
21398 vector float vec_extract_fp32_from_shortl (vector unsigned short);
21399 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
21400 vec_extract_fp_from_shortl): Add defines for the two builtins.
21401 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
21402 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
21404 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
21405 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
21406 * doc/extend.texi: Update the built-in documentation file for the
21407 new built-in function.
21409 2017-07-24 Jakub Jelinek <jakub@redhat.com>
21412 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
21414 * doc/generic.texi: Likewise.
21415 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21416 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21418 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
21420 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
21421 (aarch64_mls_elt_merge<mode>): Likewise.
21423 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
21425 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21426 having __cxa_atexit.
21428 2017-07-23 Michael Collison <michael.collison@arm.com>
21430 * config/arm/arm.c (arm_option_override): Deprecate
21431 use of -mstructure-size-boundary.
21432 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
21433 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
21435 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21438 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21439 Reduce cost estimate for direct moves.
21441 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
21444 * config/i386/i386.c (ix86_option_override_internal): Disable
21445 BMI, BMI2 and TBM instructions for -m16.
21447 2017-07-21 Carl Love <cel@us.ibm.com>
21449 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21450 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21451 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21452 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21453 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21454 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21455 VMULOSW): New enum "unspec" values.
21456 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21457 altivec_vmulosw): New patterns.
21458 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21459 VMULOSW): Add definitions.
21461 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
21463 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
21464 (qdf24xx): Likewise.
21465 * config/aarch64/aarch64-options-extensions.def (rdma); New.
21466 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
21467 (AARCH64_FL_V8_1): Renumber.
21468 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
21469 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
21470 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
21471 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
21472 rdma to feature modifiers list.
21474 2017-07-21 Yury Gribov <tetra2005@gmail.com>
21476 PR middle-end/56727
21477 * ipa-visibility (function_and_variable_visibility): Convert
21478 recursive PLT call to direct call if appropriate.
21480 2017-07-21 Andrew Pinski <apinski@cavium.com>
21482 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
21483 operand 1 to see if the types precision matches.
21484 * fold-const.c (operand_equal_p): Likewise.
21486 2017-07-21 Richard Biener <rguenther@suse.de>
21488 PR tree-optimization/81303
21489 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
21490 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
21491 (vect_peeling_hash_get_lowest_cost): Adjust.
21492 (vect_enhance_data_refs_alignment): Likewise. Use
21493 vect_get_peeling_costs_all_drs to compute the penalty for no
21494 peeling to match up costs.
21496 2017-07-21 Richard Biener <rguenther@suse.de>
21498 PR tree-optimization/81500
21499 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
21500 we didn't identify a reduction path.
21502 2017-07-21 Tom de Vries <tom@codesourcery.com>
21503 Cesar Philippidis <cesar@codesourcery.com>
21505 PR gcov-profile/81442
21506 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
21509 2017-07-21 Tom de Vries <tom@codesourcery.com>
21512 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
21514 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
21515 nvptx_override_options_after_change.
21517 2017-07-21 Ulrich Drepper <drepper@redhat.com>
21519 * dwarf2out.c (output_file_names): Avoid double testing for
21520 dwarf_version >= 5.
21522 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
21524 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
21526 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21528 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
21530 (try_crossjump_to_edge): Do not punt on partitioned functions.
21532 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21534 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
21535 Put all BBs reachable only via paths crossing cold region to cold
21537 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
21539 2016-07-21 Richard Biener <rguenther@suse.de>
21541 PR tree-optimization/81303
21542 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
21543 into account prologue and epilogue iterations when raising
21544 min_profitable_iters to sth at least covering one vector iteration.
21546 2017-07-21 Tamar Christina <tamar.christina@arm.com>
21548 * config/arm/arm.c (arm_test_cpu_arch_dat):
21551 2017-07-20 Nathan Sidwell <nathan@acm.org>
21553 Remove TYPE_METHODS.
21554 * tree.h (TYPE_METHODS): Delete.
21555 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
21556 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
21557 (dbxout_type_methods): Scan TYPE_FIELDS.
21558 (dbxout_type): Don't check TYPE_METHODS here.
21559 * function.c (use_register_for_decl): Always ignore register for
21560 class types when not optimizing.
21561 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
21562 * tree.c (free_lang_data_in_type): Stitch out member functions and
21563 templates from TYPE_FIELDS.
21564 (build_distinct_type_copy, verify_type_variant,
21565 verify_type): Member fns are on TYPE_FIELDS.
21566 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
21567 * tree-pretty-print.c (dump_generic_node): Likewise.
21569 2017-07-20 Jakub Jelinek <jakub@redhat.com>
21572 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
21573 V2TImode and V4TImode.
21574 (ix86_expand_vector_extract): Likewise.
21575 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
21576 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
21577 (ssescalarmode): Handle V4TImode and V2TImode.
21578 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
21579 (*vec_extractv2ti, *vec_extractv4ti): New insns.
21580 (VEXTRACTI128_MODE): New mode iterator.
21581 (splitter for *vec_extractv?ti first element): New.
21582 (VEC_INIT_MODE): New mode iterator.
21583 (vec_init<mode>): Consolidate 3 expanders into one using
21584 VEC_INIT_MODE mode iterator.
21586 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21588 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
21589 non_spilled_static_chain_regno_p.
21591 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21593 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
21595 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
21597 * bb-reorder.c (connect_traces): Allow copying of blocks within
21600 2017-07-20 Richard Biener <rguenther@suse.de>
21602 * gimple.h (gimple_phi_result): Add gphi * overload.
21603 (gimple_phi_result_ptr): Likewise.
21604 (gimple_phi_arg): Likewise. Adjust index assert to only
21605 allow actual argument accesses rather than all slots available
21607 (gimple_phi_arg_def): Add gphi * overload.
21608 * tree-phinodes.c (make_phi_node): Initialize only actual
21610 (resize_phi_node): Clear memory not covered by old node,
21611 do not initialize excess argument slots.
21612 (reserve_phi_args_for_new_edge): Initialize new argument slot
21615 2017-07-20 Bin Cheng <bin.cheng@arm.com>
21617 PR tree-optimization/81388
21619 2016-07-21 Bin Cheng <bin.cheng@arm.com>
21621 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
21622 by removing computation of may_be_zero.
21624 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21625 Tom de Vries <tom@codesourcery.com>
21627 PR middle-end/81030
21628 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
21629 when gimple level profile disagrees with what RTL expander did.
21631 2017-07-20 Richard Biener <rguenther@suse.de>
21633 PR tree-optimization/61171
21634 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
21635 (vect_analyze_stmt): Add slp instance parameter.
21636 (vectorizable_reduction): Likewise.
21637 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
21638 (vect_is_simple_reduction): Deal with chains not detected
21639 as SLP reduction chain, specifically not properly associated
21640 chains containing a mix of plus/minus.
21641 (get_reduction_op): Remove.
21642 (get_initial_defs_for_reduction): Simplify, pass in whether
21643 this is a reduction chain, pass in the SLP node for the PHIs.
21644 (vect_create_epilog_for_reduction): Get the SLP instance as
21646 (vectorizable_reduction): Get the SLP instance as arg.
21647 During analysis remember the SLP node with the PHIs in the
21648 instance. Simplify getting at the vectorized reduction PHIs.
21649 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
21650 through SLP instance.
21651 (vect_slp_analyze_operations): Likewise.
21652 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
21653 (vect_transform_stmt): Likewise.
21655 2017-07-20 Tom de Vries <tom@codesourcery.com>
21657 PR tree-optimization/81489
21658 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
21659 read of phi arg location to before loop that modifies phi.
21661 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21663 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
21666 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21668 PR middle-end/81331
21669 * except.c (execute): Fix ordering issue.
21671 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21673 PR rtl-optimization/81423
21674 * combine.c (make_compound_operation_int): Don't try to optimize
21675 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
21677 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21679 PR rtl-optimization/81423
21680 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
21681 with a constant that is -1 in the truncated to mode.
21683 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21685 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
21686 (determine_unlikely_bbs): ... here.
21687 * predict.h (propagate_unlikely_bbs_forward): Declare.
21688 * cfgexpand.c (pass_expand::execute): Use it.
21689 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
21691 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
21692 propagate_unlikely_bbs_forward.
21694 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21696 PR middle-end/81331
21697 * except.c (maybe_add_nop_after_section_switch): New function.
21700 2017-07-19 Tom de Vries <tom@codesourcery.com>
21702 * gimple.h (gimple_phi_set_arg): Make assert more strict.
21704 2017-07-19 Tom de Vries <tom@codesourcery.com>
21706 * gimple.h (gimple_phi_arg): Make assert more strict.
21708 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
21710 * config.gcc (powerpc*-*-*): Add mmintrin.h.
21711 * config/rs6000/mmintrin.h: New file.
21712 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
21714 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21716 PR tree-optimization/81346
21717 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
21719 2017-07-19 Tom de Vries <tom@codesourcery.com>
21721 * config/nvptx/nvptx.md (VECIM): Add V2DI.
21723 2017-07-19 Tom de Vries <tom@codesourcery.com>
21725 * config/nvptx/nvptx-modes.def: Add V2DImode.
21726 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
21727 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
21728 (nvptx_output_mov_insn): Handle lack of mov.b128.
21729 (nvptx_print_operand): Handle 'H' and 'L' codes.
21730 (nvptx_vector_mode_supported): Allow V2DImode.
21731 (nvptx_preferred_simd_mode): New function.
21732 (nvptx_data_alignment): New function.
21733 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
21734 nvptx_preferred_simd_mode.
21735 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
21737 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
21739 2017-07-19 Tom de Vries <tom@codesourcery.com>
21741 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
21742 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
21743 (nvptx_vector_mode_supported): New function. Allow V2SImode.
21744 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
21745 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
21746 (mov<VECIM>_insn): New define_insn.
21747 (define_expand "mov<VECIM>): New define_expand.
21749 2017-07-19 Tom de Vries <tom@codesourcery.com>
21751 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
21753 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21755 PR tree-optimization/81346
21756 * fold-const.h (fold_div_compare, range_check_type): Declare.
21757 * fold-const.c (range_check_type): New function.
21758 (build_range_check): Use range_check_type.
21759 (fold_div_compare): No longer static, rewritten into
21760 a match.pd helper function.
21761 (fold_comparison): Don't call fold_div_compare here.
21762 * match.pd (X / C1 op C2): New optimization using fold_div_compare
21763 as helper function.
21765 2017-07-19 Nathan Sidwell <nathan@acm.org>
21767 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
21768 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
21769 * tree.c (find_decls_types_r, verify_type): Use
21770 TYPE_{MIN,MAX}_VALUE_RAW.
21771 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
21772 (hash_tree): Likewise.
21773 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21775 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21778 2017-07-18 Tom de Vries <tom@codesourcery.com>
21780 PR middle-end/81464
21781 * omp-expand.c (expand_omp_for_static_chunk): Handle
21782 equal-argument loop exit phi.
21784 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
21787 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
21788 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
21789 operand 2 predicate.
21790 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
21791 operand 2 predicate.
21792 (ror,rol -> rorx splitters): Use const_int_operand as
21793 operand 2 predicate.
21795 2017-06-18 Richard Biener <rguenther@suse.de>
21797 PR tree-optimization/81410
21798 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
21799 the gap in the ! slp_perm SLP case after each group.
21801 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21803 PR middle-end/81463
21804 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
21807 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21809 PR middle-end/81462
21810 * predict.c (set_even_probabilities): Cleanup; do not affect
21811 probabilities that are already known.
21812 (combine_predictions_for_bb): Call even when count is set.
21814 2017-07-18 Nathan Sidwell <nathan@acm.org>
21816 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
21819 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21822 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
21823 optimization for loop niter analysis.
21825 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
21828 * config/avr/avr.c (avr_optimize_casesi): Don't use
21829 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
21831 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
21833 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
21834 body_cost_vec from _vect_peel_extended_info.
21835 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
21836 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
21839 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21841 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
21843 2017-07-18 Richard Biener <rguenther@suse.de>
21845 PR tree-optimization/80620
21846 PR tree-optimization/81403
21847 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
21848 info when re-using a VN table entry.
21850 2017-07-18 Richard Biener <rguenther@suse.de>
21852 PR tree-optimization/81418
21853 * tree-vect-loop.c (vectorizable_reduction): Properly compute
21854 vectype_in. Verify that with lane-reducing reduction operations
21855 we have a single def-use cycle.
21857 2017-07-17 Carl Love <cel@us.ibm.com>
21859 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
21861 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21862 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21863 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21864 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21865 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21866 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21867 VMULOSW): New enum "unspec" values.
21868 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21869 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21870 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21871 altivec_vmulosw): New patterns.
21872 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21873 VMULOSW): Add definitions.
21875 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
21877 * config/alpha/alpha.c: Include predict.h.
21879 2017-07-17 Yury Gribov <tetra2005@gmail.com>
21881 * tree-vrp.c (compare_assert_loc): Fix comparison function
21882 to return predictable results.
21884 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21886 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
21888 (subdi3): Likewise.
21889 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
21890 * doc/invoke.texi (mexpand-adddi): Update text.
21892 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21894 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
21895 that also clobbers the CC register. The old expand code is moved
21897 (*arc_clzsi2): ... here.
21898 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
21899 the CC register. The old expand code is moved to ...
21900 (arc_ctzsi2): ... here.
21902 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21904 * config/arc/arc.opt (mindexed-loads): Use initial value
21905 TARGET_INDEXED_LOADS_DEFAULT.
21906 (mauto-modify-reg): Use initial value
21907 TARGET_AUTO_MODIFY_REG_DEFAULT.
21908 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21909 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21910 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21911 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21913 2017-07-17 Martin Liska <mliska@suse.cz>
21916 * opts.c (finish_options): Do not allow -fgnu-tm
21917 w/ -fsanitize={kernel-,}address. Say sorry.
21919 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21922 * tree-loop-distribution.c (classify_partition): Only assert on
21923 numer of iterations.
21924 (merge_dep_scc_partitions): Delete prameter. Update function call.
21925 (distribute_loop): Remove code handling loop with unknown niters.
21926 (pass_loop_distribution::execute): Skip loop with unknown niters.
21928 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21931 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
21932 function sort_partitions_by_post_order.
21934 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21936 PR tree-optimization/81374
21937 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
21938 the max index of basic blocks, rather than number of basic blocks.
21940 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21942 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
21944 (arc_legitimate_pic_operand_p): Likewise.
21945 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
21947 (arc_needs_pcl_p): Likewise.
21948 (arc_legitimate_pc_offset_p): Likewise.
21949 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
21950 function is also used in constrains.md.
21951 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
21952 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
21953 PLUS. Only return true/false in known cases, otherwise assert.
21954 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
21955 is already called in arc_legitimate_constant_p.
21956 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
21958 (LEGITIMATE_PIC_OPERAND_P): Use
21959 arc_raw_symbolic_reference_mentioned_p function.
21960 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
21965 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21966 Andrew Burgess <andrew.burgess@embecosm.com>
21968 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
21969 (arc_return_address_register): New function.
21970 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
21971 (arc_handle_fndecl_attribute): Add naked attribute.
21972 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21973 (TARGET_WARN_FUNC_RETURN): Likewise.
21974 (arc_allocate_stack_slots_for_args): New function.
21975 (arc_warn_func_return): Likewise.
21976 (machine_function): Change type fn_type.
21977 (arc_compute_function_type): Consider new naked function type,
21978 change function return type.
21979 (arc_must_save_register): Adapt to handle new
21980 arc_compute_function_type's return type.
21981 (arc_expand_prologue): Likewise.
21982 (arc_expand_epilogue): Likewise.
21983 (arc_return_address_regs): Delete.
21984 (arc_return_address_register): New function.
21985 (arc_epilogue_uses): Use above function.
21986 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
21987 (arc_function_type): Change encoding, add naked type.
21988 (ARC_INTERRUPT_P): Change to handle the new encoding.
21989 (ARC_FAST_INTERRUPT_P): Likewise.
21990 (ARC_NORMAL_P): Define.
21991 (ARC_NAKED_P): Likewise.
21992 (arc_compute_function_type): Delete prototype.
21993 * config/arc/arc.md (in_ret_delay_slot): Use
21994 arc_return_address_register function.
21995 (simple_return): Likewise.
21996 (p_return_i): Likewise.
21998 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22000 PR tree-optimization/81428
22001 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22002 can't be built for those types.
22004 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22006 Remove stuff dead since r239246.
22008 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22009 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22010 (avr_inform_devices): Remove dead stuff.
22012 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22014 * config/arm/arm_neon.h: Fix softp typo.
22016 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22018 PR tree-optimization/81365
22019 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22020 aggregate moves onto bb predecessor edges, make sure there are no
22021 loads that could alias the lhs in between the start of bb and the
22024 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22027 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22028 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22029 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22031 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22033 PR tree-optimization/81396
22034 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22035 (init_symbolic_number): Initialize it to 1.
22036 (perform_symbolic_merge): Add n_ops from both operands into the new
22038 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22039 without base_addr as useless if they need more than one operation.
22040 (bswap_replace): Handle !bswap case for NULL base_addr.
22042 2017-07-17 Tom de Vries <tom@codesourcery.com>
22045 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
22048 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
22050 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
22051 conditional builtin define __FIX_LEON3FT_B2BST.
22053 2017-07-17 Daniel Cederman <cederman@gaisler.com>
22055 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
22056 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
22059 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22061 PR rtl-optimization/81424
22062 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
22063 to remove potential trapping from operands if -fnon-call-exceptions.
22065 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22067 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
22068 profile_proability for scalling.
22069 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
22071 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22073 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
22075 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22077 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
22078 fixpoint arithmetics.
22080 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22082 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
22083 fixpoint arithmetics.
22085 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22087 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
22088 fixpoint arithmetics.
22090 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22092 * profile-count.h (profile_probability::from_reg_br_prob_note,
22093 profile_probability::to_reg_br_prob_note): New functions.
22094 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
22095 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
22096 * predict.c (probability_reliable_p): Update.
22097 (edge_probability_reliable_p): Update.
22098 (br_prob_note_reliable_p): Update.
22099 (invert_br_probabilities): Update.
22100 (add_reg_br_prob_note): New function.
22101 (combine_predictions_for_insn): Update.
22102 * asan.c (asan_clear_shadow): Update.
22103 * cfgbuild.c (compute_outgoing_frequencies): Update.
22104 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22105 (update_br_prob_note): Update.
22106 (rtl_verify_edges): Update.
22107 (purge_dead_edges): Update.
22108 (fixup_reorder_chain): Update.
22109 * emit-rtl.c (try_split): Update.
22110 * ifcvt.c (cond_exec_process_insns): Update.
22111 (cond_exec_process_if_block): Update.
22112 (dead_or_predicable): Update.
22113 * internal-fn.c (expand_addsub_overflow): Update.
22114 (expand_neg_overflow): Update.
22115 (expand_mul_overflow): Update.
22116 * loop-doloop.c (doloop_modify): Update.
22117 * loop-unroll.c (compare_and_jump_seq): Update.
22118 * optabs.c (emit_cmp_and_jump_insn_1): Update.
22119 * predict.h: Update.
22120 * reorg.c (mostly_true_jump): Update.
22122 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
22123 * config/alpha/alpha.c (emit_unlikely_jump): Update.
22124 * config/arc/arc.c: (emit_unlikely_jump): Update.
22125 * config/arm/arm.c: (emit_unlikely_jump): Update.
22126 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
22127 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
22128 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
22129 (ix86_print_operand): Update.
22130 (ix86_split_fp_branch): Update.
22131 (predict_jump): Update.
22132 * config/ia64/ia64.c (ia64_print_operand): Update.
22133 * config/mmix/mmix.c (mmix_print_operand): Update.
22134 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
22135 (rs6000_expand_split_stack_prologue): Update.
22136 * config/rs6000/rs6000.c: Update.
22137 * config/s390/s390.c (s390_expand_vec_strlen): Update.
22138 (s390_expand_vec_movstr): Update.
22139 (s390_expand_cs_tdsi): Update.
22140 (s390_expand_split_stack_prologue): Update.
22141 * config/sh/sh.c (sh_print_operand): Update.
22142 (expand_cbranchsi4): Update.
22143 (expand_cbranchdi4): Update.
22144 * config/sparc/sparc.c (output_v9branch): Update.
22145 * config/spu/spu.c (get_branch_target): Update.
22146 (ea_load_store_inline): Update.
22147 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
22148 * config/tilepro/tilepro.c: Update.
22150 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22152 * gimplify.c (mostly_copy_tree_r): Revert latest change.
22153 (gimplify_save_expr): Likewise.
22155 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22157 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
22159 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22161 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
22163 * timevar.def (TV_IPA_FNSUMMARY): Define.
22165 2017-07-16 Daniel Cederman <cederman@gaisler.com>
22167 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
22168 to back store errata sensitive sequence from being generated.
22169 (sqrtdf2_fix): Likewise.
22171 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22173 * tree-ssa-threadupdate.c (compute_path_counts,
22174 update_joiner_offpath_counts): Use profile_probability.
22176 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
22179 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22181 * config/arm/arm-c.c (arm_cpu_builtins): Define
22182 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22184 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22186 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22187 array entries to represent __ieee128 versions of the
22188 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
22189 scalar_extract_sig, and scalar_insert_exp built-in functions.
22190 (altivec_resolve_overloaded_builtin): Add special case handling
22191 for the __builtin_scalar_insert_exp function, as represented by
22192 the P9V_BUILTIN_VEC_VSIEDP constant.
22193 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
22194 exponent support for __ieee128 argument.
22195 (VSESQP): Add scalar extract signature support for __ieee128
22197 (VSTDCNQP): Add scalar test negative support for __ieee128
22199 (VSIEQP): Add scalar insert exponent support for __int128 argument
22200 with __ieee128 result.
22201 (VSIEQPF): Add scalar insert exponent support for __ieee128
22202 argument with __ieee128 result.
22203 (VSTDCQP): Add scalar test data class support for __ieee128
22205 (VSTDCNQP): Add overload support for scalar test negative with
22206 __ieee128 argument.
22207 (VSTDCQP): Add overload support for scalar test data class
22208 __ieee128 argument.
22209 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
22210 UNSPEC_VSX_SXSIGDP.
22211 (UNSPEC_VSX_SIEXPQP): New constant.
22212 (xsxexpqp): New insn for VSX scalar extract exponent quad
22214 (xsxsigqp): New insn for VSX scalar extract significand quad
22216 (xsiexpqpf): New insn for VSX scalar insert exponent quad
22217 precision with floating point argument.
22218 (xststdcqp): New expand for VSX scalar test data class quad
22220 (xststdcnegqp): New expand for VSX scalar test negative quad
22222 (xststdcqp): New insn to match expansions for VSX scalar test data
22223 class quad precision and VSX scalar test negative quad precision.
22224 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
22225 special case operand checking to enforce that second operand of
22226 VSX scalar test data class with quad precision argument is a 7-bit
22228 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22229 prototypes and descriptions of __ieee128 versions of
22230 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
22231 scalar_test_data_class, and scalar_test_neg built-in functions.
22233 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22235 PR tree-optimization/81162
22236 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
22237 replace a negate with an add.
22239 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
22241 * doc/invoke.texi (arm/-mcpu): Document +crypto.
22243 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22245 * config/arm/arm-c.c (arm_cpu_builtins): Define
22246 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22248 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22250 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
22251 (armv8-r): Set ARM Cortex-R52 as default CPU.
22252 * config/arm/arm-tables.opt: Regenerate.
22253 * config/arm/arm-tune.md: Regenerate.
22254 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22256 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
22257 extension for -mcpu=cortex-r52.
22259 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22261 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
22262 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
22263 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
22264 (fp-armv8): Define it as FP_ARMv8 only.
22265 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
22266 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
22268 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
22269 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
22270 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
22271 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
22272 than TARGET_FPU_ARMV8.
22273 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
22274 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
22275 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
22277 * config/arm/neon.md (neon_vrint): Likewise.
22278 (neon_vcvt): Likewise.
22279 (neon_<fmaxmin_op><mode>): Likewise.
22280 (<fmaxmin><mode>3): Likewise.
22281 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
22282 * config/arm/predicates.md (arm_cond_move_operator): Check against
22283 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
22285 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
22287 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
22288 to top of function.
22290 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22292 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
22293 loop in comment with memset.
22295 2017-07-14 Martin Liska <mliska@suse.cz>
22297 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
22298 * dwarf2out.c (is_java): Remove the function.
22299 (output_pubname): Remove usage of the function.
22300 (lower_bound_default): Remove usage of DW_LANG_Java.
22301 (gen_compile_unit_die): Likewise.
22302 * gcc.c: Remove compiler defaults for .java and .zip files.
22303 * gimple-expr.c (remove_suffix): Change as there's no longer
22304 extension than 4-letter one.
22305 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
22306 (gimplify_save_expr): Likewise.
22307 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
22308 as it's possible even for other languages than Java.
22309 * langhooks.h (struct lang_hooks): Remove Java from a comment.
22310 * lto-opts.c (lto_write_options): Remove reference to Java.
22311 * opts.c (strip_off_ending): Update file extension handling.
22312 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
22313 * tree-eh.c (lower_resx): Likewise.
22314 * tree.c (free_lang_data_in_type): Remove dead code.
22315 (find_decls_types_r): Likewise.
22316 (build_common_builtin_nodes): Remove Java from a comment.
22317 (verify_type): Remove dead code.
22318 * varasm.c (assemble_external): Remove Java from a comment.
22320 2017-07-14 Martin Liska <mliska@suse.cz>
22322 * opts.c (finish_options): Add quotes.
22323 (common_handle_option): Likewise.
22325 2017-07-14 Martin Liska <mliska@suse.cz>
22327 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
22328 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
22329 Remove N_SO_PASCAL.
22330 * dwarf2out.c (lower_bound_default): Do not handle
22332 (gen_compile_unit_die): Likewise.
22333 * gcc.c: Remove default extension binding for GNU Pascal.
22334 * stmt.c: Remove Pascal language from a comment.
22335 * xcoffout.c: Likewise.
22337 2017-07-13 David Malcolm <dmalcolm@redhat.com>
22340 * diagnostic-show-locus.c (fixit_cmp): New function.
22341 (layout::layout): Sort m_fixit_hints.
22342 (column_range::column_range): Assert that the values are valid.
22343 (struct char_span): New struct.
22344 (correction::overwrite): New method.
22345 (struct source_line): New struct.
22346 (line_corrections::add_hint): Add assertions. Reimplement memcpy
22347 calls in terms of classes source_line and char_span, and
22348 correction::overwrite.
22349 (selftest::test_overlapped_fixit_printing_2): New function.
22350 (selftest::diagnostic_show_locus_c_tests): Call it.
22352 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
22354 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
22355 early if there is no lhs.
22357 2017-07-13 Martin Liska <mliska@suse.cz>
22359 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
22360 (gen_reference_type_die): Likewise.
22361 * stor-layout.c: Remove Pascal-related comment.
22363 2017-07-13 Martin Liska <mliska@suse.cz>
22365 * opts.c (finish_options): Add quotes to error messages.
22366 (parse_sanitizer_options): Likewise.
22368 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
22370 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
22372 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
22374 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
22376 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22378 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
22380 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
22382 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22385 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
22386 provides the hardware capability bits, define the macro
22387 __BUILTIN_CPU_SUPPORTS__.
22388 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
22389 if GLIBC does not provide the hardware capability bits. Add a
22390 gcc_unreachable call if the built-in cpu function is neither
22391 __builtin_cpu_is nor __builtin_cpu_supports.
22392 (rs6000_get_function_versions_dispatcher): Change the warning
22393 that an old GLIBC is used which does not export the capability
22394 bits to be an error.
22395 * doc/extend.texi (target_clones attribute): Document the
22396 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
22397 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
22398 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
22399 the macros defined by GCC if the newer GLIBC is available.
22401 2017-07-12 Jeff Law <law@redhat.com>
22403 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
22404 remaining includes slightly.
22405 * config/riscv/riscv-builtins.c: Include profile-count.h.
22407 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22410 * config/avr/avr.c (avr_set_current_function): In diagnostic
22411 messages: Quote keywords and (parts of) identifiers.
22412 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
22415 2017-07-12 Carl Love <cel@us.ibm.com>
22417 * config/rs6000/rs6000-c.c: Add support for built-in functions
22418 vector bool char vec_revb (vector bool char);
22419 vector bool short vec_revb (vector short char);
22420 vector bool int vec_revb (vector bool int);
22421 vector bool long long vec_revb (vector bool long long);
22422 * doc/extend.texi: Update the built-in documentation file for the
22423 new built-in functions.
22425 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22427 * config/s390/s390.md: Remove movcc splitter.
22429 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22431 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
22432 load/store on condition.
22434 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22437 * config/avr/avr.c (avr_encode_section_info)
22438 [progmem && !TREE_READONLY]: Error if progmem object needs
22441 2017-07-11 Michael Collison <michael.collison@arm.com>
22443 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
22446 2017-07-11 Carl Love <cel@us.ibm.com>
22448 * config/rs6000/rs6000-c.c: Add support for builtins
22449 vector unsigned int vec_parity_lsbb (vector signed int);
22450 vector unsigned int vec_parity_lsbb (vector unsigned int);
22451 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
22452 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
22453 vector unsigned long long vec_parity_lsbb (vector signed long long);
22454 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
22455 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
22456 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
22457 * doc/extend.texi: Update the built-in documentation file for the
22458 new built-in functions.
22460 2017-07-11 David Malcolm <dmalcolm@redhat.com>
22462 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
22463 (layout::m_primary_loc): New field.
22464 (layout::layout): Initialize new field. Move location filtering
22465 logic from here to...
22466 (layout::maybe_add_location_range): ...this new method. Add
22467 support for filtering to just the lines already specified by other
22469 (layout::will_show_line_p): New method.
22470 (gcc_rich_location::add_location_if_nearby): New method.
22471 (selftest::test_add_location_if_nearby): New test function.
22472 (selftest::diagnostic_show_locus_c_tests): Call it.
22473 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
22476 2017-07-11 Tom de Vries <tom@codesourcery.com>
22478 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
22479 (bb_first_real_insn): New function.
22480 (nvptx_single): Add extra initialization of broadcasted condition
22483 2017-07-11 Nathan Sidwell <nathan@acm.org>
22485 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
22487 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
22489 * doc/extend.texi (AVR Function Attributes): Remove weblink to
22490 Binutils doc as TEXI will mess them up.
22491 * doc/invoke.texi (AVR Options): Same here.
22493 2017-07-11 Daniel Cederman <cederman@gaisler.com>
22495 * config/sparc/sparc.opt (mfix-ut700): New option.
22496 (mfix-gr712rc): Likewise.
22497 (sparc_fix_b2bst): New variable.
22498 * doc/invoke.texi (SPARC options): Document them.
22499 (ARM options): Fix warnings.
22500 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
22501 instructions to prevent sequences that can trigger the store-store
22502 errata for certain LEON3FT processors.
22503 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
22504 (sparc_option_override): Set sparc_fix_b2bst appropriately.
22505 * config/sparc/sparc.md (fix_b2bst): New attribute.
22506 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
22508 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
22511 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
22513 (*rsqrtsf2_sse): Ditto.
22515 (div<mode>3): Macroize insn from divdf3 and divsf3
22516 using MODEF mode iterator.
22518 2017-07-10 Martin Sebor <msebor@redhat.com>
22520 PR tree-optimization/80397
22521 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
22522 instead of testing for equality to INTEGER_TYPE.
22524 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
22526 * config.gcc: Remove uclibc from arc target spec.
22528 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
22530 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
22532 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22535 * lto-wrapper.c (remove_option): New function.
22536 (merge_and_complain): Merge PIC/PIE options more realistically.
22538 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22540 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
22544 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
22545 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
22546 * config.in: Regenerate.
22547 * configure: Regenerate.
22548 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
22549 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
22550 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
22551 (TARGET_GASISR_PROLOGUES): ...target mask.
22552 * common/config/avr/avr-common.c
22553 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
22554 Set -mgas-isr-prologues.
22555 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
22556 INSERT_PASS_BEFORE for it.
22557 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
22558 * config/avr/avr.c (avr_option_override)
22559 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
22560 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
22561 (avr_attribute_table) <no_gccisr>: Add new function attribute.
22562 (avr_set_current_function) <is_no_gccisr>: Init machine field.
22563 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
22565 (make_avr_pass_pre_proep): New function.
22566 (emit_push_sfr) <treg>: Add argument to function and use it
22567 instead of TMP_REG.
22568 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
22569 and set machine->gasisr.yes.
22570 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
22571 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
22572 __gcc_isr.n_pushed to .L__stack_usage.
22573 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
22574 (avr_asm_final_postscan_insn): ...this new static function.
22575 * config/avr/avr.h (machine_function)
22576 <is_no_gccisr, use_L__stack_usage>: New fields.
22577 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
22578 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
22579 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
22580 (gasisr, *gasisr): New expander and insn.
22581 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22582 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
22583 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
22585 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
22587 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
22590 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22592 Move jump-tables out of .text again.
22595 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
22596 (ASM_OUTPUT_ADDR_VEC): New function.
22597 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
22598 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
22599 INSN_ADDRESSes as asm comment.
22600 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
22601 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
22602 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
22603 * config/avr/avr.md (*tablejump): Adjust comment.
22604 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
22605 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
22607 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
22608 (avr_output_addr_vec): New proto.
22609 (avr_log_t) <insn_addresses>: New field.
22611 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
22614 * config/i386/i386.c (ix86_function_arg_advance): Set
22615 outgoing_args_on_stack to true if there are outgoing arguments
22617 (ix86_function_arg): Likewise.
22618 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
22619 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
22620 * config/i386/i386.h (machine_function): Add
22621 outgoing_args_on_stack.
22623 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
22625 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22626 supporting pthreds.
22627 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
22629 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
22631 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
22632 (REAL_H): Remove $(MACHMODE_H).
22633 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
22635 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
22636 $(MACHMODE_H) and double-int.h.
22637 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
22639 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
22642 2017-07-07 Andrew Pinski <apinski@cavium.com>
22644 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
22645 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
22647 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22649 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
22650 Add warning if GCC was not configured to link against a GLIBC that
22651 exports the hardware capability bits.
22652 (make_resolver_func): Make resolver function private and not a
22653 COMDAT function. Create the name with clone_function_name instead
22654 of make_unique_name.
22657 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
22658 correct operand in doing the split.
22660 2017-07-07 Carl Love <cel@us.ibm.com>
22662 * config/rs6000/rs6000-c: Add support for built-in function
22663 vector unsigned short vec_pack_to_short_fp32 (vector float,
22665 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
22666 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
22667 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
22668 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
22669 (convert_4f32_8i16): Add define_expand.
22670 * doc/extend.texi: Update the built-in documentation file for the
22671 new built-in function.
22673 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22675 * config/sparc/m8.md: New file.
22676 * config/sparc/sparc.md: Include m8.md.
22678 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22680 * config/sparc/sparc.opt: New option -mvis4b.
22681 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
22682 (sparc_option_override): Handle VIS4B.
22683 (enum sparc_builtins): Define
22684 SPARC_BUILTIN_DICTUNPACK{8,16,32},
22685 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
22686 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
22687 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
22688 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
22689 (check_constant_argument): New function.
22690 (sparc_vis_init_builtins): Define builtins
22691 __builtin_vis_dictunpack{8,16,32},
22692 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
22693 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
22694 __builtin_vis_fpcmpde{8,16,32}shl and
22695 __builtin_vis_fpcmpur{8,16,32}shl.
22696 (sparc_expand_builtin): Check that the constant operands to
22697 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
22698 constant and in range.
22699 * config/sparc/sparc-c.c (sparc_target_macros): Handle
22701 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
22702 (SPARC_IMM5_P): Likewise.
22703 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
22704 (enabled): Handle vis4b.
22705 (UNSPEC_DICTUNPACK): New unspec.
22706 (UNSPEC_FPCMPSHL): Likewise.
22707 (UNSPEC_FPUCMPSHL): Likewise.
22708 (UNSPEC_FPCMPDESHL): Likewise.
22709 (UNSPEC_FPCMPURSHL): Likewise.
22710 (cpu_feature): New CPU feature `vis4b'.
22711 (dictunpack{8,16,32}): New insns.
22712 (FPCSMODE): New mode iterator.
22713 (fpcscond): New code iterator.
22714 (fpcsucond): Likewise.
22715 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
22716 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
22717 (fpcmpde{8,16,32}{si,di}shl): Likewise.
22718 (fpcmpur{8,16,32}{si,di}shl): Likewise.
22719 * config/sparc/constraints.md: Define constraints `q' for unsigned
22720 2-bit integer constants and `t' for unsigned 5-bit integer
22722 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
22724 (imm5_operand_dictunpack16): Likewise.
22725 (imm5_operand_dictunpack32): Likewise.
22726 (imm2_operand): Likewise.
22727 * doc/invoke.texi (SPARC Options): Document -mvis4b.
22728 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
22729 ditunpack* and fpcmp*shl builtins.
22731 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22733 * config.gcc: Handle m8 in --with-{cpu,tune} options.
22734 * config.in: Add HAVE_AS_SPARC6 define.
22735 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
22737 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
22739 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
22740 (CPP_CPU_SPEC): Handle m8.
22741 (ASM_CPU_SPEC): Likewise.
22742 * config/sparc/sparc-opts.h (enum processor_type): Add
22744 * config/sparc/sparc.c (m8_costs): New struct.
22745 (sparc_option_override): Handle TARGET_CPU_m8.
22746 (sparc32_initialize_trampoline): Likewise.
22747 (sparc64_initialize_trampoline): Likewise.
22748 (sparc_issue_rate): Likewise.
22749 (sparc_register_move_cost): Likewise.
22750 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
22751 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
22752 (ASM_CPU64_DEFAULT_SPEC): Likewise.
22753 (CPP_CPU_SPEC): Handle M8.
22754 (ASM_CPU_SPEC): Likewise.
22755 (AS_M8_FLAG): Define.
22756 * config/sparc/sparc.md: Add m8 to the cpu attribute.
22757 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
22758 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
22760 * configure: Regenerate.
22761 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
22764 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22766 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
22768 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
22769 ("*movdi_insn_sp32"): Do not set v3pipe.
22770 ("*movsi_insn"): Likewise.
22771 ("*movdi_insn_sp64"): Likewise.
22772 ("*movsf_insn"): Likewise.
22773 ("*movdf_insn_sp32"): Likewise.
22774 ("*movdf_insn_sp64"): Likewise.
22775 ("*zero_extendsidi2_insn_sp64"): Likewise.
22776 ("*sign_extendsidi2_insn"): Likewise.
22777 ("*mov<VM32:mode>_insn"): Likewise.
22778 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22779 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22780 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22781 ("<vlop:code><VL:mode>3"): Likewise.
22782 ("*not_<vlop:code><VL:mode>3"): Likewise.
22783 ("*nand<VL:mode>_vis"): Likewise.
22784 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
22785 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
22786 ("one_cmpl<VL:mode>2"): Likewise.
22787 ("faligndata<VM64:mode>_vis"): Likewise.
22788 ("alignaddrsi_vis"): Likewise.
22789 ("alignaddrdi_vis"): Likweise.
22790 ("alignaddrlsi_vis"): Likewise.
22791 ("alignaddrldi_vis"): Likewise.
22792 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22793 ("bmaskdi_vis"): Likewise.
22794 ("bmasksi_vis"): Likewise.
22795 ("bshuffle<VM64:mode>_vis"): Likewise.
22796 ("cmask8<P:mode>_vis"): Likewise.
22797 ("cmask16<P:mode>_vis"): Likewise.
22798 ("cmask32<P:mode>_vis"): Likewise.
22799 ("pdistn<P:mode>_vis"): Likewise.
22800 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22802 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22804 * config/sparc/sparc.md ("subtype"): New insn attribute.
22805 ("*wrgsr_sp64"): Set insn subtype.
22806 ("*rdgsr_sp64"): Likewise.
22807 ("alignaddrsi_vis"): Likewise.
22808 ("alignaddrdi_vis"): Likewise.
22809 ("alignaddrlsi_vis"): Likewise.
22810 ("alignaddrldi_vis"): Likewise.
22811 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22812 ("fexpand_vis"): Likewise.
22813 ("fpmerge_vis"): Likewise.
22814 ("faligndata<VM64:mode>_vis"): Likewise.
22815 ("bshuffle<VM64:mode>_vis"): Likewise.
22816 ("cmask8<P:mode>_vis"): Likewise.
22817 ("cmask16<P:mode>_vis"): Likewise.
22818 ("cmask32<P:mode>_vis"): Likewise.
22819 ("fchksm16_vis"): Likewise.
22820 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
22821 ("fmean16_vis"): Likewise.
22822 ("fp<plusminus_insn>64_vis"): Likewise.
22823 ("<plusminus_insn>v8qi3"): Likewise.
22824 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22825 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
22826 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
22827 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
22828 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
22829 ("*movqi_insn"): Likewise.
22830 ("*movhi_insn"): Likewise.
22831 ("*movsi_insn"): Likewise.
22832 ("movsi_pic_gotdata_op"): Likewise.
22833 ("*movdi_insn_sp32"): Likewise.
22834 ("*movdi_insn_sp64"): Likewise.
22835 ("movdi_pic_gotdata_op"): Likewise.
22836 ("*movsf_insn"): Likewise.
22837 ("*movdf_insn_sp32"): Likewise.
22838 ("*movdf_insn_sp64"): Likewise.
22839 ("*zero_extendhisi2_insn"): Likewise.
22840 ("*zero_extendqihi2_insn"): Likewise.
22841 ("*zero_extendqisi2_insn"): Likewise.
22842 ("*zero_extendqidi2_insn"): Likewise.
22843 ("*zero_extendhidi2_insn"): Likewise.
22844 ("*zero_extendsidi2_insn_sp64"): Likewise.
22845 ("ldfsr"): Likewise.
22846 ("prefetch_64"): Likewise.
22847 ("prefetch_32"): Likewise.
22848 ("tie_ld32"): Likewise.
22849 ("tie_ld64"): Likewise.
22850 ("*tldo_ldub_sp32"): Likewise.
22851 ("*tldo_ldub1_sp32"): Likewise.
22852 ("*tldo_ldub2_sp32"): Likewise.
22853 ("*tldo_ldub_sp64"): Likewise.
22854 ("*tldo_ldub1_sp64"): Likewise.
22855 ("*tldo_ldub2_sp64"): Likewise.
22856 ("*tldo_ldub3_sp64"): Likewise.
22857 ("*tldo_lduh_sp32"): Likewise.
22858 ("*tldo_lduh1_sp32"): Likewise.
22859 ("*tldo_lduh_sp64"): Likewise.
22860 ("*tldo_lduh1_sp64"): Likewise.
22861 ("*tldo_lduh2_sp64"): Likewise.
22862 ("*tldo_lduw_sp32"): Likewise.
22863 ("*tldo_lduw_sp64"): Likewise.
22864 ("*tldo_lduw1_sp64"): Likewise.
22865 ("*tldo_ldx_sp64"): Likewise.
22866 ("*mov<VM32:mode>_insn"): Likewise.
22867 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22868 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22870 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22872 * config/sparc/sparc.md ("type"): New insn type viscmp.
22873 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
22875 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
22876 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
22877 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22878 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
22880 ("n7_vis_logical_11cycle"): Likewise.
22881 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
22882 * config/sparc/niagara2.md ("niag3_vis": Likewise.
22883 * config/sparc/niagara.md ("niag_vis"): Likewise.
22884 * config/sparc/ultra3.md ("us3_fga"): Likewise.
22885 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
22887 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22889 * config/sparc/sparc.md: New instruction type `bmask'.
22890 (bmaskdi_vis): Use the `bmask' type.
22891 (bmasksi_vis): Likewise.
22892 * config/sparc/ultra3.md (us3_array): Likewise.
22893 * config/sparc/niagara7.md (n7_array): Likewise.
22894 * config/sparc/niagara4.md (n4_array): Likewise.
22895 * config/sparc/niagara2.md (niag2_vis): Likewise.
22896 (niag3_vis): Likewise.
22897 * config/sparc/niagara.md (niag_vis): Likewise.
22899 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22901 * ipa-comdats.c: Remove optimize check from gate.
22902 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
22903 for functions not optimized.
22904 (ipa_fn_summary_read): Skip optimize check.
22905 (ipa_fn_summary_write): Likewise.
22906 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
22908 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
22910 (can_inline_edge_p): Check flag_pcc_struct_return for match.
22911 (check_callers): Give up on caller which is not optimized.
22912 (inline_small_functions): Likewise.
22913 (ipa_inline): Do not give up when not optimizing.
22914 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
22915 away unoptimizes cdtors.
22916 (whole_program_function_and_variable_visibility): Do
22917 ipa_discover_readonly_nonaddressable_vars in LTO mode.
22918 * ipa.c (process_references): Do not check optimize.
22919 (symbol_table::remove_unreachable_nodes): Update optimize check.
22920 (set_writeonly_bit): Update optimize check.
22921 (pass_ipa_cdtor_merge::gate): Do not check optimize.
22922 (pass_ipa_single_use::gate): Remove.
22924 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22926 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
22927 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
22928 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
22929 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
22930 permute_load, permute_store, adjust_extract, adjust_splat,
22931 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
22932 replace_swap_with_copy, dump_swap_insn_table,
22933 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
22934 recombine_lvx_pattern, recombine_stvx_pattern,
22935 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
22936 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
22937 to file rs6000-p8swap.c.
22938 * config/rs6000/rs6000-p8swap.c: New file.
22939 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
22940 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
22941 and rs6000*-*-* targets.
22943 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22945 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
22947 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22949 * lto-wrapper.c (merge_and_complain): Do not merge
22950 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
22951 fsigned_zeros, ftrapping_math, fwrapv.
22952 (append_compiler_options): Do not track these options.
22953 (append_linker_options): Likewie
22955 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22957 * cgraphunit.c (cgraph_node::finalize_function): When
22958 !flag_toplevel_reorde set no_reorder flag.
22959 (varpool_node::finalize_decl): Likewise.
22960 (symbol_table::compile): Drop no toplevel reorder path.
22962 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22964 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
22965 edges; zero probability is not better than uninitialized.
22967 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22969 * asan.h (asan_sanitize_allocas_p): Declare.
22970 * asan.c (asan_sanitize_allocas_p): New function.
22971 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
22972 (handle_builtin_alloca): Likewise.
22973 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
22974 if !asan_sanitize_allocas_p.
22975 * params.def (asan-instrument-allocas): Add new option.
22976 * params.h (ASAN_PROTECT_ALLOCAS): Define.
22977 * opts.c (common_handle_option): Disable allocas sanitization for
22980 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22982 * asan.c: Include gimple-fold.h.
22983 (get_last_alloca_addr): New function.
22984 (handle_builtin_stackrestore): Likewise.
22985 (handle_builtin_alloca): Likewise.
22986 (asan_emit_allocas_unpoison): Likewise.
22987 (get_mem_refs_of_builtin_call): Add new parameter, remove const
22988 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
22989 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
22990 (instrument_builtin_call): Pass gimple iterator to
22991 get_mem_refs_of_builtin_call.
22992 (last_alloca_addr): New global.
22993 * asan.h (asan_emit_allocas_unpoison): Declare.
22994 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
22995 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
22996 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
22997 if function calls alloca.
22998 * gimple-fold.c (replace_call_with_value): Remove static keyword.
22999 * gimple-fold.h (replace_call_with_value): Declare.
23000 * internal-fn.c: Include asan.h.
23001 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23002 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23004 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23006 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23007 (C_SELFTEST_FLAGS): New.
23008 (CPP_SELFTEST_FLAGS): New.
23009 (SELFTEST_DEPS): New, from deps of s-selftest.
23010 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23011 (CPP_SELFTEST_DEPS): New.
23012 (selftest): Add dependency on s-selftest-c++.
23013 (s-selftest): Rename to...
23014 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23015 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23016 than SELFTEST_FLAGS.
23017 (selftest-gdb): Rename to...
23018 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23020 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23021 (selftest-valgrind): Rename to...
23022 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23024 (selftest-valgrind): Reintroduce as an alias for
23025 selftest-c-valgrind.
23026 (s-selftest-c++): New.
23027 (selftest-c++-gdb): New.
23028 (selftest-c++-valgrind): New.
23030 2017-07-06 Olivier Hainque <hainque@adacore.com>
23032 * gcc.c (process_command): When deciding if undefined variables
23033 should be ignored when processing specs, accept "gcc -v" as well.
23035 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23037 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23038 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23040 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23042 * config/arm/arm-cpus.in (armv8-r): Add new entry.
23043 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
23044 * config/arm/arm-tables.opt: Regenerate.
23045 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
23047 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
23049 2017-07-06 Carl Love <cel@us.ibm.com>
23051 * ChangeLog: Clean up from mid air collision
23053 2017-07-06 Carl Love <cel@us.ibm.com>
23055 * config/rs6000/rs6000-c.c: Add support for built-in functions
23056 vector signed int vec_subc (vector signed int, vector signed int);
23057 vector signed __int128 vec_subc (vector signed __int128,
23058 vector signed __int128);
23059 vector unsigned __int128 vec_subc (vector unsigned __int128,
23060 vector unsigned __int128);
23061 vector signed int vec_sube (vector signed int, vector signed int,
23062 vector signed int);
23063 vector unsigned int vec_sube (vector unsigned int,
23064 vector unsigned int,
23065 vector unsigned int);
23066 vector signed __int128 vec_sube (vector signed __int128,
23067 vector signed __int128,
23068 vector signed__int128);
23069 vector unsigned __int128 vec_sube (vector unsigned __int128,
23070 vector unsigned __int128,
23071 vector unsigned __int128);
23072 vector signed int vec_subec (vector signed int, vector signed int,
23073 vector signed int);
23074 vector unsigned int vec_subec (vector unsigned int,
23075 vector unsigned int,
23076 vector unsigned int);
23077 vector signed __int128 vec_subec (vector signed __int128,
23078 vector signed __int128,
23079 vector signed__int128);
23080 vector unsigned __int128 vec_subec (vector unsigned __int128,
23081 vector unsigned __int128,
23082 vector unsigned __int128);
23083 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
23084 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
23085 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
23086 BU_ALTIVEC_OVERLOAD_X definitions.
23087 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
23088 * doc/extend.texi: Update the built-in documentation file for the new
23089 built-in functions.
23091 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23094 * diagnostic-show-locus.c (layout::layout): Use start and finish
23095 spelling location for the start and finish of each range.
23096 * genmatch.c (linemap_client_expand_location_to_spelling_point):
23097 Add unused aspect param.
23098 * input.c (expand_location_1): Add "aspect" param, and use it
23099 to access the correct part of the location.
23100 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
23102 (expand_location_to_spelling_point): Likewise.
23103 (linemap_client_expand_location_to_spelling_point): Add "aspect"
23104 param, and pass it to expand_location_1.
23106 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
23108 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
23109 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
23110 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
23111 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
23112 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
23113 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
23114 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
23115 _mm_maskz_getmant_ss): New intrinsics.
23116 (__builtin_ia32_getexpss128_mask): Changed to ...
23117 __builtin_ia32_getexpss128_round ... this.
23118 (__builtin_ia32_getexpsd128_mask): Changed to ...
23119 __builtin_ia32_getexpsd128_round ... this.
23120 * config/i386/i386-builtin-types.def
23121 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
23122 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
23123 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
23124 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
23125 __builtin_ia32_getmantss_mask_round): New builtins.
23126 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
23127 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
23128 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
23129 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
23130 * config/i386/sse.md
23131 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
23132 avx512f_sgetexp<mode><mask_scalar_name>
23133 <round_saeonly_scalar_name> ... this.
23134 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
23135 %0, %1, %2<round_saeonly_op3>}): Changed to ...
23136 vgetexp<ssescalarmodesuffix>
23137 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23138 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
23139 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
23140 avx512f_vgetmant<mode><mask_scalar_name>
23141 <round_saeonly_scalar_name> ... this.
23142 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23143 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
23144 vgetmant<ssescalarmodesuffix>
23145 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
23146 %0<mask_scalar_operand4>, %1, %2
23147 <round_saeonly_scalar_mask_op4>, %3} ... this.
23148 * config/i386/subst.md (mask_scalar_operand4,
23149 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
23150 round_saeonly_scalar_nimm_predicate): New subst attributes.
23152 2017-07-06 Julia Koval <julia.koval@intel.com>
23154 * config/i386/i386.c (ix86_erase_embedded_rounding):
23155 Remove code for old rounding pattern.
23157 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
23159 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
23161 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
23163 * doc/sourcebuild.texi (Test Directives, Variants of
23164 dg-require-support): Add documentation for dg-require-stack-check.
23166 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
23168 * config/i386/subst.md (mask_scalar, round_scalar,
23169 round_saeonly_scalar): New meta-templates.
23170 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
23171 round_scalar_mask_operand3, round_scalar_mask_op3,
23172 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
23173 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
23174 round_saeonly_scalar_constraint,
23175 round_saeonly_scalar_prefix): New subst attribute.
23176 * config/i386/sse.md
23177 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
23178 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
23179 <round_scalar_name> ... this.
23180 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
23181 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
23182 <round_scalar_name> ... this.
23183 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
23184 <sse>_vm<code><mode>3<mask_scalar_name>
23185 <round_saeonly_scalar_name> ... this.
23186 (v<plusminus_mnemonic><ssescalarmodesuffix>
23187 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23188 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23189 v<plusminus_mnemonic><ssescalarmodesuffix>
23190 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23191 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23192 (v<multdiv_mnemonic><ssescalarmodesuffix>
23193 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23194 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23195 v<multdiv_mnemonic><ssescalarmodesuffix>
23196 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23197 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23198 (v<maxmin_float><ssescalarmodesuffix>
23199 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
23200 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
23201 v<maxmin_float><ssescalarmodesuffix>
23202 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23203 %0<mask_scalar_operand3>, %1, %<iptr>2
23204 <round_saeonly_scalar_mask_op3>} ... this.
23206 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
23208 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
23209 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
23211 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
23212 Alan Hayward <alan.hayward@arm.com>
23213 David Sherwood <david.sherwood@arm.com>
23215 * combine.c (simplify_if_then_else): Remove "enum" before
23217 * compare-elim.c (can_eliminate_compare): Likewise.
23218 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
23220 (aarch64_lookup_simd_builtin_type): Likewise.
23221 (aarch64_simd_builtin_type): Likewise.
23222 (aarch64_init_simd_builtin_types): Likewise.
23223 (aarch64_simd_expand_args): Likewise.
23224 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
23226 (aarch64_reverse_mask): Likewise.
23227 (aarch64_simd_emit_reg_reg_move): Likewise.
23228 (aarch64_gen_adjusted_ldpstp): Likewise.
23229 (aarch64_ccmp_mode_to_code): Likewise.
23230 (aarch64_operands_ok_for_ldpstp): Likewise.
23231 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23232 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
23234 (aarch64_min_divisions_for_recip_mul): Likewise.
23235 (aarch64_reassociation_width): Likewise.
23236 (aarch64_get_condition_code_1): Likewise.
23237 (aarch64_simd_emit_reg_reg_move): Likewise.
23238 (aarch64_simd_attr_length_rglist): Likewise.
23239 (aarch64_reverse_mask): Likewise.
23240 (aarch64_operands_ok_for_ldpstp): Likewise.
23241 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23242 (aarch64_gen_adjusted_ldpstp): Likewise.
23243 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
23245 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
23246 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
23247 (arm_lookup_simd_builtin_type): Likewise.
23248 (arm_simd_builtin_type): Likewise.
23249 (arm_init_simd_builtin_types): Likewise.
23250 (arm_expand_builtin_args): Likewise.
23251 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
23252 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
23253 (ft32_setup_incoming_varargs): Likewise.
23254 (ft32_function_arg): Likewise.
23255 (ft32_function_arg_advance): Likewise.
23256 (ft32_pass_by_reference): Likewise.
23257 (ft32_arg_partial_bytes): Likewise.
23258 (ft32_valid_pointer_mode): Likewise.
23259 (ft32_addr_space_pointer_mode): Likewise.
23260 (ft32_addr_space_legitimate_address_p): Likewise.
23261 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
23263 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
23264 (ix86_emit_outlined_ms2sysv_restore): Likewise.
23265 (iamcu_alignment): Likewise.
23266 (canonicalize_vector_int_perm): Likewise.
23267 (ix86_noce_conversion_profitable_p): Likewise.
23268 (ix86_mpx_bound_mode): Likewise.
23269 (ix86_operands_ok_for_move_multiple): Likewise.
23270 * config/microblaze/microblaze-protos.h
23271 (microblaze_expand_conditional_branch_reg): Likewise.
23272 * config/microblaze/microblaze.c
23273 (microblaze_expand_conditional_branch_reg): Likewise.
23274 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
23276 (rs6000_reassociation_width): Likewise.
23277 (rs6000_invalid_binary_op): Likewise.
23278 (fusion_p9_p): Likewise.
23279 (emit_fusion_p9_load): Likewise.
23280 (emit_fusion_p9_store): Likewise.
23281 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
23283 (riscv_hard_regno_mode_ok_p): Likewise.
23284 (riscv_address_insns): Likewise.
23285 (riscv_split_symbol): Likewise.
23286 (riscv_legitimize_move): Likewise.
23287 (riscv_function_value): Likewise.
23288 (riscv_hard_regno_nregs): Likewise.
23289 (riscv_expand_builtin): Likewise.
23290 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
23291 (riscv_build_integer): Likewise.
23292 (riscv_split_integer): Likewise.
23293 (riscv_legitimate_constant_p): Likewise.
23294 (riscv_cannot_force_const_mem): Likewise.
23295 (riscv_regno_mode_ok_for_base_p): Likewise.
23296 (riscv_valid_base_register_p): Likewise.
23297 (riscv_valid_offset_p): Likewise.
23298 (riscv_valid_lo_sum_p): Likewise.
23299 (riscv_classify_address): Likewise.
23300 (riscv_legitimate_address_p): Likewise.
23301 (riscv_address_insns): Likewise.
23302 (riscv_load_store_insns): Likewise.
23303 (riscv_force_binary): Likewise.
23304 (riscv_split_symbol): Likewise.
23305 (riscv_force_address): Likewise.
23306 (riscv_legitimize_address): Likewise.
23307 (riscv_move_integer): Likewise.
23308 (riscv_legitimize_const_move): Likewise.
23309 (riscv_legitimize_move): Likewise.
23310 (riscv_address_cost): Likewise.
23311 (riscv_subword): Likewise.
23312 (riscv_output_move): Likewise.
23313 (riscv_canonicalize_int_order_test): Likewise.
23314 (riscv_emit_int_order_test): Likewise.
23315 (riscv_function_arg_boundary): Likewise.
23316 (riscv_pass_mode_in_fpr_p): Likewise.
23317 (riscv_pass_fpr_single): Likewise.
23318 (riscv_pass_fpr_pair): Likewise.
23319 (riscv_get_arg_info): Likewise.
23320 (riscv_function_arg): Likewise.
23321 (riscv_function_arg_advance): Likewise.
23322 (riscv_arg_partial_bytes): Likewise.
23323 (riscv_function_value): Likewise.
23324 (riscv_pass_by_reference): Likewise.
23325 (riscv_setup_incoming_varargs): Likewise.
23326 (riscv_print_operand): Likewise.
23327 (riscv_elf_select_rtx_section): Likewise.
23328 (riscv_save_restore_reg): Likewise.
23329 (riscv_for_each_saved_reg): Likewise.
23330 (riscv_register_move_cost): Likewise.
23331 (riscv_hard_regno_mode_ok_p): Likewise.
23332 (riscv_hard_regno_nregs): Likewise.
23333 (riscv_class_max_nregs): Likewise.
23334 (riscv_memory_move_cost): Likewise.
23335 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
23336 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
23337 (rl78_addr_space_address_mode): Likewise.
23338 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23340 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
23341 (rs6000_reassociation_width): Likewise.
23342 (rs6000_invalid_binary_op): Likewise.
23343 (fusion_p9_p): Likewise.
23344 (emit_fusion_p9_load): Likewise.
23345 (emit_fusion_p9_store): Likewise.
23346 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
23347 (ok_for_simple_move_operands): Likewise.
23348 (ok_for_simple_move_strict_operands): Likewise.
23349 (ok_for_simple_arith_logic_operands): Likewise.
23350 (visium_legitimize_reload_address): Likewise.
23351 (visium_select_cc_mode): Likewise.
23352 (output_cbranch): Likewise.
23353 (visium_split_double_move): Likewise.
23354 (visium_expand_copysign): Likewise.
23355 (visium_expand_int_cstore): Likewise.
23356 (visium_expand_fp_cstore): Likewise.
23357 * config/visium/visium.c (visium_pass_by_reference): Likewise.
23358 (visium_function_arg): Likewise.
23359 (visium_function_arg_advance): Likewise.
23360 (visium_libcall_value): Likewise.
23361 (visium_setup_incoming_varargs): Likewise.
23362 (visium_legitimate_constant_p): Likewise.
23363 (visium_legitimate_address_p): Likewise.
23364 (visium_legitimize_address): Likewise.
23365 (visium_secondary_reload): Likewise.
23366 (visium_register_move_cost): Likewise.
23367 (visium_memory_move_cost): Likewise.
23368 (prepare_move_operands): Likewise.
23369 (ok_for_simple_move_operands): Likewise.
23370 (ok_for_simple_move_strict_operands): Likewise.
23371 (ok_for_simple_arith_logic_operands): Likewise.
23372 (visium_function_value_1): Likewise.
23373 (rtx_ok_for_offset_p): Likewise.
23374 (visium_legitimize_reload_address): Likewise.
23375 (visium_split_double_move): Likewise.
23376 (visium_expand_copysign): Likewise.
23377 (visium_expand_int_cstore): Likewise.
23378 (visium_expand_fp_cstore): Likewise.
23379 (visium_split_cstore): Likewise.
23380 (visium_select_cc_mode): Likewise.
23381 (visium_split_cbranch): Likewise.
23382 (output_cbranch): Likewise.
23383 (visium_print_operand_address): Likewise.
23384 * expmed.c (flip_storage_order): Likewise.
23385 * expmed.h (emit_cstore): Likewise.
23386 (flip_storage_order): Likewise.
23387 * genrecog.c (validate_pattern): Likewise.
23388 * hsa-gen.c (gen_hsa_addr): Likewise.
23389 * internal-fn.c (expand_arith_overflow): Likewise.
23390 * ira-color.c (allocno_copy_cost_saving): Likewise.
23391 * lra-assigns.c (find_hard_regno_for_1): Likewise.
23392 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
23393 (process_invariant_for_inheritance): Likewise.
23394 * lra-eliminations.c (move_plus_up): Likewise.
23395 * omp-low.c (lower_oacc_reductions): Likewise.
23396 * simplify-rtx.c (simplify_subreg): Likewise.
23397 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
23398 (TARGET_CHKP_BOUND_MODE): Likewise..
23399 * targhooks.c (default_chkp_bound_mode): Likewise.
23400 (default_setup_incoming_vararg_bounds): Likewise.
23401 * targhooks.h (default_chkp_bound_mode): Likewise.
23402 (default_setup_incoming_vararg_bounds): Likewise.
23403 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
23404 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
23405 (have_whole_vector_shift): Likewise.
23406 * tree-vect-stmts.c (vectorizable_load): Likewise.
23407 * doc/tm.texi: Regenerate.
23409 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23411 Graceful degrade if Binutils PR21472 is not available.
23414 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
23415 .rodata in flash test fails.
23416 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
23417 * confgure: Regenerate.
23418 * config.in: Regenerate.
23419 * config/avr/avr.c (avr_asm_named_section)
23420 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
23421 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
23422 (avr_asm_init_sections): Same.
23424 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23426 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
23427 (fma<VH:mode>4_intrinsic): Likewise.
23428 (*fmsub<VCVTF:mode>4): Likewise.
23429 (*fmsub<VH:mode>4_intrinsic): Likewise.
23431 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23434 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
23435 Don't depend on "optimize > 0".
23436 (out_movhi_r_mr, out_movqi_mr_r): Same.
23437 (out_movhi_mr_r, out_movqi_r_mr): Same.
23438 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
23439 io_address_operand on "optimize > 0".
23441 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23443 * tree-loop-distribution.c: Add general explanantion on the pass.
23444 (generate_loops_for_partition): Mark distributed loop.
23445 (pg_add_dependence_edges): New parameter. Handle alias data
23446 dependence specially and record it in the parameter if asked.
23447 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
23448 (init_partition_graph_vertices, add_partition_graph_edge): New.
23449 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
23450 (free_partition_graph_vdata, build_partition_graph): New.
23451 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
23452 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
23453 (data_ref_segment_size, latch_dominated_by_data_ref): New.
23454 (compute_alias_check_pairs, version_loop_by_alias_check): New.
23455 (version_for_distribution_p, finalize_partitions): New.
23456 (distribute_loop): Handle alias data dependence specially. Factor
23457 out loop fusion code as functions and call these functions.
23459 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23461 * tree-loop-distribution.c (classify_partition): New parameter and
23462 better handle reduction statement.
23463 (rdg_build_partitions): Revise comment.
23464 (distribute_loop): Compute statements in all partitions and pass it
23465 to classify_partition.
23467 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23469 * tree-loop-distribution.c (enum partition_type): New.
23470 (struct partition): New field type.
23471 (partition_merge_into): Add parameter. Update partition type.
23472 (data_dep_in_cycle_p, update_type_for_merge): New functions.
23473 (build_rdg_partition_for_vertex): Compute partition type.
23474 (rdg_build_partitions): Dump partition type.
23475 (distribute_loop): Update calls to partition_merge_into.
23477 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23479 * tree-loop-distribution.c (struct ddr_hasher): New.
23480 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
23482 (classify_partition): Call get_data_dependence.
23483 (pg_add_dependence_edges): Ditto.
23484 (distribute_loop): Release data dependence hash table.
23486 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23488 * tree-loop-distribution.c (ref_base_address): Delete.
23489 (similar_memory_accesses): Rename ...
23490 (share_memory_accesses): ... to this. Check if partitions access
23491 the same memory reference.
23492 (distribute_loop): Call share_memory_accesses.
23494 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23496 * tree-loop-distribution.c (struct partition): New field recording
23497 its data reference.
23498 (partition_alloc, partition_free): Init and release data refs.
23499 (partition_merge_into): Merge data refs.
23500 (build_rdg_partition_for_vertex): Collect data refs for partition.
23501 (pg_add_dependence_edges): Change parameters from vector to bitmap.
23503 (distribute_loop): Remve data refs from vertice data of partition
23506 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23508 * tree-loop-distribution.c (params.h): Include header file.
23509 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
23510 (datarefs_vec): New global var.
23511 (create_rdg_vertices): Use datarefs_vec directly.
23512 (free_rdg): Don't free data references.
23513 (build_rdg): Update use. Don't free data references.
23514 (distribute_loop): Compute global variable for data references.
23515 Bail out if there are too many data references.
23517 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23519 * tree-loop-distribution.c (loop_nest): New global var.
23520 (build_rdg): Use loop directly, rather than loop nest.
23521 (pg_add_dependence_edges): Remove loop nest parameter. Use global
23523 (distribute_loop): Compute global variable loop nest. Update use.
23525 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23527 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
23528 (partition_merge_into): New parameter. Dump reason for fusion.
23529 (distribute_loop): Update use of partition_merge_into.
23531 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23533 * tree-loop-distribution.c (bb_top_order_index): New.
23534 (bb_top_order_index_size, bb_top_order_cmp): New.
23535 (stmts_from_loop): Use topological order.
23536 (pass_loop_distribution::execute): Compute and release topological
23537 order for basic blocks.
23539 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23541 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
23544 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23546 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
23547 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
23548 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
23549 * internal-fn.def (LOOP_DIST_ALIAS): New.
23550 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
23551 (fold_loop_internal_call): ... this.
23552 (vect_loop_dist_alias_call): New function.
23553 (set_uid_loop_bbs): Call fold_loop_internal_call.
23554 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
23557 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23560 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
23561 Require dead FLAGS_REG at the beginning of a peephole.
23563 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23566 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
23567 arguments in the call to __builtin_ia32_sbb_u32.
23568 (_subborrow_u64): Swap _X and _Y arguments in the call to
23569 __builtin_ia32_sbb_u64.
23571 2017-07-04 Jakub Jelinek <jakub@redhat.com>
23574 * tree-vrp.c (compare_assert_loc): Turn into a function template
23575 with stable template parameter. Only test if a->e is NULL,
23576 !a->e == !b->e has been verified already. Use e == NULL or
23577 e != NULL instead of e or ! e tests. If stable is true, don't use
23578 iterative_hash_expr, on the other side allow a or b or both NULL
23579 and sort the NULLs last.
23580 (process_assert_insertions): Sort using compare_assert_loc<false>
23581 instead of compare_assert_loc, later sort using
23582 compare_assert_loc<true> before calling process_assert_insertions_for
23583 in a loop. Use break instead of continue once seen NULL pointer.
23585 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23587 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23588 Cortex-R7 and Cortex-R8 processors.
23590 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23592 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
23593 uninitialized while src is not.
23595 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
23597 * common/config/arm/arm-common.c: Adjust include path for
23599 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
23600 (arm-cpu.h): Create in build directory. Adjust dependency rules.
23601 (arm-cpu-data.h): Likewise.
23602 (arm-cpu-cdata.h): Likewise.
23603 * config/arm/arm-cpu.h: Delete.
23604 * config/arm/arm-cpu-cdata.h: Delete.
23605 * config/arm/arm-cpu-data.h: Delete.
23607 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
23609 * config/arm/arm-cpus.in (cortex-a55): New.
23610 (cortex-a75): Likewise.
23611 (cortex-a75.cortex-a55): Likewise.
23612 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
23614 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
23615 * config/arm/arm-cpu-cdata.h: Regenerate.
23616 * config/arm/arm-cpu-data.h: Regenerate.
23617 * config/arm/arm-cpu.h: Regenerate.
23618 * config/arm/arm-tables.opt: Regenerate.
23619 * config/arm/arm-tune.md: Regenerate.
23621 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23623 * haifa-sched.c (sched_create_recovery_edges): Update profile.
23625 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23627 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
23630 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
23632 PR tree-optimization/81292
23633 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
23634 full_string_p, also call adjust_related_strinfos if the adjustment
23635 is simple, otherwise invalidate related strinfos.
23637 2017-07-04 Martin Liska <mliska@suse.cz>
23640 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
23641 newly created variable as DECL_IGNORED_P.
23643 2017-07-04 Martin Liska <mliska@suse.cz>
23646 * ipa-inline.c (inline_small_functions):
23647 Use xstrdup_for_dump.
23649 2017-07-04 Tom de Vries <tom@codesourcery.com>
23651 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
23653 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23656 * config/darwin.c (darwin_function_switched_text_sections):
23659 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
23661 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
23663 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23665 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
23667 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23669 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
23670 min_profitable_iters, and th as inclusive lower bounds.
23671 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
23672 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
23673 for min_profitable_iters and min_profitable_estimate.
23674 (vect_transform_loop): Treat th as an inclusive lower bound.
23675 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23677 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23680 * config/darwin.c (darwin_function_switched_text_sections):
23681 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
23682 in two pieces, and suppress the use of buf.
23684 2017-07-03 Nathan Sidwell <nathan@acm.org>
23686 * hash-table.h (hash_table_mod1): Fix indentation.
23688 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23690 PR middle-end/81290
23691 * predict.c (force_edge_cold): Be more careful about propagation
23693 * profile-count.h (profile_probability::guessed,
23694 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
23696 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
23698 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
23700 * doc/invoke.texi (rcpc architecture extension): Document it.
23702 2017-07-03 Richard Biener <rguenther@suse.de>
23704 PR tree-optimization/60510
23705 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
23706 the scalar reduction PHI and use it.
23707 (vectorizable_reduction): Properly guard the single_defuse_cycle
23708 path for non-SLP reduction chains where we cannot use it.
23709 Rework reduc_def/index and vector type deduction. Rework
23710 vector operand gathering during reduction op code-gen.
23711 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
23712 chains dissolve the chain and leave it to non-SLP reduction
23715 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23717 * tree-data-ref.h (dr_alignment): Declare.
23718 * tree-data-ref.c (dr_alignment): New function.
23719 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
23720 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23722 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
23724 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23726 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
23727 and base_misalignment fields.
23728 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
23729 * tree-data-ref.c: Include builtins.h.
23730 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
23731 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
23732 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
23733 * tree-vect-data-refs.c: Include tree-cfg.h.
23734 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
23735 fields instead of calculating an alignment here.
23736 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
23737 innermost_loop_behavior fields.
23739 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23741 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
23743 (DR_STEP_ALIGNMENT): New macro.
23744 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
23745 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
23746 (create_data_ref): Print it.
23747 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
23748 to tell whether the step preserves vector (mis)alignment.
23749 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23750 Move the check for an integer step and generalise to all INTEGER_CST.
23751 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
23752 Print the outer step alignment.
23754 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23756 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
23757 with offset_alignment.
23758 (DR_ALIGNED_TO): Delete.
23759 (DR_OFFSET_ALIGNMENT): New macro.
23760 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
23761 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
23762 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
23763 (create_data_ref): Likewise.
23764 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23765 (vect_analyze_data_refs): Likewise.
23766 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
23767 creating dummy innermost behavior.
23769 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23771 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
23772 with a "innermost_loop_behavior *" and refeence tree.
23773 * tree-data-ref.c (dr_analyze_innermost): Likewise.
23774 (create_data_ref): Update call accordingly.
23775 * tree-predcom.c (find_looparound_phi): Likewise.
23777 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23779 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
23780 fields with dr_wrt_vec_loop.
23781 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
23782 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
23783 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
23784 (vect_dr_behavior): New function.
23785 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23786 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23787 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
23788 track whether the step preserves the misalignment.
23789 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23790 Use vect_dr_behavior.
23791 (vect_setup_realignment): Update call accordingly.
23792 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
23793 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
23794 call to vect_create_addr_base_for_vector_ref.
23795 (vect_create_cond_for_align_checks): Likewise.
23796 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
23797 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
23798 (vect_recog_mask_conversion_pattern): Likewise.
23799 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
23800 (new_stmt_vec_info): Remove redundant zeroing.
23802 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23804 * common/config/arm/arm-common.c (arm_be8_option): New function.
23805 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
23806 (ISA_ARMv6): Add isa_bit_be8.
23807 * config/arm/arm.h (arm_be8_option): Add prototype.
23808 (BE8_SPEC_FUNCTION): New define.
23809 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
23810 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
23811 (mlittle-endian): Similarly.
23812 (mbe8, mbe32): New options.
23813 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
23814 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
23816 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23818 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
23820 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23822 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
23823 (cleanup_tree_cfg_bb): Use it.
23824 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
23826 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
23828 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23831 * loop-doloop.c (add_test): Update profile.
23833 2017-07-03 Martin Liska <mliska@suse.cz>
23836 * sanopt.c (rewrite_usage_of_param): New function.
23837 (sanitize_rewrite_addressable_params): Likewise.
23838 (pass_sanopt::execute): Call rewrite_usage_of_param.
23840 2017-07-03 Richard Biener <rguenther@suse.de>
23842 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
23843 back to using VIEW_CONVERT_EXPR.
23845 2017-07-03 Martin Liska <mliska@suse.cz>
23848 * doc/extend.texi: Document when a resolver function is
23849 generated for target_clones.
23851 2017-07-03 Martin Liska <mliska@suse.cz>
23853 * asan.c (asan_emit_stack_protection): Unpoison just red zones
23854 and shadow memory of auto variables which are subject of
23855 use-after-scope sanitization.
23856 (asan_expand_mark_ifn): Add do set only when is_poison.
23858 2016-07-03 Richard Biener <rguenther@suse.de>
23860 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
23862 (vect_force_simple_reduction): Record reduction def -> phi mapping.
23863 (vectorizable_reduction): Perform reduction PHI creation when
23864 visiting a reduction PHI and adjust and simplify code generation
23865 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
23866 (vect_transform_loop): Visit reduction PHIs.
23867 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
23868 defs into the SLP tree.
23869 (vect_build_slp_tree): Reduction defs terminate the recursion.
23870 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
23872 (vect_get_vec_defs_for_stmt_copy): Export.
23873 (vect_get_vec_defs): Likewise.
23874 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
23876 (vect_get_vec_defs_for_stmt_copy): Declare.
23877 (vect_get_vec_defs): Likewise.
23879 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23881 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
23882 parameter with a "loop" parameter and use it instead of the
23883 loop containing DR_STMT. Don't check simple_iv when doing
23884 BB analysis. Describe the two analysis modes in the comment.
23886 2017-07-03 Tom de Vries <tom@codesourcery.com>
23888 PR tree-optimization/69468
23889 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
23890 (find_same_succ_bb): Handle ignore_edge_flags.
23892 2017-07-03 Tom de Vries <tom@codesourcery.com>
23894 PR tree-optimization/81192
23895 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
23897 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
23899 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
23901 2017-07-03 Tom de Vries <tom@codesourcery.com>
23903 PR tree-optimization/81192
23904 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
23905 BB_SAME_SUCC (bb) == NULL.
23907 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23909 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
23912 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23914 * dumpfile.c: Include profile-count.h
23915 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
23917 (insert_cond_bb): Update profile.
23918 * tree-cfg.h (insert_cond_bb): Update prototype.
23919 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
23920 * tree-dump.c: Do not include tree-cfg.
23922 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23924 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
23926 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23928 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
23931 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23933 * tree-complex.c (expand_complex_div_wide): update profile.
23935 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23936 Alan Hayward <alan.hayward@arm.com>
23937 David Sherwood <david.sherwood@arm.com>
23939 * Makefile.in (MACHMODE_H): Remove insn-modes.h
23940 (CORETYPES_H): New define.
23941 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
23942 (insn-modes-inline.h, s-modes-inline-h): New rules.
23943 (generated_files): Add insn-modes-inline.h.
23944 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
23945 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
23946 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
23947 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
23948 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
23949 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
23950 (build/gencodes.o, build/genconditions.o): Likewise.
23951 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
23952 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
23953 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
23954 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
23955 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
23956 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
23957 * coretypes.h: Include everything up to real.h for generators.
23958 Include insn-modes.h first. Include wide-int-print.h after
23959 wide-int.h. Include insn-modes-inline.h and then machmode.h.
23960 * machmode.h: Don't include insn-modes.h here.
23961 * function-tests.c: Remove includes of signop.h, machmode.h,
23962 double-int.h and wide-int.h.
23964 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
23966 * optc-save-gen.awk: Likewise.
23967 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
23968 * godump.c: Remove include of wide-int-print.h.
23969 * pretty-print.h: Likewise.
23970 * wide-int-print.cc: Likewise.
23971 * wide-int.cc: Likewise.
23972 * hash-map-tests.c: Remove include of signop.h.
23973 * hash-set-tests.c: Likewise.
23974 * rtl-tests.c: Likewise.
23975 * mkconfig.sh: Remove include of machmode.h.
23976 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
23978 (emit_insn_modes_inline_h): ...this new function. Emit the code
23979 into an insn-modes-inline.h header file, adding appropriate
23980 include guards and end comments.
23981 (emit_insn_modes_c_header): Remove include of machmode.h.
23982 (emit_min_insn_modes_c_header): Include coretypes.h rather than
23984 (main): Handle -i flag and call emit_insn_modes_inline_h when
23987 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23989 * tree-ssa-strlen.c (strinfo): Rename the length field to
23990 nonzero_chars. Add a full_string_p field.
23991 (compare_nonzero_chars, zero_length_string_p): New functions.
23992 (get_addr_stridx): Add an offset_out parameter.
23993 Use compare_nonzero_chars.
23994 (get_stridx): Update accordingly. Use compare_nonzero_chars.
23995 (new_strinfo): Update after above changes to strinfo.
23996 (set_endptr_and_length): Set full_string_p.
23997 (get_string_length): Update after above changes to strinfo.
23998 (unshare_strinfo): Update call to new_strinfo.
23999 (maybe_invalidate): Likewise.
24000 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24001 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24002 as a uhwi instead of an shwi. Update after above changes to
24003 strinfo and new_strinfo.
24004 (zero_length_string): Assert that chainsi contains full strings.
24005 Use zero_length_string_p. Update call to new_strinfo.
24006 (adjust_related_strinfos): Update after above changes to strinfo.
24007 Copy full_string_p from origsi.
24008 (adjust_last_stmt): Use zero_length_string_p.
24009 (handle_builtin_strlen): Update after above changes to strinfo and
24010 new_strinfo. Install the lhs as the string length if the previous
24011 entry didn't describe a full string.
24012 (handle_builtin_strchr): Update after above changes to strinfo
24014 (handle_builtin_strcpy): Likewise.
24015 (handle_builtin_strcat): Likewise.
24016 (handle_builtin_malloc): Likewise.
24017 (handle_pointer_plus): Likewise.
24018 (handle_builtin_memcpy): Likewise. Track nonzero characters
24019 that aren't necessarily followed by a nul terminator.
24020 (handle_char_store): Likewise.
24022 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24024 PR tree-optimization/80769
24025 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24026 for malloc and calloc. Document the new invariant that all related
24027 strinfos have delayed lengths or none do.
24028 (verify_related_strinfos): Move earlier in file.
24029 (set_endptr_and_length): New function, split out from...
24030 (get_string_length): ...here. Also set the lengths of related
24032 (zero_length_string): Assert that chainsi has known (rather than
24034 (adjust_related_strinfos): Likewise.
24036 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24038 PR tree-optimization/81136
24039 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24040 assert that two references with the same misalignment have the same
24041 compile-time misalignment if those compile-time misalignments
24044 2017-07-01 Andi Kleen <ak@linux.intel.com>
24046 * print-tree.c (print_node): Print all attributes.
24048 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24050 * cfg.c (scale_bbs_frequencies): New function.
24051 * cfg.h (scale_bbs_frequencies): Declare it.
24052 * cfgloopanal.c (single_likely_exit): Cleanup.
24053 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
24055 (scale_loop_profile): Likewise.
24056 (loop_version): Likewise.
24057 (create_empty_loop_on_edge): Update.
24058 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
24059 scale_loop_frequencies, scale_loop_profile, loopify,
24060 loop_version): Update prototypes.
24061 * modulo-sched.c (sms_schedule): Update.
24062 * predict.c (unlikely_executed_edge_p): Also check probability.
24063 (probably_never_executed_edge_p): Fix typo.
24064 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24065 * tree-parloops.c (gen_parallel_loop): Update.
24066 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
24067 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24068 * tree-ssa-loop-split.c (split_loop): Update.
24069 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24070 * tree-vect-loop-manip.c (vect_do_peeling): Update.
24071 (vect_loop_versioning): Update.
24072 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24074 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24076 * trans-mem.c (split_bb_make_tm_edge): Update profile.
24078 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24080 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
24081 to keep profile consistent.
24083 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24085 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
24086 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
24087 * profile-count.h (max_safe_multiplier): Make unsigned.
24088 (profile_count::guessed_zero): New.
24090 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24092 * bb-reorder.c (fix_up_crossing_landing_pad,
24093 fix_crossing_conditional_branches): Use make_single_succ_edge
24094 to keep profile consistent.
24096 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24098 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
24101 2017-07-01 Jakub Jelinek <jakub@redhat.com>
24104 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
24105 the right scopes, make sure cond_jump isn't preserved between multiple
24106 iterations. Search for fallthru edge whenever there are 3+ edges and
24107 use find_fallthru_edge for it.
24109 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24111 Patch by Alexander Monakov <amonakov@ispras.ru>
24112 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
24113 probabilities consistently.
24115 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24117 * pa.c (pa_expand_compare_and_swap_loop): Update call of
24118 emit_cmp_and_jump_insns.
24120 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24123 * tree-inline.c (expand_call_inline): Combine profile statuses.
24125 2017-06-30 Andrew Pinski <apinski@cavium.com>
24127 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
24128 fold_stmt returned true.
24130 2017-06-30 Nathan Sidwell <nathan@acm.org>
24132 * ggc.h (empty_string): Delete.
24133 * cfgexpand.c (expand_asm_stmt): Use plain "".
24134 * optabs.c (expand_asm_memory_barrier): Likewise.
24135 * stringpool.c (empty_string): Delete.
24136 (digit_vector, digit_string): Delete.
24137 (ggc_alloc_string): Use plain "", don't optimize single digit
24138 strings. Use ggc_alloc_atomic.
24140 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
24142 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
24143 comparison set and one other set, use the cost of the non-comparison
24146 2017-06-30 Nathan Sidwell <nathan@acm.org>
24148 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
24151 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
24153 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
24154 loops. Remove now unneeded calls to gimple_switch_set_label() that
24155 just set removed labels to NULL_TREE.
24157 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
24159 * tree-ssanames.c (set_range_info_raw): Abstract from ...
24160 (set_range_info): ...here. Only call set_range_info_raw if domain
24162 (set_nonzero_bits): Call set_range_info_raw.
24163 * tree-ssanames.h (set_range_info_raw): New.
24165 2017-06-30 Jakub Jelinek <jakub@redhat.com>
24168 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
24169 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
24170 of nonimmediate_operand and <store_mask_constraint> instead of m
24171 for the input operand. For V8FI iterator, always split if input
24172 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
24173 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
24174 <store_mask_predicate> instead of register_operand and
24175 <store_mask_constraint> instead of v for the input operand. Make
24176 sure both operands aren't MEMs for if not <mask_applied>.
24178 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
24180 * lto-wrapper.c (copy_file) Close both file descriptors before
24183 2017-06-30 Martin Liska <mliska@suse.cz>
24186 * multiple_target.c (create_dispatcher_calls): Make ifunc
24187 also for function that don't have calls or are not referenced.
24189 2017-06-30 Richard Biener <rguenther@suse.de>
24191 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
24192 analyze the first scalar stmt. Move vector type computation
24193 for the BB case here from ...
24194 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
24195 live operation processing in the SLP case properly.
24197 2017-06-30 Richard Biener <rguenther@suse.de>
24199 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
24201 2017-06-30 Martin Liska <mliska@suse.cz>
24204 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
24205 before BUILT_IN_UNWIND_RESUME when ASAN is used.
24207 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
24209 * doc/invoke.texi (AArch64): Add missing options and remove redundant
24212 2017-06-30 Richard Biener <rguenther@suse.de>
24214 PR tree-optimization/81249
24215 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
24216 condition reduction result to original scalar type.
24218 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24220 * profile-count.h (enum profile_quality): Fix typos and whitespace
24223 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24225 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
24226 type for branch probabilities.
24228 2017-06-29 Julian Brown <julian@codesourcery.com>
24229 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24231 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
24232 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
24233 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
24234 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
24236 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24238 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
24239 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
24240 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
24241 CC usage from generic code to here.
24242 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
24243 CC usage into the target macros.
24245 2017-06-29 Maya Rashish <coypu@sdf.org>
24247 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
24250 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24252 * arm/arm-builtins.c: Include profile-count.h
24253 * except.c (sjlj_emit_function_enter): Use
24254 profile_probability::unlikely.
24256 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24258 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
24259 and tocrel_offset be pointer args rather than implicitly using
24261 (legitimate_constant_pool_address_p, rs6000_emit_move,
24262 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
24263 tocrel_offset and use in toc_relative_expr_p call.
24264 (print_operand, print_operand_address): Use static tocrel_base_oac
24265 and tocrel_offset_oac.
24266 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
24269 2017-06-29 Maya Rashish <coypu@sdf.org>
24271 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
24273 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
24275 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
24276 objects, take into account only the alignment of 'op0' and 'mode1' if
24279 2017-06-29 Steve Ellcey <sellcey@cavium.com>
24281 * ccmp.c (ccmp_tree_comparison_p): New function.
24282 (ccmp_candidate_p): Update to use above function.
24283 (get_compare_parts): New function.
24284 (expand_ccmp_next): Update to use new functions.
24285 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
24287 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
24288 take mode as argument.
24289 * ccmp.h (expand_ccmp_expr): Add mode as argument.
24290 * expr.c (expand_expr_real_1): Pass mode as argument.
24292 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
24294 * combine.c (combine_instructions): Print insns to dump_file, together
24297 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24299 * asan.c (asan_emit_stack_protection): Update.
24300 (create_cond_insert_point): Update.
24301 * auto-profile.c (afdo_propagate_circuit): Update.
24302 * basic-block.h (struct edge_def): Turn probability to
24303 profile_probability.
24304 (EDGE_FREQUENCY): Update.
24305 * bb-reorder.c (find_traces_1_round): Update.
24306 (better_edge_p): Update.
24307 (sanitize_hot_paths): Update.
24308 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
24309 (make_single_succ_edge): Update.
24310 (check_bb_profile): Update.
24311 (dump_edge_info): Update.
24312 (update_bb_profile_for_threading): Update.
24313 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
24314 probabilitycount to 0.
24315 * cfgbuild.c (compute_outgoing_frequencies): Update.
24316 * cfgcleanup.c (try_forward_edges): Update.
24317 (outgoing_edges_match): Update.
24318 (try_crossjump_to_edge): Update.
24319 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
24320 (expand_gimple_tailcall): Update.
24321 (construct_init_block): Use make_single_succ_edge.
24322 (construct_exit_block): Use make_single_succ_edge.
24323 * cfghooks.c (verify_flow_info): Update.
24324 (redirect_edge_succ_nodup): Update.
24325 (split_edge): Update.
24326 (account_profile_record): Update.
24327 * cfgloopanal.c (single_likely_exit): Update.
24328 * cfgloopmanip.c (scale_loop_profile): Update.
24329 (set_zero_probability): Remove.
24330 (duplicate_loop_to_header_edge): Update.
24331 * cfgloopmanip.h (loop_version): Update prototype.
24332 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
24333 (force_nonfallthru_and_redirect): Update.
24334 (update_br_prob_note): Update.
24335 (rtl_verify_edges): Update.
24336 (purge_dead_edges): Update.
24337 (rtl_lv_add_condition_to_bb): Update.
24338 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
24339 * cgraphunit.c (init_lowered_empty_function): Update.
24340 (cgraph_node::expand_thunk): Update.
24341 * cilk-common.c: Include profile-count.h
24342 * dojump.c (inv): Remove.
24343 (jumpifnot): Update.
24344 (jumpifnot_1): Update.
24345 (do_jump_1): Update.
24347 (do_jump_by_parts_greater_rtx): Update.
24348 (do_compare_rtx_and_jump): Update.
24349 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
24350 do_jump_1. do_compare_rtx_and_jump): Update prototype.
24351 * dwarf2cfi.c: Include profile-count.h
24352 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
24353 (sjlj_emit_dispatch_table): Likewise.
24354 * explow.c: Include profile-count.h
24355 * expmed.c (emit_store_flag_force): Update.
24356 (do_cmp_and_jump): Update.
24357 * expr.c (compare_by_pieces_d::generate): Update.
24358 (compare_by_pieces_d::finish_mode): Update.
24359 (emit_block_move_via_loop): Update.
24360 (store_expr_with_bounds): Update.
24361 (store_constructor): Update.
24362 (expand_expr_real_2): Update.
24363 (expand_expr_real_1): Update.
24364 * expr.h (try_casesi, try_tablejump): Update prototypes.
24365 * gimple-pretty-print.c (dump_probability): Update.
24366 (dump_profile): New.
24367 (dump_gimple_label): Update.
24368 (dump_gimple_bb_header): Update.
24369 * graph.c (draw_cfg_node_succ_edges): Update.
24370 * hsa-gen.c (convert_switch_statements): Update.
24371 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
24372 (find_if_case_1): Update.
24373 (find_if_case_2): Update.
24374 * internal-fn.c (expand_arith_overflow_result_store): Update.
24375 (expand_addsub_overflow): Update.
24376 (expand_neg_overflow): Update.
24377 (expand_mul_overflow): Update.
24378 (expand_vector_ubsan_overflow): Update.
24379 * ipa-cp.c (good_cloning_opportunity_p): Update.
24380 * ipa-split.c (split_function): Use make_single_succ_edge.
24381 * ipa-utils.c (ipa_merge_profiles): Update.
24382 * loop-doloop.c (add_test): Update.
24383 (doloop_modify): Update.
24384 * loop-unroll.c (compare_and_jump_seq): Update.
24385 (unroll_loop_runtime_iterations): Update.
24386 * lra-constraints.c (lra_inheritance): Update.
24387 * lto-streamer-in.c (input_cfg): Update.
24388 * lto-streamer-out.c (output_cfg): Update.
24389 * mcf.c (adjust_cfg_counts): Update.
24390 * modulo-sched.c (sms_schedule): Update.
24391 * omp-expand.c (expand_omp_for_init_counts): Update.
24392 (extract_omp_for_update_vars): Update.
24393 (expand_omp_ordered_sink): Update.
24394 (expand_omp_for_ordered_loops): Update.
24395 (expand_omp_for_generic): Update.
24396 (expand_omp_for_static_nochunk): Update.
24397 (expand_omp_for_static_chunk): Update.
24398 (expand_cilk_for): Update.
24399 (expand_omp_simd): Update.
24400 (expand_omp_taskloop_for_outer): Update.
24401 (expand_omp_taskloop_for_inner): Update.
24402 * omp-simd-clone.c (simd_clone_adjust): Update.
24403 * optabs.c (expand_doubleword_shift): Update.
24404 (expand_abs): Update.
24405 (emit_cmp_and_jump_insn_1): Update.
24406 (expand_compare_and_swap_loop): Update.
24407 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
24408 * predict.c (predictable_edge_p): Update.
24409 (edge_probability_reliable_p): Update.
24410 (set_even_probabilities): Update.
24411 (combine_predictions_for_insn): Update.
24412 (combine_predictions_for_bb): Update.
24413 (propagate_freq): Update.
24414 (estimate_bb_frequencies): Update.
24415 (force_edge_cold): Update.
24416 * profile-count.c (profile_count::dump): Add missing space into dump.
24417 (profile_count::debug): Add newline.
24418 (profile_count::differs_from_p): Explicitly convert to unsigned.
24419 (profile_count::stream_in): Update.
24420 (profile_probability::dump): New member function.
24421 (profile_probability::debug): New member function.
24422 (profile_probability::differs_from_p): New member function.
24423 (profile_probability::differs_lot_from_p): New member function.
24424 (profile_probability::stream_in): New member function.
24425 (profile_probability::stream_out): New member function.
24426 * profile-count.h (profile_count_quality): Rename to ...
24427 (profile_quality): ... this one.
24428 (profile_probability): New.
24429 (profile_count): Update.
24430 * profile.c (compute_branch_probabilities): Update.
24431 * recog.c (peep2_attempt): Update.
24432 * sched-ebb.c (schedule_ebbs): Update.
24433 * sched-rgn.c (find_single_block_region): Update.
24434 (compute_dom_prob_ps): Update.
24435 (schedule_region): Update.
24436 * sel-sched-ir.c (compute_succs_info): Update.
24437 * stmt.c (struct case_node): Update.
24438 (do_jump_if_equal): Update.
24439 (get_outgoing_edge_probs): Update.
24440 (conditional_probability): Update.
24441 (emit_case_dispatch_table): Update.
24442 (expand_case): Update.
24443 (expand_sjlj_dispatch_table): Update.
24444 (emit_case_nodes): Update.
24445 * targhooks.c: Update.
24446 * tracer.c (better_p): Update.
24447 (find_best_successor): Update.
24448 * trans-mem.c (expand_transaction): Update.
24449 * tree-call-cdce.c: Update.
24450 * tree-cfg.c (gimple_split_edge): Upate.
24451 (move_sese_region_to_fn): Upate.
24452 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
24453 * tree-eh.c (lower_resx): Upate.
24454 (cleanup_empty_eh_move_lp): Upate.
24455 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24456 * tree-inline.c (copy_edges_for_bb): Update.
24457 (copy_cfg_body): Update.
24458 * tree-parloops.c (gen_parallel_loop): Update.
24459 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
24460 (gimple_gen_time_profiler): Update.
24461 * tree-ssa-dce.c (remove_dead_stmt): Update.
24462 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24463 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
24464 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
24465 (unloop_loops): Update.
24466 (try_peel_loop): Update.
24467 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24468 * tree-ssa-loop-split.c (connect_loops): Update.
24469 (split_loop): Update.
24470 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24471 (hoist_guard): Update.
24472 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
24473 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
24474 (value_replacement): Update.
24475 * tree-ssa-reassoc.c (branch_fixup): Update.
24476 * tree-ssa-tail-merge.c (replace_block_by): Update.
24477 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
24478 (create_edge_and_update_destination_phis): Update.
24479 (compute_path_counts): Update.
24480 (recompute_probabilities): Update.
24481 (update_joiner_offpath_counts): Update.
24482 (freqs_to_counts_path): Update.
24483 (duplicate_thread_path): Update.
24484 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
24485 (struct switch_conv_info): Update.
24486 (gen_inbound_check): Update.
24487 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24488 (vect_do_peeling): Update.
24489 (vect_loop_versioning): Update.
24490 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24491 (optimize_mask_stores): Update.
24492 * ubsan.c (ubsan_expand_null_ifn): Update.
24493 * value-prof.c (gimple_divmod_fixed_value): Update.
24494 (gimple_divmod_fixed_value_transform): Update.
24495 (gimple_mod_pow2): Update.
24496 (gimple_mod_pow2_value_transform): Update.
24497 (gimple_mod_subtract): Update.
24498 (gimple_mod_subtract_transform): Update.
24499 (gimple_ic): Update.
24500 (gimple_stringop_fixed_value): Update.
24501 (gimple_stringops_transform): Update.
24502 * value-prof.h: Update.
24504 2017-06-29 Carl Love <cel@us.ibm.com>
24506 * config/rs6000/rs6000-c.c: Add support for built-in functions
24507 vector signed int vec_signed (vector float);
24508 vector signed long long vec_signed (vector double);
24509 vector signed int vec_signed2 (vector double, vector double);
24510 vector signed int vec_signede (vector double);
24511 vector signed int vec_signedo (vector double);
24512 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
24513 instruction generator.
24514 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24515 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
24516 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
24518 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
24519 vunsignede_v2df): Add define_expands.
24520 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
24521 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
24522 VEC_UNSIGNEDO): Add definitions.
24523 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24524 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
24525 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
24526 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
24527 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
24528 * config/rs6000/altivec.h (vec_signed, vec_signed2,
24529 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
24530 vec_unsignede, vec_unsignedo): Add builtin defines.
24531 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
24533 * doc/extend.texi: Update the built-in documentation file for the
24534 new built-in functions.
24536 2017-06-29 Richard Biener <rguenther@suse.de>
24538 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
24539 reduction chains to LOOP_VINFO_REDUCTIONS.
24540 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
24541 SLP reductions after processing reduction chains.
24543 2017-06-29 Nathan Sidwell <nathan@acm.org>
24545 * builtins.c (fold_builtin_FUNCTION): Use
24546 lang_hooks.decl_printable_name.
24548 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
24550 PR middle-end/81194
24551 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
24552 with only one label.
24553 * stmt.c (expand_case): Assert NCASES is greater than one.
24555 2017-06-29 Richard Biener <rguenther@suse.de>
24557 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
24559 (group_case_labels): Likewise.
24560 (find_taken_edge): Push sanity checking on val to workers...
24561 (find_taken_edge_cond_expr): ... here
24562 (find_taken_edge_switch_expr): ... and here, handle cases
24563 with just a default label.
24564 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
24565 (group_case_labels): Likewise.
24566 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
24567 group_case_labels does anything cleanup the CFG again.
24569 2017-06-29 Bin Cheng <bin.cheng@arm.com>
24571 PR tree-optimization/81196
24572 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
24573 exit condition comparing two IVs.
24575 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
24577 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
24578 profile to the dummy entry at the end of the list of architectures.
24579 * config/arm/arm-cpu-cdata.h: Regenerated.
24581 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24582 Michael Collison <michael.collison@arm.com>
24585 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
24587 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
24588 (*aarch64_reg_<mode>3_minus_mask): New pattern.
24589 (*aarch64_<optab>_reg_di3_mask2): New pattern.
24590 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
24591 of shift when the shift amount is masked with constant equal to
24592 the size of the mode.
24593 * config/aarch64/predicates.md (subreg_lowpart_operator): New
24596 2017-06-29 Martin Liska <mliska@suse.cz>
24598 * config/i386/i386.opt: Change range from [1,5] to [0,5].
24600 2017-06-29 Yury Gribov <tetra2005@gmail.com>
24603 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
24606 (edge_growth_cache_entry::edge_growth_cache_entry): New
24608 (reset_edge_growth_cache): Update to use constructor.
24610 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24612 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
24613 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24614 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24616 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
24618 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
24619 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
24621 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
24623 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
24624 (*-linux-uclibc*): Add t-uclibc tmake_file.
24625 * config/t-musl: New.
24626 * config/t-uclibc: New.
24628 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
24630 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
24632 (gen_comm_data): Emit architectural setting of arch_prof.
24633 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
24635 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
24636 (armv8-m.base, armv8-m.main): Likewise.
24637 * arm-protos.h (arm_build_target): Add profile field.
24638 (arch_option): Likewise.
24639 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
24641 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
24642 arm_active_target.profile.
24644 2017-06-28 Richard Biener <rguenther@suse.de>
24646 PR middle-end/81227
24647 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
24648 TYPE_OVERFLOW_WRAPS.
24649 * match.pd (negate_expr_p): Likewise.
24650 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
24651 fold_build2, not fold_binary.
24653 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24655 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
24656 Convert memory address to Pmode.
24657 (aarch64_print_operand): Assert MEM operands are always Pmode.
24659 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24662 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
24663 Remove redundant if.
24664 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
24665 * config/arm/aarch-common-protos.h
24666 (aarch_forward_to_shift_is_not_shifted_re): Remove.
24667 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
24669 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
24672 * multiple_target.c (create_dispatcher_calls): Set the default
24673 clone to be static, not public.
24675 2017-06-28 Richard Biener <rguenther@suse.de>
24677 * tree-vect-loop.c (vectorizable_reduction): Move special
24678 cond reduction IV var creation ...
24679 (vect_create_epilog_for_reduction): ... here. Remove induction_index
24680 parameter. Use STMT_VINFO_VECTYPE.
24681 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
24684 2017-06-28 Martin Liska <mliska@suse.cz>
24687 * ipa-visibility.c (non_local_p): Handle visibility.
24689 2017-06-28 Martin Liska <mliska@suse.cz>
24692 * common.opt: Add IntegerRange to various options.
24693 * opt-functions.awk (integer_range_info): New function.
24694 * optc-gen.awk: Add integer_range_info to cl_options struct.
24695 * opts-common.c (decode_cmdline_option): Handle
24696 CL_ERR_INT_RANGE_ARG.
24697 (cmdline_handle_error): Likewise.
24698 * opts.c (print_filtered_help): Show valid interval in
24699 when --help is provided.
24700 * opts.h (struct cl_option): Add range_min and range_max fields.
24701 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
24703 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
24705 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
24706 (x * C EQ/NE y * C): New transformation.
24708 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
24710 * genmultilib (combination_space): Accept '+' in option names.
24712 2017-06-28 Martin Liska <mliska@suse.cz>
24715 * asan.c (instrument_derefs): Bail out inner references
24716 that are hard register variables.
24718 2017-06-28 Jakub Jelinek <jakub@redhat.com>
24721 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
24722 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
24724 2017-06-28 Richard Biener <rguenther@suse.de>
24726 * tree-vectorizer.h (vect_get_vec_defs): Remove.
24727 (vect_get_slp_defs): Adjust.
24728 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
24730 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
24732 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
24733 get_initial_defs_for_reduction instead of vect_get_vec_defs.
24734 (vectorizable_reduction): Adjust.
24735 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
24737 (vect_get_slp_defs): Likewise.
24738 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
24739 (vectorizable_bswap): Adjust.
24740 (vectorizable_call): Likewise.
24741 (vectorizable_conversion): Likewise.
24742 (vectorizable_assignment): Likewise.
24743 (vectorizable_shift): Likewise.
24744 (vectorizable_operation): Likewise.
24745 (vectorizable_store): Likewise.
24746 (vectorizable_condition): Likewise.
24747 (vectorizable_comparison): Likewise.
24749 2017-06-28 Michael Collison <michael.collison@arm.com>
24752 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
24754 (arm_gen_movmemqi): Removed unused variable 'i'.
24755 Convert 'for' loop into 'while' loop.
24756 (arm_expand_prologue): Remove last unnecessary set of insn.
24757 (thumb_pop): Remove unused variable 'pushed_words'.
24758 (thumb_exit): Remove last unnecessary set of regs_to_pop.
24760 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24762 * config/s390/predicates.md: Use s390_rel_address_ok_p.
24763 * config/s390/s390-protos.h: Add prototype of
24764 s390_rel_address_ok_p.
24765 * config/s390/s390.c (s390_got_symbol): New function.
24766 (s390_rel_address_ok_p): New function.
24767 (legitimize_pic_address): Use s390_rel_address_ok_p.
24768 (s390_load_got): Use s390_got_symbol.
24769 (s390_option_override): Issue error if
24770 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
24771 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
24773 * config/s390/s390.opt: New option mpic-data-is-text-relative.
24775 2017-06-27 Andrew Pinski <apinski@cavium.com>
24777 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
24778 (X * copysign (1.0, X)): New pattern.
24779 (X * copysign (1.0, -X)): New pattern.
24780 (copysign (-1.0, CST)): New pattern.
24782 2017-06-27 Joseph Myers <joseph@codesourcery.com>
24784 * genmultilib (combination_space): Remove variable.
24785 Validate reuse rules against regular expression for any sequence
24786 of multilib options in any order.
24788 2017-06-27 Michael Collison <michael.collison@arm.com>
24790 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
24791 call aarch64_split_simd_combine.
24792 * (aarch64_combine_internal<mode>): Delete pattern.
24793 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
24794 Allow register and subreg operands.
24796 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24798 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
24799 specific need, just fallback on defaults.
24800 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
24802 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24803 Olivier Hainque <hainque@adacore.com>
24805 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
24807 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
24808 targets. Pick a default if no particular attempt applied.
24809 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
24812 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24814 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
24815 (x86_64-wrs-vxworks7): Likewise.
24817 2017-06-27 Marek Polacek <polacek@redhat.com>
24820 * ubsan.c (instrument_null): Check get_base_address's result for null.
24822 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24824 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
24826 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24828 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
24829 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
24830 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
24831 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
24832 New function types.
24833 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
24834 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
24835 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
24836 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
24837 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
24838 BUILT_IN_FEUPDATEENV): New builtins.
24839 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
24840 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
24841 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
24842 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
24844 (builtin_structptr_types): Adjust size.
24845 * tree.c (builtin_structptr_types): Add four entries.
24847 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24848 Olivier Hainque <hainque@adacore.com>
24850 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
24851 (TLS_SYM): New local macro, forcing reference to __tls__ on
24852 link command lines for VxWorks 7 RTPs, triggering initialization
24854 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
24855 OS features TLS support, true for RTPs on VxWorks 7.
24856 * config/vxworks.c (vxworks_override_options): Setup emutls
24859 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24861 * predict.c (test_prediction_value_range): Use -1U instead of -1
24862 to avoid narrowing conversion warning.
24863 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
24864 to avoid narrowing conversion warning.
24865 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
24867 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
24869 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24871 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
24872 64bit configurations.
24873 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
24874 (SIZE_TYPE): Likewise.
24875 * config/vxworks.c (vxworks_emutls_var_fields): Use
24876 long_unsigned_type_node instead of unsigned_type_node as the offset
24877 field type, which is "pointer" mode in emutls.c.
24879 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24882 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
24884 PR middle-end/81207
24885 * gimple-fold.c (replace_call_with_call_and_fold): Handle
24886 gimple_vuse copying separately from gimple_vdef copying.
24888 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24890 * value-prof.c (free_hist): Remove call to memset and the enclosing if
24893 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24894 Olivier Hainque <hainque@adacore.com>
24896 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
24897 for all vxworks7 targets.
24898 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
24899 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
24900 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
24901 variations for VX6/VX7 and 32/64bits later on in ...
24902 (VXWORKS_LIB_SPEC): Leverage new macros.
24903 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
24904 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
24906 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24908 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
24909 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
24911 2017-06-26 Carl Love <cel@us.ibm.com>
24913 * config/rs6000/rs6000-c.c: Add support for built-in functions
24914 vector bool char vec_reve (vector bool char);
24915 vector signed char vec_reve (vector signed char);
24916 vector unsigned char vec_reve (vector unsigned char);
24917 vector bool int vec_reve (vector bool int);
24918 vector signed int vec_reve (vector signed int);
24919 vector unsigned int vec_reve (vector unsigned int);
24920 vector bool long long vec_reve (vector bool long long);
24921 vector signed long long vec_reve (vector signed long long);
24922 vector unsigned long long vec_reve (vector unsigned long long);
24923 vector bool short vec_reve (vector bool short);
24924 vector signed short vec_reve (vector signed short);
24925 vector double vec_reve (vector double);
24926 vector float vec_reve (vector float);
24927 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
24928 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
24929 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
24930 (altivec_vreve): New pattern.
24931 * config/rs6000/altivec.h (vec_reve): New define.
24932 * doc/extend.texi (vec_rev): Update the built-in documentation file
24933 for the new built-in functions.
24935 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24937 PR tree-optimization/71815
24938 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
24940 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
24942 (slsr_process_phi): Likewise.
24943 (replace_uncond_cands_and_profitable_phis): Don't replace a
24944 multiply candidate with a stride of 1 (copy or cast).
24945 (phi_incr_cost): Call uses_consumed_by_stmt rather than
24947 (lowest_cost_path): Likewise.
24948 (total_savings): Likewise.
24950 2017-06-26 Richard Biener <rguenther@suse.de>
24953 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
24954 Use def_builtin_pure for all gather builtins.
24956 2017-06-26 Richard Biener <rguenther@suse.de>
24958 PR tree-optimization/81203
24959 * tree-tailcall.c (find_tail_calls): Do not move stmts into
24960 non-dominating BBs.
24962 2017-06-26 Marek Polacek <polacek@redhat.com>
24965 * doc/invoke.texi: Document -Wmultistatement-macros.
24967 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
24969 * doc/sourcebuild.texi (ARM-specific attributes): Document new
24970 arm_neon_ok_no_float_abi effective target.
24972 2017-06-26 Richard Biener <rguenther@suse.de>
24974 PR tree-optimization/80928
24975 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
24976 (copy_bbs): Set BB_DUPLICATED flag early.
24977 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
24979 (execute_on_shrinking_pred): Likewise.
24980 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
24981 BB_DUPLICATED blocks.
24982 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
24983 iterate over all PHIs considering removal of *gsi.
24985 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
24987 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
24990 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24992 * config/rs6000/rs6000-string.c: (expand_block_clear,
24993 do_load_for_compare, select_block_compare_mode,
24994 compute_current_alignment, expand_block_compare,
24995 expand_strncmp_align_check, expand_strn_compare,
24996 expand_block_move, rs6000_output_load_multiple)
24997 Move functions related to string/block move/compare
24998 to a separate file.
24999 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25000 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25001 for this function which is now used in two files.
25002 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25003 * config.gcc: Add rs6000-string.o to extra_objs for
25004 targets powerpc*-*-* and rs6000*-*-*.
25006 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25009 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25010 32-bit, since indexed is not valid for DImode.
25011 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25012 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25013 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25014 (define_peephole2 for Altivec d-form store): Likewise.
25017 * multiple_target.c (create_dispatcher_calls): Only create the
25018 dispatcher call if the function is the default clone of a
25019 versioned function.
25021 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25023 PR middle-end/80902
25024 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25025 a call, force the call to not be a tail call.
25027 2017-06-23 Jeff Law <law@redhat.com>
25029 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25032 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25034 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25035 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25036 handling for early expansion of vector shifts (sl,sr,sra,rl).
25037 (builtin_function_type): Add vector shift right instructions
25038 to the unsigned argument list.
25040 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
25042 rtl-optimizatoin/79286
25043 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
25044 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
25045 trap. PIC register plus a const unspec without offset can never trap.
25047 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
25049 * tree.h (builtin_structptr_type): New type.
25050 (builtin_structptr_types): Declare new array.
25051 * tree.c (builtin_structptr_types): New array.
25052 (free_lang_data, build_common_tree_nodes): Use it.
25054 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
25057 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
25060 2017-06-22 Matt Turner <mattst88@gmail.com>
25062 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
25063 Lake models to skylake case. Assume skylake for unknown
25064 models with clflushopt.
25066 2017-06-22 Jeff Law <law@redhat.com>
25068 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
25069 frame sizes that do not satisfy aarch64_uimm12_shift.
25071 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
25073 * profile-count.h (apply_probability,
25074 apply_scale, probability_in): Fix checks for zero.
25076 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25078 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
25079 * doc/cppdiropts.texi (-I @var{dir}): Document it.
25081 2016-06-22 Richard Biener <rguenther@suse.de>
25083 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25084 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
25085 REDUC_MAX_EXPR support.
25086 (vectorizable_reduction): Likewise.
25087 (vect_create_epilog_for_reduction): Likewise.
25089 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
25091 * match.pd (A / (1 << B) -> A >> B): New.
25092 * generic-match-head.c: Include optabs-tree.h.
25093 * gimple-match-head.c: Likewise.
25094 * optabs-tree.h (target_supports_op_p): New.
25095 * optabs-tree.c (target_supports_op_p): New.
25097 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25099 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
25100 $gcc_cv_ld --help output.
25101 (gcc_cv_ld_demangle): Likewise.
25102 (gcc_cv_ld_eh_frame_hdr): Likewise.
25103 (gcc_cv_ld_pie): Likewise.
25104 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
25105 (gcc_cv_ld_buildid): Likewise.
25106 (gcc_cv_ld_sysroot): Likewise.
25107 (ld_bndplt_support): Likewise.
25108 (ld_pushpopstate_support): Likewise.
25109 * configure: Regenerate.
25110 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
25112 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25115 * config/i386/sse.md (round<mode>2): Renumber match_dup and
25116 operands indexes to avoid gap between operands and match_dups.
25118 2017-06-21 Andrew Pinski <apinski@cavium.com>
25120 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
25121 Increment Arith_shift and Arith_shift_reg by 1.
25122 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
25124 * config/aarch64/aarch64.c (thunderx_tunings): Enable
25125 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
25126 (aarch64_strip_extend): Add new argument and test for it.
25127 (aarch64_cheap_mult_shift_p): New function.
25128 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
25129 add a cost if it is true.
25130 Update calls to aarch64_strip_extend.
25131 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
25133 2017-06-21 Andrew Pinski <apinski@cavium.com>
25135 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
25137 (thunderxt88): Likewise.
25138 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
25139 (thunderx_prefetch_tune): New variable.
25140 (thunderx2t99_prefetch_tune): Update for the correct values.
25141 (thunderxt88_tunings): New variable.
25142 (thunderx_tunings): Use thunderx_prefetch_tune instead of
25143 generic_prefetch_tune.
25144 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
25146 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25148 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
25149 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
25150 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
25151 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
25152 (aarch64_atomic_cas<mode>, GPI): Likewise.
25154 2017-06-21 Martin Liska <mliska@suse.cz>
25156 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
25157 statements on cold and hot labels.
25158 * predict.c (tree_estimate_probability_bb): Remove the
25159 prediction from this place.
25161 2017-06-21 Martin Liska <mliska@suse.cz>
25163 PR tree-optimization/79489
25164 * gimplify.c (maybe_add_early_return_predict_stmt): New
25166 (gimplify_return_expr): Call the function.
25167 * predict.c (tree_estimate_probability_bb): Remove handling
25169 * predict.def: Update comment about early return predictor.
25170 * gimple-predict.h (is_gimple_predict): New function.
25171 * predict.def: Change default value of early return to 66.
25172 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
25174 * passes.def: Put pass_strip_predict_hints to the beginning of
25177 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
25179 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
25180 FUNCTION_DECL declarations.
25181 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
25183 (dwaf2out_decl): Likewise.
25184 * godump.c (go_early_global_decl): Skip call to the real debug hook
25185 for FUNCTION_DECL declarations.
25186 * passes.c (rest_of_decl_compilation): Skip call to the
25187 early_global_decl debug hook for FUNCTION_DECL declarations, unless
25188 -fdump-go-spec is passed.
25190 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25192 * config/i386/i386.c (struct builtin_isa): New field pure_p.
25193 Reorder for compactness.
25194 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
25195 (def_builtin_pure, def_builtin_pure2): New functions.
25196 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
25198 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25200 * match.pd (nop_convert): New predicate.
25201 ((A +- CST1) +- CST2): Allow some NOP conversions.
25203 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25206 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
25207 with ctors/dtors if GOVD_SHARED is set.
25209 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25211 * config/aarch64/aarch64.md (movti_aarch64):
25212 Emit mov rather than orr.
25213 (movtf_aarch64): Likewise.
25214 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
25215 Emit mov rather than orr.
25217 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25219 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
25220 Swap alternatives, make integer dup more expensive.
25222 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25224 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25225 Return true for non-tls symbols.
25227 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
25229 * config/aarch64/aarch64-cores.def (cortex-a55): New.
25230 (cortex-a75): Likewise.
25231 (cortex-a75.cortex-a55): Likewise.
25232 * config/aarch64/aarch64-tune.md: Regenerate.
25233 * doc/invoke.texi (-mtune): Document new values for -mtune.
25235 2017-06-21 Tom de Vries <tom@codesourcery.com>
25237 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
25238 stack_size feature.
25239 (Effective-Target Keywords, Other attributes): Suggest using
25240 dg-add-options stack_size feature to get stack limit in stack_size
25241 effective target documentation.
25243 2017-06-21 Julian Brown <julian@codesourcery.com>
25244 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25246 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
25247 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
25248 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
25250 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
25251 attribute type list for neon_multiply.
25252 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
25253 attribute type list for neon_multiply.
25254 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
25255 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
25256 attribute type list for neon_multiply.
25257 * config/arm/types.md (crypto_pmull): Add.
25258 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
25259 attribute type list.
25261 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
25263 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
25266 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25268 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
25269 to make sure not to dereference a NULL cost_classes_ptr pointer.
25271 2017-06-20 Carl Love <cel@us.ibm.com>
25273 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25274 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
25275 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
25276 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
25277 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
25278 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
25279 VMULOSW): New enum "unspec" values.
25280 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
25281 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
25282 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
25283 altivec_vmulosw): New patterns.
25284 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
25285 VMULOSW): Add definitions.
25287 2017-06-20 Julia Koval <julia.koval@intel.com>
25289 * config/i386/i386.c: Fix rounding expand for new pattern.
25290 * config/i386/subst.md: Fix pattern (parallel -> unspec).
25292 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25294 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
25295 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
25297 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25299 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
25302 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25304 * config/aarch64/aarch64-cores.def: Rearrange to sort by
25305 architecture, then by implementer ID.
25306 * config/aarch64/aarch64-tune.md: Regenerate.
25308 2017-06-20 Richard Biener <rguenther@suse.de>
25310 PR middle-end/81097
25311 * fold-const.c (split_tree): Fold to type before negating.
25313 2017-06-20 David Malcolm <dmalcolm@redhat.com>
25315 * diagnostic-show-locus.c
25316 (selftest::test_fixit_deletion_affecting_newline): New function.
25317 (selftest::diagnostic_show_locus_c_tests): Call it.
25319 2017-06-20 Andreas Schwab <schwab@suse.de>
25322 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
25325 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25327 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
25328 __ARM_FEATURE_COPROC according to support.
25330 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25332 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25333 Rewritten to avoid overflow for > 32-bit pointers.
25336 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
25337 by removing enum keyword.
25338 (ubsan_type_descriptor): Likewise. Formatting fix.
25341 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
25342 splitter): Require TARGET_SSE2 in the condition.
25344 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
25347 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
25348 for doing vector set of SFmode on ISA 3.0.
25349 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
25350 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
25352 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
25353 SFmode value into a V4SF variable that was extracted from another
25354 V4SF variable without converting the element to double precision
25355 and back to single precision vector format.
25356 (vsx_insert_extract_v4sf_p9_2): Likewise.
25358 2017-06-19 Jakub Jelinek <jakub@redhat.com>
25360 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
25361 in UWHI to avoid undefined overflow.
25364 * ubsan.h (enum ubsan_encode_value_phase): New.
25365 (ubsan_encode_value): Change second argument to
25366 enum ubsan_encode_value_phase with default value of
25367 UBSAN_ENCODE_VALUE_GENERIC.
25368 * ubsan.c (ubsan_encode_value): Change second argument to
25369 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
25370 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
25371 create_tmp_var_raw instead of create_tmp_var and use a
25373 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
25374 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
25375 ubsan_encode_value callers.
25378 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
25379 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
25380 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
25382 2017-06-19 Richard Biener <rguenther@suse.de>
25384 PR middle-end/81118
25385 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
25386 estimates if we changed anything.
25388 2017-06-19 Richard Biener <rguenther@suse.de>
25390 PR tree-optimization/80887
25391 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
25392 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
25393 simplified lookups, then reset mprts_hook.
25394 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
25396 (try_to_simplify): Likewise.
25398 2017-06-19 Martin Liska <mliska@suse.cz>
25401 * gimplify.c (gimplify_switch_expr):
25402 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
25404 2017-06-19 Martin Liska <mliska@suse.cz>
25406 * doc/install.texi: Document that PGO runs in 4 stages.
25408 2017-06-19 Martin Liska <mliska@suse.cz>
25411 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
25412 to dispatcher function name.
25413 * multiple_target.c (replace_function_decl): New function.
25414 (create_dispatcher_calls): Redirect both edges and references.
25416 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
25418 * profile-count.c (profile_count::dump): Dump quality.
25419 (profile_count::differs_from_p): Update for unsigned val.
25420 * profile-count.h (profile_count_quality): New enum.
25421 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
25423 2017-06-19 Richard Biener <rguenther@suse.de>
25425 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
25426 struct function as arg.
25427 (estimate_numbers_of_iterations): Export overload with loop arg.
25428 (free_numbers_of_iterations_estimates_loop): Use an overload of
25429 free_numbers_of_iterations_estimates instead.
25430 * tree-cfg.c (remove_bb): Adjust.
25431 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
25432 * tree-parloops.c (gen_parallel_loop): Likewise.
25433 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
25435 (tree_unroll_loops_completely): Likewise.
25436 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
25437 Use an overload instead and export.
25438 (estimated_loop_iterations): Adjust.
25439 (max_loop_iterations): Likewise.
25440 (likely_max_loop_iterations): Likewise.
25441 (estimate_numbers_of_iterations): Take struct function as arg
25443 (loop_exits_before_overflow): Adjust.
25444 (free_numbers_of_iterations_estimates_loop): Use an overload.
25445 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
25446 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
25448 2017-06-19 Richard Biener <rguenther@suse.de>
25451 * ipa-prop.c (find_constructor_constant_at_offset): Handle
25452 RANGE_EXPR conservatively.
25454 2017-06-16 Carl Love <cel@us.ibm.com>
25456 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25457 definitions for vec_float, vec_float2, vec_floato,
25458 vec_floate built-ins.
25459 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
25460 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
25462 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
25463 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
25464 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
25465 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
25466 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
25467 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
25468 vec_floato): Add builtin defines.
25469 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
25470 Update the built-in documentation file for the new built-in
25473 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25475 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
25476 (mthumb): Mark as the negative of -marm.
25478 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25480 * doc/invoke.texi (ARM Options, -mcpu): Document supported
25482 (ARM Options, -mtune): Document that this accepts the same
25483 extension options as -mcpu.
25484 (ARM Options, -mfpu): Document addition of -mfpu=auto.
25486 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25488 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
25489 permitted extensions.
25491 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25493 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
25494 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
25495 (armv8-m.main): Add option +nodsp.
25496 * config/arm/arm-cpu-cdata.h: Regenerated.
25498 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25500 * config/arm/t-fuchsia: New file.
25501 * config.gcc (arm*-*-fuchsia*): Use it.
25503 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25505 * config/arm/t-symbian: Rewrite for new option infrastructure.
25507 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25509 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
25510 (MULTILIB_REQUIRED): Likewise.
25512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25514 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
25515 (MULTILIB_RESUE): Likewise.
25516 (MULTILIB_MATCHES): Likewise.
25517 (MULTLIB_REQUIRED): Likewise.
25519 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25521 * config/arm/t-rtems: Rewrite for new option framework.
25523 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25525 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
25526 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
25527 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
25528 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
25529 * config/arm/t-multilib: ... here.
25530 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
25531 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
25532 armv7-a and armv8*-a when A-profile libraries have not been built.
25533 * config/arm/t-rmprofile: Rewrite.
25535 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25537 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
25538 with a backslash. Remove the backslash after substituting unescaped
25540 * doc/fragments.texi (MULTILIB_REUSE): Document it.
25542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25544 * config.gcc: (arm*-*-*): When building a-profile libraries, force
25545 the driver to pass through the default setting of -mfloat-abi.
25546 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
25548 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
25549 (all_feat_combs): New rule.
25550 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
25552 * config/arm/t-aprofile: Rewrite.
25554 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25556 * config/arm/arm.h (FPUTYPE_AUTO): Define.
25557 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
25558 fpu is not specified by the user/command-line.
25559 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
25560 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
25561 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
25562 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
25563 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
25564 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
25566 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25568 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
25569 * config/arm/t-arm-elf: Rewritten.
25571 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25573 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
25574 have some floating-point instructions.
25575 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
25576 (TARGET_MAYBE_HARD_FLOAT): New macro.
25577 * config/arm/arm-builtins.c (arm_init_builtins): Use
25578 TARGET_MAYBE_HARD_FLOAT.
25579 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
25581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25583 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
25584 (configargs.h): Include it.
25585 (arm_print_hint_for_fpu_option): New function.
25586 (arm_parse_fpu_option): New function.
25587 (candidate_extension): New class.
25588 (arm_canon_for_multilib): New function.
25589 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
25590 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
25591 (ARCH_CANONICAL_SPECS): New macro.
25592 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
25594 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25596 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
25597 are set after handling multilib fragments. Set target_cpu_default2
25600 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25602 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
25604 (arm*-*-*): Set target_cpu_default2 to a quoted string.
25605 * config/arm/parsecpu.awk (check_cpu): Validate any extension
25607 (check_arch): Likewise.
25608 * config/arm/arm.c (arm_configure_build_target): Handle
25609 TARGET_CPU_DEFAULT being a string constant. Scan any feature
25610 options in the default.
25612 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25614 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
25615 when an option is an alias of another.
25616 * config/arm/parsecpu.awk (optalias): New parser token.
25617 (gen_comm_data): Mark non-alias options as such. Emit entries
25618 for extension aliases.
25619 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
25620 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
25621 (armv6kz, armv6zk, armv6t2): Likewise.
25622 (armv7): Make vfpv3-d16 an alias.
25623 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
25625 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
25626 Sort in canonical order.
25627 (armv8-a): Sort in canonical order.
25628 (armv8.1-a, armv8.2-a): Likewise.
25629 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
25631 (cortex-a9): Sort in canonical order.
25632 * config/arm/arm.c (selftests.h): Include it.
25633 (arm_test_cpu_arch_data): New function.
25634 (arm_run_self_tests): New function.
25635 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
25636 (targetm): Move declaration to the end of the file.
25637 * arm-cpu-cdata.h: Regenerated.
25639 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25641 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
25642 call to target_mode_check describing the type of option passed.
25643 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
25644 (arm_target_thumb_only): Use arm_parse_arch_option_name or
25645 arm_parse_cpu_option_name to match parameters against list of
25647 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
25648 arm_arch_core_flags data structure.
25649 * config/arm/arm-cpu_cdata.h: Regenerated.
25651 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25653 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
25655 (arm_print_hint_for_cpu_option): Likewise.
25656 (arm_print_hint_for_arch_option): Likewise.
25657 (arm_parse_cpu_option_name): Likewise.
25658 (arm_parse_arch_option_name): Likewise.
25659 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
25660 of entries in the all_fpus list.
25661 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
25662 (arm_parse_cpu_option_name): Declare.
25663 (arm_parse_arch_option_name): Declare.
25664 (arm_parse_option_features): Declare.
25665 (arm_intialize_isa): Declare.
25666 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
25668 (gen_comm_data): ... here. Make definitions non-static.
25669 * config/arm/arm-cpu-data.h: Regenerated.
25670 * config/arm/arm-cpu-cdata.h: Regenerated.
25672 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25674 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
25675 (cpu_arch_extension): New structure.
25676 (cpu_arch_option, arch_option, cpu_option): New structures.
25677 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
25678 architecture types.
25679 (gen_data): Generate new format data tables.
25680 * config/arm/arm.c (cpu_tune): New structure.
25681 (cpu_option, processors): Delete.
25682 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
25683 (arm_print_hint_for_cpu_option): ... this and ...
25684 (arm_print_hint_for_arch_option): ... this.
25685 (arm_parse_arch_cpu_name): Delete. Replace with ...
25686 (arm_parse_cpu_option_name): ... this and ...
25687 (arm_parse_arch_option_name): ... this.
25688 (arm_unrecognized_feature): Change type of target parameter to
25690 (arm_parse_arch_cpu_features): Delete. Replace with ...
25691 (arm_parse_option_features): ... this.
25692 (arm_configure_build_target): Rework to use new configuration data
25694 (arm_print_tune_info): Rework for new configuration data tables.
25695 * config/arm/arm-cpu-data.h: Regenerated.
25696 * config/arm/arm-cpu.h: Regenerated.
25698 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25700 * Makefile.in (OBJS): Move sbitmap.o from here ...
25701 (OBJS-libcommon): ... to here.
25703 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25705 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
25706 (ISA_ALL_CRYPTO): New macro.
25707 (ISA_ALL_SIMD): New macro
25708 (ISA_ALL_FP): New macro.
25709 * config/arm/arm.c (fpu_bitlist): Update initializer.
25710 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
25712 (arm9e): Add fpu. Add option for nofp
25713 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
25714 (arm926ej-s, arm1026ej-s): Likewise.
25715 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
25716 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
25717 neon-fp16, neon-vfpv4, nofp and nosimd.
25718 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
25719 (cortex-a8): Add fpu. Add option for nofp.
25720 (cortex-a9): Add fpu. Add options for nosimd and nofp.
25721 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
25722 (cortex-r4f): Add fpu.
25723 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
25724 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
25726 (cortex-r8): Likewise.
25727 (cortex-m4): Add fpu. Add option for nofp.
25728 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
25729 (cortex-a17.cortex-a7): Likewise.
25730 (cortex-a32): Add fpu. Add options for crypto and nofp.
25731 (cortex-a35, cortex-a53): Likewise.
25732 (cortex-a57): Add fpu. Add option for crypto.
25733 (cortex-a72, cortex-a73): Likewise.
25734 (exynos-m1): Likewise.
25735 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
25736 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
25737 (cortex-m33): Add fpu. Add option for nofp.
25738 * config/arm/arm-cpu-cdata.h: Regenerated
25739 * config/arm/arm-cpu-data.h: Regenerated.
25741 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25743 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
25744 (armv5te, armv5tej): Likewise.
25745 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
25746 (armv7): Add options fp and vfpv3-d16.
25747 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
25748 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
25750 (armv7ve): Likewise.
25751 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
25752 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
25753 (armv8-a): Add nocrypto option.
25754 (armv8.1-a, armv8.2-a): Likewise.
25755 (armv8-m.main): add options fp, fp.dp and nofp.
25757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25759 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
25761 (armv8-a+crc): Delete.
25762 (armv8.1-a): Add options simd, crypto and nofp.
25763 (armv8.2-a): Add options fp16, simd, crypto and nofp.
25764 (armv8.2-a+fp16): Delete.
25765 (armv8-m.main): Add option dsp.
25766 (armv8-m.main+dsp): Delete.
25767 (cortex-a8): Add fpu. Add nofp option.
25768 (cortex-a9): Add fpu. Add nofp and nosimd options.
25769 * config/arm/parsecpu.awk (gen_data): Generate option tables and
25770 link to main cpu and architecture data structures.
25771 (gen_comm_data): Only put isa attributes from the main architecture
25773 (option): New statement for architecture and CPU entries.
25774 * arm.c (struct cpu_option): New structure.
25775 (struct processors): Add entry for options.
25776 (arm_unrecognized_feature): New function.
25777 (arm_parse_arch_cpu_name): Ignore any characters after the first
25779 (arm_parse_arch_cpu_feature): New function.
25780 (arm_configure_build_target): Separate out any CPU and architecture
25781 features and parse separately. Don't error out if -mfpu=auto is
25782 used with only an architecture string.
25783 (arm_print_asm_arch_directives): New function.
25784 (arm_file_start): Call it.
25785 * config/arm/arm-cpu-cdata.h: Regenerated.
25786 * config/arm/arm-cpu-data.h: Likewise.
25787 * config/arm/arm-tables.opt: Likewise.
25789 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25791 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
25792 assembler when it is not -mfpu=auto.
25794 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25796 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
25797 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
25798 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
25799 (ASM_CPU_SPEC): Rewrite.
25800 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
25801 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
25802 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
25803 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
25804 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
25805 copied string is NUL-terminated. Also strip any characters prefixed
25807 (arm_rewrite_selected_arch): New function.
25808 (arm_rewrite_march): New function.
25810 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25812 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
25813 (x_arm_cpu_string, x_arm_tune_string): Likewise.
25814 (march, mcpu, mtune): Convert to string-based options.
25815 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
25816 (arm_parse_arch_cpu_name): New function.
25817 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
25818 identify selected architecture or CPU.
25819 (arm_option_save): New function.
25820 (TARGET_OPTION_SAVE): Redefine.
25821 (arm_option_restore): Restore string options.
25822 (arm_option_print): Print string options.
25824 2017-06-16 Martin Sebor <msebor@redhat.com>
25826 PR tree-optimization/80933
25827 PR tree-optimization/80934
25828 * builtins.c (fold_builtin_3): Do not handle bcmp here.
25829 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
25830 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
25831 (gimple_fold_builtin): Call them.
25833 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25835 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
25836 as unlikely; update profile.
25838 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25840 * predict.c (force_edge_cold): Handle declaring edges impossible
25843 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25845 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
25847 (try_unroll_loop_completely): Fix reporting.
25849 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25851 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
25853 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
25856 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
25857 if given a non-constant argument for an intrinsic which requires a
25860 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25862 * profile.c (compare_freqs): New function.
25863 (branch_prob): Sort edge list.
25864 (find_spanning_tree): Assume that the list is priority sorted.
25866 2017-06-16 Richard Biener <rguenther@suse.de>
25868 PR tree-optimization/81090
25869 * passes.def (pass_record_bounds): Remove.
25870 * tree-pass.h (make_pass_record_bounds): Likewise.
25871 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
25872 make_pass_record_bounds): Likewise.
25873 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
25874 not free niter estimates at the beginning but at the end.
25875 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
25877 2017-06-16 Richard Biener <rguenther@suse.de>
25879 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
25880 initializer to workaround ICE in host GCC 4.8.
25882 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25884 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
25886 (clone_inlined_nodes): Update.
25888 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25890 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
25891 prefetch settings, and enable prefetching by default at -O3.
25893 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25895 * config/aarch64/aarch64.c (aarch64_override_options_internal):
25896 Set flag_prefetch_loop_arrays according to tuning data.
25898 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25900 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
25901 New tune structure.
25902 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
25903 [Unrelated to main purpose of the patch] Place the pointer field last
25904 to enable type checking errors when tune structure are wrongly merged.
25905 * config/aarch64/aarch64.c (generic_prefetch_tune,)
25906 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
25907 (thunderx2t99_prefetch_tune): New tune constants.
25908 (tune_params *_tunings): Update all tunings (no functional change).
25909 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
25910 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
25911 from tunings structures.
25913 2017-06-16 Jakub Jelinek <jakub@redhat.com>
25916 * ubsan.c (instrument_null): Add T argument, use it instead
25917 of computing it based on IS_LHS.
25918 (instrument_object_size): Likewise.
25919 (pass_ubsan::execute): Adjust instrument_null and
25920 instrument_object_size callers to pass gimple_get_lhs or
25921 gimple_assign_rhs1 result to it. Use instrument_null instead of
25922 calling get_base_address and instrument_mem_ref. Handle
25923 aggregate call arguments for object-size sanitization.
25925 2017-06-16 Yury Gribov <tetra2005@gmail.com>
25927 PR tree-optimization/81089
25928 * tree-vrp.c (is_masked_range_test): Validate operands of
25931 2017-06-15 Martin Sebor <msebor@redhat.com>
25934 * dumpfile.c (dump_register): Avoid calling memset to initialize
25935 a class with a default ctor.
25936 * gcc.c (struct compiler): Remove const qualification.
25937 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
25938 * hash-table.h: Ditto.
25939 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
25941 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
25942 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
25944 * params.h (struct param_info): Make struct members non-const.
25945 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
25946 with default initialization.
25947 * vec.h (vec_copy_construct, vec_default_construct): New helper
25949 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
25950 with vec_copy_construct.
25951 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
25952 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
25953 * doc/invoke.texi (-Wclass-memaccess): Document.
25955 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25957 * emit-rtl.h (is_leaf): Update comment about local
25958 register allocator.
25960 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25963 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
25964 for a variable to have a section before checking if the section has a
25966 Set section to.persistent if persistent attribute is set.
25967 Warn if .persistent attribute is used on an automatic variable.
25969 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
25971 PR rtl-optimization/80474
25972 * reorg.c (update_block): Do not ignore instructions in a delay slot.
25974 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
25976 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
25979 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
25981 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
25982 (casesi): Emit bounds checking as RTL.
25983 (casesi_internal_mips16_<mode>): Remove bounds checking.
25985 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
25987 * config/xtensa/xtensa.c (xtensa_option_override): Append
25988 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
25989 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
25990 xtensa_doloop_hooks): Define unconditionally.
25991 (xtensa_reorg_loops): Only call reorg_loops in the presence of
25993 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
25994 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
25995 for it in xtensa_option_override.
25996 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
25997 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
25999 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26001 * doc/cppopts.texi: Document '-' special value to -MF.
26003 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26005 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26006 (cortex_a53_fconst): Likewise.
26007 (cortex_a53_fpmul): Likewise.
26008 (cortex_a53_f_load_64): Likewise.
26009 (cortex_a53_f_load_many): Likewise.
26010 (cortex_a53_advsimd_alu): Likewise.
26011 (cortex_a53_advsimd_alu_q): Likewise.
26012 (cortex_a53_advsimd_mul): Likewise.
26013 (cortex_a53_advsimd_mul_q): Likewise.
26014 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26015 Add missing fmul, r2f_cvt and fconst cases.
26017 2017-06-14 Richard Biener <rguenther@suse.de>
26019 PR middle-end/81088
26020 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26022 (fold_binary_loc): When associating do not treat pre-existing
26023 TREE_OVERFLOW on literal constants as a reason to allow
26024 TREE_OVERFLOW on associated literal constants.
26026 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26028 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26029 (MASK_FEATURES): New macro.
26030 * config/sparc/sparc.c (sparc_option_override): Remove the special
26031 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26033 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26035 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26036 a division of 0 if non-call exceptions are enabled.
26038 2017-06-14 Andrew Pinski <apinski@cavium.com>
26039 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26042 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
26043 Improve vector initialization code gen for only variable case.
26045 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26047 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
26049 2017-06-14 Richard Biener <rguenther@suse.de>
26051 PR tree-optimization/81083
26052 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
26055 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26057 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
26058 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
26059 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
26060 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
26061 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
26062 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
26064 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26066 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
26067 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
26069 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26071 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
26073 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26075 * config/rs6000/t-rtems: Don't handle SPE.
26077 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26079 * config/rs6000/t-linux: Don't handle SPE.
26081 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26083 * config/rs6000/eabispe.h: Delete file.
26085 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26087 * config/rs6000/t-spe: Delete file.
26089 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26091 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
26092 (rs6000_legitimate_offset_address_p): Return false for anything in
26093 V2SImode or V2SFmode.
26095 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26097 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
26098 except V2SF and V2SI. Rearrange the vector modes, and add comments.
26099 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
26101 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
26102 (rs6000_legitimate_offset_address_p): Ditto.
26103 (rs6000_emit_move): Ditto.
26104 (rs6000_init_builtins): Remove V4HI_type_node.
26106 2017-06-13 Martin Liska <mliska@suse.cz>
26109 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
26110 (gate_asan): Likewise.
26111 * asan.h (asan_no_sanitize_address_p): Remove the function.
26112 (sanitize_flags_p): New function.
26113 * builtins.def: Fix coding style.
26114 * common.opt: Use renamed enum value.
26115 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
26116 * doc/extend.texi: Document no_sanitize attribute.
26117 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
26118 to SANITIZE_UNDEFINED_NONDEFAULT.
26119 * gcc.c (sanitize_spec_function): Use the renamed enum value.
26120 * gimple-fold.c (optimize_atomic_compare_exchange_p):
26121 Use sanitize_flags_p.
26122 * gimplify.c (gimplify_function_tree): Likewise.
26123 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
26124 * opts.c (parse_no_sanitize_attribute): New function.
26125 (common_handle_option): Use renamed enum value.
26126 * opts.h (parse_no_sanitize_attribute): Declare.
26127 * tree.c (sanitize_flags_p): New function.
26128 * tree.h: Declared here.
26129 * tsan.c: Use sanitize_flags_p.
26130 * ubsan.c (ubsan_expand_null_ifn): Likewise.
26131 (instrument_mem_ref): Likewise.
26132 (instrument_bool_enum_load): Likewise.
26133 (do_ubsan_in_current_function): Remove the function.
26134 (pass_ubsan::execute): Use sanitize_flags_p.
26135 * ubsan.h: Remove do_ubsan_in_current_function
26136 * tree-cfg.c (print_no_sanitize_attr_value): New function.
26137 (dump_function_to_file): Use it here.
26139 2017-06-13 Martin Jambor <mjambor@suse.cz>
26141 PR tree-optimization/80803
26142 PR tree-optimization/81063
26143 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
26144 (propagate_subaccesses_across_link): Enqueue subtree whenever
26145 necessary instead of relying on the caller.
26147 2017-06-13 Martin Jambor <mjambor@suse.cz>
26149 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
26150 that have a first_link.
26151 (sort_and_splice_var_accesses): Do not check first_link before
26153 (subtree_mark_written_and_enqueue): Likewise.
26154 (propagate_all_subaccesses): Likewise and do not stop at first
26155 parent with a first_link.
26157 2017-06-13 Martin Jambor <mjambor@suse.cz>
26159 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
26162 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26164 * match.pd: New pattern.
26166 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26168 * tree-vrp.c (is_masked_range_test): New function.
26169 (register_edge_assert_for): Determine ranges for
26172 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26174 PR tree-optimization/67328
26175 * fold-const.c (maskable_range_p): New function.
26176 (build_range_check): Generate bittests if possible.
26178 2017-06-13 Martin Liska <mliska@suse.cz>
26180 * gimple-pretty-print.c (dump_probability): Add new argument.
26181 (dump_edge_probability): Dump both probability and count.
26182 (dump_gimple_label): Likewise.
26183 (dump_gimple_bb_header): Likewise.
26185 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
26188 * config/avr/avr-devices.c: Fix indentation.
26189 * config/avr/gen-avr-mmcu-specs.c: Dito.
26191 2017-06-13 Richard Biener <rguenther@suse.de>
26193 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
26194 instead get vector type from stmt_info.
26195 (vectorizable_reduction): Adjust. Remove dead code.
26197 2017-06-13 Richard Biener <rguenther@suse.de>
26199 PR middle-end/81065
26200 * fold-const.c (extract_muldiv_1): Remove bogus distribution
26201 case of C * (x * C2 + C3).
26202 (fold_addr_of_array_ref_difference): Properly fold index difference.
26204 2017-06-12 David S. Miller <davem@davemloft.net>
26207 * config/sparc/sparc.md (return expander): Emit frame blockage if
26208 function uses alloca.
26210 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
26212 * combine.c (make_field_assignment): Check len rather than the mode
26213 precision when calling force_to_mode.
26215 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
26217 Support multilibs and devices that see flash in RAM address range.
26220 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
26221 (avr_mcu_t) <flash_pm_offset>: New field.
26222 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
26223 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
26224 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
26225 (AVR_TINY_PM_OFFSET): Remove macro.
26226 * config/avr/avr.opt (-mshort-calls): New option.
26227 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
26228 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
26229 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
26230 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
26231 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
26232 instead of avr_arch->have_jmp_call.
26233 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
26234 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
26235 avr_arch->flash_pm_offset to define.
26236 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
26237 new field flash_pm_offset. Add entry for avrxmega3.
26238 (avr_texinfo): Add entry for avrxmega3.
26239 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
26240 attiny212, attiny214,
26241 attiny412, attiny414, attiny416, attiny417,
26242 attiny814, attiny816, attiny817,
26243 attiny1614, attiny1616, attiny1617,
26244 attiny3214, attiny3216, attiny3217.
26245 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
26246 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
26247 (avr_print_operand_address) [AVR_TINY]: Same.
26248 (avr_asm_init_sections) <readonly_data_section>: Only patch
26249 callback if avr_arch->flash_pm_offset = 0.
26250 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
26251 for rodata if avr_arch->flash_pm_offset != 0.
26252 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
26253 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
26254 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
26255 (m_options): Append opt_rcall.
26256 (m_dirnames): Append dir_rcall.
26257 * config/avr/t-multilib: Regenerate.
26259 * configure.ac [target=avr]: Check whether avrxmega3 default
26260 linker description file works as needed.
26261 * configure: Regenerate.
26262 * doc/avr-mmcu.texi: Regenerate.
26263 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
26264 <__AVR_ARCH__>: Document avrxmega3 and 103.
26265 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
26266 <__AVR_SHORT_CALLS__>: Document it.
26267 <__AVR_PM_BASE_ADDRESS__>: Document it.
26268 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
26269 (AVR Variable Attributes) <progmem>: Document this is
26270 not needed for avrxmega3.
26271 (AVR Named Address Spaces) <__flash>: Dito.
26273 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
26275 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
26277 2017-06-12 Doug Rupp <rupp@adacore.com>
26279 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
26280 Append vxworks-stdint.h to the tm_file list.
26281 * config/vxworks-stdint.h: New file.
26283 2017-06-12 Martin Liska <mliska@suse.cz>
26285 PR tree-optimization/81041
26286 * tree-profile.c (gimple_gen_ic_func_profiler):
26287 Create an extra BB in profile-generate
26288 (gimple_gen_time_profiler): Likewise.
26290 2017-06-12 Jakub Jelinek <jakub@redhat.com>
26292 PR tree-optimization/81003
26293 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
26294 (update_range_test): Use it instead of force_gimple_operand_gsi.
26296 2017-06-12 Richard Biener <rguenther@suse.de>
26298 PR tree-optimization/81053
26299 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
26300 with backedge value not defined in loop. Simplify def stmt
26303 2017-06-11 Tom de Vries <tom@codesourcery.com>
26306 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
26308 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
26309 nvptx_cannot_force_const_mem.
26311 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26313 * opts.c (finish_options): Move test for flag_split_stack after
26314 it has been initialized.
26316 2017-06-11 Jason Merrill <jason@redhat.com>
26318 * tree.h (id_equal): New.
26319 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
26320 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
26321 instead of strcmp of IDENTIFIER_POINTER.
26323 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26325 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
26326 (mark_all_inlined_calls_cdtor): Fix formating.
26327 (inline_transform): Rescale profile before inlining.
26329 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26331 * cgraph.h (cgraph_edge::clone): Update prototype.
26332 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
26333 (cgraph_node::create_clone): Update.
26334 (cgraph_node::create_version_clone): Update.
26335 * tree-inline.c (copy_bb): Update.
26336 (expand_call_inline): Update.
26338 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26340 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
26341 factored out from ...
26342 (rs6000_emit_prologue): ... here.
26344 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26346 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
26347 factored out from ...
26348 (rs6000_emit_prologue): ... here.
26350 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26352 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
26354 (handle_missing_profiles): Fix computation of tp_first_run.
26355 (counts_to_freqs): Do not touch freqs when count is 0.
26357 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26359 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
26362 2017-06-10 Tom de Vries <tom@codesourcery.com>
26364 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26365 attributes): Document signal effective target.
26367 2017-06-10 Tom de Vries <tom@codesourcery.com>
26369 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26370 Document effective target stack_size.
26372 2017-06-09 David Malcolm <dmalcolm@redhat.com>
26374 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
26375 to the edit_context if they can be auto-applied.
26377 2017-06-9 Ian Lance Taylor <iant@golang.org>
26379 * opts.c (finish_options): If -fsplit-stack, disable implicit
26380 -forder-blocks-and-partition.
26381 * doc/invoke.texi (Optimize Options): Document that when using
26382 -fsplit-stack -forder-blocks-and-partition is not implicitly
26385 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26387 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
26388 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
26389 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
26390 * builtins.def (abort, trap, unreachable): Declare cold.
26391 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
26392 * tree-core.h (ECF_COLD): New.
26393 * tree.c (set_call_expr_flags): Handle ECF_COLD.
26394 (build_common_builtin_nodes): Mark unreachable and abort as cold.
26396 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26398 * predict.c (unlikely_executed_stmt_p): Cleanup.
26400 2017-06-09 Richard Biener <rguenther@suse.de>
26402 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
26403 model if the ref is always written to.
26405 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26407 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
26409 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26411 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
26414 2017-06-09 Tom de Vries <tom@codesourcery.com>
26417 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
26418 "target cannot support label values" when encountering LABEL_REF.
26420 2017-06-09 Martin Liska <mliska@suse.cz>
26422 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
26423 (gimple_gen_ic_func_profiler): Emit direct comparison
26424 of __gcov_indirect_call_callee with NULL.
26425 (gimple_gen_time_profiler): Change probability from
26426 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
26428 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26430 * profile.c (edge_gcov_counts): Turn to pointer.
26431 (compute_branch_probabilities, compute_branch_probabilities): Update.
26432 (branch_prob): Do not clear edge_gcov_count.
26433 * profile.h (edge_gcov_counts): Turn to pointer.
26434 (edge_gcov_count): Update.
26436 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26438 * gimple.h (gimple_check_failed): Mark cold.
26440 2017-06-09 Richard Biener <rguenther@suse.de>
26442 PR tree-optimization/66623
26443 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
26444 refactor check_reduction into two parts, properly computing
26445 whether we have to check reduction validity for outer loop
26448 2017-06-09 Richard Biener <rguenther@suse.de>
26450 PR tree-optimization/79483
26451 * graphite-scop-detection.c (order): New global.
26452 (get_order): Compute bb to order mapping that satisfies code
26453 generation constraints.
26454 (cmp_pbbs): New helper.
26455 (build_scops): Start domwalk at entry block, sort generated
26458 2017-06-09 Richard Biener <rguenther@suse.de>
26460 PR middle-end/81007
26461 * ipa-polymorphic-call.c
26462 (ipa_polymorphic_call_context::restrict_to_inner_class):
26463 Skip FIELD_DECLs with error_mark_node type.
26464 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
26467 2017-06-09 Martin Liska <mliska@suse.cz>
26469 * predict.c (struct branch_predictor): New struct.
26470 (test_prediction_value_range): New test.
26471 (predict_c_tests): New function.
26472 * selftest-run-tests.c (selftest::run_tests): Run the function.
26473 * selftest.h: Declare new tests.
26475 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
26478 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
26479 gen_add3_insn did not fail.
26480 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
26481 r0, construct that number in a temporary reg and add that reg to r0.
26482 If asked to put the result in r0 as well, fail.
26484 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
26486 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26487 for early expansion of vec_eqv.
26489 2017-06-08 Jakub Jelinek <jakub@redhat.com>
26491 PR middle-end/81005
26492 * ubsan.c (instrument_null): Avoid pointless code temporary.
26493 (pass_ubsan::execute): Instrument aggregate arguments of calls.
26495 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
26499 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
26502 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
26505 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26507 * predict.c (unlikely_executed_edge_p): Move ahead.
26508 (probably_never_executed_edge_p): Use it.
26510 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26512 PR middle-end/79988
26513 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
26514 gimple_call_builtin_p call.
26516 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26518 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
26519 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
26520 rtl_check_failed_type2, rtl_check_failed_code1,
26521 rtl_check_failed_code2, rtl_check_failed_code_mode,
26522 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
26523 rtvec_check_failed_bounds, rtl_check_failed_flag,
26524 _fatal_insn_not_found, _fatal_insn): Likewise.
26525 * tree.h (tree_contains_struct_check_failed,
26526 tree_check_failed, tree_not_check_failed,
26527 tree_class_check_failed, tree_range_check_failed,
26528 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
26529 tree_vec_elt_check_failed, phi_node_elt_check_failed,
26530 tree_operand_check_failed, omp_clause_check_failed,
26531 omp_clause_operand_check_failed, omp_clause_range_check_failed):
26534 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26536 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
26537 flag_branch_probabilities.
26538 * ipa-inline.c (edge_badness): Likewise.
26539 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
26540 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
26541 * predict.c (maybe_hot_frequency_p): Likewise.
26542 (probably_never_executed): Likewise.
26543 * sched-ebb.c (schedule_ebbs): Likewise.
26544 * sched-rgn.c (find_single_block_region): Likewise.
26545 * tracer.c (tail_duplicate): Likewise.
26547 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26549 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
26550 longer requires x_flag_profile_use.
26552 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26554 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
26555 instead of flag_reorder_blocks_and_partition.
26556 * dbxout.c (dbxout_function_end): Likewise.
26557 * dwarf2out.c (gen_subprogram_die): Likewise.
26558 * haifa-sched.c (sched_create_recovery_edges): Likewise.
26559 * hw-doloop.c (reorg_loops): Likewise.
26560 * varasm.c (assemble_start_function,
26561 assemble_end_function): Likewise.
26562 (decide_function_section): Do not check for
26563 flag_reorder_blocks_and_partition.
26565 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26567 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26569 (chkp_get_hard_register_fake_addr_expr): Ditto.
26570 (chkp_build_addr_expr): Add check for hard reg case.
26571 (chkp_parse_array_and_component_ref): Ditto.
26572 (chkp_find_bounds_1): Ditto.
26573 (chkp_process_stmt): Don't generate bounds store for
26576 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26578 * predict.c (maybe_hot_bb_p): Do not check profile status.
26579 (maybe_hot_edge_p): Likewise.
26580 (probably_never_executed): Check for zero counts even if profile
26582 (unlikely_executed_edge_p): New function.
26583 (unlikely_executed_stmt_p): New function.
26584 (unlikely_executed_bb_p): New function.
26585 (set_even_probabilities): Use unlikely predicates.
26586 (combine_predictions_for_bb): Likewise.
26587 (predict_paths_for_bb): Likewise.
26588 (predict_paths_leading_to_edge): Likewise.
26589 (determine_unlikely_bbs): New function.
26590 (estimate_bb_frequencies): Use it.
26591 (compute_function_frequency): Use zero counts even if profile is
26593 * profile-count.h: Fix typo.
26595 2017-08-08 Julia Koval <julia.koval@intel.com>
26597 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
26598 _mm512_mask_cvtsepi16_storeu_epi8,
26599 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
26600 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
26601 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
26602 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
26603 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
26604 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
26605 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
26606 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
26607 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
26608 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
26609 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
26610 __builtin_ia32_pmovuswb256mem_mask,
26611 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
26612 __builtin_ia32_pmovwb512mem_mask): New builtins.
26614 2017-08-08 Julia Koval <julia.koval@intel.com>
26616 PR target/73350,80862
26617 * config/i386/subst.md (round): Fix round pattern.
26618 * config/i386/i386.c (ix86_erase_embedded_rounding):
26619 Fix erasing rounding for the fixed pattern.
26621 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26623 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
26625 2017-06-08 Martin Liska <mliska@suse.cz>
26627 PR gcov-profile/80911
26628 * gcov.c (block_info::block_info): New constructor.
26630 2017-06-07 Carl Love <cel@us.ibm.com>
26632 * config/rs6000/rs6000-c: The return type of the following
26633 built-in functions was implemented as int not long long. Fix sign
26634 of return value for the unsigned version of vec_mulo and vec_mule.
26635 vector unsigned long long vec_bperm (vector unsigned long long,
26636 vector unsigned char)
26637 vector signed long long vec_mule (vector signed int,
26639 vector unsigned long long vec_mule (vector unsigned int,
26640 vector unsigned int)
26641 vector signed long long vec_mulo (vector signed int,
26643 vector unsigned long long vec_mulo (vector unsigned int,
26644 vector unsigned int)
26645 * doc/extend.texi: Fix the documentation for the built-in
26648 2017-06-07 Carl Love <cel@us.ibm.com>
26651 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
26654 2017-06-07 Carl Love <cel@us.ibm.com>
26656 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
26657 support, Generate doublehv for signed int/float for BE case only.
26659 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
26661 * doc/invoke.texi (mcx16): Rewrite.
26663 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26665 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
26666 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
26667 *mov<mode>_softfloat, and an anonymous splitter): Use
26668 nonimmediate_operand instead of rs6000_nonimmediate_operand.
26670 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26672 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
26674 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26675 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
26676 (rs6000_debug_reg_global): Adjust.
26677 (rs6000_init_hard_regno_mode_ok): Adjust.
26678 (rs6000_dbx_register_number): Adjust.
26679 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
26680 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26681 Remove SPE_ACC and SPEFSCR.
26682 (REG_ALLOC_ORDER): Ditto.
26683 (FRAME_POINTER_REGNUM): Change to 111.
26684 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
26685 (REG_CLASS_NAMES): Ditto.
26686 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
26687 (REGISTER_NAMES): Ditto.
26688 (ADDITIONAL_REG_NAMES): Ditto.
26689 (rs6000_reg_names): Ditto.
26690 * config/rs6000/rs6000.md: Renumber some register number
26693 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26695 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
26697 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26698 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
26700 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
26701 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26702 Delete the SPE high registers.
26703 (REG_ALLOC_ORDER): Ditto.
26704 (enum reg_class): Remove SPE_HIGH_REGS.
26705 (REG_CLASS_NAMES): Ditto.
26706 (REG_CLASS_CONTENTS): Delete the SPE high registers.
26707 (REGISTER_NAMES): Ditto.
26708 (rs6000_reg_names): Ditto.
26709 * doc/tm.texi.in: Remove SPE as example.
26710 * doc/tm.texi: Regenerate.
26712 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26714 * config/rs6000/8540.md (ppc8540_brinc): Delete.
26715 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
26716 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
26717 * config/rs6000/rs6000.md (type): Remove "brinc".
26719 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26721 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
26722 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
26723 * config/rs6000/linuxspe.h: Delete file.
26724 * config/rs6000/rs6000.md: Don't include spe.md.
26725 * config/rs6000/spe.h: Delete file.
26726 * config/rs6000/spe.md: Delete file.
26727 * config/rs6000/t-rs6000: Remove spe.md.
26729 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26731 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
26732 (reg_or_none500mem_operand): Delete.
26733 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
26734 instead of reg_or_none500mem_operand.
26736 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26738 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26739 handling of SPE flags.
26740 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
26742 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26744 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
26746 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
26747 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
26748 paired_divv2sf3): Similar.
26749 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
26750 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
26751 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
26753 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
26754 Rename the paired_* instruction patterns.
26755 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
26757 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
26758 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
26759 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
26760 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
26761 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
26762 PAIRED_VECTOR_MODE.
26763 (struct machine_function): Delete field spe_insn_chain_scanned_p.
26764 (spe_func_has_64bit_regs_p): Delete.
26765 (spe_expand_predicate_builtin): Delete.
26766 (spe_expand_evsel_builtin): Delete.
26767 (TARGET_DWARF_REGISTER_SPAN): Do not define.
26768 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
26769 (invalid_e500_subreg): Delete.
26770 (rs6000_legitimize_address): Always force_reg op2 as well, for
26771 paired single memory accesses.
26772 (rs6000_member_type_forces_blk): Delete.
26773 (rs6000_spe_function_arg): Delete.
26774 (rs6000_expand_unop_builtin): Delete SPE handling.
26775 (rs6000_expand_binop_builtin): Ditto.
26776 (spe_expand_stv_builtin): Delete.
26777 (bdesc_2arg_spe): Delete.
26778 (spe_expand_builtin): Delete.
26779 (spe_expand_predicate_builtin): Delete.
26780 (spe_expand_evsel_builtin): Delete.
26781 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
26782 (spe_init_builtins): Delete.
26783 (spe_func_has_64bit_regs_p): Delete.
26784 (savres_routine_name): Delete "info" parameter. Adjust callers.
26785 (rs6000_emit_stack_reset): Ditto.
26786 (rs6000_dwarf_register_span): Delete.
26787 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
26788 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
26789 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
26791 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
26793 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
26794 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
26795 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
26796 mulv2sf3, divv2sf3): Delete expanders.
26798 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26800 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
26802 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26804 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
26805 * config/rs6000/rs6000.c: Ditto.
26807 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26809 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
26810 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
26811 comparison_operator.
26813 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26815 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
26816 * config/rs6000/rs6000.opt: Ditto.
26817 * config/rs6000/t-rtems: Ditto.
26819 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26821 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
26822 TARGET_E500_SINGLE by 0, simplify.
26823 * config/rs6000/rs6000.c: Ditto.
26824 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
26825 (spe_build_register_parallel): Delete.
26826 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
26827 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
26828 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
26829 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
26830 (E500_CONVERT): Delete.
26831 * config/rs6000/spe.md: Remove many patterns and all define_constants.
26833 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26835 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
26836 * config/rs6000/dfp.md: Ditto.
26837 (negdd2, *negdd2_fpr): Merge.
26838 (absdd2, *absdd2_fpr): Merge.
26839 (negtd2, *negtd2_fpr): Merge.
26840 (abstd2, *abstd2_fpr): Merge.
26841 * config/rs6000/e500.h: Delete file.
26842 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
26843 TARGET_FPRS by 1 and simplify.
26844 * config/rs6000/rs6000-c.c: Ditto.
26845 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
26846 TARGET_DF_SPE by 0.
26847 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
26849 * config/rs6000/rs6000.md: Ditto.
26850 (floatdidf2, *floatdidf2_fpr): Merge.
26851 (move_from_CR_gt_bit): Delete.
26852 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
26853 (E500_CR_IOR_COMPARE): Delete.
26854 (All patterns that require !TARGET_FPRS): Delete.
26855 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
26857 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26859 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
26861 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26863 * graphds.c (add_edge): Intitialize edge's attached data.
26864 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
26865 pointer parameter. Call pointed function on each edge during
26866 graph traversing. Skip traversing the edge when the function
26868 (graphds_dfs, graphds_scc): Ditto.
26869 (for_each_edge): New parameter. Pass the new parameter to callback
26871 * graphds.h (skip_edge_callback): New function pointer type.
26872 (graphds_dfs, graphds_scc): New function pointer parameter.
26873 (graphds_edge_callback, for_each_edge): New parameter.
26875 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26877 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
26878 out code checking if runtime alias check is possible to below ...
26879 Call the new function.
26880 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
26881 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
26883 2017-06-07 Marek Polacek <polacek@redhat.com>
26886 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
26887 TYPE_OVERFLOW_WRAPS check.
26889 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26891 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
26892 if versioning is required.
26893 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
26894 peeling with the check for versioning.
26896 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26898 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
26899 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
26900 Set true to new parameter if new ssa variable is defined.
26901 (vect_gen_vector_loop_niters): Refactor. Set range information
26902 for the new vector loop bound variable.
26903 (vect_do_peeling): Ditto.
26905 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26907 * tree-affine.c (ssa.h): Include header file.
26908 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
26909 has wrapping overflow behavior.
26911 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26913 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
26915 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26917 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
26918 (tree_to_aff_combination): ... here.
26920 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26922 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
26923 reg_pressure model function.
26924 (ivopts_global_cost_for_size): Delete.
26925 (determine_set_costs, iv_ca_recount_cost): Call new model function
26926 ivopts_estimate_reg_pressure.
26928 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26930 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
26931 expensive than udiv. Remove floating point cases from mod.
26933 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26935 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
26936 Increase idiv cost.
26938 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26940 * config/aarch64/aarch64.md
26941 (copysignsf3): Fix mask generation.
26943 2017-06-07 Jakub Jelinek <jakub@redhat.com>
26945 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
26947 (class dump_manager): Add register_dumps method.
26948 * dumpfile.c: Include langhooks.h.
26949 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
26950 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
26951 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
26952 (dump_manager::dump_register): Start with 512 entries instead of 32.
26953 (dump_manager::register_dumps): New method.
26954 * toplev.c (general_init): Instead of invoking register_dumps
26955 langhook, invoke register_dumps method on the dump manager.
26956 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
26959 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
26961 * doc/md.texi: Clarify the restrictions on a define_insn condition.
26962 Say that # requires an associated define_split to exist, and that
26963 the define_split must be suitable for use after register allocation.
26965 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26967 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
26968 (compute_outgoing_frequencies): Also initialize zero counts.
26969 (find_many_sub_basic_blocks): Do not produce uninitialized profile
26970 around loops; preserve more of profile when nothing changes.
26972 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
26974 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
26976 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
26977 * config/arm/arm-cpu-cdata.h: Regenerate.
26978 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
26979 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
26980 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
26981 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
26982 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
26984 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
26986 * config/arm/t-rmprofile: Likewise.
26987 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
26989 2017-06-06 David S. Miller <davem@davemloft.net>
26992 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
26993 blockage if function uses alloca.
26995 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26997 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
26998 New "uid" fields to hold pretty-print IDs of group and ref.
26999 Memory references are now identified as <group_id>:<ref_id>
27000 instead of using [random] addresses.
27001 (dump_mem_details): Simplify, no functional change.
27002 (dump_mem_ref): Simplify and make output more concise.
27003 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27004 (find_or_create_group): Initialize group uid.
27005 (record_ref): Initialize ref uid. Improve debug output.
27006 (prune_group_by_reuse, should_issue_prefetch_p,)
27007 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27008 (mark_nontemporal_store, determine_loop_nest_reuse):
27009 Improve debug output.
27011 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27013 * dbgcnt.def (prefetch): New debug counter.
27014 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27015 (schedule_prefetches): Stop issueing prefetches if debug counter
27018 2017-06-06 Tom de Vries <tom@codesourcery.com>
27020 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27021 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27023 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27025 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27026 Use aarch64_reg_or_zero predicate for operand 4.
27027 (aarch64_compare_and_swap<mode> define_insn_and_split):
27028 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27029 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27031 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27033 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27034 (arm_compute_save_core_reg_mask): This.
27035 (thumb1_compute_save_reg_mask): Rename into ...
27036 (thumb1_compute_save_core_reg_mask): This.
27037 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27038 (arm_compute_frame_layout): Likewise.
27040 2017-06-06 Richard Biener <rguenther@suse.de>
27042 PR tree-optimization/80974
27043 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
27044 keep or clear leaders SSA info.
27046 2017-06-06 Tom de Vries <tom@codesourcery.com>
27048 * config/nvptx/nvptx.c (split_mode_p): New function.
27049 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
27051 2017-06-06 Tom de Vries <tom@codesourcery.com>
27053 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
27055 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27058 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
27061 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27063 * shrink-wrap.c (handle_simple_exit): Update profile.
27064 (try_shrink_wrapping): Upate profile.
27066 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27068 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
27069 (tree_guess_outgoing_edge_probabilities): New.
27070 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
27071 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
27073 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27075 * ipa-split.c (split_function): Initialize return bb profile.
27077 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27079 * profile.c (compute_branch_probabilities): Also initialize
27080 EXIT_BLOCK profile.
27082 2017-06-06 Richard Biener <rguenther@suse.de>
27084 PR tree-optimization/80928
27085 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
27086 (vect_analyze_loop_operations): Properly guard analysis for
27088 (vect_transform_loop): Likewise.
27089 (vect_analyze_loop_2): Also reset SLP type on PHIs.
27090 (vect_model_induction_cost): Do not cost for pure SLP.
27091 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
27092 of induction in inner loop vectorization.
27093 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
27094 (vect_get_and_check_slp_defs): Handle vect_induction_def.
27095 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
27097 (vect_analyze_slp_cost_1): Cost induction.
27098 (vect_detect_hybrid_slp_stmts): Handle PHIs.
27099 (vect_get_slp_vect_defs): Likewise.
27100 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
27101 (vect_transform_stmt): Handle SLP reductions.
27102 * tree-vectorizer.h (vectorizable_induction): Adjust.
27104 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27106 * config/rs6000/rs6000.c (make_resolver_func): Update
27107 init_lowered_empty_function call.
27109 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
27111 * doc/invoke.texi: Document the -fprofile-abs-path option.
27112 * common.opt (fprofile-abs-path): New option.
27113 * gcov-io.h (gcov_write_filename): Declare.
27114 * gcov-io.c (gcov_write_filename): New function.
27115 * coverage.c (coverage_begin_function): Use gcov_write_filename.
27116 * profile.c (output_location): Likewise.
27118 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27120 * shring-wrap.c: Revert accidental commit.
27122 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
27124 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
27126 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27128 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
27130 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
27131 profile in callgraph edge.
27132 * profile-count.h (apply_probability): If THIS is 0, then result is 0
27133 (apply_scale): Likewise.
27134 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
27135 Also scale profile when inlining function with zero profile.
27136 (initialize_cfun): Update exit block profile even when it is zero.
27137 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
27138 when profile is read.
27140 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27142 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
27143 (CLONE_*): New constants to define the processors we can generate
27144 code for with the target_clone attribute.
27145 (rs6000_clone_map): New array to identify which clone processors
27146 the current program is running on.
27147 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
27148 target_clone attribute.
27149 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
27150 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
27151 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
27152 (cpu_expand_builtin): Add support for target_clone attribute.
27153 (rs6000_valid_attribute_p): Allow "default" attribute.
27154 (get_decl_name): New debug function to simplify printing the
27155 current function name in debugging statements.
27156 (rs6000_clone_priority): New functions to support the target_clone
27157 attribute, and be able to generate code to switch between ISA 2.05
27158 through ISA 3.0 (power6 through power9).
27159 (rs6000_compare_version_priority): Likewise.
27160 (rs6000_get_function_versions_dispatcher): Likewise.
27161 (make_resolver_func): Likewise.
27162 (add_condition_to_bb): Likewise.
27163 (dispatch_function_versions): Likewise.
27164 (rs6000_generate_version_dispatcher_body): Likewise.
27165 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
27166 (fusion_gpr_load_p): Fix a spacing issue.
27167 * doc/extend.texi (Common Function Attributes): Document that the
27168 PowerPC supports the target_clone attribute.
27170 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27172 * config/arm/arm.h: explain F symbol found in description of ARM
27173 register allocation in its legend.
27175 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27177 * config/mips/frame-header-opt.c: Include profile-count.h.
27178 * config/riscv/riscv.c: Include profile-count.h
27180 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27182 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
27184 (sm_set_flag_if_changed): Add bbs field.
27185 (execute_sm_if_changed_flag_set): Pass BBS.
27186 (execute_sm): Update.
27188 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27190 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
27193 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27195 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
27196 (peephole2): New peephole2 to emit the above.
27197 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
27199 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27201 * config/aarch64/aarch64.c (define_peephole2 above
27202 *sub_<shift>_<mode>): New peephole.
27204 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27206 * config/i386/i386.c (make_resolver_func): Update.
27207 * Makefile.in: Add profile-count.h and profile-count.o
27208 * auto-profile.c (afdo_indirect_call): Update to new API.
27209 (afdo_set_bb_count): Update.
27210 (afdo_propagate_edge): Update.
27211 (afdo_propagate_circuit): Update.
27212 (afdo_calculate_branch_prob): Update.
27213 (afdo_annotate_cfg): Update.
27214 * basic-block.h: Include profile-count.h
27215 (struct edge_def): Turn count to profile_count.
27216 (struct basic_block_def): Likewie.
27217 (REG_BR_PROB_BASE): Move to profile-count.h
27218 (RDIV): Move to profile-count.h
27219 * bb-reorder.c (max_entry_count): Turn to profile_count.
27220 (find_traces): Update.
27221 (rotate_loop):Update.
27222 (connect_traces):Update.
27223 (sanitize_hot_paths):Update.
27224 * bt-load.c (migrate_btr_defs): Update.
27225 * cfg.c (RDIV): Remove.
27226 (init_flow): Use alloc_block.
27227 (alloc_block): Uninitialize count.
27228 (unchecked_make_edge): Uninitialize count.
27229 (check_bb_profile): Update.
27230 (dump_edge_info): Update.
27231 (dump_bb_info): Update.
27232 (update_bb_profile_for_threading): Update.
27233 (scale_bbs_frequencies_int): Update.
27234 (scale_bbs_frequencies_gcov_type): Update.
27235 (scale_bbs_frequencies_profile_count): New.
27236 * cfg.h (update_bb_profile_for_threading): Update.
27237 (scale_bbs_frequencies_profile_count): Declare.
27238 * cfgbuild.c (compute_outgoing_frequencies): Update.
27239 (find_many_sub_basic_blocks): Update.
27240 * cfgcleanup.c (try_forward_edges): Update.
27241 (try_crossjump_to_edge): Update.
27242 * cfgexpand.c (expand_gimple_tailcall): Update.
27243 (construct_exit_block): Update.
27244 * cfghooks.c (verify_flow_info): Update.
27245 (dump_bb_for_graph): Update.
27246 (split_edge): Update.
27247 (make_forwarder_block): Update.
27248 (duplicate_block): Update.
27249 (account_profile_record): Update.
27250 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
27251 (get_estimated_loop_iterations): Update.
27252 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
27253 (single_likely_exit): Update.
27254 * cfgloopmanip.c (scale_loop_profile): Update.
27256 (set_zero_probability): Update.
27257 (lv_adjust_loop_entry_edge): Update.
27258 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
27259 (purge_dead_edges): Update.
27260 (rtl_account_profile_record): Update.
27261 * cgraph.c (cgraph_node::create): Uninitialize count.
27262 (symbol_table::create_edge): Uninitialize count.
27263 (cgraph_update_edges_for_call_stmt_node): Update.
27264 (cgraph_edge::dump_edge_flags): Update.
27265 (cgraph_node::dump): Update.
27266 (cgraph_edge::maybe_hot_p): Update.
27267 * cgraph.h: Include profile-count.h
27268 (create_clone), create_edge, create_indirect_edge): Update.
27269 (cgraph_node): Turn count to profile_count.
27270 (cgraph_edge0: Likewise.
27271 (make_speculative, clone): Update.
27272 (create_edge): Update.
27273 (init_lowered_empty_function): Update.
27274 * cgraphclones.c (cgraph_edge::clone): Update.
27275 (duplicate_thunk_for_node): Update.
27276 (cgraph_node::create_clone): Update.
27277 * cgraphunit.c (cgraph_node::analyze): Update.
27278 (cgraph_node::expand_thunk): Update.
27279 * final.c (dump_basic_block_info): Update.
27280 * gimple-streamer-in.c (input_bb): Update.
27281 * gimple-streamer-out.c (output_bb): Update.
27282 * graphite.c (print_global_statistics): Update.
27283 (print_graphite_scop_statistics): Update.
27284 * hsa-brig.c: Include basic-block.h.
27285 * hsa-dump.c: Include basic-block.h.
27286 * hsa-gen.c (T sum_slice): Update.
27287 (convert_switch_statements):Update.
27288 * hsa-regalloc.c: Include basic-block.h.
27289 * ipa-chkp.c (chkp_produce_thunks): Update.
27290 * ipa-cp.c (struct caller_statistics): Update.
27291 (init_caller_stats): Update.
27292 (gather_caller_stats): Update.
27293 (ipcp_cloning_candidate_p): Update.
27294 (good_cloning_opportunity_p): Update.
27295 (get_info_about_necessary_edges): Update.
27296 (dump_profile_updates): Update.
27297 (update_profiling_info): Update.
27298 (update_specialized_profile): Update.
27299 (perhaps_add_new_callers): Update.
27300 (decide_about_value): Update.
27301 (ipa_cp_c_finalize): Update.
27302 * ipa-devirt.c (struct odr_type_warn_count): Update.
27303 (struct decl_warn_count): Update.
27304 (struct final_warning_record): Update.
27305 (possible_polymorphic_call_targets): Update.
27306 (ipa_devirt): Update.
27307 * ipa-fnsummary.c (redirect_to_unreachable): Update.
27308 * ipa-icf.c (sem_function::merge): Update.
27309 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
27310 * ipa-inline.c (compute_uninlined_call_time): Update.
27311 (compute_inlined_call_time): Update.
27312 (want_inline_small_function_p): Update.
27313 (want_inline_self_recursive_call_p): Update.
27314 (edge_badness): Update.
27315 (lookup_recursive_calls): Update.
27316 (recursive_inlining): Update.
27317 (inline_small_functions): Update.
27318 (dump_overall_stats): Update.
27319 (dump_inline_stats): Update.
27320 * ipa-profile.c (ipa_profile_generate_summary): Update.
27321 (ipa_propagate_frequency): Update.
27322 (ipa_profile): Update.
27323 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27324 * ipa-utils.c (ipa_merge_profiles): Update.
27325 * loop-doloop.c (doloop_modify): Update.
27326 * loop-unroll.c (report_unroll): Update.
27327 (unroll_loop_runtime_iterations): Update.
27328 * lto-cgraph.c (lto_output_edge): Update.
27329 (lto_output_node): Update.
27330 (input_node): Update.
27331 (input_edge): Update.
27332 (merge_profile_summaries): Update.
27333 * lto-streamer-in.c (input_cfg): Update.
27334 * lto-streamer-out.c (output_cfg): Update.
27335 * mcf.c (create_fixup_graph): Update.
27336 (adjust_cfg_counts): Update.
27337 (sum_edge_counts): Update.
27338 * modulo-sched.c (sms_schedule): Update.
27339 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
27340 * predict.c (maybe_hot_count_p): Update.
27341 (probably_never_executed): Update.
27342 (dump_prediction): Update.
27343 (combine_predictions_for_bb): Update.
27344 (propagate_freq): Update.
27345 (handle_missing_profiles): Update.
27346 (counts_to_freqs): Update.
27347 (rebuild_frequencies): Update.
27348 (force_edge_cold): Update.
27349 * predict.h: Include profile-count.h
27350 (maybe_hot_count_p, counts_to_freqs): UPdate.
27351 * print-rtl-function.c: Do not include cfg.h
27352 * print-rtl.c: Include basic-block.h
27353 * profile-count.c: New file.
27354 * profile-count.h: New file.
27355 * profile.c (is_edge_inconsistent): Update.
27356 (correct_negative_edge_counts): Update.
27357 (is_inconsistent): Update.
27358 (set_bb_counts): Update.
27359 (read_profile_edge_counts): Update.
27360 (compute_frequency_overlap): Update.
27361 (compute_branch_probabilities): Update; Initialize and deinitialize
27363 (branch_prob): Update.
27364 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
27365 (edge_gcov_count): New.
27366 (bb_gcov_count): New.
27367 * shrink-wrap.c (try_shrink_wrapping): Update.
27368 * tracer.c (better_p): Update.
27369 * trans-mem.c (expand_transaction): Update.
27370 (ipa_tm_insert_irr_call): Update.
27371 (ipa_tm_insert_gettmclone_call): Update.
27372 * tree-call-cdce.c: Update.
27373 * tree-cfg.c (gimple_duplicate_sese_region): Update.
27374 (gimple_duplicate_sese_tail): Update.
27375 (gimple_account_profile_record): Update.
27376 (execute_fixup_cfg): Update.
27377 * tree-inline.c (copy_bb): Update.
27378 (copy_edges_for_bb): Update.
27379 (initialize_cfun): Update.
27380 (freqs_to_counts): Update.
27381 (copy_cfg_body): Update.
27382 (expand_call_inline): Update.
27383 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
27384 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
27385 (try_unroll_loop_completely): Update.
27386 (try_peel_loop): Update.
27387 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
27388 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
27389 * tree-ssa-loop-split.c (connect_loops): Update.
27390 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
27391 * tree-ssa-reassoc.c (branch_fixup): Update.
27392 * tree-ssa-tail-merge.c (replace_block_by): Update.
27393 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
27394 (compute_path_counts): Update.
27395 (update_profile): Update.
27396 (recompute_probabilities): Update.
27397 (update_joiner_offpath_counts): Update.
27398 (estimated_freqs_path): Update.
27399 (freqs_to_counts_path): Update.
27400 (clear_counts_path): Update.
27401 (ssa_fix_duplicate_block_edges): Update.
27402 (duplicate_thread_path): Update.
27403 * tree-switch-conversion.c (case_bit_test_cmp): Update.
27404 (struct switch_conv_info): Update.
27405 * tree-tailcall.c (decrease_profile): Update.
27406 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
27407 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
27408 * value-prof.c (check_counter): Update.
27409 (gimple_divmod_fixed_value): Update.
27410 (gimple_mod_pow2): Update.
27411 (gimple_mod_subtract): Update.
27412 (gimple_ic_transform): Update.
27413 (gimple_stringop_fixed_value): Update.
27414 * value-prof.h (gimple_ic): Update.
27416 2017-06-02 Carl Love <cel@us.ibm.com>
27418 * config/rs6000/rs6000-c: Add support for built-in functions
27419 vector double vec_doublee (vector signed int);
27420 vector double vec_doublee (vector unsigned int);
27421 vector double vec_doublee (vector float);
27422 vector double vec_doubleh (vector signed int);
27423 vector double vec_doubleh (vector unsigned int);
27424 vector double vec_doubleh (vector float);
27425 vector double vec_doublel (vector signed int);
27426 vector double vec_doublel (vector unsigned int);
27427 vector double vec_doublel (vector float);
27428 vector double vec_doubleo (vector signed int);
27429 vector double vec_doubleo (vector unsigned int);
27430 vector double vec_doubleo (vector float);.
27431 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
27432 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
27434 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
27435 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
27436 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
27438 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
27439 vec_doublel, vec_doubleh.
27440 * doc/extend.texi: Update the built-in documentation file for the
27441 new built-in functions.
27443 2017-06-02 David Malcolm <dmalcolm@redhat.com>
27446 * ipa-inline-analysis.c (free_growth_caches): Set
27447 edge_removal_hook_holder to NULL after removing it.
27449 2017-06-02 Sudakshina Das <sudi.das@arm.com>
27451 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
27452 comparision with zero.
27454 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27455 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27456 for early expansion of vec_min and vec_max builtins.
27457 (builtin_function_type): Add min/max unsigned variants to those
27458 identified as having unsigned arguments.
27460 2017-06-02 Olivier Hainque <hainque@adacore.com>
27462 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
27464 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27466 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
27467 Use VALL_F16 iterator rather than VALL.
27469 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27471 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
27472 Emit CBNZ inside loop when doing a strong exchange and comparing
27473 against zero. Generate the CC flags after the loop.
27475 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
27477 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
27478 (dl_section_ref): New.
27479 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
27480 On AIX, append an expression to subtract the size of the
27481 section length to dl_section_ref.
27483 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27485 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27486 for early expansion of vector absolute builtins.
27488 2017-06-02 Richard Biener <rguenther@suse.de>
27490 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
27491 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
27493 2017-06-02 Richard Biener <rguenther@suse.de>
27495 PR tree-optimization/80948
27496 * tree-tailcall.c (find_tail_calls): Track stmts to move in
27497 stmt order as well.
27499 2017-06-02 Richard Biener <rguenther@suse.de>
27501 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
27503 * tree-vect-stmts.c (process_use): Do not mark backedge defs
27504 for inductions as relevant.
27506 2017-06-02 Richard Biener <rguenther@suse.de>
27508 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
27509 (vectorizable_induction): ... this. Remove dead code.
27511 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
27513 * builtins. (expand_builtin_alloca): Remove second parameter and
27514 infer its value from the first parameter instead.
27515 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
27517 2017-06-02 Jakub Jelinek <jakub@redhat.com>
27519 PR rtl-optimization/80903
27520 * loop-doloop.c (add_test): Unshare sequence.
27522 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
27524 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
27526 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
27528 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
27530 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
27531 xlogue_layout::get_instance, logue_layout::xlogue_layout,
27532 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
27533 (xlogue_layout::get_stub_rtx): Make static.
27534 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
27535 (xlogue_layout::compute_stub_managed_regs): Rename to...
27536 (xlogue_layout::count_stub_managed_regs): ...this.
27537 (xlogue_layout::is_stub_managed_reg): New function.
27538 (xlogue_layout::m_stub_names): Rename to...
27539 (xlogue_layout::s_stub_names): ...this, make static.
27540 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
27541 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
27542 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
27543 xlogue_layout::s_stub_names): Instantiate statics.
27544 (stub_managed_regs): Remove.
27545 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
27546 (disable_call_ms2sysv_xlogues): Rename to...
27547 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
27548 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
27550 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
27551 change after reload_completed.
27552 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
27554 (ix86_expand_prologue): Likewise.
27555 (ix86_expand_epilogue): Likewise.
27556 (ix86_expand_split_stack_prologue): Likewise.
27557 (ix86_compute_frame_layout): Remove frame parameter ...
27558 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
27559 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
27561 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
27562 (ix86_frame): Move from here ...
27563 * config/i386/i386.h (ix86_frame): ... to here.
27564 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
27565 complete ix86_frame data structure instead. Remove some_ld_name.
27567 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27569 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
27570 symbols that hold a DECL_VALUE_EXPR.
27572 2017-06-01 Martin Jambor <mjambor@suse.cz>
27574 PR tree-optimization/80898
27575 * tree-sra.c (process_subtree_disqualification): Removed.
27576 (disqualify_candidate): Do not acll
27577 process_subtree_disqualification.
27578 (subtree_mark_written_and_enqueue): New function.
27579 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
27580 RHS has been disqualified and re-queue LHS if necessary. Apart
27581 from that, ignore disqualified RHS.
27583 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27585 * config/s390/s390.c (s390_emit_epilogue): Disable early return
27586 address fetch for z10 or later.
27588 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27590 * config/arc/arc.md (tst_movb): Add guard when splitting.
27592 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27594 * config/arc/arc.c (arc_can_eliminate): Test against
27595 arc_frame_pointer_needed.
27597 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27599 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
27600 to prevent store reordering.
27601 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
27602 (type): Add block type.
27603 (stack_tie): Define special instruction to be used in
27606 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27608 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
27609 constraint. It is not valid for the pattern.
27610 (noncommutative_binary_comparison): Likewise.
27612 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27614 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
27617 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27619 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
27620 be used by the reg-alloc.
27622 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27624 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
27625 reg-alloc when having mul64 or mul32x16 instructions.
27626 (mulsidi3): Likewise.
27627 (umulsidi3): Likewise.
27628 (mulsi32x16): New pattern.
27629 (mulsi64): Likewise.
27630 (mulsidi64): Likewise.
27631 (umulsidi64): Likewise.
27632 (MUL32x16_REG): Define.
27633 (mul64_600): Use MUL32x16_REG.
27634 (mac64_600): Likewise.
27635 (umul64_600): Likewise.
27636 (umac64_600): Likewise.
27638 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27640 * config/arc/arc.md (mulsi3_700): Make it commutative.
27642 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
27644 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
27646 (*sign_extendsidi2_insn): Likewise for movstosw.
27648 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27650 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
27651 the type of the input discriminant value. Convert the
27652 discriminant value of signedness vary.
27654 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
27656 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
27657 Add to -Wall section.
27659 2017-06-01 Richard Biener <rguenther@suse.de>
27661 PR middle-end/66313
27662 * fold-const.c (fold_plusminus_mult_expr): If the factored
27663 factor may be zero use a wrapping type for the inner operation.
27664 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
27665 and handle moved defs.
27666 (process_assignment): Properly guard the unary op case. Return a
27667 tri-state indicating that moving the stmt before the call may allow
27668 to continue. Pass through to_move.
27669 (find_tail_calls): Handle moving unrelated defs before
27672 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
27675 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
27676 splitter result in the canonical way.
27678 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
27680 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
27681 also for 32bit target. Update insn attributes.
27682 (zero-extendsidi2 splitter): Allow all registers for operand 1.
27684 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
27686 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
27687 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
27688 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
27689 (_mm_maskz_min_ss): New intrinsics.
27691 2017-05-31 Martin Liska <mliska@suse.cz>
27693 * tree-vect-loop.c (vect_create_epilog_for_reduction):
27694 Change comment style to one we normally use.
27695 (vectorizable_reduction): Likewise.
27696 (vectorizable_induction): Likewise.
27697 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
27698 (vectorizable_call): Likewise.
27699 (vectorizable_simd_clone_call): Likewise.
27700 (vectorizable_conversion): Likewise.
27701 (vectorizable_assignment): Likewise.
27702 (vectorizable_shift): Likewise.
27703 (vectorizable_operation): Likewise.
27704 (vectorizable_store): Likewise.
27705 (vectorizable_load): Likewise.
27706 * tree-vectorizer.h: Likewise.
27708 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
27710 * passes.c (emergency_dump_function): New.
27711 * tree-pass.h (emergency_dump_function): Declare.
27712 * plugin.c (plugins_internal_error_function): Remove.
27713 * plugin.h (plugins_internal_error_function): Remove declaration.
27714 * toplev.c (internal_error_function): New static function. Use it...
27715 (general_init): ...here.
27717 2017-05-31 Graham Markall <graham.markall@embecosm.com>
27719 * config/arc/arc.c (arc_print_operand): Handle constant operands.
27720 (arc_rtx_costs): Add costs for new patterns.
27721 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
27722 * config/arc/predicates.md: Add _1_2_3_operand predicate.
27724 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27726 * tree-ssa-strlen.c (get_next_strinfo): New function.
27727 (get_stridx_plus_constant): Use it.
27728 (zero_length_string): Likewise.
27729 (adjust_related_strinfos): Likewise.
27730 (adjust_last_stmt): Likewise.
27732 2017-05-31 Richard Biener <rguenther@suse.de>
27735 * config/i386/i386.c (ix86_expand_builtin): Remove assert
27736 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
27738 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27740 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
27741 loop_vinfo argument and use of dependence distance vectors.
27742 Check instead whether the two references differ only in their
27743 initial value and assume that they have the same alignment if the
27744 difference is a multiple of the vector alignment.
27745 (vect_analyze_data_refs_alignment): Update call accordingly.
27747 2017-05-31 Martin Liska <mliska@suse.cz>
27750 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
27752 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27754 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
27755 (create_intersect_range_checks): Move from ...
27756 * tree-data-ref.c (create_intersect_range_checks_index)
27757 (create_intersect_range_checks): ... to here.
27758 (create_runtime_alias_checks): New function factored from ...
27759 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
27760 here. Call above function.
27761 * tree-data-ref.h (create_runtime_alias_checks): New function.
27763 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27765 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
27766 segment length for dr_b and compute it in wide_int.
27768 2017-05-31 Richard Biener <rguenther@suse.de>
27770 PR tree-optimization/80906
27771 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
27772 and pass through iv_map.
27773 (copy_bb_and_scalar_dependences): Adjust.
27774 (translate_pending_phi_nodes): Likewise.
27775 (copy_loop_close_phi_args): Handle code-generating IVs instead
27778 2017-05-30 David Malcolm <dmalcolm@redhat.com>
27780 * diagnostic-color.c (color_dict): Add "type-diff".
27781 (parse_gcc_colors): Update comment.
27782 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
27783 -fdiagnostics-show-template-tree and -fno-elide-type.
27784 (GCC_COLORS): Add type-diff to example.
27786 (-fdiagnostics-show-template-tree): New.
27787 (-fno-elide-type): New.
27788 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
27789 the pp_format_decoder callback. Call any m_format_postprocessor's
27791 (pretty_printer::pretty_printer): Initialize
27792 m_format_postprocessor.
27793 (pretty_printer::~pretty_printer): Delete any
27794 m_format_postprocessor.
27795 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
27796 (class format_postprocessor): New class.
27797 (struct pretty_printer::format_decoder): Document the new parameters.
27798 (struct pretty_printer::m_format_postprocessor): New field.
27799 * tree-diagnostic.c (default_tree_printer): Update for new
27800 bool and const char ** params.
27801 * tree-diagnostic.h (default_tree_printer): Likewise.
27803 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
27805 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
27806 (lwa_operand): Delete rs6000_gen_cell_microcode test.
27807 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27808 rs6000_gen_cell_microcode code.
27809 (rs6000_final_prescan_insn): Delete.
27810 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
27811 "warn-cell-microcode" entries.
27812 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
27813 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
27814 throughout. Change cc_reg_not_micro_cr0_operand to
27815 cc_reg_not_cr0_operand throughout.
27816 (*extendhi<mode>2_noload): Delete.
27817 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
27818 (mwarn-cell-microcode): Delete.
27819 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
27820 -mgen-cell-microcode and -mwarn-cell-microcode.
27822 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
27825 * config/i386/constraints.md (Yd): New constraint.
27827 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
27828 and (?Yd, r) alternatives. Update insn attributes.
27829 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
27830 and (?*Yd, r) alternatives. Update insn attributes.
27831 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
27833 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
27835 * gimplify.c (gimplify_modify_expr): Don't create a
27836 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
27839 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
27841 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
27843 2017-05-30 Richard Biener <rguenther@suse.de>
27845 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
27846 and reduc_def fields.
27847 (STMT_VINFO_REDUC_TYPE): New define.
27848 (STMT_VINFO_REDUC_DEF): Likewise.
27849 (vect_force_simple_reduction): Adjust prototype.
27850 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
27851 (vect_is_simple_reduction): Remove check_reduction argument.
27852 (vect_force_simple_reduction): Adjust and set
27853 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27854 (vectorizable_reduction): Do not re-do reduction analysis
27855 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27856 * tree-parloops.c (gather_scalar_reductions): Adjust.
27858 2017-05-30 Richard Biener <rguenther@suse.de>
27860 PR middle-end/80901
27861 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
27864 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
27866 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
27867 Introduce unknown_misalignment parameter and remove vf.
27868 (vect_peeling_hash_get_lowest_cost):
27869 Pass unknown_misalignment parameter.
27870 (vect_enhance_data_refs_alignment):
27871 Fix unsupportable data ref treatment.
27873 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27875 * tree-vect-data-refs.c (vect_get_data_access_cost):
27876 Workaround for SLP handling.
27877 (vect_enhance_data_refs_alignment):
27878 Compute costs for doing no peeling at all, compare to the best
27879 peeling costs so far and avoid peeling if cheaper.
27881 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27883 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
27884 Return peeling info and set costs to zero for unlimited cost
27886 (vect_enhance_data_refs_alignment): Also inspect all datarefs
27887 with unknown misalignment. Compute and costs for unknown
27888 misalignment, compare them to the costs for known misalignment
27889 and choose the cheapest for peeling.
27891 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27893 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
27894 (vect_get_peeling_costs_all_drs): Create function.
27895 (vect_peeling_hash_get_lowest_cost):
27896 Use vect_get_peeling_costs_all_drs.
27897 (vect_peeling_supportable): Create function.
27898 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
27900 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27902 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
27903 DR_HAS_NEGATIVE_STEP.
27904 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
27905 (vect_enhance_data_refs_alignment): Use.
27906 (vect_duplicate_ssa_name_ptr_info): Use.
27907 * tree-vectorizer.h (dr_misalignment): Use.
27908 (known_alignment_for_access_p): Use.
27910 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27913 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
27915 (has_section_name): New function.
27917 2017-05-30 Martin Liska <mliska@suse.cz>
27920 * auto-profile.c (get_function_decl_from_block): Fix
27923 2017-05-30 Richard Biener <rguenther@suse.de>
27925 PR middle-end/80876
27926 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
27928 2017-05-30 Martin Liska <mliska@suse.cz>
27930 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
27931 * dumpfile.h (struct dump_file_info): Remove ctors.
27933 2017-05-30 Martin Liska <mliska@suse.cz>
27935 * predict.def: Fix GNU coding style.
27937 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27939 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27940 Mark 'to' argument with ATTRIBUTE_UNUSED.
27942 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27944 * config/xtensa/xtensa.c (xtensa_emit_call): Use
27945 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
27946 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
27949 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
27951 * doc/install.texi (Options specification): Restore entry of
27952 --enable-sjlj-exceptions.
27954 2017-05-27 Michael Eager <eager@eagercon.com>
27957 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
27959 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
27961 * config/microblaze/microblaze.h
27962 (FIXED_REGISTERS): Update in macro.
27963 (CALL_USED_REGISTERS): Update in macro.
27965 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
27967 * doc/install.texi: Add links to macOS binary distributions.
27969 2017-05-27 Jakub Jelinek <jakub@redhat.com>
27973 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27975 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27977 2017-05-26 Martin Liska <mliska@suse.cz>
27979 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
27981 2017-05-26 Martin Liska <mliska@suse.cz>
27983 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
27984 always leading ';; '.
27985 (dump_bb_info): Likewise.
27986 (brief_dump_cfg): Likewise.
27987 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
27988 * dumpfile.c: Remove usage of TDF_VERBOSE.
27989 * dumpfile.h (enum dump_kind): Likewise.
27990 (dump_gimple_bb_header): Do not use TDF_COMMENT.
27991 * print-tree.c (debug_verbose): Remove.
27992 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
27993 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
27994 * tree-diagnostic.c (default_tree_printer): Replace
27995 TDF_DIAGNOSTIC with TDF_SLIM.
27997 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27999 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28000 in parameter loop, rather than loop_vinfo.
28001 (create_intersect_range_checks): Ditto.
28002 (vect_create_cond_for_alias_checks): Update call to above functions.
28004 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28006 PR tree-optimization/80815
28007 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28008 for merging runtime alias checks. Handle negative DR_STEPs.
28010 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28012 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28014 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28015 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28016 out code pruning runtime alias checks.
28017 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28018 factored out from above.
28019 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28021 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28023 (prune_runtime_alias_test_list): New decalaration.
28025 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28027 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28028 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28029 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28030 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28031 (operator==, comp_dr_with_seg_len_pair): Ditto.
28032 (vect_prune_runtime_alias_test_list): Ditto.
28034 2017-05-26 Martin Liska <mliska@suse.cz>
28037 * params.def: Bound partial-inlining-entry-probability param.
28039 2017-05-26 Marek Polacek <polacek@redhat.com>
28042 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
28045 2017-05-26 Richard Biener <rguenther@suse.de>
28047 PR tree-optimization/80842
28048 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
28051 2017-05-26 Richard Biener <rguenther@suse.de>
28053 PR tree-optimization/80844
28054 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
28056 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
28058 * doc/md.texi (Machine Constraints): Update x86 family
28059 machine constraints section to match 'config/i386/constraints.md'.
28061 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
28063 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
28065 2017-05-25 Nathan Sidwell <nathan@acm.org>
28067 * doc/invoke.texi (--enable-languages): Update documentation.
28069 2017-05-25 Martin Liska <mliska@suse.cz>
28071 * dumpfile.c: Add TDF_FOLDING.
28072 * dumpfile.h (enum dump_kind): Likewise.
28073 * genmatch.c (dt_simplify::gen_1): Use it.
28075 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28077 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
28079 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28081 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28082 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
28084 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28086 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
28087 * match.pd (X == C): Rewrite it here.
28088 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
28089 with_certain_nonzero_bits2): New predicates.
28090 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
28092 2017-05-24 Nathan Sidwell <nathan@acm.org>
28094 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
28097 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
28100 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28102 * config/powerpcspe: New port. Files are copied from the rs6000
28103 port, with "rs6000" in filenames replaced by "powerpcspe".
28105 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
28107 PR rtl-optimization/80754
28108 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
28110 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
28112 * config/sparc/sparc.md (length): Return the correct value for -mflat
28113 sibcalls to match output_sibcall.
28115 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28119 * config/rs6000/rs6000.c (struct machine_function): Add new field
28121 (rs6000_get_separate_components): Init that field, use it.
28122 (rs6000_components_for_bb): Use the field.
28124 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
28126 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
28128 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
28130 PR middle-end/80823
28131 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
28133 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28136 * config/s390/s390.c (s390_check_qrst_address): Check incoming
28137 address against address_operand predicate.
28138 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
28140 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
28142 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
28143 parameters passed indirectly.
28145 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
28147 * config/i386/i386.md (*movdi_internal): Remove SSE4
28148 alternative 18 (?r, *v). Update insn attributes.
28149 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
28150 Update insn attributes.
28151 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
28152 Update insn attributes.
28153 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
28154 alternative 1 (r, v). Remove isa attribute.
28155 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
28156 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
28157 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
28159 2017-05-23 Tom de Vries <tom@codesourcery.com>
28161 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
28164 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28166 * cgraphunit.c (symbol_table::process_new_functions): Update.
28167 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
28168 (inline_generate_summary): Rename to ...
28169 (ipa_fn_summary_generate): ... this one.
28170 (inline_read_summary): Rename to ...
28171 (ipa_fn_summary_read): ... this one.
28172 (inline_write_summary): Rename to ...
28173 (ipa_fn_summary_write): ... this one.
28174 (inline_free_summary): Rename to ...
28175 (ipa_free_fn_summary): ... this one.
28176 (pass_data_local_fn_summary, pass_local_fn_summary,
28177 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
28178 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
28179 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
28180 make_pass_ipa_fn_summary): New.
28181 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
28182 inline_write_summary, inline_free_summary): Remove.
28183 (ipa_free_fn_summary) : New.
28184 * ipa-inline.c (ipa_inline): Update.
28185 (pass_ipa_inline): Do not generate summaries.
28186 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
28188 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
28189 and add pass_ipa_fn_summary.
28190 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
28192 (make_pass_inline_parameters): Remove.
28194 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
28196 * omp-low.c (struct omp_context): Remove "default_kind" member.
28199 * omp-offload.c (execute_oacc_device_lower): Remove the
28200 parallelism dimensions function attributes for unparallelized
28201 OpenACC kernels constructs.
28203 2017-05-23 Martin Liska <mliska@suse.cz>
28205 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
28207 (cgraph_edge::make_speculative): Likewise.
28208 (cgraph_edge::resolve_speculation): Likewise.
28209 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28210 (cgraph_node::dump): Likewise.
28211 * cgraph.h: Likewise.
28212 * cgraphunit.c (analyze_functions): Likewise.
28213 (symbol_table::compile): Likewise.
28214 * ipa-cp.c (print_all_lattices): Likewise.
28215 (determine_versionability): Likewise.
28216 (initialize_node_lattices): Likewise.
28217 (ipcp_verify_propagated_values): Likewise.
28218 (estimate_local_effects): Likewise.
28219 (update_profiling_info): Likewise.
28220 (create_specialized_node): Likewise.
28221 (perhaps_add_new_callers): Likewise.
28222 (decide_about_value): Likewise.
28223 (decide_whether_version_node): Likewise.
28224 (identify_dead_nodes): Likewise.
28225 (ipcp_store_bits_results): Likewise.
28226 * ipa-devirt.c (dump_targets): Likewise.
28227 (ipa_devirt): Likewise.
28228 * ipa-icf.c (sem_item::dump): Likewise.
28229 (sem_function::equals): Likewise.
28230 (sem_variable::equals): Likewise.
28231 (sem_item_optimizer::read_section): Likewise.
28232 (sem_item_optimizer::execute): Likewise.
28233 (congruence_class::dump): Likewise.
28234 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
28235 (dump_inline_summary): Likewise.
28236 (estimate_node_size_and_time): Likewise.
28237 (inline_analyze_function): Likewise.
28238 * ipa-inline-transform.c (inline_call): Likewise.
28239 * ipa-inline.c (report_inline_failed_reason): Likewise.
28240 (want_early_inline_function_p): Likewise.
28241 (edge_badness): Likewise.
28242 (update_edge_key): Likewise.
28243 (inline_small_functions): Likewise.
28244 * ipa-profile.c (ipa_profile): Likewise.
28245 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
28246 (ipa_make_edge_direct_to_target): Likewise.
28247 (remove_described_reference): Likewise.
28248 (ipa_impossible_devirt_target): Likewise.
28249 (propagate_controlled_uses): Likewise.
28250 (ipa_print_node_params): Likewise.
28251 (ipcp_transform_function): Likewise.
28252 * ipa-pure-const.c (pure_const_read_summary): Likewise.
28253 (propagate_pure_const): Likewise.
28254 * ipa-reference.c (generate_summary): Likewise.
28255 (read_write_all_from_decl): Likewise.
28256 (propagate): Likewise.
28257 (ipa_reference_read_optimization_summary): Likewise.
28258 * ipa-utils.c (ipa_merge_profiles): Likewise.
28259 * ipa.c (walk_polymorphic_call_targets): Likewise.
28260 (symbol_table::remove_unreachable_nodes): Likewise.
28261 (ipa_single_use): Likewise.
28262 * passes.c (execute_todo): Likewise.
28263 * predict.c (drop_profile): Likewise.
28264 * symtab.c (symtab_node::get_dump_name): New function.
28265 (symtab_node::dump_name): Likewise.
28266 (symtab_node::dump_asm_name): Likewise.
28267 (symtab_node::dump_references): Likewise.
28268 (symtab_node::dump_referring): Likewise.
28269 (symtab_node::dump_base): Likewise.
28270 (symtab_node::debug_symtab): Likewise.
28271 * tree-sra.c (convert_callers_for_node): Likewise.
28272 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
28273 * value-prof.c (init_node_map): Likewise.
28275 2017-05-23 Martin Liska <mliska@suse.cz>
28277 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
28278 and symtab_node::debug_symtab to symbol_table::debug.
28279 * cgraphunit.c (analyze_functions): Use the renamed function.
28280 (symbol_table::compile): Likewise.
28281 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
28282 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
28283 * passes.c (execute_todo): Likewise.
28284 * symtab.c (symbol_table::dump): New function.
28285 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
28287 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28289 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
28290 that nonconst implies exec.
28292 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28294 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28295 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28296 (inline_edge_summary_vec): Turn into ...
28297 (ipa_call_summaries): ... this one.
28298 (redirect_to_unreachable, edge_set_predicate,
28299 evaluate_properties_for_edge, inline_summary_alloc,
28300 reset_ipa_call_summary, reset_inline_summary,
28301 inline_summary_t::duplicate): Update.
28302 (inline_edge_duplication_hook): Turn to ...
28303 (ipa_call_summary_t::duplicate): ... this one.
28304 (inline_edge_removal_hook): Turn to ...
28305 (ipa_call_summary_t::remove): ... this one.
28306 (dump_inline_edge_summary): Turn to ...
28307 (dump_ipa_call_summary): ... this one.
28308 (estimate_function_body_sizes): Update.
28309 (inline_update_callee_summaries): Update.
28310 (remap_edge_change_prob): Update.
28311 (remap_edge_summaries): Update.
28312 (inline_merge_summary): Update.
28313 (do_estimate_edge_time): Update.
28314 (inline_generate_summary): Update.
28315 (inline_read_section): Update.
28316 (inline_read_summary): Update.
28317 (inline_free_summary): Update.
28318 * ipa-inline.c (can_inline_edge_p): Update.
28319 (compute_inlined_call_time): Update.
28320 (want_inline_small_function_p): Update.
28321 (edge_badness): Update.
28322 (early_inliner): Update.
28323 * ipa-inline.h (inline_edge_summary): Turn to ...
28324 (ipa_call_summary): ... this one.
28325 (ipa_call_summary_t): New class.
28326 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28327 (ipa_call_summaries): New.
28328 (inline_edge_summary): Remove.
28329 (estimate_edge_growth): Update.
28330 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28331 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28332 * ipa-split.c (execute_split_functions): Update.
28333 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28335 2017-05-23 Tom de Vries <tom@codesourcery.com>
28337 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28338 attributes): Document rdrand effective target.
28340 2017-05-23 Tom de Vries <tom@codesourcery.com>
28342 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28343 attributes): Sort alphabetically.
28345 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
28347 * config/avr/genmultilib.awk: Use gsub instead of gensub.
28349 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
28352 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
28353 V2DF/V2DI splat into two separate patterns, one that handles
28354 registers, and the other that only handles memory. Drop support
28355 for splatting from a GPR on ISA 2.07 and then splitting the
28356 splat into direct move and splat.
28357 (vsx_splat_<mode>_reg): Likewise.
28358 (vsx_splat_<mode>_mem): Likewise.
28360 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
28362 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
28364 2017-05-22 Jakub Jelinek <jakub@redhat.com>
28366 PR middle-end/80809
28367 * omp-low.c (finish_taskreg_remap): New function.
28368 (finish_taskreg_scan): If unit size of ctx->record_type
28369 is non-constant, unshare the size expression and replace
28370 decls in it with possible outer var refs.
28372 PR middle-end/80809
28373 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
28374 GOVD_SHARED rather than GOVD_PRIVATE with it.
28375 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
28376 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
28378 PR middle-end/80853
28379 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
28380 as last argument to build_outer_var_ref for pointer bases of array
28381 section reductions.
28383 2017-05-19 Martin Sebor <msebor@redhat.com>
28385 * print-tree.c (print_node): Print DECL_READ_P flag.
28387 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28389 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
28390 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
28391 * cgraph.c: Likewise.
28392 * cgraphunit.c: Likewise.
28393 * gengtype.c: Likewise.
28394 * ipa-cp.c: Likewise.
28395 * ipa-devirt.c: Likewise.
28396 * ipa-icf.c: Likewise.
28397 * ipa-predicate.c: Likewise.
28398 * ipa-profile.c: Likewise.
28399 * ipa-prop.c: Likewise.
28400 * ipa-split.c: Likewise.
28402 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
28403 edge_predicate_pool, dump_inline_hints,
28404 inline_summary::account_size_time, redirect_to_unreachable,
28405 edge_set_predicate, set_hint_predicate,
28406 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
28407 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
28408 inline_summary_t::remove, remap_hint_predicate_after_duplication,
28409 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
28410 ipa_call_summary_t::remove, initialize_growth_caches,
28411 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
28412 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
28413 mark_modified, unmodified_parm_1, unmodified_parm,
28414 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
28415 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
28416 compute_bb_predicates, will_be_nonconstant_expr_predicate,
28417 will_be_nonconstant_predicate, record_modified_bb_info,
28418 get_minimal_bb, record_modified, param_change_prob,
28419 phi_result_unknown_predicate, predicate_for_phi_result,
28420 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
28421 estimate_function_body_sizes, compute_inline_parameters,
28422 compute_inline_parameters_for_curren, pass_data_inline_parameters,
28423 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
28424 inline_update_callee_summaries, remap_edge_change_prob,
28425 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
28426 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
28427 inline_analyze_function, inline_summary_t::insert,
28428 inline_generate_summary, read_ipa_call_summary, inline_read_section,
28429 inline_read_summary, write_ipa_call_summary, inline_write_summary,
28430 inline_free_summary): Move to ipa-fnsummary.h
28431 (predicate_t): Remove.
28432 * ipa-fnsummary.c: New file.
28433 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
28434 (enum inline_hints_vals, inline_hints, agg_position_info,
28435 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
28436 inline_summaries, ipa_call_summary, ipa_call_summary_t,
28437 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
28438 dump_inline_summary, dump_inline_hints, inline_generate_summary,
28439 inline_read_summary, inline_write_summary, inline_free_summary,
28440 inline_analyze_function, initialize_inline_failed,
28441 inline_merge_summary, inline_update_overall_summary,
28442 compute_inline_parameters): Move to ipa-fnsummary.h
28443 * ipa-fnsummary.h: New file.
28444 * ipa-inline-transform.h: Include ipa-inline.h.
28445 * ipa-inline.c: LIkewise.
28447 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28449 * ipa-inline.c (edge_badness): Use inlined_time instead of
28450 inline_summaries->get.
28452 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28454 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
28456 2017-05-22 Nathan Sidwell <nathan@acm.org>
28458 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
28459 (fdump-lang): Document 'raw' option.
28460 * dumpfile.h (TDI_tu): Delete.
28461 * dumpfile.c (dump_files): Remove translation-unit.
28462 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
28464 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
28466 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
28467 command option from $(AWK) call.
28468 * config/avr/genmultilib.awk: Simplify and rewrite so that it
28469 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
28470 [FORMAT]: Remove handling of variable.
28471 * config/avr/t-multilib: Regenerate.
28473 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28475 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
28477 (dump_inline_summary): Do not print self_time.
28478 (estimate_function_body_sizes): Do not set self_time.
28479 (compute_inline_parameters): Likewise.
28480 (inline_read_section, inline_write_summary): Do not stream self_time.
28481 * ipa-inline.h (inline_summary): Drop self_time.
28483 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28485 * ipa-inline-analysis.c (account_size_time): Rename to ...
28486 (inline_summary::account_size_time): ... this one.
28487 (reset_ipa_call_summary): Turn to ...
28488 (ipa_call_summary::reset): ... this one.
28489 (reset_inline_summary): Turn to ...
28490 (inline_summary::reset): ... this one.
28491 (inline_summary_t::remove): Update.
28492 (inline_summary_t::duplicate): Update.
28493 (ipa_call_summary_t::remove): Update.
28494 (dump_inline_summary): Update.
28495 (estimate_function_body_sizes): Update.
28496 (compute_inline_parameters): Update.
28497 (estimate_node_size_and_time): Update.
28498 (inline_merge_summary): Update.
28499 (inline_update_overall_summary): Update.
28500 (inline_read_section): Update.
28501 (inline_write_summary): Update.
28502 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
28503 add account_size_time and reset member functions.
28504 (ipa_call_summary): Add reset function.
28505 * ipa-predicate.h (predicate::operator &): Constify.
28507 2017-05-22 Richard Biener <rguenther@suse.de>
28509 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
28511 2017-05-19 Jason Merrill <jason@redhat.com>
28513 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
28515 2017-05-19 Marek Polacek <polacek@redhat.com>
28518 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
28519 TYPE_OVERFLOW_WRAPS checks.
28521 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
28523 * tree-core.h (enum omp_clause_default_kind): Add
28524 "OMP_CLAUSE_DEFAULT_PRESENT".
28525 * tree-pretty-print.c (dump_omp_clause): Handle it.
28526 * gimplify.c (enum gimplify_omp_var_data): Add
28527 "GOVD_MAP_FORCE_PRESENT".
28528 (gimplify_adjust_omp_clauses_1): Map it to
28529 "GOMP_MAP_FORCE_PRESENT".
28530 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
28532 * gimplify.c (oacc_default_clause): Clarify.
28534 2017-05-19 Nathan Sidwell <nathan@acm.org>
28536 LANG_HOOK_REGISTER_DUMPS
28537 * toplev.c (general_init): Call register dump lang hook.
28538 * doc/invoke.texi: Document -fdump-lang option family.
28539 * dumpfile.c (dump_files): Remove class dump here.
28540 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
28541 * dumpfile.h (tree_dump_index): Remove TDI_class.
28542 * langhooks-def.h (lhd_register_dumps): Declare.
28543 (LANG_HOOKS_REGISTER_DUMPS): Define.
28544 (LANG_HOOKS_INITIALIZER): Add it.
28545 * langhooks.c (lhd_register_dumps): Define.
28546 * langhooks.h (struct lang_hooks): Add register_dumps.
28548 2017-05-19 Nathan Sidwell <nathan@acm.org>
28550 * context.h (context::set_passes): New.
28551 * context.c (context::context): Do not create pass manager.
28552 * toplev.c (general_init): Create pass manager here.
28554 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
28556 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
28557 use this splitter if two add or or instructions would also work for
28558 the constant we want to generate.
28560 2017-05-19 Richard Biener <rguenther@suse.de>
28563 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
28564 predicate evaluation.
28566 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28568 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
28570 * ipa-inline.c (want_inline_small_function_p): Do not cast to
28573 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28575 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28576 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28577 (inline_edge_summary_vec): Turn into ...
28578 (ipa_call_summaries): ... this one.
28579 (redirect_to_unreachable, edge_set_predicate,
28580 evaluate_properties_for_edge, inline_summary_alloc,
28581 reset_ipa_call_summary, reset_inline_summary,
28582 inline_summary_t::duplicate): Update.
28583 (inline_edge_duplication_hook): Turn to ...
28584 (ipa_call_summary_t::duplicate): ... this one.
28585 (inline_edge_removal_hook): Turn to ...
28586 (ipa_call_summary_t::remove): ... this one.
28587 (dump_inline_edge_summary): Turn to ...
28588 (dump_ipa_call_summary): ... this one.
28589 (estimate_function_body_sizes): Update.
28590 (inline_update_callee_summaries): Update.
28591 (remap_edge_change_prob): Update.
28592 (remap_edge_summaries): Update.
28593 (inline_merge_summary): Update.
28594 (do_estimate_edge_time): Update.
28595 (inline_generate_summary): Update.
28596 (inline_read_section): Update.
28597 (inline_read_summary): Update.
28598 (inline_free_summary): Update.
28599 * ipa-inline.c (can_inline_edge_p): Update.
28600 (compute_inlined_call_time): Update.
28601 (want_inline_small_function_p): Update.
28602 (edge_badness): Update.
28603 (early_inliner): Update.
28604 * ipa-inline.h (inline_edge_summary): Turn to ...
28605 (ipa_call_summary): ... this one.
28606 (ipa_call_summary_t): New class.
28607 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28608 (ipa_call_summaries): New.
28609 (inline_edge_summary): Remove.
28610 (estimate_edge_growth): Update.
28611 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28612 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28613 * ipa-split.c (execute_split_functions): Update.
28614 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28616 2017-05-19 Richard Biener <rguenther@suse.de>
28618 PR middle-end/80764
28619 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
28621 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
28623 * config/rs6000/rs6000.c (struct machine_function): Add field
28624 fpr_is_wrapped_separately.
28625 (rs6000_get_separate_components): Use 64 components. Handle the
28626 new FPR components.
28627 (rs6000_components_for_bb): Handle the FPR components.
28628 (rs6000_emit_prologue_components): Handle the FPR components.
28629 (rs6000_emit_epilogue_components): Handle the FPR components.
28630 (rs6000_set_handled_components): Handle the FPR components.
28631 (rs6000_emit_prologue): Don't output prologue code for those FPRs
28632 that are already separately shrink-wrapped.
28633 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
28634 that are already separately shrink-wrapped.
28636 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28639 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
28642 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
28643 (define_peephole2 for Altivec d-form load): Add peepholes to catch
28644 cases where the register allocator uses a move and an offsettable
28645 memory operation to/from a FPR register on ISA 2.06/2.07.
28646 (define_peephole2 for Altivec d-form store): Likewise.
28648 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
28651 * config/i386/mmx.md (*mov<mode>_internal): Enable
28652 alternatives 11, 12, 13 and 14 also for 32bit targets.
28653 Remove alternatives 15, 16, 17 and 18.
28654 * config/i386/sse.md (vec_concatv2di): Change
28655 alternative (!x, *y) to (x, ?!*Yn).
28657 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
28659 * dumpfile.h (enum dump_kind): Remove stray comma.
28661 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28663 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
28664 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
28665 predicate::num_conditions
28666 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
28667 (CHANGED): turn into predicate::changed.
28668 (agg_position_info): Move to ipa-predicate.h
28669 (add_condition, predicate::add_clause, predicate::operator &=,
28670 predicate::or_with, predicate::evaluate, predicate::probability,
28671 dump_condition, dump_clause, predicate::dump,
28672 predicate::remap_after_duplication, predicate::remap_after_inlining,
28673 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
28674 (evaluate_conditions_for_known_args): Update.
28675 (set_cond_stmt_execution_predicate): Update.
28676 * ipa-inline.h: Include ipa-predicate.h
28677 (condition, inline_param_summary, conditions, agg_position_info,
28678 predicate): Move to ipa-predicate.h
28679 * ipa-predicate.c: New file.
28680 * ipa-predicate.h: New file.
28682 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
28684 * final.c (leaf_function_p): Check we are not in a sequence.
28686 2017-05-18 Martin Liska <mliska@suse.cz>
28688 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
28689 * dumpfile.c (dump_register): Use new enum dump_kind.
28690 (get_dump_file_name): Likewise.
28691 (dump_enable_all): Likewise.
28692 (dump_switch_p_1): Likewise.
28693 (enable_rtl_dump_file): Remove usage of TDF_RTL.
28694 * dumpfile.h (enum dump_kind): New enum type.
28695 (struct dump_file_info): Create constructor and
28696 format fields and comments.
28697 * passes.c (pass_manager::register_one_dump_file):
28699 * statistics.c (statistics_early_init): Likewise.
28700 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
28701 TDF_TREE with TDF_SLIM.
28702 (gather_memory_references_ref): Likewise.
28704 2017-05-18 Martin Liska <mliska@suse.cz>
28706 * vec.h (struct vnull): Use it.
28708 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28710 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
28711 (true_predicate, false_predicate, true_predicate_p,
28712 false_predicate_p): Remove.
28713 (single_cond_predicate, not_inlined_predicate): Turn to member function
28715 (add_condition): Update.
28716 (add_clause): Turn to...
28717 (predicate::add_clause): ... this one; update; allow passing NULL
28719 (and_predicates): Turn to ...
28720 (predicate::operator &=): ... this one.
28721 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
28722 (or_predicates): Turn to ...
28723 (predicate::or_with): ... this one.
28724 (evaluate_predicate): Turn to ...
28725 (predicate::evaluate): ... this one.
28726 (predicate_probability): Turn to ...
28727 (predicate::probability): ... this one.
28728 (dump_condition): Update.
28729 (dump_predicate): Turn to ...
28730 (predicate::dump): ... this one.
28731 (account_size_time): Update.
28732 (edge_set_predicate): Update.
28733 (set_hint_predicate): UPdate.
28734 (evaluate_conditions_for_known_args): Update.
28735 (evaluate_properties_for_edge): Update.
28736 (remap_predicate_after_duplication): Turn to...
28737 (predicate::remap_after_duplication): ... this one.
28738 (remap_hint_predicate_after_duplication): Update.
28739 (inline_summary_t::duplicate): UPdate.
28740 (dump_inline_edge_summary): Update.
28741 (dump_inline_summary): Update.
28742 (set_cond_stmt_execution_predicate): Update.
28743 (set_switch_stmt_execution_predicate): Update.
28744 (compute_bb_predicates): Update.
28745 (will_be_nonconstant_expr_predicate): Update.
28746 (will_be_nonconstant_predicate): Update.
28747 (phi_result_unknown_predicate): Update.
28748 (predicate_for_phi_result): Update.
28749 (array_index_predicate): Update.
28750 (estimate_function_body_sizes): Update.
28751 (estimate_node_size_and_time): Update.
28752 (estimate_ipcp_clone_size_and_time): Update.
28753 (remap_predicate): Rename to ...
28754 (predicate::remap_after_inlining): ... this one.
28755 (remap_hint_predicate): Update.
28756 (inline_merge_summary): Update.
28757 (inline_update_overall_summary): Update.
28758 (estimate_size_after_inlining): Update.
28759 (read_predicate): Rename to ...
28760 (predicate::stream_in): ... this one.
28761 (read_inline_edge_summary): Update.
28762 (write_predicate): Rename to ...
28763 (predicate::stream_out): ... this one.
28764 (write_inline_edge_summary): Update.
28765 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
28766 (clause_t): Turn to uint32_t
28767 (predicate): Turn to class; implement constructor and operators
28769 (size_time_entry): Update.
28770 (inline_summary): Update.
28771 (inline_edge_summary): Update.
28773 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
28775 * fold-const.c (fold_binary_loc): Move transformation...
28776 * match.pd (C - X CMP X): ... here.
28778 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
28780 * config/sparc/sparc.c (sparc_option_override): Set function
28781 alignment for -mcpu=niagara7 to 64 to match the I$ line.
28782 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
28784 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
28786 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
28788 2017-05-18 Marek Polacek <polacek@redhat.com>
28791 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
28792 (pass_ubsan::execute): Call gimple_assign_single_p instead of
28793 gimple_assign_load_p.
28795 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
28797 PR middle-end/80692
28798 * real.c (do_compare): Give decimal_do_compare preference over
28799 comparing just the signs.
28801 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
28803 * doc/md.texi (Canonicalization of Instructions): Describe the
28804 canonical form of instructions that inherently set a condition
28807 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
28809 PR middle-end/80775
28810 * tree-cfg.c: Move deletion of unreachable case statements to after
28811 the merging of consecutive case labels.
28813 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
28815 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
28816 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
28817 restoring of callee-saved registers.
28819 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
28821 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
28822 * config/visium/visium.c (single_set_and_flags): Likewise.
28823 * config/visium/visium.md (Substitutions): Likewise.
28825 2017-05-17 Martin Liska <mliska@suse.cz>
28827 * cfg.c: Introduce dump_flags_t type and
28828 use it instead of int type.
28830 * cfghooks.c: Likewise.
28831 * cfghooks.h (struct cfg_hooks): Likewise.
28832 * cfgrtl.c: Likewise.
28833 * cfgrtl.h: Likewise.
28834 * cgraph.c (cgraph_node::get_body): Likewise.
28835 * coretypes.h: Likewise.
28836 * domwalk.c: Likewise.
28837 * domwalk.h: Likewise.
28838 * dumpfile.c (struct dump_option_value_info): Likewise.
28839 (dump_enable_all): Likewise.
28840 (dump_switch_p_1): Likewise.
28841 (opt_info_switch_p): Likewise.
28842 * dumpfile.h (enum tree_dump_index): Likewise.
28843 (struct dump_file_info): Likewise.
28844 * genemit.c: Likewise.
28845 * generic-match-head.c: Likewise.
28846 * gengtype.c (open_base_files): Likewise.
28847 * gimple-pretty-print.c: Likewise.
28848 * gimple-pretty-print.h: Likewise.
28849 * graph.c (print_graph_cfg): Likewise.
28850 * graphite-scop-detection.c (dot_all_sese): Likewise.
28851 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
28852 * loop-unroll.c (report_unroll): Likewise.
28853 * passes.c (pass_manager::register_one_dump_file): Likewise.
28854 * print-tree.c: Likewise.
28855 * statistics.c: Likewise.
28856 * tree-cfg.c: Likewise.
28857 * tree-cfg.h: Likewise.
28858 * tree-dfa.c: Likewise.
28859 * tree-dfa.h: Likewise.
28860 * tree-dump.c (dump_function): Likewise.
28861 * tree-dump.h (struct dump_info): Likewise.
28862 * tree-pretty-print.c: Likewise.
28863 * tree-pretty-print.h: Likewise.
28864 * tree-ssa-live.c: Likewise.
28865 * tree-ssa-live.h: Likewise.
28866 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
28867 * tree-vect-loop.c: Likewise.
28868 * tree-vect-slp.c: Likewise.
28870 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
28871 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28873 PR tree-optimization/80457
28874 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
28875 of all arguments to a statement as scalar_to_vec operations.
28876 (vectorizable_call): Adjust call to vect_model_simple_cost for
28878 (vectorizable_conversion): Likewise.
28879 (vectorizable_assignment): Likewise.
28880 (vectorizable_shift): Likewise.
28881 (vectorizable_operation): Likewise.
28882 (vectorizable_comparison): Likewise.
28883 (vect_is_simple_cond): Record the def types for operands.
28884 (vectorizable_condition): Likewise, call vect_model_simple_cost.
28885 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
28886 for statement argument count.
28888 2017-05-16 Carl Love <cel@us.ibm.com>
28890 * config/rs6000/rs6000-c: Add support for built-in functions
28891 vector unsigned long long vec_bperm (vector unsigned long long,
28892 vector unsigned char)
28893 vector signed long long vec_mule (vector signed int,
28895 vector unsigned long long vec_mule (vector unsigned int,
28896 vector unsigned int)
28897 vector signed long long vec_mulo (vector signed int,
28899 vector unsigned long long vec_mulo (vector unsigned int,
28900 vector unsigned int)
28901 vector signed char vec_sldw (vector signed char,
28902 vector signed char,
28904 vector unsigned char vec_sldw (vector unsigned char,
28905 vector unsigned char,
28907 vector signed short vec_sldw (vector signed short,
28908 vector signed short,
28910 vector unsigned short vec_sldw (vector unsigned short,
28911 vector unsigned short,
28913 vector signed int vec_sldw (vector signed int,
28916 vector unsigned int vec_sldw (vector unsigned int,
28917 vector unsigned int,
28919 vector signed long long vec_sldw (vector signed long long,
28920 vector signed long long,
28922 vector unsigned long long vec_sldw (vector unsigned long long,
28923 vector unsigned long long,
28925 * config/rs6000/rs6000-c: Add support for built-in functions
28926 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
28927 * config/rs6000/altivec.h: Add defintion for vec_sldw.
28928 * doc/extend.texi: Update the built-in documentation for the
28929 new built-in functions.
28931 2017-05-16 Marek Polacek <polacek@redhat.com>
28935 * tree.c (save_expr): Don't fold the expression.
28937 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
28939 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
28940 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
28941 and (?*y,m). Update insn attributes.
28943 2017-05-16 Martin Liska <mliska@suse.cz>
28945 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
28946 flags argument of print_gimple_stmt, print_gimple_expr,
28947 print_generic_stmt and print_generic_expr.
28948 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
28949 * coretypes.h: Likewise.
28950 * except.c (dump_eh_tree): Likewise.
28951 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
28952 * gimple-pretty-print.h: Likewise.
28953 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
28954 (backprop::push_to_worklist): Likewise.
28955 (backprop::pop_from_worklist): Likewise.
28956 (backprop::process_use): Likewise.
28957 (backprop::intersect_uses): Likewise.
28958 (note_replacement): Likewise.
28959 * gimple-ssa-store-merging.c
28960 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
28961 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
28962 (pass_store_merging::execute): Likewise.
28963 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
28964 (ssa_base_cand_dump_callback): Likewise.
28965 (dump_incr_vec): Likewise.
28966 (replace_refs): Likewise.
28967 (replace_mult_candidate): Likewise.
28968 (create_add_on_incoming_edge): Likewise.
28969 (create_phi_basis): Likewise.
28970 (insert_initializers): Likewise.
28971 (all_phi_incrs_profitable): Likewise.
28972 (introduce_cast_before_cand): Likewise.
28973 (replace_one_candidate): Likewise.
28974 * gimplify.c (gimplify_expr): Likewise.
28975 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
28976 (set_rename): Likewise.
28977 (rename_uses): Likewise.
28978 (copy_loop_phi_nodes): Likewise.
28979 (add_close_phis_to_merge_points): Likewise.
28980 (copy_loop_close_phi_args): Likewise.
28981 (copy_cond_phi_args): Likewise.
28982 (graphite_copy_stmts_from_block): Likewise.
28983 (translate_pending_phi_nodes): Likewise.
28984 * graphite-poly.c (print_pdr): Likewise.
28985 (dump_gbb_cases): Likewise.
28986 (dump_gbb_conditions): Likewise.
28987 (print_scop_params): Likewise.
28988 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
28989 (build_cross_bb_scalars_use): Likewise.
28990 (gather_bbs::before_dom_children): Likewise.
28991 * hsa-dump.c (dump_hsa_immed): Likewise.
28992 * ipa-cp.c (print_ipcp_constant_value): Likewise.
28993 (get_replacement_map): Likewise.
28994 * ipa-inline-analysis.c (dump_condition): Likewise.
28995 (estimate_function_body_sizes): Likewise.
28996 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
28997 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
28998 * ipa-prop.c (ipa_dump_param): Likewise.
28999 (ipa_print_node_jump_functions_for_edge): Likewise.
29000 (ipa_modify_call_arguments): Likewise.
29001 (ipa_modify_expr): Likewise.
29002 (ipa_dump_param_adjustments): Likewise.
29003 (ipa_dump_agg_replacement_values): Likewise.
29004 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29005 * ipa-pure-const.c (check_stmt): Likewise.
29006 (pass_nothrow::execute): Likewise.
29007 * ipa-split.c (execute_split_functions): Likewise.
29008 * omp-offload.c (dump_oacc_loop_part): Likewise.
29009 (dump_oacc_loop): Likewise.
29010 * trans-mem.c (tm_log_emit): Likewise.
29011 (tm_memopt_accumulate_memops): Likewise.
29012 (dump_tm_memopt_set): Likewise.
29013 (dump_tm_memopt_transform): Likewise.
29014 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29015 (print_loop): Likewise.
29016 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29017 (chkp_gather_checks_info): Likewise.
29018 (chkp_get_check_result): Likewise.
29019 (chkp_remove_check_if_pass): Likewise.
29020 (chkp_use_outer_bounds_if_possible): Likewise.
29021 (chkp_reduce_bounds_lifetime): Likewise.
29022 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29023 (chkp_mark_completed_bounds): Likewise.
29024 (chkp_register_incomplete_bounds): Likewise.
29025 (chkp_mark_invalid_bounds): Likewise.
29026 (chkp_maybe_copy_and_register_bounds): Likewise.
29027 (chkp_build_returned_bound): Likewise.
29028 (chkp_get_bound_for_parm): Likewise.
29029 (chkp_build_bndldx): Likewise.
29030 (chkp_get_bounds_by_definition): Likewise.
29031 (chkp_generate_extern_var_bounds): Likewise.
29032 (chkp_get_bounds_for_decl_addr): Likewise.
29033 * tree-chrec.c (chrec_apply): Likewise.
29034 * tree-data-ref.c (dump_data_reference): Likewise.
29035 (dump_subscript): Likewise.
29036 (dump_data_dependence_relation): Likewise.
29037 (analyze_overlapping_iterations): Likewise.
29038 * tree-inline.c (expand_call_inline): Likewise.
29039 (tree_function_versioning): Likewise.
29040 * tree-into-ssa.c (dump_defs_stack): Likewise.
29041 (dump_currdefs): Likewise.
29042 (dump_names_replaced_by): Likewise.
29043 (dump_update_ssa): Likewise.
29044 (update_ssa): Likewise.
29045 * tree-object-size.c (pass_object_sizes::execute): Likewise.
29046 * tree-parloops.c (build_new_reduction): Likewise.
29047 (try_create_reduction_list): Likewise.
29048 (ref_conflicts_with_region): Likewise.
29049 (oacc_entry_exit_ok_1): Likewise.
29050 (oacc_entry_exit_single_gang): Likewise.
29051 * tree-pretty-print.h: Likewise.
29052 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
29053 (get_scalar_evolution): Likewise.
29054 (add_to_evolution): Likewise.
29055 (get_loop_exit_condition): Likewise.
29056 (analyze_evolution_in_loop): Likewise.
29057 (analyze_initial_condition): Likewise.
29058 (analyze_scalar_evolution): Likewise.
29059 (instantiate_scev): Likewise.
29060 (number_of_latch_executions): Likewise.
29061 (gather_chrec_stats): Likewise.
29062 (final_value_replacement_loop): Likewise.
29063 (scev_const_prop): Likewise.
29064 * tree-sra.c (dump_access): Likewise.
29065 (disqualify_candidate): Likewise.
29066 (create_access): Likewise.
29067 (reject): Likewise.
29068 (maybe_add_sra_candidate): Likewise.
29069 (create_access_replacement): Likewise.
29070 (analyze_access_subtree): Likewise.
29071 (analyze_all_variable_accesses): Likewise.
29072 (sra_modify_assign): Likewise.
29073 (initialize_constant_pool_replacements): Likewise.
29074 (find_param_candidates): Likewise.
29075 (decide_one_param_reduction): Likewise.
29076 (replace_removed_params_ssa_names): Likewise.
29077 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
29078 * tree-ssa-copy.c (dump_copy_of): Likewise.
29079 (copy_prop_visit_cond_stmt): Likewise.
29080 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
29081 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
29082 (record_equivalences_from_stmt): Likewise.
29083 * tree-ssa-dse.c (compute_trims): Likewise.
29084 (delete_dead_call): Likewise.
29085 (delete_dead_assignment): Likewise.
29086 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
29087 (forward_propagate_into_cond): Likewise.
29088 (pass_forwprop::execute): Likewise.
29089 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
29090 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
29092 (move_computations_worker): Likewise.
29093 (execute_sm): Likewise.
29094 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
29095 (remove_exits_and_undefined_stmts): Likewise.
29096 (remove_redundant_iv_tests): Likewise.
29097 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
29098 (adjust_iv_update_pos): Likewise.
29099 * tree-ssa-math-opts.c (bswap_replace): Likewise.
29100 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
29101 (value_replacement): Likewise.
29102 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
29103 * tree-ssa-pre.c (print_pre_expr): Likewise.
29104 (get_representative_for): Likewise.
29105 (create_expression_by_pieces): Likewise.
29106 (insert_into_preds_of_block): Likewise.
29107 (eliminate_insert): Likewise.
29108 (eliminate_dom_walker::before_dom_children): Likewise.
29109 (eliminate): Likewise.
29110 (remove_dead_inserted_code): Likewise.
29111 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
29112 * tree-ssa-reassoc.c (get_rank): Likewise.
29113 (eliminate_duplicate_pair): Likewise.
29114 (eliminate_plus_minus_pair): Likewise.
29115 (eliminate_not_pairs): Likewise.
29116 (undistribute_ops_list): Likewise.
29117 (eliminate_redundant_comparison): Likewise.
29118 (update_range_test): Likewise.
29119 (optimize_range_tests_var_bound): Likewise.
29120 (optimize_vec_cond_expr): Likewise.
29121 (rewrite_expr_tree): Likewise.
29122 (rewrite_expr_tree_parallel): Likewise.
29123 (linearize_expr): Likewise.
29124 (break_up_subtract): Likewise.
29125 (linearize_expr_tree): Likewise.
29126 (attempt_builtin_powi): Likewise.
29127 (attempt_builtin_copysign): Likewise.
29128 (transform_stmt_to_copy): Likewise.
29129 (transform_stmt_to_multiply): Likewise.
29130 (dump_ops_vector): Likewise.
29131 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
29132 (print_scc): Likewise.
29133 (set_ssa_val_to): Likewise.
29134 (visit_reference_op_store): Likewise.
29135 (visit_use): Likewise.
29136 (sccvn_dom_walker::before_dom_children): Likewise.
29137 (run_scc_vn): Likewise.
29138 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
29140 (expr_hash_elt::print): Likewise.
29141 (const_and_copies::pop_to_marker): Likewise.
29142 (const_and_copies::record_const_or_copy_raw): Likewise.
29143 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
29144 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
29145 (dump_predicates): Likewise.
29146 (find_uninit_use): Likewise.
29147 (warn_uninitialized_phi): Likewise.
29148 (pass_late_warn_uninitialized::execute): Likewise.
29149 * tree-ssa.c (verify_vssa): Likewise.
29150 (verify_ssa): Likewise.
29151 (maybe_optimize_var): Likewise.
29152 * tree-vrp.c (dump_value_range): Likewise.
29153 (dump_all_value_ranges): Likewise.
29154 (dump_asserts_for): Likewise.
29155 (register_edge_assert_for_2): Likewise.
29156 (vrp_visit_cond_stmt): Likewise.
29157 (vrp_visit_switch_stmt): Likewise.
29158 (vrp_visit_stmt): Likewise.
29159 (vrp_visit_phi_node): Likewise.
29160 (simplify_cond_using_ranges_1): Likewise.
29161 (fold_predicate_in): Likewise.
29162 (evrp_dom_walker::before_dom_children): Likewise.
29163 (evrp_dom_walker::push_value_range): Likewise.
29164 (evrp_dom_walker::pop_value_range): Likewise.
29165 (execute_early_vrp): Likewise.
29167 2017-05-16 Richard Biener <rguenther@suse.de>
29169 * dwarf2out.c (loc_list_from_tree_1): Do not create
29170 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
29172 2017-05-16 Richard Biener <rguenther@suse.de>
29174 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
29176 (note_variable_value_in_expr): If we resolved the decl ref
29177 do not push to the stack.
29179 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
29181 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
29182 operations in fast-math mode.
29183 (vaddq_f16): Likewise.
29184 (vmul_f16): Likewise.
29185 (vmulq_f16): Likewise.
29186 (vsub_f16): Likewise.
29187 (vsubq_f16): Likewise.
29188 * config/arm/neon.md (add<mode>3): New.
29190 (fma:<VH:mode>3): New. Also remove outdated comment.
29193 2017-05-16 Martin Liska <mliska@suse.cz>
29197 * ipa-devirt.c (warn_types_mismatch): Fix typo.
29198 (odr_types_equivalent_p): Likewise.
29200 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
29202 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
29204 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
29207 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
29208 non-interunit SSE move alternatives with '?'.
29209 (zero-extendsidi peephole2): New peephole to skip intermediate
29210 general register in SSE zero-extend sequence.
29212 2017-05-15 Jeff Law <law@redhat.com>
29214 * reorg.c (relax_delay_slots): Create a new variable to hold
29215 the temporary target rather than clobbering TARGET_LABEL.
29217 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
29218 missing argument to extract_bit_field call.
29219 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
29221 2017-05-15 Martin Liska <mliska@suse.cz>
29224 * gcc.c (process_command): Do not allow empty argument of -o option.
29226 2017-05-15 Renlin Li <renlin.li@arm.com>
29228 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
29229 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
29230 * config/aarch64/constraints.md (Usf): Add long call check.
29231 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
29232 (call_value): Likewise.
29233 (sibcall): Likewise.
29234 (sibcall_value): Likewise.
29236 (call_value_insn): New.
29237 (sibcall_insn): Update rtx pattern.
29238 (sibcall_value_insn): Likewise.
29239 (call_internal): Remove.
29240 (call_value_internal): Likewise.
29241 (sibcall_internal): Likewise.
29242 (sibcall_value_internal): Likewise.
29243 (call_reg): Likewise.
29244 (call_symbol): Likewise.
29245 (call_value_reg): Likewise.
29246 (call_value_symbol): Likewise.
29248 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
29251 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
29253 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
29255 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
29256 compatible with CCGOCmode and with CCZmode.
29258 2017-05-14 Martin Sebor <msebor@redhat.com>
29260 PR middle-end/77671
29261 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
29262 (gimple_fold_builtin_snprintf): Same.
29263 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
29264 (gimple_fold_builtin_snprintf): Same.
29265 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
29266 of character types.
29267 (is_call_safe): New function.
29268 (try_substitute_return_value): Call it.
29269 (try_simplify_call): New function.
29270 (pass_sprintf_length::handle_gimple_call): Call it.
29272 2017-05-14 Martin Sebor <msebor@redhat.com>
29274 PR middle-end/80669
29275 * builtins.c (expand_builtin_stpncpy): Simplify.
29277 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
29279 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
29280 * config/i386/i386.h
29281 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
29282 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
29283 (struct machine_function): Add new members call_ms2sysv,
29284 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
29285 (struct machine_frame_state): New fields sp_realigned and
29286 sp_realigned_offset.
29287 * config/i386/i386.c
29288 (enum xlogue_stub): New enum.
29289 (enum xlogue_stub_sets): New enum.
29290 (class xlogue_layout): New class.
29291 (struct ix86_frame): New fields stack_realign_allocate_offset,
29292 stack_realign_offset and outlined_save_offset. Modify comments to
29293 detail stack layout when using out-of-line stubs.
29294 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
29295 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
29296 -mcall-ms2sysv-xlogues.
29297 (stub_managed_regs): New static variable.
29298 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
29299 registers managed by out-of-line stub.
29300 (disable_call_ms2sysv_xlogues): New function.
29301 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
29302 m->call_ms2sysv when appropriate and compute frame layout for
29304 (sp_valid_at, fp_valid_at): New inline functions.
29305 (choose_basereg): New function.
29306 (choose_baseaddr): Add align parameter, use choose_basereg and modify
29308 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
29309 Use align parameter of choose_baseaddr to generated aligned SSE movs
29311 (pro_epilogue_adjust_stack): Modify to track
29312 machine_frame_state::sp_realigned.
29313 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
29314 (ix86_nsaved_sseregs): Likewise.
29315 (ix86_emit_save_regs): Likewise.
29316 (ix86_emit_save_regs_using_mov): Likewise.
29317 (ix86_emit_save_sse_regs_using_mov): Likewise.
29318 (get_scratch_register_on_entry): Likewise.
29319 (gen_frame_set): New function.
29320 (gen_frame_load): Likewise.
29321 (gen_frame_store): Likewise.
29322 (emit_outlined_ms2sysv_save): Likewise.
29323 (emit_outlined_ms2sysv_restore): Likewise.
29324 (ix86_expand_prologue): Modify stack re-alignment code and call
29325 emit_outlined_ms2sysv_save when appropriate.
29326 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
29327 parameter rtx_insn *insn, which allows the function to be used to only
29328 generate the notes.
29329 (ix86_expand_epilogue): Modify validity checks of frame and stack
29330 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
29331 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
29332 * config/i386/predicates.md
29333 (save_multiple): New predicate.
29334 (restore_multiple): Likewise.
29335 * config/i386/sse.md
29336 (save_multiple<mode>): New pattern.
29337 (save_multiple_realign<mode>): Likewise.
29338 (restore_multiple<mode>): Likewise.
29339 (restore_multiple_and_return<mode>): Likewise.
29340 (restore_multiple_leave_return<mode>): Likewise.
29341 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
29343 2017-05-14 Julia Koval <julia.koval@intel.com>
29345 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
29346 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
29347 (__builtin_ia32_xsetbv): New builtins.
29348 * config/i386/i386.c (ix86_expand_special_args_builtin):
29350 (ix86_expand_builtin): Special expand for new intrinsics.
29351 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
29352 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
29353 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
29355 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29357 * cfganal.c (inverted_post_order_compute): Change argument type
29359 * cfganal.h (inverted_post_order_compute): Adjust prototype.
29360 * df-core.c (rest_of_handle_df_initialize): Adjust.
29361 (rest_of_handle_df_finish): Likewise.
29362 (df_analyze_1): Likewise.
29363 (df_analyze): Likewise.
29364 (loop_inverted_post_order_compute): Change argument to be a vec *.
29365 (df_analyze_loop): Adjust.
29366 (df_get_n_blocks): Likewise.
29367 (df_get_postorder): Likewise.
29368 * df.h (struct df_d): Change field to be a vec.
29369 * lcm.c (compute_laterin): Adjust.
29370 (compute_available): Likewise.
29371 * lra-lives.c (lra_create_live_ranges_1): Likewise.
29372 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
29373 * tree-ssa-pre.c (compute_antic): Likewise.
29375 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29377 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
29378 (depth_first_search::depth_first_search): Change structure init
29379 function to this constructor.
29380 (depth_first_search::add_bb): Rename function to this member.
29381 (depth_first_search::execute): Likewise.
29382 (flow_dfs_compute_reverse_finish): Adjust.
29384 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29386 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
29387 (longest_simple_path): Likewise.
29388 * shrink-wrap.c (spread_components): Likewise.
29389 (disqualify_problematic_components): Likewise.
29390 (emit_common_heads_for_components): Likewise.
29391 (emit_common_tails_for_components): Likewise.
29392 (insert_prologue_epilogue_for_components): Likewise.
29394 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29396 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
29399 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29401 * df-core.c (df_set_blocks): Start using auto_bitmap.
29402 (df_compact_blocks): Likewise.
29403 * df-problems.c (df_rd_confluence_n): Likewise.
29404 * df-scan.c (df_insn_rescan_all): Likewise.
29405 (df_process_deferred_rescans): Likewise.
29406 (df_update_entry_block_defs): Likewise.
29407 (df_update_exit_block_uses): Likewise.
29408 (df_entry_block_bitmap_verify): Likewise.
29409 (df_exit_block_bitmap_verify): Likewise.
29410 (df_scan_verify): Likewise.
29411 * lra-constraints.c (lra_constraints): Likewise.
29412 (undo_optional_reloads): Likewise.
29413 (lra_undo_inheritance): Likewise.
29414 * lra-remat.c (calculate_gen_cands): Likewise.
29415 (do_remat): Likewise.
29416 * lra-spills.c (assign_spill_hard_regs): Likewise.
29417 (spill_pseudos): Likewise.
29418 * tree-ssa-pre.c (bitmap_set_and): Likewise.
29419 (bitmap_set_subtract_values): Likewise.
29421 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29423 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
29424 management with auto_bitmap.
29425 (fix_inter_tick): Likewise.
29426 (fix_recovery_deps): Likewise.
29427 * ira.c (add_store_equivs): Likewise.
29428 (find_moveable_pseudos): Likewise.
29429 (split_live_ranges_for_shrink_wrap): Likewise.
29430 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
29431 (rtx_reuse_manager::seen_def_p): Likewise.
29432 (rtx_reuse_manager::set_seen_def): Likewise.
29433 * print-rtl.h (class rtx_reuse_manager): Likewise.
29435 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29437 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
29439 (migrate_btr_def): Likewise.
29440 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
29441 * df-core.c (loop_post_order_compute): Likewise.
29442 (loop_inverted_post_order_compute): Likewise.
29443 * hsa-common.h: Likewise.
29444 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
29445 * init-regs.c (initialize_uninitialized_regs): Likewise.
29446 * ipa-inline.c (resolve_noninline_speculation): Likewise.
29447 (inline_small_functions): Likewise.
29448 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
29449 * ira.c (combine_and_move_insns): Likewise.
29450 (build_insn_chain): Likewise.
29451 * loop-invariant.c (find_invariants): Likewise.
29452 * lower-subreg.c (propagate_pseudo_copies): Likewise.
29453 * predict.c (tree_predict_by_opcode): Likewise.
29454 (predict_paths_leading_to): Likewise.
29455 (predict_paths_leading_to_edge): Likewise.
29456 (estimate_loops_at_level): Likewise.
29457 (estimate_loops): Likewise.
29458 * shrink-wrap.c (try_shrink_wrapping): Likewise.
29459 (spread_components): Likewise.
29460 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
29461 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
29462 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
29463 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
29464 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
29465 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
29466 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
29467 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
29468 (mark_threaded_blocks): Likewise.
29469 (thread_through_all_blocks): Likewise.
29470 * tree-ssa.c (verify_ssa): Likewise.
29471 (execute_update_addresses_taken): Likewise.
29472 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
29474 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29476 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
29478 (post_order_compute): Likewise.
29479 (inverted_post_order_compute): Likewise.
29480 (pre_and_rev_post_order_compute_fn): Likewise.
29482 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29484 * genrecog.c (int_set::int_set): Explicitly construct our
29485 auto_vec base class.
29486 * vec.h (auto_vec::auto_vec): New constructor.
29488 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29490 * bitmap.h (class auto_bitmap): New constructor taking
29491 bitmap_obstack * argument.
29493 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29495 * bitmap.h (class auto_bitmap): Change type of m_bits to
29496 bitmap_head, and adjust ctor / dtor and member operators.
29498 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
29500 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
29501 when returned register mode doesn't match original mode.
29503 2017-05-12 Jeff Law <law@redhat.com>
29504 Jakub Jelinek <jakub@redhat.com>
29506 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
29507 we look for cc setter after the compare-elim changes.
29508 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
29509 within the vector to match what compare-elim now expects.
29510 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
29511 (xorsi3_flags, one_cmplsi2_flags): Likewise.
29513 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
29514 after the compare-elim changes.
29515 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
29516 the vector to match what compare-elim now expects.
29517 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
29518 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
29519 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
29520 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
29521 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
29523 * config/visium/visium.c (single_set_and_flags): Fix where
29524 we look for cc setter after the compare-elim changes.
29525 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
29526 with the vector to match what compare-elim now expects.
29527 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
29528 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
29529 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
29530 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
29531 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
29532 (neg<mode>2_insn_set_overflow): Likewise.
29534 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
29536 PR middle-end/79794
29537 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
29538 maybe_expand_insn call, set ops[0].target. If still set after call,
29539 set alt_rtl. Add extra arg to recursive calls.
29540 (extract_bit_field): Add alt_rtl argument. Pass to
29542 * expmed.h (extract_bit_field): Fix prototype.
29543 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
29544 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
29545 to extract_bit_field_calls.
29546 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
29547 Pass alt_rtl to extract_bit_field calls.
29548 * calls.c (store_unaligned_arguments_into_psuedos)
29549 load_register_parameters): Pass extra NULL to extract_bit_field calls.
29550 * optabs.c (maybe_legitimize_operand): Clear op->target when call
29552 * optabs.h (struct expand_operand): Add target bitfield.
29554 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29556 * compare-elim.c (try_eliminate_compare): Canonicalize
29557 operation with embedded compare to
29558 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
29559 (set (reg) (operation)].
29561 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
29563 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29566 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
29567 cost of adding a carry flag for ADC instruction.
29568 [case MINUS]: Ignore the cost of subtracting a carry flag
29569 for SBB instruction.
29571 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
29573 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
29575 * config/rs6000/bmiintrin.h: New file.
29576 * config/rs6000/bmi2intrin.h: New file.
29577 * config/rs6000/x86intrin.h: New file.
29579 2017-05-12 Jeff Law <law@redhat.com>
29581 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
29584 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
29586 PR middle-end/80707
29587 * tree-cfg.c: Remove cfg edges of unreachable case statements.
29589 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29591 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29592 early expansion of vector divide builtins.
29593 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
29594 builtins identified as having unsigned arguments.
29596 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29598 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
29599 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
29600 expansion of vector logical operations (and, andc, or, xor,
29603 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29605 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
29606 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
29608 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29610 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29611 early GIMPLE expansion of vector multiplies.
29613 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29615 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
29616 TARGET_HAVE_MOVT conditional.
29617 (movt splitter): Likewise.
29619 2017-05-12 Richard Biener <rguenther@suse.de>
29621 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
29622 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29623 Fold all stmts not inplace.
29625 2017-05-12 Richard Biener <rguenther@suse.de>
29627 PR tree-optimization/80713
29628 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
29629 inserted_exprs bit for not removed stmts.
29631 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
29633 PR middle-end/69921
29634 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
29635 parallelized" attribute for parallelized OpenACC kernels.
29636 * omp-offload.c (execute_oacc_device_lower): Use it.
29638 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
29639 Set "oacc kernels" attribute.
29640 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
29641 parameter. Adjust all users.
29642 (oacc_fn_attrib_kernels_p): Remove function.
29643 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
29644 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
29645 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
29646 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
29647 assert "oacc kernels" attribute is set.
29649 2017-05-11 Carl Love <cel@us.ibm.com>
29651 * config/rs6000/rs6000-c: Add support for built-in functions
29652 vector unsigned char vec_popcnt (vector signed char)
29653 vector unsigned char vec_popcnt (vector unsigned char)
29654 vector unsigned short vec_popcnt (vector signed short)
29655 vector unsigned short vec_popcnt (vector unsigned short)
29656 vector unsigned int vec_popcnt (vector signed int)
29657 vector unsigned int vec_popcnt (vector unsigned int)
29658 vector unsigned long long vec_popcnt (vector signed long long)
29659 vector unsigned long long vec_popcnt (vector unsigned long long)
29660 vector signed long long vec_slo (vector signed long long,
29661 vector signed char)
29662 vector signed long long vec_slo (vector signed long long,
29663 vector unsigned char)
29664 vector unsigned long long vec_slo (vector unsigned long long,
29665 vector signed char)
29666 vector unsigned long long vec_slo (vector unsigned long long,
29667 vector unsigned char)
29668 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
29669 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
29670 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
29671 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
29672 * doc/extend.texi: Update the built-in documentation file for the
29673 new built-in functions.
29675 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
29677 * attribs.h (sorted_attr_string): Move machine independent
29678 functions for target clone support from the i386 port to common
29679 code. Rename ix86_function_versions to common_function_versions.
29680 Rename make_name to make_unique_name.
29681 (common_function_versions): Likewise.
29682 (make_unique_name): Likewise.
29683 (make_dispatcher_decl): Likewise.
29684 (is_function_default_version): Likewise.
29685 * attribs.c (attr_strcmp): Likewise.
29686 (sorted_attr_string): Likewise.
29687 (common_function_versions): Likewise.
29688 (make_unique_name): Likewise.
29689 (make_dispatcher_decl): Likewise.
29690 (is_function_default_version): Likewise.
29691 * config/i386/i386.c (attr_strcmp): Likewise.
29692 (sorted_attr_string): Likewise.
29693 (ix86_function_versions): Likewise.
29694 (make_name): Likewise.
29695 (make_dispatcher_decl): Likewise.
29696 (is_function_default_version): Likewise.
29697 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
29699 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29702 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
29703 Account for direct move costs for vec_construct of integer
29706 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
29709 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
29710 (UNSPEC_STX_ATOMIC): Ditto.
29711 (loaddi_via_sse): New insn.
29712 (storedi_via_sse): Ditto.
29713 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
29714 Update corresponding peephole2 patterns.
29715 (atomic_storedi_fpu): Ditto.
29717 2017-05-11 Julia Koval <julia.koval@intel.com>
29719 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
29720 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
29722 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
29723 (__builtin_ia32_rsqrt14ss_mask): New builtins.
29724 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
29726 2017-05-11 Nathan Sidwell <nathan@acm.org>
29728 * graphite-poly.c: Include dumpfile.h.
29730 * dumpfle.h (dump_function): Declare here ...
29731 * tree-dump.h (dump_function): ... not here.
29732 * dumpfile.c: #include tree-cfg.h.
29733 (dump_function): Move here from ...
29734 * tree-dump.c (dump_function): ... here.
29735 * gimplify.c: #include splay-tree.h, not tree-dump.h.
29736 * graphite-poly.c: Don't include tree-dump.h.
29737 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
29738 * print-tree.c: Likewise.
29739 * stor-layout.c: Likewise.
29740 * tree-nested.c: Likewise.
29742 * dumpfile.c (dump_start): Use TDF_FLAGS.
29743 (dump_enable_all): Fix TDF_KIND check thinko.
29745 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
29747 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29748 array entries to represent two legal parameterizations of the
29749 overloaded __builtin_cmpb function, as represented by the
29750 P6_OV_BUILTIN_CMPB constant.
29751 (altivec_resolve_overloaded_builtin): Add special case handling
29752 for the __builtin_cmpb function, as represented by the
29753 P6_OV_BUILTIN_CMPB constant.
29754 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
29755 (BU_P6_64BIT_2): New macro.
29756 (BU_P6_OVERLOAD_2): New macro
29757 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
29758 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
29759 (CMPB): Add overload support to represent both 32-bit and 64-bit
29760 compare-bytes function.
29761 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
29762 support for TARGET_CMPB.
29763 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
29764 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29765 documentation of the __builtin_cmpb overloaded built-in function.
29767 2017-05-11 Richard Biener <rguenther@suse.de>
29769 PR tree-optimization/80705
29770 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
29771 bases are not vectorizable.
29773 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29775 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
29776 when counting register pressure.
29778 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29780 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
29781 (iv_ca_more_deps): Renamed to ...
29782 (iv_ca_compare_deps): ... this.
29783 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
29785 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29787 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
29789 (determine_group_iv_costs): ... here.
29790 (find_inv_vars_cb): Record inv var if it's not recorded before.
29792 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29794 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
29795 (get_shiftadd_cost): Ditto.
29797 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29799 * tree-ssa-address.c: Include header file.
29800 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
29802 (add_to_parts): Refactor.
29803 (addr_to_parts): New parameter. Update use of move_hint_to_base.
29804 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
29807 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29809 PR tree-optimization/53090
29810 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
29812 (extract_cond_operands): Detect condition with IV on both sides
29813 and return COMP_IV_EXPR_2.
29814 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
29815 (rewrite_use_compare): Simplify by removing call to function
29816 extract_cond_operands.
29818 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29820 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
29821 (extract_cond_operands): Detect condition comparing against non-
29822 invariant bound and return appropriate enum value.
29823 (find_interesting_uses_cond): Update use of extract_cond_operands.
29824 Handle its return value accordingly.
29825 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
29827 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29829 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
29830 nonlinear iv_use computation in loop invariant sensitive way.
29832 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29834 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
29835 (find_iv_candidates): Call relate_compare_use_with_all_cands.
29837 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29839 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
29840 (dump_cand): Support iv_cand.inv_exprs.
29841 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
29843 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
29846 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29848 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
29850 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
29851 as local function. Include necessary header files.
29852 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
29854 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29856 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
29858 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29860 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
29861 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
29862 RSHIFT_EXPR and BIT_NOT_EXPR.
29864 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29866 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
29867 (adjust_setup_cost): New parameter supporting round up adjustment.
29868 (struct address_cost_data): Delete.
29869 (force_expr_to_var_cost): Don't bound cost with spill_cost.
29870 (split_address_cost, ptr_difference_cost): Delete.
29871 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
29872 (struct ainc_cost_data): New struct.
29873 (get_address_cost_ainc): New function.
29874 (get_address_cost, get_computation_cost): Reimplement.
29875 (determine_group_iv_cost_address): Record inv_expr for all uses of
29877 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
29878 (iv_ca_has_deps): Reimplemented to ...
29879 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
29881 (iv_ca_extend): Call iv_ca_more_deps.
29883 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29885 * tree-ssa-address.c (struct mem_address): Move to header file.
29886 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
29887 * tree-ssa-address.h (struct mem_address): Move from C file.
29888 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
29890 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29892 * tree-affine.h (aff_combination_type): New interface.
29893 (aff_combination_zero_p): Remove static.
29894 (aff_combination_const_p): New interface.
29895 (aff_combination_singleton_var_p): New interfaces.
29897 2017-05-11 Richard Biener <rguenther@suse.de>
29899 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29900 Skip unreachable blocks and destinations.
29901 (eliminate): Move stmt removal and fixup ...
29902 (fini_eliminate): ... here. Skip inserted exprs.
29903 (pass_pre::execute): Move fini_pre after fini_eliminate.
29904 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
29905 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
29906 PRE to get rid of dead code that has invalid SSA form and
29907 split critical edges again.
29909 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29911 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29913 2017-05-11 Richard Biener <rguenther@suse.de>
29915 * passes.c (execute_function_todo): Verify loops if they are
29916 said to be up-to-date.
29917 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
29918 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
29920 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
29923 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
29924 handle calling assemble_external ourself.
29927 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
29928 modes with zero size. Enhance comment.
29930 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29932 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
29933 built-ins for vec_xl and vec_xst with short and char pointer
29936 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29938 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
29939 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
29940 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
29941 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
29942 (_mm_maskz_min_round_ss): New intrinsics.
29943 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
29944 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
29945 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
29946 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
29947 (__builtin_ia32_minss_mask_round): New builtins.
29948 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29949 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29950 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
29952 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
29953 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
29955 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
29958 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29960 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
29961 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
29962 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
29963 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
29964 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
29965 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
29966 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
29967 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29968 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29969 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
29970 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
29971 (__builtin_ia32_mulss_mask_round): New builtins.
29972 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29973 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29974 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
29976 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
29977 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29979 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29982 2017-05-10 Julia Koval <julia.koval@intel.com>
29984 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
29985 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
29986 (_mm256_setr_m128i): New intrinsics.
29988 2017-05-10 Julia Koval <julia.koval@intel.com>
29990 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
29991 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
29992 (_mm_maskz_rcp14_ss): New intrinsics.
29993 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
29994 (__builtin_ia32_rcp14ss_mask): New builtins.
29995 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
29997 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
29999 PR tree-optimization/51513
30000 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30001 (assert_unreachable_fallthru_edge_p): Use it.
30002 (group_case_labels_stmt): Likewise.
30003 * tree-cfg.h: Prototype it.
30004 * stmt.c: Include cfghooks.h and tree-cfg.h.
30005 (emit_case_dispatch_table) <gap_label>: New local variable.
30006 Use it to fill dispatch table gaps.
30007 Test for default_label before updating probabilities.
30008 (expand_case) <default_label>: Remove unneeded initialization.
30009 Test for unreachable default case statement and remove its edge.
30010 Set default_label accordingly.
30011 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30013 2017-05-10 Carl Love <cel@us.ibm.com>
30015 * config/rs6000/rs6000-c: Add support for built-in functions
30016 vector signed char vec_neg (vector signed char)
30017 vector signed short int vec_neg (vector short int)
30018 vector signed int vec_neg (vector signed int)
30019 vector signed long long vec_neg (vector signed long long)
30020 vector float vec_neg (vector float)
30021 vector double vec_neg (vector double)
30022 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30024 * config/rs6000/altivec.h: Add define for vec_neg
30025 * doc/extend.texi: Update the built-in documentation for the
30026 new built-in functions.
30028 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30030 PR tree-optimization/77644
30031 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30033 2017-05-10 Nathan Sidwell <nathan@acm.org>
30035 * dumpfile.h (TDI_lang_all): New.
30036 (TDF_KIND): New. Renumber others
30037 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30039 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30041 (get_dump_file_name): Adjust suffix generation.
30042 (dump_enable_all): Use TDF_KIND.
30043 * doc/invoke.texi (-fdump-lang-all): Document.
30045 * dumpfile.h: Tabify.
30047 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
30050 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
30051 Move member access before delete.
30053 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
30055 * tree-inline.c (expand_call_inline): Split block at stmt
30058 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
30061 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
30062 are now unused after splitting mov{sf,sd}_hardfloat.
30063 (f32_lr2): Likewise.
30064 (f32_lm): Likewise.
30065 (f32_lm2): Likewise.
30066 (f32_li): Likewise.
30067 (f32_li2): Likewise.
30068 (f32_lv): Likewise.
30069 (f32_sr): Likewise.
30070 (f32_sr2): Likewise.
30071 (f32_sm): Likewise.
30072 (f32_sm2): Likewise.
30073 (f32_si): Likewise.
30074 (f32_si2): Likewise.
30075 (f32_sv): Likewise.
30076 (f32_dm): Likewise.
30077 (f32_vsx): Likewise.
30078 (f32_av): Likewise.
30079 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
30080 For movsf, order stores so the VSX stores occur before the GPR
30081 store which encourages the register allocator to use a traditional
30082 FPR instead of a GPR. For movsd, order the stores so that the GPR
30083 store comes before the VSX stores to allow the power6 to work.
30084 This is due to the power6 not having a 32-bit integer store
30085 instruction from a FPR.
30086 (movsf_hardfloat): Likewise.
30087 (movsd_hardfloat): Likewise.
30089 2017-05-09 Martin Sebor <msebor@redhat.com>
30091 PR translation/80280
30092 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
30095 PR translation/80280
30096 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
30097 data member added in r247778.
30098 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
30100 2017-05-09 Nathan Sidwell <nathan@acm.org>
30102 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
30104 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
30107 2017-05-09 Marek Polacek <polacek@redhat.com>
30109 * doc/invoke.texi: Fix typo.
30111 2017-05-09 Richard Biener <rguenther@suse.de>
30113 * tree-vrp.c (vrp_val_is_max): Adjust comment.
30114 (vrp_val_is_min): Likewise.
30115 (set_value_range_to_value): Likewise.
30116 (set_value_range_to_nonnegative): Likewise.
30117 (gimple_assign_nonzero_p): Likewise.
30118 (gimple_stmt_nonzero_p): Likewise.
30119 (vrp_int_const_binop): Likewise. Remove unreachable case.
30120 (adjust_range_with_scev): Adjust comments.
30121 (compare_range_with_value): Likewise.
30122 (extract_range_from_phi_node): Likewise.
30123 (test_for_singularity): Likewise.
30125 2017-05-09 Richard Biener <rguenther@suse.de>
30127 * tree-vrp.c (get_single_symbol): Add assert that we don't
30128 get overflowed constants as invariant part.
30129 (compare_values_warnv): Add comment before the TREE_NO_WARNING
30130 checks. Use wi::cmp instead of recursing for integer constants.
30131 (compare_values): Just ignore whether we assumed undefined
30132 overflow instead of failing the compare.
30133 (extract_range_for_var_from_comparison_expr): Add comment before the
30134 TREE_NO_WARNING sets.
30135 (test_for_singularity): Likewise.
30136 (extract_range_from_comparison): Do not disable optimization
30137 when we assumed undefined overflow.
30138 (extract_range_basic): Remove init of unused var.
30140 2017-05-09 Richard Biener <rguenther@suse.de>
30142 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
30143 (extract_range_from_multiplicative_op_1): Adjust.
30144 (extract_range_from_binary_expr_1): Use int_const_binop.
30146 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
30149 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
30150 rs6000_store_data_bypass_p in seven define_bypass directives and
30151 in several comments.
30152 * config/rs6000/rs6000-protos.h: Add prototype for
30153 rs6000_store_data_bypass_p function.
30154 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
30155 function implements slightly different (rs6000-specific) semantics
30156 than store_data_bypass_p, returning false rather than aborting
30157 with assertion error when arguments do not satisfy the
30158 requirements of store data bypass.
30159 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
30160 rs6000_store_data_bypass_p.
30162 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
30164 * config/xtensa/xtensa-protos.h
30165 (xtensa_initial_elimination_offset): New declaration.
30166 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30167 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
30168 macro definition, add case for FRAME_POINTER_REGNUM when
30169 FRAME_GROWS_DOWNWARD.
30170 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
30171 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
30172 xtensa_initial_elimination_offset.
30174 2017-05-08 Nathan Sidwell <nathan@acm.org>
30176 * doc/invoke.texi: Alphabetize -fdump options.
30178 2017-05-08 Martin Sebor <msebor@redhat.com>
30180 PR translation/80280
30181 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
30183 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
30185 * target.def (compute_frame_layout): New optional target hook.
30186 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
30187 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
30188 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
30190 * reload1.c (verify_initial_elim_offsets): Likewise.
30191 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
30192 (use_simple_return_p): Call arm_compute_frame_layout if needed.
30193 (arm_get_frame_offsets): Split up into this ...
30194 (arm_compute_frame_layout): ... and this function.
30196 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
30198 * config/aarch64/constraints.md (Usa): New constraint.
30199 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
30201 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
30203 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
30204 with_multilib_list after it has been checked.
30206 2017-05-08 Richard Biener <rguenther@suse.de>
30208 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
30209 (bitmap_set_subtract_values): Likewise.
30211 2017-05-08 Richard Biener <rguenther@suse.de>
30213 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
30214 (gimple_assign_nonzero): ... this and remove strict_overflow_p
30216 (gimple_stmt_nonzero_warnv_p): Rename to ...
30217 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
30219 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
30220 (extract_range_basic): Adjust, do not disable propagation on
30221 strict overflow sensitive simplification.
30222 (vrp_visit_cond_stmt): Likewise.
30224 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
30226 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
30227 body size unconditionally.
30229 2017-05-07 Jeff Law <law@redhat.com>
30232 2017-05-06 Jeff Law <law@redhat.com>
30233 PR tree-optimization/78496
30234 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30237 PR tree-optimization/78496
30238 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30239 (simplify_stmt_using_ranges): Call it.
30240 (vrp_dom_walker::before_dom_children): Extract equivalences
30241 from an ASSERT_EXPR with an equality comparison against a
30244 2017-05-06 Jeff Law <law@redhat.com>
30246 PR tree-optimization/78496
30247 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30250 PR tree-optimization/78496
30251 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30252 (simplify_stmt_using_ranges): Call it.
30253 (vrp_dom_walker::before_dom_children): Extract equivalences
30254 from an ASSERT_EXPR with an equality comparison against a
30257 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30259 * lra-constraints.c (lra_copy_reg_equiv): New function.
30260 (split_reg): Use it to copy equivalence information from the
30261 original register to the spill register.
30263 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30265 PR rtl-optimization/75964
30266 * simplify-rtx.c (simplify_const_relational_operation): Remove
30267 invalid handling of comparisons of integer ABS.
30269 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
30271 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
30272 initialize to zero.
30273 (init_regs): Remove declaration.
30274 (function_arg_advance_32): Initialize error_p as boolean variable.
30276 2017-05-05 Nathan Sidwell <nathan@acm.org>
30278 * store-motion.c (remove_reachable_equiv_notes): Reformat long
30279 lines. Use for (;;).
30281 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30283 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
30284 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
30285 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
30286 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
30287 VF=2 that require versioning.
30289 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30291 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
30294 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30296 * diagnostic.h (diagnostic_override_option_index): Convert from
30297 macro to inline function.
30299 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30301 * diagnostic.c (last_module_changed_p): New function.
30302 (set_last_module): New function.
30303 (diagnostic_report_current_module): Convert macro usage to
30304 the above functions.
30305 * diagnostic.h (diagnostic_context::last_module): Strengthen
30306 from const line_map * to const line_map_ordinary *.
30307 (diagnostic_last_module_changed): Delete macro.
30308 (diagnostic_set_last_module): Delete macro.
30310 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30312 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
30313 with diagnostic_report_diagnostic.
30314 (diagnostic_n_impl_richloc): Likewise.
30315 * diagnostic.h (report_diagnostic): Delete macro.
30316 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
30317 with diagnostic_report_diagnostic.
30318 * substring-locations.c (format_warning_va): Likewise.
30320 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30322 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
30323 save/restor of format_spec. Move option-printing code to...
30324 (print_option_information): ...this new function, and
30325 reimplement by simply printing to the pretty_printer,
30326 rather than appending to the format string.
30328 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30330 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
30331 handling logic into...
30332 (update_effective_level_from_pragmas): ...this new function.
30334 2017-05-04 Andrew Waterman <andrew@sifive.com>
30336 * config/riscv/riscv.opt (mstrict-align): New option.
30337 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
30338 (SLOW_UNALIGNED_ACCESS): Define.
30339 (riscv_slow_unaligned_access): Declare.
30340 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
30342 (riscv_slow_unaligned_access): New variable.
30343 (rocket_tune_info): Set slow_unaligned_access to true.
30344 (optimize_size_tune_info): Set slow_unaligned_access to false.
30345 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
30346 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
30347 (riscv_option_override): Set riscv_slow_unaligned_access.
30348 * doc/invoke.texi: Add -mstrict-align to RISC-V.
30350 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
30352 * config/riscv/riscv.md: Unify indentation.
30354 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
30359 * config/rs6000/rs6000.md (u code attribute): Add FIX and
30361 (extendsi<mode>2): Add support for doing sign extension via
30362 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
30363 don't have ISA 3.0 instructions.
30364 (extendsi<mode>2 splitter): Likewise.
30365 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
30366 generate the normal insns since SImode can now go in vector
30367 registers. Disallow the special UNSPECs needed for previous
30368 machines to hide SImode being used. Add new insns
30369 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
30370 (fix_trunc<mode>si2_stfiwx): Likewise.
30371 (fix_trunc<mode>si2_internal): Likewise.
30372 (fixuns_trunc<mode>si2): Likewise.
30373 (fixuns_trunc<mode>si2_stfiwx): Likewise.
30374 (fctiw<u>z_<mode>_smallint): Likewise.
30375 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
30376 of floating point to 32-bit integer from doing a direct move to
30377 the GPR registers to do a store.
30378 (fctiwz_<mode>): Break long line.
30380 2017-05-05 Bin Cheng <bin.cheng@arm.com>
30382 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
30383 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
30384 (addr_list, addr_offset_valid_p): New.
30385 (split_address_groups): Check offset validity with above function.
30386 (gt-tree-ssa-loop-ivopts.h): Include header file.
30388 2017-05-05 Nathan Sidwell <nathan@acm.org>
30390 * config.gcc (arm*-*-*): Add missing 'fi'.
30392 2017-05-05 Steve Ellcey <sellcey@cavium.com>
30394 * doc/invoke.texi (-fopt-info): Explicitly say order of options
30395 included in -fopt-info does not matter.
30396 * doc/optinfo.texi (-fopt-info): Fix description of default
30397 behavour. Explicitly say order of options included in -fopt-info
30400 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
30402 * config.gcc: Allow combinations of aprofile and rmprofile values for
30403 --with-multilib-list.
30404 * config/arm/t-multilib: New file.
30405 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
30406 variables. Remove setting of ISA and floating-point ABI in
30407 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
30408 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
30409 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
30410 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
30412 * config/arm/t-rmprofile: Likewise except for the matches changes.
30413 * doc/install.texi (--with-multilib-list): Document the combination of
30414 aprofile and rmprofile values and warn about pitfalls in doing that.
30416 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30418 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
30419 (movdi_aarch64): Likewise.
30421 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30423 PR tree-optimization/80632
30424 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
30426 (build_arrays): Initialize it for virtual phis.
30427 (fix_phi_nodes): Use it for virtual phis.
30429 PR tree-optimization/80558
30430 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
30431 [x, y] op z into [x op, y op z] for op & or | if conditions
30434 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
30435 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30438 * config/arm/arm.md (use_literal_pool): Remove.
30439 (64-bit immediate split): No longer takes cost into consideration
30440 if arm_disable_literal_pool is enabled.
30441 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
30442 used when arm_disable_literal_pool is enabled.
30443 (arm_max_const_double_inline_cost): Remove use of
30444 arm_disable_literal_pool.
30445 (push_minipool_fix): Add assert.
30446 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
30447 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
30448 (no_literal_pool_sf_immediate): New.
30450 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30452 PR tree-optimization/80613
30453 * tree-ssa-dce.c (propagate_necessity): Remove cases for
30454 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
30456 2017-05-05 Richard Biener <rguenther@suse.de>
30458 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
30460 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
30462 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
30463 of this flag from insn conditions due to removal from r247495.
30465 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30467 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
30469 (arm_early_store_addr_dep_ptr): Likewise.
30470 * config/arm/aarch-common-protos.h
30471 (arm_early_load_addr_dep_ptr): Add prototype.
30472 (arm_early_store_addr_dep_ptr): Likewise.
30473 * config/arm/cortex-a53.md: Add new bypasses.
30475 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30477 * tree.c (next_type_uid): Change type to unsigned.
30478 (type_hash_canon): Decrement back next_type_uid if
30479 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
30480 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
30483 2017-05-04 Martin Sebor <msebor@redhat.com>
30485 * builtins.c: Fix a trivial typo in a comment.
30487 PR middle-end/79234
30488 * builtins.c (check_sizes): Adjust to handle reading past the end.
30489 Avoid printing excessive upper bound of ranges. Use %E to print
30490 tree nodes instead of converting them to %wu.
30491 (expand_builtin_memchr): New function.
30492 (compute_dest_size): Rename...
30493 (compute_objsize): ...to this.
30494 (expand_builtin_memcpy): Adjust.
30495 (expand_builtin_mempcpy): Adjust.
30496 (expand_builtin_strcat): Adjust.
30497 (expand_builtin_strcpy): Adjust.
30498 (check_strncat_sizes): Adjust.
30499 (expand_builtin_strncat): Adjust.
30500 (expand_builtin_strncpy): Adjust and simplify.
30501 (expand_builtin_memset): Adjust.
30502 (expand_builtin_bzero): Adjust.
30503 (expand_builtin_memcmp): Adjust.
30504 (expand_builtin): Handle memcmp.
30505 (maybe_emit_chk_warning): Check strncat just once.
30507 2017-05-04 Martin Sebor <msebor@redhat.com>
30509 PR preprocessor/79214
30510 PR middle-end/79222
30511 PR middle-end/79223
30512 * builtins.c (check_sizes): Add inlining context and issue
30513 warnings even when -Wno-system-headers is set.
30514 (check_strncat_sizes): Same.
30515 (expand_builtin_strncat): Same.
30516 (expand_builtin_memmove): New function.
30517 (expand_builtin_stpncpy): Same.
30518 (expand_builtin): Handle memmove and stpncpy.
30520 2017-05-04 Bin Cheng <bin.cheng@arm.com>
30522 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
30523 which is not used any more.
30525 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30527 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
30529 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30531 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
30532 (cortexa53_tunings): Likewise.
30533 (cortexa57_tunings): Likewise.
30534 (cortexa72_tunings): Likewise.
30535 (cortexa73_tunings): Likewise.
30537 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30539 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
30540 Set loop alignment to 8.
30542 2017-05-04 Martin Sebor <msebor@redhat.com>
30544 PR translation/80280
30545 * builtins.c (expand_builtin_object_size): Add missing quoting to
30546 %D and like directives.
30547 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
30548 (hsa_type_for_tree_type): Same.
30549 (verify_function_arguments): Same.
30550 * symtab.c (symbol_table::change_decl_assembler_name): Same.
30551 * varasm.c (get_section): Same.
30554 2017-05-04 Martin Sebor <msebor@redhat.com>
30556 PR translation/80280
30557 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
30559 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30561 * config/aarch64/aarch64.c (generic_addrcost_table):
30562 Change HI/TI mode setting.
30564 2017-05-04 Martin Jambor <mjambor@suse.cz>
30566 PR tree-optimization/80622
30567 * tree-sra.c (comes_initialized_p): New function.
30568 (build_accesses_from_assign): Only set write lazily when
30569 comes_initialized_p is false.
30570 (analyze_access_subtree): Use comes_initialized_p.
30571 (propagate_subaccesses_across_link): Assert !comes_initialized_p
30572 instead of testing for PARM_DECL.
30574 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30576 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
30577 constraint on operand 0 to allow more general addressing modes.
30578 Adjust output template.
30579 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
30581 * config/aarch64/aarch64-protos.h
30582 (aarch64_address_valid_for_prefetch_p): Declare prototype.
30583 * config/aarch64/constraints.md (Dp): New address constraint.
30584 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
30587 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
30589 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
30590 update use of estimate_ipcp_clone_size_and_time.
30591 (estimate_local_effects): Update use of
30592 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
30593 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
30594 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
30595 Return nonspecialized time.
30597 2017-05-04 Richard Biener <rguenther@suse.de>
30599 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
30600 for the last VUSE which def dominates the PHI. Directly call
30602 (get_continuation_for_phi_1): Remove.
30604 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
30606 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
30607 to explain the use of truncating division. Cap the number of
30608 iterations to the maximum given by nb_iterations_upper_bound,
30611 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30613 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
30614 * configure: Regenerate.
30615 * config.in: Regenerate.
30616 * config/i386/driver-mingw32.c: new file.
30617 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
30618 * config.host: Link driver-mingw32.o on MinGW host.
30619 * doc/install.texi: Document new --enable-mingw-wildcard configure
30622 2017-05-04 Marek Polacek <polacek@redhat.com>
30624 PR tree-optimization/80612
30625 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
30627 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30628 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
30630 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
30631 (movt splitter): Likewise.
30632 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
30633 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
30634 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
30635 block for Thumb-1 with MOVT.
30636 (thumb2_legitimate_address_p): Move code block ...
30637 (can_avoid_literal_pool_for_label_p): ... into this new function.
30638 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
30640 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
30641 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
30642 "M-profile targets with the MOVT instruction".
30644 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30646 * config/arm/arm-builtins.c (arm_init_builtins): Rename
30647 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
30648 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
30650 2017-05-04 Martin Liska <mliska@suse.cz>
30652 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
30653 variable cond_code.
30655 2017-05-04 Richard Biener <rguenther@suse.de>
30657 * tree.c (array_at_struct_end_p): Handle arrays at struct
30658 end with flexarrays more conservatively. Refactor and treat
30659 arrays of arrays or aggregates more strict. Fix
30660 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
30661 * tree.c (array_at_struct_end_p): Adjust prototype.
30662 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
30663 * gimple-fold.c (get_range_strlen): Likewise.
30664 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
30666 2017-05-04 Richard Biener <rguenther@suse.de>
30668 PR tree-optimization/31130
30669 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
30671 (supports_overflow_infinity): Likewise.
30672 (is_negative_overflow_infinity): Likewise.
30673 (is_positive_overflow_infinity): Likewise.
30674 (is_overflow_infinity): Likewise.
30675 (stmt_overflow_infinity): Likewise.
30676 (overflow_infinity_range_p): Likewise.
30677 (usable_range_p): Remove as always returning true.
30678 (make_overflow_infinity): Remove.
30679 (negative_overflow_infinity): Likewise.
30680 (positive_overflow_infinity): Likewise.
30681 (avoid_overflow_infinity): Likewise.
30682 (set_value_range): Adjust accordingly.
30683 (set_value_range_to_nonnegative): Likewise, remove now unused
30684 overflow_infinity arg.
30685 (vrp_operand_equal_p): Adjust.
30686 (update_value_range): Likewise.
30687 (range_int_cst_singleton_p): Likewise.
30688 (operand_less_p): Likewise.
30689 (compare_values_warnv): Likewise.
30690 (extract_range_for_var_from_comparison_expr): Likewise.
30691 (vrp_int_const_binop): Likewise.
30692 (zero_nonzero_bits_from_vr): Likewise.
30693 (extract_range_from_multiplicative_op_1): Likewise.
30694 (extract_range_from_binary_expr_1): Likewise.
30695 (extract_range_from_unary_expr): Likewise.
30696 (extract_range_from_comparison): Likewise.
30697 (extract_range_basic): Likewise.
30698 (adjust_range_with_scev): Likewise.
30699 (compare_ranges): Likewise.
30700 (compare_range_with_value): Likewise.
30701 (dump_value_range): Likewise.
30702 (test_for_singularity): Likewise, remove strict_overflow_p parameter
30704 (simplify_cond_using_ranges): Adjust.
30706 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30708 * brig-builtins.def: Added a builtin for class_f64.
30709 * builtin-types.def: Added a builtin type needed by class_f64.
30711 2017-05-03 Jason Merrill <jason@redhat.com>
30713 * timevar.def: Add TV_CONSTEXPR.
30715 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30717 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
30719 2017-05-03 Martin Jambor <mjambor@suse.cz>
30721 * ipa-prop.c (ipa_update_after_lto_read): Removed.
30722 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
30723 * ipa-cp.c (ipcp_propagate_stage): Do not call
30724 ipa_update_after_lto_read.
30725 * ipa-inline.c (ipa_inline): Likewise.
30727 2017-05-03 Martin Jambor <mjambor@suse.cz>
30729 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
30730 tag. Added a default constructor and a destructor.
30731 (ipa_edge_args_sum_t): New class;
30732 (ipa_edge_args_sum): Declare.
30733 (ipa_edge_args_vector): Remove declaration.
30734 (IPA_EDGE_REF): Use ipa_edge_args_sum.
30735 (ipa_free_edge_args_substructures): Remove declaration.
30736 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
30737 (ipa_edge_args_info_available_for_edge_p): Likewise.
30738 * ipa-prop.c (ipa_edge_args_vector): Removed.
30739 (edge_removal_hook_holder): Likewise.
30740 (edge_duplication_hook_holder): Likewise.
30741 (ipa_edge_args_sum): New variable.
30742 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
30743 ipa_edge_args_vector.
30744 (ipa_free_edge_args_substructures): Likewise.
30745 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
30746 ipa_edge_args_vector.
30747 (ipa_edge_removal_hook): Turned into method
30748 ipa_edge_args_sum_t::remove.
30749 (ipa_edge_duplication_hook): Turned into method
30750 ipa_edge_args_sum_t::duplicate.
30751 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
30752 registering edge hooks.
30753 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
30754 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
30755 ipa_edge_args_sum instead of ipa_edge_args_vector.
30756 * ipa-profile.c (ipa_profile): Likewise.
30758 2017-05-03 Martin Jambor <mjambor@suse.cz>
30760 * symbol-summary.h (function_summary): New method exists.
30761 (function_summary::symtab_removal): Deallocate through release.
30762 (call_summary): New class.
30763 (gt_ggc_mx): New overload.
30764 (gt_pch_nx): Likewise.
30765 (gt_pch_nx): Likewise.
30767 2017-05-03 Jeff Law <law@redhat.com>
30769 PR tree-optimization/78496
30770 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
30771 from simplify_cond_using_ranges. Split off code to walk
30772 backwards through casts into ...
30773 (simplify_cond_using_ranges_2): New function.
30774 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
30775 (execute_vrp): After identifying jump threads, call
30776 simplify_cond_using_ranges_2.
30778 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
30781 * ipa-inline.h (inline_summary): Add ctor.
30782 (create_ggc): Do not use ggc_cleared_alloc.
30784 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
30785 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30787 * gcc.c (handle_braces): Support escaping in switch matching
30789 * doc/invoke.texi (Spec Files): Document it.
30790 Remove superfluous @code markup in items.
30792 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30794 * diagnostic-show-locus.c (struct column_range): New struct.
30795 (get_affected_columns): New function.
30796 (get_printed_columns): New function.
30797 (struct correction): New struct.
30798 (correction::ensure_capacity): New function.
30799 (correction::ensure_terminated): New function.
30800 (struct line_corrections): New struct.
30801 (line_corrections::~line_corrections): New dtor.
30802 (line_corrections::add_hint): New function.
30803 (layout::print_trailing_fixits): Reimplement in terms of the new
30805 (selftest::test_overlapped_fixit_printing): New function.
30806 (selftest::diagnostic_show_locus_c_tests): Call it.
30808 2017-05-03 Nathan Sidwell <nathan@acm.org>
30810 Canonicalize canonical type hashing
30811 * tree.h (type_hash_canon_hash): Declare.
30812 * tree.c (type_hash_list, attribute_hash_list): Move into
30813 type_hash_canon_hash.
30814 (build_type_attribute_qual_variant): Break out hash code calc into
30815 type_hash_canon_hash.
30816 (type_hash_canon_hash): New. Generic type hash computation.
30817 (build_range_type_1, build_array_type_1, build_function_type,
30818 build_method_type_directly, build_offset_type, build_complex_type,
30819 make_vector_type): Call it.
30821 2017-05-03 Richard Biener <rguenther@suse.de>
30823 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
30824 When all DRs have unknown misaligned do not always peel
30825 when there is a store but apply the same costing model as if
30826 there were only loads.
30828 2017-05-03 Richard Biener <rguenther@suse.de>
30831 PR tree-optimization/80492
30832 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30833 compare_base_decls returning dont-know properly.
30835 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
30837 * config/arm/iterators.md (CCSI): New mode iterator.
30838 (arch): New mode attribute.
30839 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
30840 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
30841 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
30842 code iterator for success result mode.
30843 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
30844 the corresponding new insn generators.
30846 2017-05-03 Bin Cheng <bin.cheng@arm.com>
30849 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30850 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30852 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
30854 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
30855 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
30856 (DDR_A): Wrap DDR argument in brackets.
30857 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
30858 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
30859 (DDR_REVERSED_P): Likewise.
30861 2017-05-03 Jakub Jelinek <jakub@redhat.com>
30863 PR tree-optimization/79472
30864 * tree-switch-conversion.c (struct switch_conv_info): Add
30865 contiguous_range and default_case_nonstandard fields.
30866 (collect_switch_conv_info): Compute contiguous_range and
30867 default_case_nonstandard fields, don't clear final_bb if
30868 contiguous_range and only the default case doesn't have the required
30870 (check_all_empty_except_final): Set default_case_nonstandard instead
30871 of failing if contiguous_range and the default case doesn't have empty
30873 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
30874 and only the default case doesn't have the required constants. Skip
30876 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
30877 if default_case_nonstandard.
30878 (build_constructors): Build constant 1 just once. Assert that default
30879 values aren't inserted in between cases if contiguous_range. Skip
30881 (build_arrays): Skip virtual phis.
30882 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
30883 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
30884 Handle virtual phis.
30885 (gen_inbound_check): Handle default_case_nonstandard case.
30886 (process_switch): Adjust check_final_bb caller. Call
30887 gather_default_values with the first non-default case instead of
30888 default case if default_case_nonstandard.
30890 2017-05-02 Nathan Sidwell <nathan@acm.org>
30892 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
30893 check. Fix formatting.
30895 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
30897 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
30898 errors when comparing specialized and unspecialized times.
30900 2017-05-02 David Malcolm <dmalcolm@redhat.com>
30902 * diagnostic-show-locus.c
30903 (layout::should_print_annotation_line_p): Make private.
30904 (layout::print_annotation_line): Make private.
30905 (layout::annotation_line_showed_range_p): Make private.
30906 (layout::show_ruler): Make private.
30907 (layout::print_source_line): Make private. Pass in line and
30908 line_width, rather than calling location_get_source_line. Drop
30910 (layout::print_leading_fixits): New method.
30911 (layout::print_any_fixits): Rename to...
30912 (layout::print_trailing_fixits): ...this, and make private.
30913 Don't print newline fixits.
30914 (diagnostic_show_locus): Move logic for printing one row into...
30915 (layout::print_line): ...this new function. Move the
30916 location_get_source_line call and error-handling from
30917 print_source_line to here. Call print_leading_fixits, and rename
30918 print_any_fixits to print_trailing_fixits.
30919 (selftest::test_fixit_insert_containing_newline): Update now that
30920 newlines are partially supported.
30921 (selftest::test_fixit_insert_containing_newline_2): New test.
30922 (selftest::test_fixit_replace_containing_newline): Update comments.
30923 (selftest::diagnostic_show_locus_c_tests): Call the new test.
30924 * edit-context.c (class added_line): New class.
30925 (class edited_line): Describe newline handling in comment.
30926 (edited_line::actually_edited_p): New method.
30927 (edited_line::print_content): Delete redundant decl.
30928 (edited_line::m_predecessors): New field.
30929 (edited_file::print_content): Call edited_line::print_content.
30930 (edited_file::print_diff): Update to support newlines.
30931 (edited_file::print_diff_hunk): Likewise.
30932 (edited_file::print_run_of_changed_lines): New function.
30933 (edited_file::print_diff_line): Convert to...
30934 (print_diff_line): ...this.
30935 (edited_file::get_effective_line_count): New function.
30936 (edited_line::edited_line): Initialize new field m_predecessors.
30937 (edited_line::~edited_line): Clean up m_predecessors.
30938 (edited_line::apply_fixit): Handle newlines.
30939 (edited_line::get_effective_line_count): New function.
30940 (edited_line::print_content): New function.
30941 (edited_line::print_diff_lines): New function.
30942 (selftest::test_applying_fixits_insert_containing_newline): New
30944 (selftest::test_applying_fixits_replace_containing_newline): New
30946 (selftest::insert_line): New function.
30947 (selftest::test_applying_fixits_multiple_lines): Add example of
30949 (selftest::edit_context_c_tests): Call the new tests.
30951 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30953 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
30954 parameter cand. Update dump information.
30955 (get_computation_cost): Update uses.
30957 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30959 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
30960 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
30961 (get_computation_at, rewrite_use_address): Update use of
30962 get_computation_aff.
30964 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30966 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
30967 (get_computation): Delete.
30968 (get_computation_cost): Implement like get_computation_cost_at.
30969 Use get_computation_at.
30970 (get_computation_cost_at): Delete.
30971 (rewrite_use_nonlinear_expr): Use get_computation_at.
30972 (rewrite_use_compare, remove_unused_ivs): Ditto.
30974 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30976 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
30978 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30980 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
30981 (ivopts_global_cost_for_size): Rename parameter and update uses.
30982 (iv_ca_recount_cost): Update uses.
30983 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
30984 candidates seperately in n_invs and n_cands.
30985 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
30987 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30989 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
30990 (find_inv_vars_cb): New.
30991 (find_depends): Renamed to ...
30992 (find_inv_vars): ... this.
30993 (add_candidate_1, force_var_cost): Call find_inv_vars.
30994 (split_address_cost, determine_group_iv_cost_cond): Ditto.
30996 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30998 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
30999 inv_vars. Add inv_exprs.
31000 (struct iv_cand): Rename depends_on to inv_vars.
31001 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31002 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31003 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31004 (dump_cand): Dump inv_vars.
31005 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31006 (record_invariant, find_depends, add_candidate_1): Ditto.
31007 (set_group_iv_cost, force_var_cost): Ditto.
31008 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31009 (get_computation_cost_at, get_computation_cost): Ditto.
31010 (determine_group_iv_cost_generic): Ditto.
31011 (determine_group_iv_cost_address): Ditto.
31012 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31013 (determine_group_iv_costs): Ditto.
31014 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31015 (iv_ca_set_remove_invariants): Renamed to ...
31016 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31017 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31018 (iv_ca_set_add_invariants): Renamed to ...
31019 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31020 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31021 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31022 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31023 (create_new_ivs): Remove useless dump.
31025 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31027 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31029 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31030 (iv_ca_set_no_cp, create_new_iv): Ditto.
31032 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31034 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31036 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31038 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31039 function tree_check2.
31041 2017-05-02 Martin Liska <mliska@suse.cz>
31043 * doc/gcov.texi: Add missing preposition.
31044 * gcov.c (function_info::function_info): Properly fill up
31045 all member variables.
31047 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31049 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
31051 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31053 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
31055 2017-05-02 Martin Liska <mliska@suse.cz>
31058 * lto-streamer-in.c (lto_read_tree_1): Remove
31059 LTO_STREAMER_DEBUG.
31060 * lto-streamer.c (struct tree_hash_entry): Likewise.
31061 (struct tree_entry_hasher): Likewise.
31062 (tree_entry_hasher::hash): Likewise.
31063 (tree_entry_hasher::equal): Likewise.
31064 (lto_streamer_init): Likewise.
31065 (lto_orig_address_map): Likewise.
31066 (lto_orig_address_get): Likewise.
31067 (lto_orig_address_remove): Likewise.
31068 * lto-streamer.h: Likewise.
31069 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
31070 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
31072 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
31074 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
31075 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
31076 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
31077 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
31078 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
31079 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
31080 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
31081 (mm_maskz_sub_ss): New intrinsics.
31082 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31083 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31084 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
31085 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
31086 (__builtin_ia32_subss_mask_round): New builtins.
31087 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31088 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31089 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
31091 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
31092 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31094 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31097 2017-05-02 Martin Jambor <mjambor@suse.cz>
31099 PR tree-optimization/78687
31100 * tree-sra.c (access): New field parent.
31101 (process_subtree_disqualification): New function.
31102 (disqualify_candidate): Call it.
31103 (build_accesses_from_assign): Reset write flag if creating an
31105 (build_access_subtree): Fill in parent field and also prpagate
31106 down grp_write flag.
31107 (create_artificial_child_access): New parameter set_grp_write, set
31108 grp_write to its value.
31109 (propagate_subaccesses_across_link): Also propagate grp_write flag
31111 (propagate_all_subaccesses): Push the closest parent back to work
31112 queue if add_access_to_work_queue returned true.
31114 2017-05-02 Richard Biener <rguenther@suse.de>
31116 * common.opt (fstrict-overflow): Alias negative to fwrapv.
31117 * doc/invoke.texi (fstrict-overflow): Remove all traces of
31118 -fstrict-overflow documentation.
31119 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
31120 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
31121 flag_strict_overflow.
31122 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
31123 * lto-opts.c (lto_write_options): Do not stream it.
31124 * lto-wrapper.c (merge_and_complain): Do not handle it.
31125 * opts.c (default_options_table): Do not set -fstrict-overflow.
31126 (finish_options): Likewise do not clear it when sanitizing.
31127 * simplify-rtx.c (simplify_const_relational_operation): Do not
31128 test flag_strict_overflow.
31130 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31132 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
31133 using enabled attribute.
31134 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
31135 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
31136 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
31137 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
31138 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
31139 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
31140 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
31141 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
31142 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
31143 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
31145 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31147 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
31149 2017-05-02 Richard Biener <rguenther@suse.de>
31151 PR tree-optimization/80591
31153 2017-04-10 Richard Biener <rguenther@suse.de>
31155 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31158 2017-05-02 Richard Biener <rguenther@suse.de>
31160 PR tree-optimization/80549
31161 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
31162 (cleanup_tree_cfg_noloop): Create forwarders to known loop
31163 headers if they do not have a preheader.
31165 2017-05-02 Martin Liska <mliska@suse.cz>
31168 * common.opt: Fix typo.
31169 * doc/invoke.texi: Likewise.
31171 2017-05-01 Jan Beulich <jbeulich@suse.com>
31173 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
31174 swapping, add (x,x,m,x,n) alternative.
31176 2017-05-01 Nathan Sidwell <nathan@acm.org>
31178 * calls.c (combine_pending_stack_adjustment_and_call): Remove
31179 unnecessary unadjusted_alignment check.
31181 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31184 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
31186 * gimplify.c (gimplify_cilk_detach): New function.
31187 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
31188 * tree-core.h: Document EXPR_CILK_SPAWN.
31189 * tree.h (EXPR_CILK_SPAWN): Define.
31191 2017-05-01 David Malcolm <dmalcolm@redhat.com>
31193 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
31194 to use new fixit_hint representation, using the "replace" logic.
31195 (get_line_span_for_fixit_hint): Likewise.
31196 (layout::print_any_fixits): Likewise.
31197 (selftest::test_one_liner_many_fixits): Rename to...
31198 (selftest::test_one_liner_many_fixits_1): ...this, and update
31199 comment and expected output to reflect that the multiple fix-it
31200 hints are now consolidated into one insertion.
31201 (selftest::test_one_liner_many_fixits_2): New test.
31202 (selftest::test_diagnostic_show_locus_one_liner): Update for
31204 (selftest::test_fixit_consolidation): Update for fix-it API
31206 * diagnostic.c (print_parseable_fixits): Likewise.
31207 * edit-context.c (edited_line::m_line_events): Convert from
31208 auto_vec <line_event *> to auto_vec <line_event>.
31209 (class line_event): Convert from abstract base class to a concrete
31210 class, taking over the role of replace_event.
31211 (class insert_event): Delete.
31212 (class replace_event): Rename to class line_event. Convert to
31214 (edit_context::add_fixits): Reimplement.
31215 (edit_context::apply_insert): Delete.
31216 (edit_context::apply_replace): Rename to...
31217 (edit_context::apply_fixit): ...this. Convert to half-open range.
31218 (edited_file::apply_insert): Delete.
31219 (edited_file::apply_replace): Rename to...
31220 (edited_file::apply_fixit): ...this.
31221 (edited_line::~edited_line): Drop deletion of events.
31222 (edited_line::apply_insert): Delete.
31223 (edited_line::apply_replace): Rename to...
31224 (edited_line::apply_fixit): ...this. Convert to half-open range.
31225 Update for change to type of m_line_events.
31226 * edit-context.h (edit_context::apply_insert): Delete.
31227 (edit_context::apply_replace): Rename to...
31228 (edit_context::apply_fixit): ...this.
31230 2017-05-01 Martin Sebor <msebor@redhat.com>
31232 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
31235 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
31238 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
31239 __get_cpuid_max returns 0.
31240 (__get_cpuid_count): Ditto.
31242 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
31244 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
31245 replacement expression is another instance of one of its arguments.
31247 2017-05-01 Jakub Jelinek <jakub@redhat.com>
31250 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
31251 check for stack push/pop autoinc.
31252 * config/i386/i386.c (ix86_agi_dependent): Return false
31253 if the only reason why modified_in_p returned true is that
31254 addr is SP based and set_insn is a push or pop.
31256 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31258 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
31261 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31264 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
31265 (account_size_time): Use two predicates - exec_pred and
31267 (evaluate_conditions_for_known_args): Compute both clause and
31269 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
31270 (inline_summary_t::duplicate): Update.
31271 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
31273 (compute_inline_parameters): Likewise.
31274 (estimate_edge_size_and_time): Update caluclation of time.
31275 (estimate_node_size_and_time): Compute both time and nonspecialized
31277 (estimate_ipcp_clone_size_and_time): Update.
31278 (inline_merge_summary): Update.
31279 (do_estimate_edge_time): Update.
31280 (do_estimate_edge_size): Update.
31281 (do_estimate_edge_hints): Update.
31282 (inline_read_section, inline_write_summary): Stream both new predicates.
31283 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
31285 (compute_inlined_call_time): Cleanup.
31286 (big_speedup_p): Update.
31287 (edge_badness): Update.
31288 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
31289 (size_time_entry): Replace predicate by exec_predicate and
31290 nonconst_predicate.
31291 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
31292 (estimate_edge_time): Return also nonspec_time.
31293 (reset_edge_growth_cache): Update.
31295 2017-04-29 Jakub Jelinek <jakub@redhat.com>
31297 PR rtl-optimization/80491
31298 * ifcvt.c (noce_process_if_block): When looking for x setter
31299 with missing else_bb, don't check only the insn right before
31300 cond_earliest, but look for the last insn that x is modified in
31301 within the same bb.
31303 PR rtl-optimization/80491
31304 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
31306 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
31308 PR tree-optimization/80487
31309 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
31311 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31313 PR tree-optimization/79697
31314 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
31315 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
31316 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
31318 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
31319 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
31321 2017-04-28 Martin Sebor <msebor@redhat.com>
31323 PR tree-optimization/80523
31324 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
31325 (init_target_to_host_charmap, target_to_host, target_strtol10): New
31327 (maybe_warn, format_directive, parse_directive): Use new functions.
31328 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
31330 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
31332 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
31334 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
31336 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
31337 target_header_dir): Set correctly.
31338 * configure: Regenerated.
31339 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31340 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31341 instead of SYSTEM_HEADER_DIR.
31343 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
31345 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
31346 (estimate_local_effects): Likewise.
31347 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
31348 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
31349 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
31350 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
31351 do_estimate_edge_time, estimate_edge_time): Likewise.
31352 * ipa-inline-analysis.c (estimate_node_size_and_time,
31353 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
31354 (estimate_time_after_inlining): Remove.
31356 2017-04-28 Martin Liska <mliska@suse.cz>
31358 * doc/gcov.texi: Enhance documentation of gcov.
31360 2017-04-28 Martin Liska <mliska@suse.cz>
31362 * doc/gcov.texi: Sort options in alphabetic order.
31363 * doc/gcov-dump.texi: Likewise.
31364 * doc/gcov-tool.texi: Likewise.
31365 * gcov.c (print_usage): Likewise.
31366 * gcov-dump.c (print_usage): Likewise.
31367 * gcov-tool.c (print_merge_usage_message): Likewise.
31368 (print_rewrite_usage_message): Likewise.
31369 (print_overlap_usage_message): Likewise.
31371 2017-04-28 Martin Liska <mliska@suse.cz>
31373 PR gcov-profile/53915
31374 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
31376 2017-04-28 Martin Liska <mliska@suse.cz>
31378 PR gcov-profile/79891
31379 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
31380 is marked by compiler as living on a line.
31381 (get_cycles_count): Remove usage of the union.
31382 (output_intermediate_file): Likewise.
31383 (find_source): Fix GNU coding style.
31384 (accumulate_line_counts): Remove old non-all block mode.
31385 (output_lines): Remove usage of the union.
31386 * profile.c (output_location): Include all BBs, even if
31387 belonging to a same line (and file) as a previous BB.
31389 2017-04-28 Martin Liska <mliska@suse.cz>
31391 * gcov.c (process_args): Handle new argument 'w'.
31392 (read_graph_file): Assign ID to BBs.
31393 (output_branch_count): Display BB # if verbose flag is set.
31394 (output_lines): Likewise for arcs.
31395 (print_usage): Add '--verbose' option help.
31396 * doc/gcov.texi: Document --verbose (-w) option.
31398 2017-04-28 Martin Liska <mliska@suse.cz>
31400 * gcov.c (struct block_location_info): New struct.
31401 (process_file): Fill up the new structure.
31402 (read_graph_file): Replace usage of encoding by the newly added
31404 (add_line_counts): Likewise.
31405 (accumulate_line_counts): Remove usage of the union.
31406 (function_info::function_info): New function.
31407 (function_info::~function_info): Likewise.
31408 (process_file): Call delete instead of release_function.
31409 (release_function): Release the function.
31410 (release_structures): Call delete instead of release_function.
31411 (solve_flow_graph): Replace usage of num_blocks.
31412 (find_exception_blocks): Likewise.
31413 (output_lines): Fix GNU coding style.
31415 2017-04-28 Martin Liska <mliska@suse.cz>
31418 * coverage.c (coverage_remove_note_file): New function.
31419 * coverage.h: Declare the function.
31420 * toplev.c (finalize): Clean if an error has been seen.
31422 2017-04-28 Martin Liska <mliska@suse.cz>
31424 PR gcov-profile/80031
31425 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
31426 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
31427 * gcov.c (read_graph_file): Read just number of blocks.
31428 * profile.c (branch_prob): Do not stream 0 flags per a basic
31431 2017-04-28 Martin Liska <mliska@suse.cz>
31433 * gcov-dump.c (tag_*): Add new argument to declarations.
31434 (dump_gcov_file): Likewise.
31435 (tag_blocks): Add and use new argument depth.
31436 (tag_arcs): Likewise.
31437 (tag_lines): Likewise.
31438 (tag_counters): Likewise.
31439 (tag_summary): Likewise.
31440 (dump_working_sets): Use depth to do a proper indentation.
31442 2017-04-28 Jakub Jelinek <jakub@redhat.com>
31445 * cgraph.h (symtab_node::debug_symtab): No longer inline.
31446 * symtab.c (symtab_node::debug_symtab): Move definition here.
31448 2017-04-28 Richard Biener <rguenther@suse.de>
31450 * lto-streamer.h (LTO_major_version): Bump to 7.
31452 2017-04-28 Richard Biener <rguenther@suse.de>
31454 * tree-vrp.c (assert_info): New struct.
31455 (add_assert_info): New helper.
31456 (register_edge_assert_for_2): Refactor to add asserts to a vector
31458 (register_edge_assert_for_1): Likewise.
31459 (register_edge_assert_for): Likewise.
31460 (finish_register_edge_assert_for): New helper actually registering
31461 asserts where live on edge.
31462 (find_conditional_asserts): Adjust.
31463 (find_switch_asserts): Likewise.
31464 (evrp_dom_walker::try_find_new_range): Generalize.
31465 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
31467 2017-04-27 Marek Polacek <polacek@redhat.com>
31470 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
31471 arg10 and arg11 to itype.
31473 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
31475 * doc/extend.texi (Object Size Checking): Improve grammar.
31477 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
31480 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
31481 that the logic for permitting reciprocal estimates matches that
31484 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31487 * tree.c (type_cache_hasher::equal): Only compare
31488 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
31489 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
31490 non-aggregate element types.
31491 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
31492 about the flag on ARRAY_TYPEs in the comment, formatting fix.
31494 2017-04-27 Richard Biener <rguenther@suse.de>
31496 PR middle-end/80533
31497 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
31498 stripping ARRAY_REFs from MEM_EXPR make sure we're not
31499 keeping a reference to a trailing array.
31501 2017-04-27 Richard Biener <rguenther@suse.de>
31503 PR middle-end/80539
31504 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
31505 being in loop-closed SSA form conservatively.
31506 (chrec_fold_multiply_poly_poly): Likewise.
31508 2017-04-27 Tamar Christina <tamar.christina@arm.com>
31510 PR middle-end/79665
31511 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
31512 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
31514 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31517 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
31518 (aarch64_function_arg_alignment): Return unsigned int again, but still
31519 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
31520 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31521 Don't emit -Wpsabi note.
31522 (aarch64_function_arg_boundary): Likewise.
31523 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31526 2017-04-26 Nathan Sidwell <nathan@acm.org>
31528 * tree.h (crc32_unsigned_n): Declare.
31529 (crc32_unsigned, crc32_unsigned): Make inline.
31530 * tree.c (crc32_unsigned_bits): Replace with ...
31531 (crc32_unsigned_n): ... this.
31532 (crc32_unsigned, crc32_byte): Remove.
31533 (crc32_string): Remove unnecessary braces.
31535 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31537 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
31538 * ipa-inline-analysis.c (MAX_TIME): Remove.
31539 (account_size_time): Use sreal for time.
31540 (dump_inline_summary): Update.
31541 (estimate_function_body_sizes): Update.
31542 (estimate_edge_size_and_time): Update.
31543 (estimate_calls_size_and_time): Update.
31544 (estimate_node_size_and_time): Update.
31545 (inline_merge_summary): Update.
31546 (inline_update_overall_summary): Update.
31547 (estimate_time_after_inlining): Update.
31548 (inline_read_section): Update.
31549 (inline_write_summary): Update.
31550 * ipa-inline.c (compute_uninlined_call_time): Update.
31551 (compute_inlined_call_time): Update.
31552 (recursive_inlining): Update.
31553 (inline_small_functions): Update.
31554 (dump_overall_stats): Update.
31555 * ipa-inline.h: Include sreal.h.
31556 (size_time_entry): Turn time to sreal.
31557 (inline_summary): Turn self_time nad time to sreal.
31559 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31561 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
31563 (sreal::stream_out, sreal::stream_in): New.
31564 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
31566 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31568 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
31571 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
31574 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
31576 (dimode_scalar_chain::compute_convert_gain): Ditto.
31577 (dimode_scalar_chain::make_vector_copies): Ditto.
31578 (dimode_scalar_chain::convert_reg): Ditto.
31579 (dimode_scalar_chain::convert_insn): Ditto.
31580 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
31581 (VI248_AVX512BW_1): New mode iterator.
31582 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
31583 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
31586 2017-04-25 Martin Sebor <msebor@redhat.com>
31588 PR tree-optimization/80497
31589 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
31590 constants are representable in HOST_WIDE_INT.
31591 (parse_directive): Ditto.
31593 2017-04-25 Martin Sebor <msebor@redhat.com>
31596 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
31597 (new_zero_array): Adjust signature.
31598 (dom_info::dom_init): Used unsigned rather that size_t.
31599 (dom_info::dom_info): Same.
31601 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31602 Jakub Jelinek <jakub@redhat.com>
31605 * config/arm/arm.c: Include gimple.h.
31606 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31607 returns negative, increment ncrn only if it returned positive.
31608 (arm_needs_doubleword_align): Return int instead of bool,
31609 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
31610 members, but if there is any such non-FIELD_DECL
31611 > PARM_BOUNDARY aligned decl, return -1 instead of false.
31612 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31613 returns negative, increment nregs only if it returned positive.
31614 (arm_setup_incoming_varargs): Likewise.
31615 (arm_function_arg_boundary): Emit -Wpsabi note if
31616 arm_needs_doubleword_align returns negative, return
31617 DOUBLEWORD_ALIGNMENT only if it returned positive.
31619 2017-04-25 Marek Polacek <polacek@redhat.com>
31622 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31623 first argument to type.
31625 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
31628 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
31629 type checks to test for compatibility instead of equality.
31631 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31632 Jakub Jelinek <jakub@redhat.com>
31635 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
31637 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
31638 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
31639 the alignment computation, but return their maximum in warn_alignment.
31640 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31641 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
31643 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
31644 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31647 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31649 * config/arc/simdext.md (dmpyh): Fix typo.
31651 2017-04-25 Richard Biener <rguenther@suse.de>
31653 PR tree-optimization/80492
31654 * alias.c (compare_base_decls): Handle registers with asm
31655 specification conservatively.
31656 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31657 compare_base_decls returning dont-know properly.
31659 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31661 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
31662 (legitimate_offset_address_p): New function.
31663 (arc_legitimate_address_p): Use above function.
31665 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31667 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
31669 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31671 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
31672 ACCH registers whenever they are available.
31674 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31676 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
31677 double regs fix when not used.
31679 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31681 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
31683 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
31684 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
31686 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31688 * config/arc/arc.c (arc_output_addsi): Check for h-register class
31689 when emitting short ADD instructions.
31691 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31693 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
31695 (cmpsi_cc_c_insn): Likewise.
31696 (cbranchsi4_scratch): Compute proper instruction length using
31697 compact_hreg_operand.
31698 * config/arc/predicates.md (compact_hreg_operand): New predicate.
31700 2017-04-25 Richard Biener <rguenther@suse.de>
31702 PR middle-end/80509
31703 * passes.c (pass_manager::pass_manager): Initialize
31704 m_name_to_pass_map.
31706 2017-04-25 Richard Biener <rguenther@suse.de>
31708 PR tree-optimization/79201
31709 * tree-ssa-sink.c (statement_sink_location): Handle calls.
31711 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31714 * config/s390/vector.md: Split MEM->GPR vector moves for
31715 non-s_operand addresses.
31717 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31720 * config/s390/predicates.md (reload_const_wide_int_operand): New
31722 * config/s390/s390.md ("movti"): Remove d/P alternative.
31723 ("movti_bigconst"): New pattern definition.
31725 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31728 * s390-protos.h (s390_expand_cs_hqi): Removed.
31729 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
31730 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
31731 modes as well as CCZ1mode and CCZmode.
31732 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
31733 signature of s390_emit_compare_and_swap.
31734 (s390_expand_cs_hqi): Likewise, make static.
31735 (s390_expand_cs_tdsi): Generate an explicit compare before trying
31736 compare-and-swap, in some cases.
31737 (s390_expand_cs): Wrapper function.
31738 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
31740 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
31741 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
31742 patterns for small and large integers. Forbid symref memory operands.
31743 Move expander to s390.c. Require cc register.
31744 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
31745 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
31746 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
31747 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
31748 symref memory operands. Remove CC mode and call s390_match_ccmode
31750 ("atomic_exchange<mode>"): Allow and implement all integer modes.
31752 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31754 * config/s390/s390.md (define_peephole2): New peephole to help
31755 combining the load-and-test pattern with volatile memory.
31757 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31759 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
31760 with CCZmode for TARGET_Z196.
31762 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31764 PR rtl-optimization/80501
31765 * combine.c (make_compound_operation_int): Set subreg_code to SET
31766 even for AND with mask of the sign bit of mode.
31768 PR rtl-optimization/80500
31769 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
31770 sum's initial value.
31772 2017-04-25 Julian Brown <julian@codesourcery.com>
31773 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31775 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
31777 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
31779 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
31781 2017-04-25 Julian Brown <julian@codesourcery.com>
31782 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31784 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
31785 (thunderx2t99_sha): New Reservation.
31787 2017-04-25 Julian Brown <julian@codesourcery.com>
31788 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31790 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
31791 type for 1-element load.
31793 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
31795 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
31797 2017-04-24 Martin Jambor <mjambor@suse.cz>
31799 PR tree-optimization/80293
31800 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
31801 char arrays not totally scalarizable if it is false.
31802 (analyze_all_variable_accesses): Pass correct value in the new
31803 parameter. Add a statistics counter.
31805 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
31807 PR middle-end/79931
31808 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
31810 2017-04-24 Richard Biener <rguenther@suse.de>
31812 PR tree-optimization/80494
31813 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
31814 out for complex types.
31816 2017-04-24 Richard Biener <rguenther@suse.de>
31818 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
31819 * tree-ssa-sccvn.c (print_scc): Print SCC size.
31820 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
31821 (DFS): Adjust and never fail.
31822 (sccvn_dom_walker::fail): Remove.
31823 (sccvn_dom_walker::before_dom_children): Adjust.
31824 (run_scc_vn): Likewise and never fail.
31825 * tree-ssa-pre.c (pass_pre::execute): Adjust.
31826 (pass_fre::execute): Likewise.
31828 2017-04-24 Richard Biener <rguenther@suse.de>
31830 PR tree-optimization/79725
31831 * tree-ssa-sink.c (statement_sink_location): Return whether
31832 failure reason was zero uses. Move that check later.
31833 (sink_code_in_bb): Deal with zero uses by removing the stmt
31836 2017-04-24 Richard Biener <rguenther@suse.de>
31839 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
31840 pointer-based references.
31842 2017-04-24 Richard Biener <rguenther@suse.de>
31845 * pass_manager.h (pass_manager::operator new): Remove.
31846 (pass_manager::operator delete): Likewise.
31847 * passes.c (pass_manager::operator new): Remove.
31848 (pass_manager::operator delete): Likewise.
31849 (pass_manager::pass_manager): Zero individual pass members.
31851 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
31854 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
31855 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
31856 Check "XEXP (src, 1)" operand here.
31857 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
31858 Check "XEXP (src, 1)" operand here.
31859 (dimode_scalar_chain::make_vector_copies): Detect count register
31860 of a shift instruction. Zero extend count register from QImode
31861 to DImode to satisfy vector shift pattern count operand predicate.
31862 Substitute vector shift count operand with a DImode copy.
31863 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
31866 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
31868 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
31869 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31870 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
31871 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31872 (UNSPEC_NOREX_MEM): Remove definition.
31874 2017-04-21 Richard Biener <rguenther@suse.de>
31876 PR tree-optimization/79547
31877 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
31878 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
31879 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
31880 without any constraints.
31882 2017-04-21 Richard Biener <rguenther@suse.de>
31884 PR tree-optimization/78847
31885 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
31887 2017-04-21 Richard Biener <rguenther@suse.de>
31889 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
31890 (build_distinct_type_copy): Likewise.
31891 (build_variant_type_copy): Likewise.
31892 * tree.c (build_qualified_type): Pass down mem-stat info.
31893 (build_distinct_type_copy): Likewise.
31894 (build_variant_type_copy): Likewise.
31896 2017-04-21 Richard Biener <rguenther@suse.de>
31898 PR tree-optimization/80237
31899 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
31901 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
31902 for a simplified result.
31904 2016-04-21 Richard Biener <rguenther@suse.de>
31906 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
31907 sth as strict as a simple_iv but a chrec without symbols and an
31908 operand defined in the loop we are peeling (and not some subloop).
31909 (propagate_constants_for_unrolling): Propagate all constants.
31911 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31914 * config/i386/i386.c (print_reg): Remove assert for disalowed
31915 regno values, call output_operand_lossage instead.
31917 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31920 * config/i386/constraints.md (Yc): New register constraint.
31921 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
31922 Use Yc constraint for alternative 2 of operand 0. Remove
31923 preferred_for_speed attribute.
31925 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31927 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
31928 lastprivate clauses in SIMT case.
31930 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
31932 * doc/invoke.texi (-Wextra-semi): Document new warning option.
31934 2017-04-20 Richard Biener <rguenther@suse.de>
31936 PR tree-optimization/57796
31937 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
31938 as N scalar stores.
31939 (vect_model_load_cost): Cost gathers as N scalar loads.
31941 2017-04-20 Richard Biener <rguenther@suse.de>
31943 * ggc-page.c (ggc_allocated_p): Rename to ...
31944 (safe_lookup_page_table_entry): ... this and return the lookup
31946 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
31948 2017-04-20 Richard Biener <rguenther@suse.de>
31950 PR tree-optimization/80453
31951 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
31952 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
31953 from the conditions.
31954 (vn_phi_eq): Pass them down.
31955 (vn_phi_lookup): Record them.
31956 (vn_phi_insert): Likewise.
31958 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
31960 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
31961 uninitialized variable warning to avoid buffer overrun.
31963 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31966 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
31967 is suppressed for '{ 0 }' in C.
31969 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31971 * BASE-VER: Set to 8.0.0.
31973 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
31975 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
31976 priority .init_array and .fini_array section with SECTION_NOTYPE
31979 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31981 PR middle-end/80423
31982 * tree.h (build_array_type): Add typeless_storage default argument.
31983 * tree.c (type_cache_hasher::equal): Also compare
31984 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
31985 (build_array_type): Add typeless_storage argument, set
31986 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
31988 (build_nonshared_array_type): Adjust build_array_type_1 caller.
31989 (build_array_type): Likewise. Add typeless_storage argument.
31991 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31992 Jakub Jelinek <jakub@redhat.com>
31994 PR tree-optimization/80426
31995 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
31996 operation on symbolic operands, also compute the overflow for the
31997 invariant part when the operation degenerates into a negation.
31999 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32002 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32003 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32006 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32008 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32011 * config/avr/avr.c (tree.h): Include it.
32012 (cgraph.h): Include it.
32013 (avr_encode_section_info): Don't warn for uninitialized progmem
32014 variable if it's just an alias.
32016 2017-04-19 Richard Biener <rguenther@suse.de>
32019 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32020 when needed by AutoPGO.
32022 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32025 * doc/lto.texi: Remove an extra 'that'.
32027 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32029 PR rtl-optimization/80429
32030 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32031 are only used in debug insns.
32033 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32034 Vladimir Makarov <vmakarov@redhat.com>
32036 * config/sparc/predicates.md (input_operand): Add comment. Return
32037 true for any memory operand when LRA is in progress.
32038 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32040 2017-04-18 Jeff Law <law@redhat.com>
32043 * mips.md ({return,simple_return}_internal): Do not overwrite
32046 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32048 PR tree-optimization/80443
32049 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
32050 instead of adding 1, subtract -1 and similarly instead of subtracting
32053 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
32055 PR rtl-optimization/80357
32056 * haifa-sched.c (tmp_bitmap): New variable.
32057 (model_recompute): Handle duplicate use records.
32058 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
32059 (free_global_sched_pressure_data): Free it.
32061 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
32064 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32065 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32066 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32067 instead of SYSTEM_HEADER_DIR.
32069 2017-04-18 Jeff Law <law@redhat.com>
32071 PR middle-end/80422
32072 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
32073 predecessors after walking up the insn chain.
32075 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32078 * dwarf2out.c (modified_type_die): Try harder not to emit internal
32079 sizetype type into debug info.
32081 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32084 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
32085 unneeded test for TARGET_UPPER_REGS_SF.
32086 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
32088 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32091 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
32092 instead of gsi_after_labels.
32094 2017-04-18 Jeff Law <law@redhat.com>
32096 * regcprop.c (maybe_mode_change): Avoid creating copies of the
32100 2017-04-13 Jeff Law <law@redhat.com>
32101 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32102 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32104 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
32107 * config/avr/avr.c (intl.h): Include it.
32108 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
32110 2017-04-18 Martin Liska <mliska@suse.cz>
32112 PR gcov-profile/78783
32113 * gcov-tool.c (gcov_output_files): Validate that destination
32114 file is either removed by the tool or by a user.
32116 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32117 Guy Benyei <guybe@mellanox.com>
32119 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
32120 block, and do not negate it, the stored id is already negative.
32122 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32124 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
32126 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32129 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
32130 masks of options that should be turned off if the VSX vector
32131 options are turned off.
32132 (OTHER_P8_VECTOR_MASKS): Likewise.
32133 (OTHER_VSX_VECTOR_MASKS): Likewise.
32134 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
32135 rs6000_disable_incompatible_switches to validate no type switches
32137 (rs6000_incompatible_switch): New function to disallow turning on
32138 other vector options if -mno-vsx, -mno-power8-vector, or
32139 -mno-power9-vector are specified.
32141 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32143 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
32145 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32147 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
32148 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
32149 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
32150 (ARG_POINTER_CFA_OFFSET): Likewise.
32152 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32154 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
32155 conditions to take advantage of various optimizations.
32157 2017-04-13 Jeff Law <law@redhat.com>
32159 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32160 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32161 (zero_extendsidi2_dext): Likewise.
32163 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32166 * fold-const.c (fold_ternary_loc): Revert
32167 use op0 instead of fold_convert_loc (loc, type, arg0) part of
32170 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
32172 PR rtl-optimization/80343
32173 * lra-remat.c (update_scratch_ops): Assign original hard reg to
32174 new scratch pseudo.
32176 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
32179 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
32180 to ubsan_encode_value.
32182 2017-04-13 Jeff Law <law@redhat.com>
32184 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
32185 appearing in DEBUG_INSNs.
32187 2017-04-13 Martin Liska <mliska@suse.cz>
32189 PR gcov-profile/80413
32190 * gcov-io.c (gcov_write_string): Copy to buffer just when
32191 allocated size is greater than zero.
32193 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32196 * dwarf2out.c (decls_for_scope): Ignore declarations of
32197 current_function_decl in BLOCK_NONLOCALIZED_VARS.
32199 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
32202 * ipa-visibility.c (non_local_p): Fix typos.
32203 (localize_node): When localizing symbol in same comdat group,
32204 dissolve the group only when we know external symbols are going
32206 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
32208 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32210 PR tree-optimization/79390
32211 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
32212 order does not result in usable sequence, retry with reversed operand
32218 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
32219 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
32220 op0 instead of fold_convert_loc (loc, type, arg0).
32222 2017-04-12 Jeff Law <law@redhat.com>
32224 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
32225 has a delay slot in the generated code.
32227 * config/cris/cris.md (cris_preferred_reload_class): Return
32228 GENNONACR_REGS rather than GENERAL_REGS.
32230 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32233 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
32234 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
32235 signedness of the result type.
32237 2017-04-12 Richard Biener <rguenther@suse.de>
32238 Jeff Law <law@redhat.com>
32240 PR tree-optimization/80359
32241 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
32242 trim stores to TARGET_MEM_REFs.
32244 2017-04-12 Richard Biener <rguenther@suse.de>
32246 PR tree-optimization/79390
32247 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
32248 threading case even more.
32250 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
32253 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
32254 for quad_address_p for TImode, instead of just not indexed_address.
32256 2017-04-12 Richard Biener <rguenther@suse.de>
32257 Bernd Edlinger <bernd.edlinger@hotmail.de>
32259 PR middle-end/79671
32260 * alias.c (component_uses_parent_alias_set_from): Handle
32261 TYPE_TYPELESS_STORAGE.
32262 (get_alias_set): Likewise.
32263 * tree-core.h (tree_type_common): Add typeless_storage flag.
32264 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
32265 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
32266 for types containing members with TYPE_TYPELESS_STORAGE.
32267 (place_field): Likewise.
32268 (layout_type): Likewise for ARRAY_TYPE.
32269 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
32270 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
32271 TYPE_TYPELESS_STORAGE.
32272 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
32274 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32277 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32278 first argument to type.
32280 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32284 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
32285 CONST0_RTX (mode) rather than const0_rtx where appropriate.
32286 (rs6000_expand_binop_builtin): Likewise.
32287 (rs6000_expand_ternop_builtin): Likewise; also add missing
32288 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
32289 vshasigma built-ins.
32290 * doc/extend.texi: Document that vec_xxpermdi's third argument
32291 must be a constant.
32293 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
32295 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
32296 Use shift_const cost parameter when calculating gain of STV shifts.
32298 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
32300 PR rtl-optimization/70478
32301 * lra-constraints.c (process_alt_operands): Check memory for
32302 disfavoring memory insn operand.
32304 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32306 PR middle-end/80100
32307 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
32308 left shift in unsigned HOST_WIDE_INT type.
32310 PR rtl-optimization/80385
32311 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
32312 (not (neg X)) into (plus X -1) for complex or non-integral modes.
32315 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
32316 if they have any depend clauses.
32318 2017-04-11 Martin Liska <mliska@suse.cz>
32321 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
32322 * ipa-split.c (split_function): Create a local comdat symbol
32323 if caller is in a comdat group.
32325 2017-04-11 Martin Liska <mliska@suse.cz>
32328 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
32331 2017-04-11 Martin Sebor <msebor@redhat.com>
32333 PR middle-end/80364
32334 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
32335 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
32337 (directive::set_width, directive::set_precision, format_character):
32339 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
32342 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
32345 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
32346 conflict, set target->arch_name instead of target->cpu_name.
32348 2017-04-11 Richard Biener <rguenther@suse.de>
32350 PR tree-optimization/80374
32351 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
32352 build_zero_cst, remove fold_convertible_p check again.
32354 2017-04-11 Martin Liska <mliska@suse.cz>
32357 * ubsan.c (instrument_object_size): Do not instrument register
32360 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32363 * config/i386/i386-builtin-types.def
32364 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
32365 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
32366 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
32367 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
32368 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
32369 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
32370 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
32371 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
32372 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
32373 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
32374 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
32375 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
32376 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
32377 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
32378 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
32379 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
32380 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
32381 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
32382 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
32383 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
32384 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
32385 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
32386 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
32387 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
32388 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
32389 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
32390 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
32391 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
32392 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
32393 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
32394 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
32395 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
32396 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
32397 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
32398 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
32399 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
32400 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
32401 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
32402 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
32403 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
32404 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
32405 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
32406 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
32407 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
32408 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
32409 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
32411 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
32412 flag to second_arg_count, handle 4 argument function type _COUNT
32413 aliases, handle second_arg_count on second argument rather than last.
32415 2017-04-10 Jeff Law <law@redhat.com>
32417 PR tree-optimization/80374
32418 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
32419 record anything if we can not convert integer_zero_node to the
32422 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32425 * config/rs6000/rs6000.c (rs6000_option_override_internal):
32426 Enhance special handling given to the TARGET_P9_MINMAX option in
32427 relation to certain other options.
32429 2017-04-10 Bin Cheng <bin.cheng@arm.com>
32431 PR tree-optimization/80153
32432 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
32433 remove POINTER_PLUS_EXPR's base part directly, rather than through
32436 2017-04-10 Richard Biener <rguenther@suse.de>
32437 Bin Cheng <bin.cheng@arm.com>
32439 PR tree-optimization/80153
32440 * tree-affine.c (aff_combination_to_tree): Get base pointer from
32441 the first element of pointer type aff_tree. Build result expr in
32443 (add_elt_to_tree): Convert to type unconditionally. Remove other
32444 fold_convert calls.
32445 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
32446 (rewrite_use_nonlinear_expr): Check invariant using iv information.
32448 2017-04-10 Richard Biener <rguenther@suse.de>
32450 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32453 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
32455 PR rtl-optimization/70478
32456 * lra-constraints.c (curr_small_class_check): New.
32457 (update_and_check_small_class_inputs): New.
32458 (process_alt_operands): Update curr_small_class_check. Disfavor
32459 alternative insn memory operands. Check available regs for small
32462 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32465 * config/mips/mips.opt (-mvirt): Update description.
32466 * doc/invoke.texi (-mvirt): Likewise.
32468 2017-04-10 Richard Biener <rguenther@suse.de>
32470 PR middle-end/80362
32471 * fold-const.c (fold_binary_loc): Look at unstripped ops when
32472 looking for NEGATE_EXPR in -A / -B to A / B folding.
32474 2017-04-10 Martin Liska <mliska@suse.cz>
32476 PR gcov-profile/80224
32477 * gcov.c (print_usage): Fix usage string.
32478 (get_gcov_intermediate_filename): Remove.
32479 (output_gcov_file): Use both for normal and intermediate format.
32480 (generate_results): Do not initialize special file for
32481 intermediate format.
32483 2017-04-10 Richard Biener <rguenther@suse.de>
32485 PR tree-optimization/80304
32486 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
32489 2017-04-10 Nathan Sidwell <nathan@acm.org>
32492 * config/rs6000/rs6000.c (rs6000_vector_type): New.
32493 (rs6000_init_builtins): Use it.
32495 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32497 * config/arm/arm.md (<mrc>): Add mode to SET source.
32498 (<mrrc>): Likewise.
32500 2017-04-10 Richard Biener <rguenther@suse.de>
32502 PR middle-end/80344
32503 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
32505 2017-04-10 Jakub Jelinek <jakub@redhat.com>
32508 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
32509 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
32510 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
32511 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
32512 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
32513 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
32514 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
32515 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
32516 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
32517 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
32518 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
32519 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
32520 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
32521 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
32522 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
32523 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
32524 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
32525 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
32526 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
32527 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
32528 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
32529 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
32530 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
32532 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
32534 PR rtl-optimization/70478
32535 * lra-constraints.c: Reverse the last patch.
32537 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
32539 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
32540 Add comment for WCHAR_T.
32542 2017-04-08 Martin Liska <mliska@suse.cz>
32545 2017-04-07 Martin Liska <mliska@suse.cz>
32548 * ipa-split.c (split_function): Add function part to a same comdat
32551 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32554 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
32556 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
32558 * rs6000/rs6000.c (vec_load_pendulum): Rename...
32559 (vec_pairing): ...to this.
32560 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
32561 (rs6000_sched_init): Adjust for name change.
32562 (struct rs6000_sched_context): Likewise.
32563 (rs6000_init_sched_context): Likewise.
32564 (rs6000_set_sched_context): Likewise.
32566 2017-04-07 Jakub Jelinek <jakub@redhat.com>
32572 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
32574 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
32575 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
32576 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
32578 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
32580 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
32582 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32584 PR rtl-optimization/70703
32585 * ira-color.c (update_conflict_hard_regno_costs): Use
32586 int64_t instead of HOST_WIDE_INT.
32588 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32590 PR rtl-optimization/70478
32591 * lra-constraints.c (process_alt_operands): Disfavor alternative
32592 insn memory operands.
32594 2017-04-07 Jeff Law <law@redhat.com>
32596 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
32597 CALL and NOTE_INSN_CALL_ARG_LOCATION.
32599 2017-04-07 Martin Liska <mliska@suse.cz>
32602 * config/aarch64/aarch64.c (aarch64_process_target_attr):
32603 Show error message instead of an ICE.
32605 2017-04-07 Martin Liska <mliska@suse.cz>
32608 * ipa-split.c (split_function): Add function part to a same comdat
32611 2017-04-07 Richard Biener <rguenther@suse.de>
32613 PR middle-end/80341
32614 * tree.c (get_unwidened): Also handle ! for_type case for
32616 * convert.c (do_narrow): Split out from ...
32617 (convert_to_integer_1): ... here. Do not pass final truncation
32618 type to get_unwidened for TRUNC_DIV_EXPR.
32620 2017-04-07 Richard Biener <rguenther@suse.de>
32622 * tree-affine.c (wide_int_ext_for_comb): Take type rather
32624 (aff_combination_const): Adjust.
32625 (aff_combination_scale): Likewise.
32626 (aff_combination_add_elt): Likewise.
32627 (aff_combination_add_cst): Likewise.
32628 (aff_combination_convert): Likewise.
32629 (add_elt_to_tree): Likewise. Remove unused argument.
32630 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
32632 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
32634 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
32636 * config/arm/arm.c (arm_default_short_enums): Use
32637 ARM_DEFAULT_SHORT_ENUMS.
32638 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
32640 2017-04-06 Jakub Jelinek <jakub@redhat.com>
32643 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
32644 members with redundant out-of-class redeclaration.
32646 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32649 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
32650 * config/i386/i386.md (*zero_extendsidi2):
32651 Add (?*x,*x) and (?*v,*v) alternatives.
32653 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32656 * config/i386/i386.c (ix86_expand_builtin)
32657 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
32658 mode from insn data. Convert operands to insn operand mode.
32659 Copy operands that don't satisfy insn predicate to a register.
32661 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
32663 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
32666 2017-04-06 Richard Biener <rguenther@suse.de>
32668 PR tree-optimization/80334
32669 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
32670 preserve alignment of accesses.
32672 2017-04-06 Richard Biener <rguenther@suse.de>
32674 PR tree-optimization/80262
32675 * tree-sra.c (build_ref_for_offset): Preserve address-space
32677 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
32678 Drop useless address-space information on MEM_REF offsets.
32680 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
32682 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
32684 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32686 PR rtl-optimization/70703
32687 * ira-color.c (update_conflict_hard_regno_costs): Use
32688 HOST_WIDE_INT instead of long.
32690 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
32693 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
32694 not defined for x86_64 target. Add -mmmx target option when __SSE2__
32696 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
32697 for x86_64 target. Handle -m3dnowa option.
32699 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32701 PR rtl-optimization/70703
32702 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
32703 (update_conflict_hard_regno_costs): Use long instead of unsigned
32704 arithmetic for cost calculation.
32706 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32707 Bernd Edlinger <bernd.edlinger@hotmail.de>
32710 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
32713 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
32716 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
32717 ptr_mode with Pmode throughout.
32718 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
32719 into probe_stack_range and use DImode.
32721 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
32724 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
32725 call_eh_return is true.
32727 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32729 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
32730 Initialize last_match_fntype_index.
32732 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32735 * tree-nvr.c: Include internal-fn.h.
32736 (pass_return_slot::execute): Ignore internal calls without
32739 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32740 Richard Biener <rguenther@suse.de>
32743 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
32744 captures used multiple times, except for the last use.
32745 * generic-match-head.c: Include gimplify.h.
32747 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32749 PR tree-optimization/79390
32750 * target.h (struct noce_if_info): Declare.
32751 * targhooks.h (default_noce_conversion_profitable_p): Declare.
32752 * target.def (noce_conversion_profitable_p): New target hook.
32753 * ifcvt.h (struct noce_if_info): New type, moved from ...
32754 * ifcvt.c (struct noce_if_info): ... here.
32755 (noce_conversion_profitable_p): Renamed to ...
32756 (default_noce_conversion_profitable_p): ... this. No longer
32758 (noce_try_store_flag_constants, noce_try_addcc,
32759 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
32760 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
32761 instead of noce_conversion_profitable_p.
32762 * config/i386/i386.c: Include ifcvt.h.
32763 (ix86_option_override_internal): Don't override
32764 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
32765 (ix86_noce_conversion_profitable_p): New function.
32766 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
32767 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
32768 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
32769 * doc/tm.texi: Regenerated.
32771 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32773 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
32776 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32779 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
32780 instructions for small multiply cores.
32782 2017-04-04 Jeff Law <law@redhat.com>
32784 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
32786 (mips_expand_vec_perm_const): Initialize elements in orig_perm
32787 that are not set by the loop over the elements.
32789 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32792 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
32793 int mode, convert_modes it to mode as unsigned, otherwise use
32794 lowpart_subreg to mode rather than SImode.
32795 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
32796 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
32797 Use DImode instead of SImode for the shift count operand.
32798 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
32801 2017-04-04 Richard Biener <rguenther@suse.de>
32803 PR middle-end/80281
32804 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
32805 arithmetic done for the negate or the plus. Simplify.
32806 (A - (-B) -> A + B): Likewise.
32807 * fold-const.c (split_tree): Make sure to not negate pointers.
32809 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
32811 PR rtl-optimization/60818
32812 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
32813 a compare of comparisons with the thing compared if this results
32814 in a different machine mode.
32816 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
32818 * alias.c (base_alias_check): Fix typo in comment.
32819 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
32820 * cgraphunit.c (symbol_table::compile): Likewise.
32821 * collect2.c (maybe_run_lto_and_relink): Likewise.
32822 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
32823 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
32824 * config/avr/avr.c (avr_map_op_t): Likewise.
32825 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
32826 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
32827 * config/epiphany/epiphany.md (movcc): Likewise.
32828 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
32829 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
32831 * config/mips/mips.c (mips_save_restore_reg): Likewise.
32832 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
32833 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
32834 * config/sh/sh.c (sh_rtx_costs): Likewise.
32835 * fold-const.c (fold_truth_andor): Likewise.
32836 * genautomata.c (collapse_flag): Likewise.
32837 * gengtype.h (struct type::u::s): Likewise.
32838 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
32839 * input.c (FORMAT_AMOUNT): Likewise.
32840 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
32841 (known_aggs_to_agg_replacement_list): Likewise.
32842 * ipa-inline-analysis.c: Likewise.
32843 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
32844 * ipa-polymorphic-call.c
32845 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
32846 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
32847 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
32849 * modulo-sched.c (apply_reg_moves): Likewise.
32850 * omp-expand.c (build_omp_regions_1): Likewise.
32851 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
32852 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
32853 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
32854 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32855 * value-prof.c: Likewise.
32856 * var-tracking.c (val_reset): Likewise.
32858 2017-04-03 Richard Biener <rguenther@suse.de>
32860 PR tree-optimization/80275
32861 * fold-const.c (split_address_to_core_and_offset): Handle
32864 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
32866 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
32867 descriptors is at least equal to that of functions.
32869 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32871 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
32873 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32876 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
32877 (mov<IMOD4:mode>): New expander.
32878 (*mov<IMOD4:mode>_internal): New insn and split pattern.
32880 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
32882 PR rtl-optimization/79405
32883 * fwprop.c (propagations_left): New variable.
32884 (forward_propagate_into): Decrement it.
32885 (fwprop_init): Initialize it.
32886 (fw_prop): If the variable has reached zero, stop propagating.
32887 (fwprop_addr): Ditto.
32889 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32892 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
32893 a FUNCTION_DECL, pass it as decl instead of origin to
32896 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
32898 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
32901 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
32904 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
32905 TARGET_VSX_SMALL_INTEGER.
32907 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32909 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32910 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
32912 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32914 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
32915 extraction from odd-numbered MSA register.
32917 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32919 PR middle-end/80173
32920 * expmed.c (store_bit_field_1): Don't attempt to create
32921 a word subreg out of hard registers wider than word if they
32922 have HARD_REGNO_NREGS of 1 for their mode.
32924 PR middle-end/80163
32925 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
32926 conversions to integer types wider than word and pointer.
32929 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
32930 (rtx_equal_for_cselib_p): Pass 0 to it.
32931 * cselib.c (cselib_hasher::equal): Likewise.
32932 (rtx_equal_for_cselib_1): Add depth argument. If depth
32933 is 128, don't look up VALUE locs and punt. Increment
32934 depth in recursive calls when walking VALUE locs.
32936 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
32938 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
32939 (make_gcov_file_name): Use the canonical path name for generating
32941 (read_line): Fix handling of files with ascii null bytes.
32943 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
32945 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
32946 to initialise a vector register instead
32947 of using a const_int.
32949 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32951 PR translation/80189
32952 * gimplify.c (omp_default_clause): Use %qs instead of %s in
32953 diagnostic messages.
32955 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
32958 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
32959 (dfp_diex_<mode>): Update mode of operand 1.
32960 * doc/extend.texi (dxex, dxexq): Document change to return type.
32961 (diex, diexq): Document change to argument type.
32963 2017-03-30 Martin Jambor <mjambor@suse.cz>
32966 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
32967 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
32968 it reflects the signature changes performed at the callee side.
32969 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
32970 to cgraph_build_function_type_skip_args.
32971 (build_function_decl_skip_args): Adjust call to the above function.
32973 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32976 * config/i386/sse.md
32977 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
32978 register as dest whenever it is a MEM not rtx_equal_p to the
32979 corresponding dup operand, and when forcing into reg move the
32980 reg into the memory afterwards.
32981 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
32982 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
32983 for the force_reg mode.
32984 (avx512vl_vextractf128<mode>): Use register as dest either
32985 always when a MEM, or when it is a MEM not rtx_equal_p to the
32986 corresponding dup operand, or even not when it is a CONST_VECTOR
32987 depending on the mode and lo vs. hi.
32988 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
32990 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
32991 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
32992 Likewise. Require that operands[2] is even.
32993 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
32994 Remove extraneous parens. Require that operands[2] is a multiple
32996 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
32997 operands[0] is a MEM if <mask_applied>, the predicates/constraints
32998 disallow memory then.
33000 2017-03-30 Richard Biener <rguenther@suse.de>
33002 PR tree-optimization/77498
33003 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33004 to non-constants over backedges.
33006 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33008 PR rtl-optimization/80233
33009 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33010 as last_combined_insn. Do not test for BARRIER_P separately.
33012 2017-03-29 Andreas Schwab <schwab@suse.de>
33015 * calls.c (prepare_call_address): Convert funexp to Pmode before
33016 copying to temp reg.
33018 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33020 PR tree-optimization/80158
33021 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33022 Handle possible future case of more than one alternate
33024 (replace_rhs_if_not_dup): Likewise.
33025 (replace_one_candidate): Likewise.
33027 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33029 PR rtl-optimization/80193
33030 * ira.c (ira): Do not check allocation for LRA.
33032 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33034 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33035 (nvptx_output_simt_exit): Declare.
33036 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33037 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33038 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33039 (nvptx_declare_function_name): ...here. Emit declaration of local
33040 memory space buffer for omp_simt_enter insn.
33041 (nvptx_output_unisimt_switch): New.
33042 (nvptx_output_softstack_switch): New.
33043 (nvptx_output_simt_enter): New.
33044 (nvptx_output_simt_exit): New.
33045 * config/nvptx/nvptx.h (struct machine_function): New fields
33046 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
33047 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
33048 (UNSPECV_SIMT_EXIT): Ditto.
33049 (omp_simt_enter_insn): New insn.
33050 (omp_simt_enter): New expansion.
33051 (omp_simt_exit): New insn.
33052 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
33054 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
33055 (expand_GOMP_SIMT_ENTER_ALLOC): New.
33056 (expand_GOMP_SIMT_EXIT): New.
33057 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
33058 (GOMP_SIMT_ENTER_ALLOC): Ditto.
33059 (GOMP_SIMT_EXIT): Ditto.
33060 * target-insns.def (omp_simt_enter): New insn.
33061 (omp_simt_exit): Ditto.
33062 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
33064 (lower_rec_simd_input_clauses): Implement SIMT privatization.
33065 (lower_rec_input_clauses): Likewise.
33066 (lower_lastprivate_clauses): Handle SIMT privatization.
33068 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
33069 (ompdevlow_adjust_simt_enter): New.
33070 (find_simtpriv_var_op): New.
33071 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
33072 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
33074 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
33075 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
33076 (copy_decl_for_dup_finish): Ditto.
33078 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
33080 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
33083 * config/i386/i386.c (ix86_option_override_internal): Always
33084 allow -mpreferred-stack-boundary=3 for 64-bit targets.
33086 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33088 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
33090 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33092 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
33093 mark new edge's irreducible flag accordign to it.
33094 (vect_do_peeling): Check loop preheader edge's irreducible flag
33095 and pass it to function slpeel_add_loop_guard.
33097 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
33099 PR tree-optimization/80218
33100 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33101 Update block frequencies and counts.
33103 2017-03-28 Richard Biener <rguenther@suse.de>
33105 PR tree-optimization/78644
33106 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
33107 of a simplification result we may not use it at all.
33109 2017-03-28 Richard Biener <rguenther@suse.de>
33112 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
33113 without arguments, generate default definition of a SSA name.
33115 2017-03-28 Richard Biener <rguenther@suse.de>
33117 PR middle-end/80222
33118 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
33119 TYPE_REF_CAN_ALIAS_ALL references.
33120 * fold-const.c (fold_indirect_ref_1): Likewise.
33122 2017-03-28 Martin Liska <mliska@suse.cz>
33125 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
33126 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
33128 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33129 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
33131 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
33132 (EXTRA_SPECS): Define.
33133 (SUBTARGET_EXTRA_SPECS): Likewise.
33134 (SUBTARGET_CPP_SPEC): Likewise.
33135 * config/arc/elf.h (EXTRA_SPECS): Renamed to
33136 SUBTARGET_EXTRA_SPECS.
33137 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
33139 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33141 * config/arc/simdext.md (vst64_insn): Update pattern.
33142 (vld32wh_insn): Likewise.
33143 (vld32wl_insn): Likewise.
33144 (vld64_insn): Likewise.
33145 (vld32_insn): Likewise.
33147 2017-03-28 Marek Polacek <polacek@redhat.com>
33150 * fold-const.c (fold_comparison): Use protected_set_expr_location
33151 instead of SET_EXPR_LOCATION.
33153 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
33155 * tree.c (add_expr): Avoid name lookup warning.
33157 2017-03-27 Jeff Law <law@redhat.com>
33159 PR tree-optimization/80216
33160 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
33161 function name. Limit recursion depth.
33162 (record_temporary_equivalences): Corresponding changes.
33164 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
33166 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
33169 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33172 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
33174 * cfgcleanup.c (reg_note_cfa_p): New array.
33175 (insns_have_identical_cfa_notes): New function.
33176 (old_insns_match_p): Don't cross-jump in between /f
33177 and non-/f instructions. If both i1 and i2 are frame related,
33178 verify all CFA notes, their order and content.
33180 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
33183 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
33184 HImode and SImode with zero extend to DImode to one insn.
33185 (bswap<mode>2_extenddi): Likewise.
33186 (bswapsi2_extenddi): Likewise.
33187 (bswaphi2_extendsi): Likewise.
33188 (bswaphi2): Combine bswap HImode and SImode into one insn.
33189 Separate memory insns from swapping register.
33190 (bswapsi2): Likewise.
33191 (bswap<mode>2): Likewise.
33192 (bswaphi2_internal): Delete, no longer used.
33193 (bswapsi2_internal): Likewise.
33194 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
33195 store, and gpr<-gpr swap insns.
33196 (bswap<mode>2_store): Likewise.
33197 (bswaphi2_reg): Register only splitter, combine with the splitter.
33198 (bswaphi2 splitter): Likewise.
33199 (bswapsi2_reg): Likewise.
33200 (bswapsi2 splitter): Likewise.
33201 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
33202 the insns into load, store, and register/register insns.
33203 (bswapdi2_ldbrx): Likewise.
33204 (bswapdi2_load): Likewise.
33205 (bswapdi2_store): Likewise.
33206 (bswapdi2_reg): Likewise.
33208 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
33210 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
33211 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
33213 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
33216 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
33218 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33219 special handling for target option conflicts between dform
33220 options (-mpower9-dform, -mpower9-dform-vector,
33221 -mpower9-dform-scalar) and -mno-direct-move.
33223 2017-03-27 Richard Biener <rguenther@suse.de>
33225 PR tree-optimization/80181
33226 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
33228 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33230 * config/arc/predicates.md (move_double_src_operand): Replace the
33231 call to move_double_src_operand with a call to address_operand.
33233 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33235 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
33236 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
33237 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
33239 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33241 * config/arc/predicates.md (long_immediate_loadstore_operand):
33242 Consider scaled addresses cases.
33244 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33246 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
33247 restored when in interrupt.
33248 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
33249 doesn't have delay slot.
33251 2017-03-27 Richard Biener <rguenther@suse.de>
33254 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
33255 inlined thunk clones.
33257 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33260 * asan.c (instrument_derefs): Copy over last operand from
33261 original COMPONENT_REF to the new COMPONENT_REF with
33262 DECL_BIT_FIELD_REPRESENTATIVE.
33263 * ubsan.c (instrument_object_size): Likewise.
33265 2017-03-27 Richard Biener <rguenther@suse.de>
33267 PR tree-optimization/80170
33268 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
33269 sure DR/SCEV didnt fold in constants we do not see when looking
33270 at the reference base alignment.
33272 2017-03-27 Richard Biener <rguenther@suse.de>
33274 PR middle-end/80171
33275 * gimple-fold.c (fold_ctor_reference): Properly guard against
33276 NULL return value from canonicalize_constructor_val.
33278 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
33281 * config/i386/i386.c (ix86_expand_builtin)
33282 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
33283 flags reg setting and flags reg using instructions.
33284 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
33285 clobbering instructions to zero extend op2.
33287 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
33289 * doc/install.texi (Configuration) <--with-aix-soname>:
33290 Update link to AIX ld.
33292 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
33294 PR rtl-optimization/80160
33295 PR rtl-optimization/80159
33296 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
33297 reg_alternate_class into account.
33299 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
33302 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
33303 to consider in curr_insn_transform.
33305 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33307 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
33308 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
33309 and emit_mode_inner.
33311 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33313 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
33314 argument to the overloaded builtin variants. Use the new flag to
33315 deprecate certain builtin variants.
33316 * config/s390/s390-builtin-types.def: Add new builtin types.
33317 * config/s390/s390-builtins.h: Support new flags field for
33318 overloaded builtins.
33319 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
33320 (s390_macro_to_expand): Enable vector float data type.
33321 (s390_cpu_cpp_builtins_internal): Indicate support of the new
33322 builtins by incrementing the __VEC__ version number.
33323 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
33325 (s390_resolve_overloaded_builtin): Emit error messages depending
33326 on the builtin flags.
33327 * config/s390/s390.c (s390_expand_builtin): Support additional
33328 flags argument. Change error message to match the messages
33329 emitted in s390-c.c.
33330 * config/s390/s390.md: New UNSPEC_* constants.
33331 (op_type): Add new instruction types.
33332 * config/s390/vecintrin.h: Add new builtins and test data class
33334 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
33335 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
33336 (VEC_INEXACT, VEC_NOINEXACT): New constants.
33337 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
33338 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
33339 ("vec_mergel<mode>"): V_HW -> VEC_HW.
33341 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
33342 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
33343 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
33344 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
33346 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
33347 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
33348 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
33349 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
33351 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
33352 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
33353 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
33354 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
33355 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
33356 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
33357 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
33359 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
33360 ("vec_scatter_element<V_HW_4:mode>_DI")
33361 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
33362 ("vec_fpint<mode>", "vflls")
33363 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
33364 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
33365 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
33366 ("*vec_cmphe<mode>_cc"): ... these.
33368 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
33369 mode constant instead of magic value.
33371 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33373 * config/s390/s390.c (s390_expand_vec_compare): Support other
33374 vector floating point modes than just V2DF.
33375 (s390_expand_vcond): Likewise.
33376 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
33377 (s390_cannot_change_mode_class): Prevent mode changes between TF
33378 and V1TF in vector registers.
33379 * config/s390/s390.md (DF, SF): New mode attributes.
33380 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
33381 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
33382 SFmode support for VRs.
33383 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
33385 (VFT, VF_HW): New mode iterators.
33386 (vw, sdx): New mode attributes.
33387 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
33388 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
33389 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
33390 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
33391 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
33392 also the new vector floating point modes. Renaming to ...
33394 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
33395 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
33396 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
33397 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
33398 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
33399 ("vec_unordered<mode>"): ... these.
33401 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
33402 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
33403 ("*vec_extendv2df"): New insn definitions.
33405 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33407 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
33408 ("mulditi3_2", "*muldi3_sign"): New patterns.
33409 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
33410 rename the pattern definition.
33412 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33414 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
33416 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
33418 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33420 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
33421 instruction if possible.
33422 * config/s390/vector.md (vec_halfnumelts): New mode
33424 ("*vec_vllezlf<mode>"): New pattern.
33426 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33428 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
33429 ("popcountv4si2", "popcountv2di2"): Rename to ...
33430 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
33431 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
33433 ("popcount<mode>2_vxe"): New pattern.
33435 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33437 * common/config/s390/s390-common.c (processor_flags_table): Add
33439 * config.gcc: Add arch12.
33440 * config/s390/driver-native.c (s390_host_detect_local_cpu):
33441 Default to arch12 for unknown CPU model numbers.
33442 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
33443 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
33444 PROCESSOR_max sanity check.
33445 * config/s390/s390-opts.h (enum processor_type): Add
33447 * config/s390/s390.c (processor_table): Add arch12.
33448 (s390_expand_builtin): Add check for B_VXE flag.
33449 (s390_issue_rate): Add PROCESSOR_ARCH12.
33450 (s390_get_sched_attrmask): Likewise.
33451 (s390_get_unit_mask): Likewise.
33452 (s390_sched_score): Enable z13 scheduling for arch12.
33453 (s390_sched_reorder): Likewise.
33454 (s390_sched_variable_issue): Likewise.
33455 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
33457 (s390_tune_attr): Use z13 scheduling also for arch12.
33458 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
33459 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
33460 (TARGET_VXE_P): New macros.
33461 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
33462 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
33463 * config/s390/s390.opt: Add arch12 as processor_type.
33465 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33467 * config/s390/s390.md
33468 ("fixuns_truncdddi2", "fixuns_trunctddi2")
33469 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
33470 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
33472 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
33473 Rename expanders to ...
33475 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
33476 ("fixuns_truncdddi2_emu"): ... these.
33478 ("fixuns_trunc<mode>si2_emu"): New expander.
33480 ("*fixuns_truncdfdi2_z13"): Rename to ...
33481 ("*fixuns_truncdfdi2_vx"): ... this.
33483 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33485 * config/s390/2964.md: Remove the single element vector compare
33486 instructions which are no longer used.
33487 * config/s390/s390.c (s390_select_ccmode): Remove handling of
33489 (s390_canonicalize_comparison): Remove handling of DFmode
33491 (s390_expand_vec_compare_scalar): Remove function.
33492 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
33493 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
33495 ("*cmp<mode>_ccs"): Add wfcdb instruction.
33497 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33499 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
33501 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
33502 will anyway by matched by mov<mode>_64dfp.
33504 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33506 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
33507 vlef/vstef. Add missing operand to vleif.
33509 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33511 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
33512 pair for all vector types with 64 bit elements.
33513 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
33514 * config/s390/vector.md (V_HW_64): ... here.
33515 (V_128_NOSINGLE): New mode iterator.
33516 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
33517 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
33518 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
33519 ("*vec_load_pairv2di"): Change to ...
33520 ("*vec_load_pair<mode>"): ... this one.
33522 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33524 * config/s390/constraints.md: Add comments.
33525 (jKK): Reject element sizes > 8 bytes.
33526 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
33528 * config/s390/s390.md: Add the s_operand checks formerly in
33529 s390_split_ok_p to various splitters where they are still
33531 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
33532 for 128 bit vectors. Plus two splitters.
33534 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33536 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
33539 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33542 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
33543 error if the boundary argument is not constant.
33545 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33547 PR rtl-optimization/80112
33548 * loop-doloop.c (doloop_condition_get): Don't check condition
33549 if cmp isn't SET with IF_THEN_ELSE src.
33551 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33553 PR tree-optimization/80158
33554 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
33555 replacing a candidate statement, also replace it for the
33556 candidate's alternate interpretation.
33557 (replace_rhs_if_not_dup): Likewise.
33558 (replace_one_candidate): Likewise.
33560 2017-03-24 Richard Biener <rguenther@suse.de>
33562 PR tree-optimization/80167
33563 * graphite-isl-ast-to-gimple.c
33564 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
33566 (translate_isl_ast_to_gimple::get_rename): Likewise.
33568 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33570 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
33571 handling of certain combinations of target options, including the
33572 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
33573 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
33575 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33578 * config/arm/arm.md (*load_multiple): Add reload_completed to
33579 matching condition.
33581 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33582 Richard Biener <rguenth@suse.de>
33584 PR tree-optimization/79908
33585 PR tree-optimization/80136
33586 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33587 been cast away, gimplify_and_add suffices.
33589 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
33591 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
33593 2017-03-23 Richard Biener <rguenther@suse.de>
33595 PR tree-optimization/80032
33596 * gimplify.c (gimple_push_cleanup): Forced unconditional
33597 cleanups still have to go to the conditional_cleanups
33600 2017-03-22 Jakub Jelinek <jakub@redhat.com>
33602 PR tree-optimization/80072
33603 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
33605 (next_operand_entry_id): Change type to unsigned int.
33606 (sort_by_operand_rank): Make sure to return the right return value
33607 even if unsigned fields are bigger than INT_MAX.
33608 (struct oecount): Change cnt and id type to unsigned int.
33609 (oecount_hasher::equal): Formatting fix.
33610 (oecount_cmp): Make sure to return the right return value
33611 even if unsigned fields are bigger than INT_MAX.
33612 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
33615 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
33616 TREE_READONLY on result if writing it more than once.
33619 * doc/invoke.texi (-fsanitize=thread): Document that with
33620 -fnon-call-exceptions atomics are not able to throw
33624 * tsan.c: Include tree-eh.h.
33625 (instrument_builtin_call): Call maybe_clean_eh_stmt or
33626 maybe_clean_or_replace_eh_stmt where needed.
33627 (instrument_memory_accesses): Add cfg_changed argument.
33628 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
33629 if it returned true.
33630 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
33632 PR rtl-optimization/63191
33633 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
33634 wrapper function, moved the whole old content into ...
33635 (ix86_delegitimize_address_1): ... this. New inline function.
33636 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
33637 true as last argument instead of ix86_delegitimize_address.
33639 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33641 * config/aarch64/aarch64.c (generic_branch_cost): Copy
33642 cortexa57_branch_cost.
33644 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33646 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
33648 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33651 * doc/md.texi (Constraints): Document wA constraint.
33652 * config/rs6000/constraints.md (wA): New.
33653 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
33654 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
33655 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
33656 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
33658 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
33661 * gimplify.c (is_oacc_declared): New function.
33662 (oacc_default_clause): Use it to set default flags for acc declared
33663 variables inside parallel regions.
33664 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
33665 declared variables.
33666 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
33667 declare attribute to any decl as necessary.
33669 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
33672 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
33673 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
33674 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
33675 (arm_arch_lpae): This.
33676 * config/arm/arm.c (arm_arch7ve): Rename into ...
33677 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
33678 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
33681 2017-03-22 Martin Liska <mliska@suse.cz>
33684 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
33685 error message instead of an ICE.
33687 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33689 * doc/extend.texi (6.11 Additional Floating Types): Revise.
33691 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
33693 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33695 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33698 2017-03-21 Martin Sebor <msebor@redhat.com>
33700 * doc/extend.texi: Use "cannot" instead of "can't."
33701 * doc/hostconfig.texi: Same.
33702 * doc/install.texi: Same.
33703 * doc/invoke.texi: Same.
33704 * doc/loop.texi: Same.
33705 * doc/md.texi: Same.
33706 * doc/objc.texi: Same.
33707 * doc/rtl.texi: Same.
33708 * doc/tm.texi: Same.
33709 * doc/tm.texi.in: Same.
33710 * doc/trouble.texi: Same.
33712 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
33715 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
33716 (collect_checksum_attributes): Set it.
33717 (die_checksum_ordered): Use it.
33719 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33721 PR tree-optimization/79908
33722 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
33723 change: For a VA_ARG whose LHS has been cast away, use
33724 force_gimple_operand to construct the side effects.
33726 2017-03-21 David Malcolm <dmalcolm@redhat.com>
33728 PR translation/80001
33729 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
33730 more amenable to translation.
33731 (oacc_loop_auto_partitions): Likewise.
33733 2017-03-21 Marek Polacek <polacek@redhat.com>
33734 Martin Sebor <msebor@redhat.com>
33736 PR tree-optimization/80109
33737 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
33738 on INTEGRAL_TYPE_P.
33740 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33741 Segher Boessenkool <segher@kernel.crashing.org>
33744 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
33745 check reg_used_between_p between insn and one of succ or succ2
33746 depending on if succ is artificial insn not inserted into insn
33749 2017-03-21 Martin Liska <mliska@suse.cz>
33751 PR gcov-profile/80081
33752 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
33753 * doc/gcc.texi: Include gcov-dump stuff.
33754 * doc/gcov-dump.texi: New file.
33756 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
33758 PR rtl-optimization/79150
33759 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
33760 conditional jump, if the jump is the last insn of the loop.
33762 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33763 Richard Biener <rguenth@suse.de>
33765 PR tree-optimization/79908
33766 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33767 been cast away, use force_gimple_operand to construct the side
33770 2017-03-21 Martin Liska <mliska@suse.cz>
33772 PR libfortran/79956
33773 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
33776 2017-03-21 Brad Spengler <spender@grsecurity.net>
33779 * plugin.c (htab_hash_plugin): New function.
33780 (add_new_plugin): Use it and adjust.
33781 (parse_plugin_arg_opt): Adjust.
33782 (init_one_plugin): Likewise.
33784 2017-03-21 Richard Biener <rguenther@suse.de>
33786 PR tree-optimization/80032
33787 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
33788 if set force the cleanup to happen unconditionally.
33789 (gimplify_target_expr): Push inserted clobbers with force_uncond
33790 to avoid them being removed by control-dependent DCE.
33792 2017-03-21 Richard Biener <rguenther@suse.de>
33794 PR tree-optimization/80122
33795 * tree-inline.c (copy_bb): Do not expans va-arg packs or
33796 va_arg_pack_len when the inlined call stmt requires pack
33798 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
33800 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33803 * tsan.c (instrument_builtin_call): If the memory model argument
33804 is not a constant, assume it is valid.
33807 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
33810 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33812 PR rtl-optimization/79910
33813 * combine.c (can_combine_p): Do not allow combining an I0 or I1
33814 if its dest is used by an insn before I2 (other than the combined
33815 insns themselves, which are properly handled already).
33817 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33820 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33822 * combine.c (record_used_regs): New static function.
33823 (try_combine): Handle situations where there is an additional
33824 instruction between I2 and I3 which needs to have a LOG_LINK
33828 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33830 * combine.c (try_combine): Delete redundant i1 test. Call
33831 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33833 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33836 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
33837 alternatives 13/14.
33839 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33841 PR tree-optimization/80054
33842 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
33843 the optimization if a PHI or any of its arguments is not dominated
33844 by the candidate's basis. Use gphi* rather than gimple* as
33846 (replace_profitable_candidates): Clean up a gimple* variable that
33847 should be a gphi* variable.
33849 2017-03-20 Martin Sebor <msebor@redhat.com>
33852 * doc/extend.texi (attribute constructor): Document present limitation.
33854 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
33857 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
33858 __POWER9_VECTOR__ #ifdef control, change template definition to
33859 use Power9-specific built-in function.
33860 (vec_any_eq): Likewise.
33861 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
33862 to control outcomes from this test.
33863 (vector_ae_<mode>p): For VEC_F modes, likewise.
33865 2017-03-20 Ian Lance Taylor <iant@google.com>
33867 * config/i386/i386.c (ix86_function_regparm): Save an extra
33868 register for -fsplit-stack with DECL_STATIC_CHAIN.
33870 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33873 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
33874 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
33876 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33878 * config/riscv/riscv.c (riscv_print_operand): Use "fence
33880 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
33883 2017-03-20 Marek Polacek <polacek@redhat.com>
33886 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
33888 2017-03-20 Richard Biener <rguenther@suse.de>
33890 PR tree-optimization/80113
33891 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
33892 allocate extra SSA name for PHI def.
33893 (add_close_phis_to_outer_loops): Likewise.
33894 (add_close_phis_to_merge_points): Likewise.
33895 (copy_loop_close_phi_args): Likewise.
33896 (copy_cond_phi_nodes): Likewise.
33898 2017-03-20 Martin Liska <mliska@suse.cz>
33900 PR middle-end/79753
33901 * tree-chkp.c (chkp_build_returned_bound): Do not build
33902 returned bounds for a LHS that's not a BOUNDED_P type.
33904 2017-03-20 Martin Liska <mliska@suse.cz>
33908 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
33909 COMPLEX_CST and VECTOR_CST.
33911 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33914 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
33915 target operand. A new splitter adds the clobber statement in case
33916 the target operand is dead anyway.
33918 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
33920 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
33921 to age-old versions of binutils and glibc.
33923 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
33925 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
33927 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33929 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
33931 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33933 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
33934 requirement for binutils 2.13.
33936 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33938 * combine.c (try_combine): Delete redundant i1 test. Call
33939 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33941 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
33943 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
33944 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
33946 <riscv64-*-elf>: Re-arrange section
33947 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
33948 <riscv32-*-linux>: Likewise.
33949 <riscv64-*-elf>: Likewise
33950 <riscv64-*-linux>: Likewise.
33952 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
33955 * aarch64.opt(verbose-cost-dump): Fix typo.
33957 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
33960 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
33961 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
33963 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33965 * reload.c (find_reloads): When reloading a nonoffsettable address,
33966 use RELOAD_OTHER for it and its address reloads.
33968 PR rtl-optimization/79910
33969 * combine.c (record_used_regs): New static function.
33970 (try_combine): Handle situations where there is an additional
33971 instruction between I2 and I3 which needs to have a LOG_LINK
33974 2017-03-17 Jeff Law <law@redhat.com>
33976 PR tree-optimization/71437
33977 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
33978 conditional in the hash table first.
33979 (vrp_dom_walker::before_dom_children): Extract condition from
33980 ASSERT_EXPR. Record condition, its inverion and any implied
33981 conditions as well.
33983 2017-03-17 Marek Polacek <polacek@redhat.com>
33984 Markus Trippelsdorf <markus@trippelsdorf.de>
33986 PR tree-optimization/80079
33987 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
33990 2017-03-17 Richard Biener <rguenther@suse.de>
33992 PR middle-end/80075
33993 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
33994 Properly verify the LHS before the RHS possibly claims to be
33996 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
33999 2017-03-17 Martin Jambor <mjambor@suse.cz>
34001 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34002 (List of -O2 options): Likewise.
34003 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34006 2017-03-17 Tom de Vries <tom@codesourcery.com>
34008 * gcov-dump.c (print_usage): Print bug_report_url.
34010 2017-03-17 Richard Biener <rguenther@suse.de>
34012 PR middle-end/80050
34013 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34014 (parser::peek): Likewise.
34016 2017-03-17 Richard Biener <rguenther@suse.de>
34018 PR tree-optimization/80048
34019 * sese.c (free_sese_info): Properly release rename_map and
34020 copied_bb_map elements.
34022 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34024 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34025 Add linked-list forward and backlinks. Insert on
34026 construction, remove on destruction.
34027 (class pass_store_merging): Add m_stores_head field.
34028 (pass_store_merging::terminate_and_process_all_chains):
34029 Iterate over m_stores_head list.
34030 (pass_store_merging::terminate_all_aliasing_chains):
34032 (pass_store_merging::execute): Check for debug stmts first.
34033 Push new chains onto the m_stores_head stack.
34035 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34038 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34039 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34040 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
34042 2017-03-16 Jeff Law <law@redhat.com>
34044 PR tree-optimization/71437
34045 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
34046 member function. Implementation moved into after_dom_children
34047 member function and into the threader's thread_outgoing_edges
34049 (dom_opt_dom_walker::after_dom_children): Simplify by moving
34050 some code into new thread_outgoing_edges.
34051 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
34052 definition. Simplify marker handling (do it here). Assume we always
34053 have the available expression and the const/copies tables.
34054 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
34056 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
34057 * tree-vrp.c (equiv_stack): No longer file scoped.
34058 (vrp_dom_walker): New class.
34059 (vrp_dom_walker::before_dom_children): New member function.
34060 (vrp_dom_walker::after_dom_children): Likewise.
34061 (identify_jump_threads): Setup domwalker. Use it rather than
34062 walking edges in a random order by hand. Simplify setup/finalization.
34063 (finalize_jump_threads): Remove.
34064 (vrp_finalize): Do not call identify_jump_threads here.
34065 (execute_vrp): Do it here instead and call thread_through_all_blocks
34068 PR tree-optimization/71437
34069 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
34071 (simplify_stmt_for_jump_threading): Add basic_block argument. All
34073 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
34074 (dom_opt_dom_walker::thread_across_edge): Remove
34075 handle_dominating_asserts argument. All callers changed.
34076 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
34077 changes. Remove calls to lhs_of_dominating_assert. Other
34078 uses of handle_dominating_asserts turn into unconditional code
34079 (simplify_control_stmt_condition_1): Likewise.
34080 (simplify_control_stmt_condition): Likewise.
34081 (thread_through_normal_block, thread_across_edge): Likewise.
34082 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
34083 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
34084 object if it is not an SSA_NAME.
34085 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
34086 before calling into the VRP specific simplifiers.
34087 (identify_jump_threads): Remove handle_dominating_asserts
34090 2017-03-16 Jakub Jelinek <jakub@redhat.com>
34093 * tree-diagnostic.c (default_tree_printer): No longer static.
34094 * tree-diagnostic.h (default_tree_printer): New prototype.
34096 2017-03-16 Tamar Christina <tamar.christina@arm.com>
34098 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
34099 Change ins into fmov.
34101 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34103 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
34104 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
34105 Use h_con constraint for operand 1.
34106 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
34107 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
34109 2017-03-15 Jeff Law <law@redhat.com>
34111 PR tree-optimization/71437
34112 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
34113 (record_temporary_equivalences): Use it.
34115 PR tree-optimization/71437
34116 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
34117 tree-ssa-scopedtables.
34118 (lookup_avail_expr, build_and_record_new_cond): Likewise.
34119 (record_conditions, record_cond, vuse_eq): Likewise.
34120 (record_edge_info): Adjust to API tweak of record_conditions.
34121 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
34122 (record_temporary_equivalences, optimize_stmt): Likewise.
34123 (eliminate_redundant_computations): Likewise.
34124 (record_equivalences_from_stmt): Likewise.
34125 * tree-ssa-scopedtables.c: Include options.h and params.h.
34126 (vuse_eq): New function, moved from tree-ssa-dom.c
34127 (build_and_record_new_cond): Likewise.
34128 (record_conditions): Likewise. Accept vector of conditions rather
34129 than edge_equivalence structure for first argument.
34130 for the first argument.
34131 (avail_exprs_stack::lookup_avail_expr): New member function, moved
34132 from tree-ssa-dom.c.
34133 (avail_exprs_stack::record_cond): Likewise.
34134 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
34135 from tree-ssa-dom.c.
34136 (avail_exprs_stack): Add new member functions lookup_avail_expr
34138 (record_conditions): Declare.
34140 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
34143 * lra-constraints.c (process_alt_operands): Increase reject for
34144 reloading an input/output operand.
34146 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
34149 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
34150 insns to convert from signed/unsigned char/short to IEEE 128-bit
34152 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
34154 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
34157 * config/i386/i386.c (ix86_vector_duplicate_value): Create
34158 subreg of inner mode for values already in registers.
34160 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
34162 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
34163 iteration reg is used after the loop.
34165 2017-03-14 Martin Sebor <msebor@redhat.com>
34167 PR tree-optimization/79800
34168 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
34169 precision in negative-positive range.
34170 (format_floating): Call non-const overload with adjusted precision.
34172 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34175 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
34178 2017-03-14 Martin Sebor <msebor@redhat.com>
34180 PR middle-end/80020
34181 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
34182 * builtins.def (aligned_alloc): Use it.
34185 * Makefile.in (GTFILES): Add calls.c.
34186 * calls.c: Include "gt-calls.h".
34188 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
34190 PR rtl-optimization/79728
34191 * regs.h (struct target_regs): New field
34192 x_contains_allocatable_regs_of_mode.
34193 (contains_allocatable_regs_of_mode): New macro.
34194 * reginfo.c (init_reg_sets_1): Initialize it, and change
34195 contains_reg_of_mode so it includes global regs as well.
34196 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
34197 rather than contains_regs_of_mode.
34199 2017-03-14 Martin Liska <mliska@suse.cz>
34201 * doc/invoke.texi: Document options that can't be combined with
34202 -fcheck-pointer-bounds.
34204 2017-03-14 Martin Liska <mliska@suse.cz>
34206 PR middle-end/79831
34207 * doc/invoke.texi (-Wchkp): Document the option.
34209 2017-03-14 Martin Liska <mliska@suse.cz>
34211 * Makefile.in: Install gcov-dump.
34213 2017-03-14 Martin Liska <mliska@suse.cz>
34215 * multiple_target.c (expand_target_clones): Bail out for
34216 an invalid attribute.
34218 2017-03-14 Richard Biener <rguenther@suse.de>
34220 * alias.c (struct alias_set_entry): Pack properly.
34221 * cfgloop.h (struct loop): Likewise.
34222 * cse.c (struct set): Likewise.
34223 * ipa-utils.c (struct searchc_env): Likewise.
34224 * loop-invariant.c (struct invariant): Likewise.
34225 * lra-remat.c (struct cand): Likewise.
34226 * recog.c (struct change_t): Likewise.
34227 * rtl.h (struct address_info): Likewise.
34228 * symbol-summary.h (function_summary): Likewise.
34229 * tree-loop-distribution.c (struct partition): Likewise.
34230 * tree-object-size.c (struct object_size_info): Likewise.
34231 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
34232 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
34233 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
34234 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
34235 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
34236 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
34237 (struct _stmt_vec_info): Likewise.
34239 2017-03-14 Martin Liska <mliska@suse.cz>
34242 * multiple_target.c (create_dispatcher_calls): Check that
34243 a target can create a function dispatcher.
34245 2017-03-14 Martin Liska <mliska@suse.cz>
34248 * multiple_target.c (expand_target_clones): Drop local.local
34249 flag for default implementation.
34251 2017-03-14 Richard Biener <rguenther@suse.de>
34253 PR tree-optimization/80030
34254 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
34256 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
34258 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
34259 gcc_fallthrough() instead of __attribute__((fallthrough));
34261 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34263 * doc/gcc.texi: Remove "up" link to (DIR).
34264 * doc/gccint.texi: Ditto.
34266 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34268 * doc/install.texi (Specific) <avr>: Remove reference to
34271 2017-03-13 Jeff Law <law@redhat.com>
34273 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
34274 attribute rather than comments.
34276 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
34277 match_scratch operand is highest.
34279 2017-03-13 Martin Liska <mliska@suse.cz>
34281 PR middle-end/78339
34282 * ipa-pure-const.c (warn_function_noreturn): If the declarations
34283 is a CHKP clone, use original declaration.
34285 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34287 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
34288 (arc_conditional_register_usage): Use a different allocation order
34289 when optimizing for size.
34290 * common/config/arc/arc-common.c (arc_option_optimization_table):
34291 Section anchors default on when optimizing for size.
34293 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34295 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
34297 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34299 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
34300 * config/arc/arc.md (cpu_facility): Add cd variant.
34301 (*movqi_insn): Add code density variant.
34302 (*movhi_insn): Likewise.
34303 (*movqi_insn): Likewise.
34304 (*addsi3_mixed): Likewise.
34305 (subsi3_insn): Likewise.
34307 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34309 * config/arc/arc.md (movsi_cond_exec): Update constraint.
34311 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34313 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
34314 expressions with MINUS and UNARY ops.
34316 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34319 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
34321 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
34322 between vec_select and vector argument.
34323 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
34324 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
34325 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
34326 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
34327 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
34328 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
34330 2017-03-13 Richard Biener <rguenther@suse.de>
34333 * params.def (vect-max-peeling-for-alignment): Fix typo.
34335 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34337 * doc/install.texi (Specific) <mips-*-*>: Remove description of
34338 issue that only occurred with binutils below 2.18.
34340 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34342 * doc/install.texi (Specific) <cris-axis-elf>: No longer
34343 refer to binutils 2.11/2.12 minimum.
34345 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34347 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
34348 ftp.kernel.org and simplify binutils requirement.
34350 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
34352 * doc/invoke.texi (Warning Options): Fix spelling of link-time
34354 (Optimize Options): Ditto. Also remove redundancy.
34356 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34358 PR translation/79848
34359 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
34361 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
34362 to G_ to avoid double translation.
34364 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34366 PR translation/79923
34367 * auto-profile.c (get_combined_location): Convert leading
34368 character of diagnostics to lower case and remove trailing period.
34369 (read_profile): Likewise for various diagnostics.
34370 * config/arm/arm.c (arm_option_override): Remove trailing period
34371 from various diagnostics.
34372 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
34373 (msp430_expand_delay_cycles): Likewise.
34375 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34378 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
34379 full command-line argument, rather than just "str".
34380 (aarch64_validate_march): Likewise.
34381 (aarch64_validate_mtune): Likewise.
34383 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
34385 PR rtl-optimization/78911
34386 * lra-assigns.c (must_not_spill_p): New function.
34387 (spill_for): Use it.
34389 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34391 PR tree-optimization/79981
34392 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
34393 ATOMIC_COMPARE_EXCHANGE ifn result.
34394 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
34395 IFN_ATOMIC_COMPARE_EXCHANGE.
34397 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34400 * opts.c (parse_sanitizer_options): Add missing question mark to
34401 "did you mean" message.
34403 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34405 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
34407 (VMULEUH_UNS): Likewise.
34408 (VMULOUB_UNS): Likewise.
34409 (VMULOUH_UNS): Likewise.
34410 * config/rs6000/rs6000.c (builtin_function_type): Remove
34411 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
34413 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34416 * read-rtl-function.c (function_reader::read_rtx_operand): Update
34417 x with result of extra_parsing_for_operand_code_0.
34418 (function_reader::extra_parsing_for_operand_code_0): Convert
34419 return type from void to rtx, returning x. When reading
34420 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
34421 larger size containing struct block_symbol.
34423 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
34425 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
34426 -mfloat128-hardware without -m64.
34428 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
34431 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
34432 entries to the case statement that marks unsigned arguments to
34433 overloaded functions.
34435 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34437 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
34438 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
34440 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
34443 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
34444 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
34446 2017-03-10 Martin Liska <mliska@suse.cz>
34450 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
34451 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
34454 2017-03-10 Olivier Hainque <hainque@adacore.com>
34456 * tree-switch-conversion (array_value_type): Start by resetting
34457 candidate type to it's main variant.
34459 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34461 PR rtl-optimization/79909
34462 * combine.c (try_combine): Use simplify_replace_rtx on individual
34463 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
34464 of the whole CALL_INSN_FUNCTION_USAGE.
34466 PR tree-optimization/79972
34467 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
34468 get_range_info on SSA_NAMEs. Formatting fixes.
34470 2017-03-10 Richard Biener <rguenther@suse.de>
34471 Jakub Jelinek <jakub@redhat.com>
34473 PR tree-optimization/77975
34474 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
34475 edge to be constant.
34476 (get_val_for): For constant x return it. Formatting fix.
34477 (loop_niter_by_eval): Avoid pointless looping if the next iteration
34478 would use the same bases as the current one.
34480 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34482 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
34483 instead of vec_select for V1TImode.
34484 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
34486 (VSX_LE_128): Add V1TI to this mode iterator.
34487 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
34488 (*vsx_le_perm_store_<mode>): Likewise.
34489 (pre-reload splitter for VSX stores): Likewise.
34490 (post-reload splitter for VSX stores): Likewise.
34491 (*vsx_xxpermdi2_le_<mode>): Likewise.
34492 (*vsx_lxvd2x2_le_<mode>): Likewise.
34493 (*vsx_stxvd2x2_le_<mode>): Likewise.
34495 2017-03-09 Michael Eager <eager@eagercon.com>
34497 Correct failures with --enable-checking=yes,rtl.
34499 * config/microblaze/microblaze.c (microblaze_expand_shift):
34500 Replace GET_CODE test with CONST_INT_P and INTVAL test with
34501 test for const0_rtx.
34502 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
34503 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
34505 2017-03-09 Richard Biener <rguenther@suse.de>
34507 PR tree-optimization/79977
34508 * graphite-scop-detection.c (scop_detection::merge_sese):
34509 Handle the case of extra exits to blocks dominating the entry.
34511 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
34513 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
34516 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
34518 PR rtl-optimization/79949
34519 * lra-constraints.c (process_alt_operands): Check memory when
34520 trying to predict a cycle. Print about the overall increase.
34522 2017-03-09 Richard Biener <rguenther@suse.de>
34524 PR middle-end/79971
34525 * gimple-expr.c (useless_type_conversion_p): Preserve
34526 TYPE_SATURATING for fixed-point types.
34528 2017-03-09 Richard Biener <rguenther@suse.de>
34531 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
34532 alignment of BLKmode params.
34534 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34537 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
34538 (VALL_NO_V2Q): Likewise.
34540 * config/aarch64/aarch64-simd.md
34541 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
34543 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
34544 VALL_NO_V2Q mode iterator.
34545 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
34547 2017-03-09 Martin Liska <mliska@suse.cz>
34549 PR tree-optimization/79631
34550 * tree-chkp-opt.c (chkp_is_constant_addr): Call
34551 tree_int_cst_sign_bit just for INTEGER constants.
34553 2017-03-09 Martin Liska <mliska@suse.cz>
34557 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
34560 2017-03-09 Marek Polacek <polacek@redhat.com>
34563 * tree.c (inchash::add_expr): Handle TREE_VEC.
34565 2017-03-09 Martin Liska <mliska@suse.cz>
34568 (chkp_narrow_size_and_offset): New function.
34569 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
34570 (void chkp_parse_bit_field_ref): New function.
34571 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
34572 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
34574 2017-03-09 Martin Liska <mliska@suse.cz>
34577 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
34578 (chkp_find_bounds_1): Remove gcc_unreachable.
34580 2017-03-09 Jakub Jelinek <jakub@redhat.com>
34583 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
34584 BUILT_IN_SYNC*, determine the access type from the size suffix and
34585 always build a MEM_REF with that type. Handle forgotten
34586 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
34589 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
34590 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
34591 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
34592 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
34593 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
34594 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
34595 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
34596 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
34597 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
34598 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
34599 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
34600 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
34601 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
34602 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
34603 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
34604 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
34605 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
34606 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
34607 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
34608 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
34609 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
34610 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
34611 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
34612 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
34613 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
34614 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
34615 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
34616 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
34617 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
34618 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
34619 definitions outside of __OPTIMIZE__ guarded section.
34622 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
34623 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
34624 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
34625 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
34628 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34630 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
34631 ("vfenez<mode>"): Add missing constraints.
34633 2017-03-08 Martin Sebor <msebor@redhat.com>
34636 * config/nds32/nds32.c (nds32_option_override):
34637 Fix misspelled diagnostic.
34639 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34642 * gimplify.c (gimplify_omp_for): Replace index var in outer
34643 taskloop statement with an artificial variable and add
34644 OMP_CLAUSE_PRIVATE clause for it.
34646 2017-03-08 Richard Biener <rguenther@suse.de>
34648 PR tree-optimization/79955
34649 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
34650 for accesses that are completely outside of the variable.
34652 2017-03-08 Andrew Haley <aph@redhat.com>
34654 PR tree-optimization/79943
34655 * tree-ssa-loop-split.c (compute_new_first_bound): When
34656 calculating the new upper bound, (END-BEG) should be added, not
34659 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34661 * config/avr/avr.md (setmemhi): Make sure match_dup
34662 operand number comes before match_scratch.
34664 2017-03-08 Richard Biener <rguenther@suse.de>
34666 PR tree-optimization/79920
34667 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
34668 with ncopies == 1 to ...
34669 (vect_transform_slp_perm_load): ... here. Properly compute
34670 all element loads by iterating VF times over the group. Do
34671 not handle ncopies (computed in a broken way) in
34672 vect_create_mask_and_perm.
34674 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34677 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
34678 is a uniform vector, use uniform_vector_p return value instead of
34679 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
34681 2017-03-07 Marek Polacek <polacek@redhat.com>
34683 PR middle-end/79809
34684 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
34685 (alloca_call_type): Likewise.
34687 2017-03-07 Martin Liska <mliska@suse.cz>
34689 * gcov.c (process_args): Put comment to correct location.
34691 2017-03-07 Martin Liska <mliska@suse.cz>
34693 PR middle-end/68270
34694 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
34695 Use array_at_struct_end_p instead of DECL_CHAIN (field).
34696 (chkp_narrow_bounds_for_field): Likewise.
34697 (chkp_parse_array_and_component_ref): Pass one more argument to
34700 2017-03-07 Richard Biener <rguenther@suse.de>
34702 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
34705 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
34707 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
34708 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
34710 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34713 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
34714 to end of description.
34715 (PARAM_MAX_STORES_TO_MERGE): Likewise.
34717 2017-03-07 Jakub Jelinek <jakub@redhat.com>
34719 PR rtl-optimization/79901
34720 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
34722 (*avx512f_<code><mode>3<mask_name>): ... this.
34723 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
34724 iterator instead of VI8_AVX2_AVX512BW.
34726 PR rtl-optimization/79901
34727 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
34728 min/max expander, expand it using expand_vec_cond_expr.
34731 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
34734 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34737 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
34738 to void * for PCH reasons.
34739 * dwarf2out.c (output_loc_operands, output_die): Cast
34740 v.val_vec.array to unsigned char *.
34742 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
34745 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
34748 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
34750 PR rtl-optimization/79571
34751 * lra-constraints.c (process_alt_operands): Calculate static
34752 reject and subtract it from overall when only addresses will be
34755 2017-03-06 Julia Koval <julia.koval@intel.com>
34758 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
34759 incoming stack boundary to 128 for 64-bit targets.
34761 2017-03-06 Richard Biener <rguenther@suse.de>
34763 PR tree-optimization/79894
34764 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
34765 to NULL after folding it.
34767 2017-03-06 Richard Biener <rguenther@suse.de>
34769 PR tree-optimization/79824
34770 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
34771 check disabling peeling for gaps.
34773 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
34775 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
34776 attributes): Document gettimeofday.
34778 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34780 * config/s390/s390.c (s390_option_override_internal): Set
34781 PARAM_MIN_VECT_LOOP_BOUND
34783 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34785 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
34786 * config/s390/s390.md: Likewise.
34788 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34791 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
34792 (<avx2_avx512>_perm<mode>): Rename to ...
34793 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
34795 (<avx512>_perm<mode>_mask): Rename to ...
34796 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
34798 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
34799 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
34800 instead of VI8F_256_512.
34801 (avx512f_perm<mode>): New define_expand.
34802 (avx512f_perm<mode>_mask): Likewise.
34803 (avx512f_perm<mode>_1<mask_name>): New define_insn.
34804 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
34806 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34808 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
34809 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
34811 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
34813 2017-03-06 Martin Liska <mliska@suse.cz>
34816 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
34817 when having a SSA NAME w/o VAR_DECL assigned to it.
34819 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34821 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
34822 msa_dpsub_<su>_d): Fix MODE for vec_select.
34824 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34826 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
34828 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
34830 2017-03-06 Richard Biener <rguenther@suse.de>
34832 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
34833 * plugin.c (register_plugin_info): Likewise.
34834 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
34836 2017-03-05 Jakub Jelinek <jakub@redhat.com>
34838 * config/i386/sse.md (sse_storehps, sse_storelps,
34839 avx_<castmode><avxsizesuffix>_<castmode>,
34840 avx512f_<castmode><avxsizesuffix>_<castmode>,
34841 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
34842 in condition that at least one operand is not a MEM.
34844 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34846 PR middle-end/79805
34847 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
34848 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
34850 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
34851 gimple_call_nothrow_p flag based on whether original builtin can throw.
34852 If it can, emit following stmts on the fallthrough edge.
34853 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
34854 don't create new bb if inserting just debug stmts on the edge, try to
34855 insert them on the fallthru bb or just reset debug stmts.
34857 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
34860 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
34861 restore recog_data (including the operand rtxes inside it) around
34862 the call to get_insn_template.
34864 2017-03-03 Martin Sebor <msebor@redhat.com>
34866 PR tree-optimization/79699
34867 * context.c (context::~context): Free MPFR caches to avoid
34868 a memory leak on program exit.
34870 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34872 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
34873 Use wide_int::ulow () instead of .elt (0).
34875 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34877 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
34878 (*pushxf): Limit oF constraint to 32bit targets and add oC
34879 constraint for 64bit targets.
34880 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
34881 (*pushdf): Change rmF constraint to rmC.
34883 2017-03-03 Martin Liska <mliska@suse.cz>
34885 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
34886 Remove unused variable.
34888 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34891 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
34892 is a memory operand, increase num_memory.
34893 (ix86_expand_args_builtin): Likewise.
34895 2017-03-03 Jan Hubicka <jh@suse.cz>
34898 * ipa-devirt.c (maybe_record_node): Properly handle
34899 __cxa_pure_virtual visibility.
34901 2017-03-03 Martin Liska <mliska@suse.cz>
34903 PR tree-optimization/79803
34904 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
34906 (pass_loop_prefetch::execute): Disabled optimization if an
34907 assumption about L1 cache size is not met.
34909 2017-03-03 Martin Liska <mliska@suse.cz>
34911 PR rtl-optimization/79574
34912 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
34913 (hash_scan_set): Likewise.
34914 (dump_hash_table): Likewise.
34915 (hoist_code): Likewise.
34917 2017-03-03 Richard Biener <rguenther@suse.de>
34919 * fixed-value.c (fixed_from_string): Restore use of elt (1)
34920 in place of uhigh ().
34921 (fixed_convert_from_real): Likewise.
34923 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34926 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
34928 2017-03-03 Richard Biener <rguenther@suse.de>
34930 PR middle-end/79818
34931 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
34932 TYPE_OVERFLOW_UNDEFINED check.
34934 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34936 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
34938 (vector_ae_<mode>_p): Likewise.
34939 (vector_nez_<mode>_p): Likewise.
34940 (vector_ne_v2di_p): Likewise.
34941 (vector_ae_v2di_p): Likewise.
34942 (vector_ne_<mode>_p): Likewise.
34943 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
34945 (vsx_tsqrt<mode>2_fe): Likewise.
34947 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
34950 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
34952 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34954 PR rtl-optimization/79780
34955 * cprop.c (one_cprop_pass): When second and further conditional trap
34956 in a single basic block is turned into an unconditional trap, turn it
34957 into a deleted note to avoid RTL verification failures.
34959 2017-03-02 Richard Biener <rguenther@suse.de>
34961 * fold-const.c (const_binop): Use ulow () instead of elt (0).
34963 2017-03-02 Richard Biener <rguenther@suse.de>
34965 PR tree-optimization/79345
34967 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
34968 param and abort the walk, returning -1 if it is hit.
34969 (walk_aliased_vdefs): Take a limit param and pass it on.
34970 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
34971 defaulting to 0 and return a signed int.
34972 * tree-ssa-uninit.c (struct check_defs_data): New struct.
34973 (check_defs): New helper.
34974 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
34975 about uninitialized memory.
34976 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
34977 bogus uninitialized warning.
34978 (fixed_convert_from_real): Likewise.
34980 2017-03-02 Bin Cheng <bin.cheng@arm.com>
34982 PR tree-optimization/66768
34983 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
34984 iv_use if base object can't be determined.
34986 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34988 PR tree-optimization/79345
34989 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
34990 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
34991 (get_pattern_stats): Initialize it.
34992 * genemit.c (gen_expand): Verify match_scratch numbers come after
34993 match_operand/match_dup numbers.
34994 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
34995 match_scratch numbers.
34996 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
34998 * config/s390/s390.md (trunctdsd2): Likewise.
35000 2017-03-02 Richard Biener <rguenther@suse.de>
35002 * wide-int.h (wide_int_storage::operator=): Implement in terms
35005 2017-03-02 Richard Biener <rguenther@suse.de>
35007 PR tree-optimization/79777
35008 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35009 the to insert expression to sth existing.
35011 2017-03-01 Martin Sebor <msebor@redhat.com>
35013 PR middle-end/79692
35014 * gimple-ssa-sprintf.c
35015 (directive::known_width_and_precision): New function.
35016 (format_integer): Use it.
35017 (get_mpfr_format_length): Consider the full range of precision
35018 when computing %g output with the # flag. Set the likely byte
35019 count to 3 rather than 1 when precision is indeterminate.
35020 (format_floating): Correct the lower bound of precision.
35022 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35024 * doc/invoke.texi: Document default code model for 64-bit Linux.
35026 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35029 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35030 udiv rather than div since input pattern is unsigned.
35032 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35034 * config/i386/i386.c (print_reg): Warn for values of
35035 unsupported size in integer register.
35037 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35040 * config/rs6000/predicates.md (current_file_function_operand): Do
35041 not allow self calls to be local if the function is replaceable.
35043 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
35046 * config/rs6000/altivec.h (vec_ctz and others): Change the
35047 preprocessor macro that controls conditional compilation from
35048 _ARCH_PWR9 to __POWER9_VECTOR__.
35049 (vec_all_ne): Change parameterization of __altivec_scalar_pred
35050 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
35051 control (instead of _ARCH_PWR9 control) so that template
35052 definition uses power9-specific function.
35053 (vec_any_eq): Likewise.
35054 (vec_all_ne): Change macro definition to use a power9-specific
35055 expansion under #ifdef __POWER9_VECTOR__ control (instead of
35056 _ARCH_PWR9 control).
35057 (vec_any_eq) Likewise.
35058 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
35059 expansion for CMPNEF to remove support for xvcmpnesp instruction.
35060 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
35061 support for xvcmpnedp instruction.
35062 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
35063 macro expansion so that Power9 implementation of vec_all_ne does
35064 not use the AltiVec predicate framework.
35065 (VCMPNEH_P): Likewise.
35066 (VCMPNEW_P): Likewise.
35067 (VCMPNED_P): Likewise.
35068 (VCMPNEFP_P): Likewise.
35069 (VCMPNEDP_P): Likewise.
35070 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
35071 implementation of vec_any_eq to not use AltiVec predicate
35073 (VCMPAEH_P): Likewise.
35074 (VCMPAEW_P): Likewise.
35075 (VCMPAED_P): Likewise.
35076 (VCMPAEFP_P): Likewise.
35077 (VCMPAEDP_P): Likewise.
35078 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
35079 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
35080 not use the AltiVec predicate framework.
35081 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
35082 of vec_any_eq to not use AltiVec predicate framework.
35083 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35084 support for predefined __POWER9_VECTOR__ macro to indicate that
35085 Power9 instruction selection is enabled.
35086 (altivec_overloaded_builtins): Remove extraneous
35087 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
35088 function argument types RS6000_BTI_bool_V16QI and
35089 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
35090 entry for overloaded function argument types RS6000_BTI_bool_V4SI
35091 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
35092 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
35093 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
35094 Power9 for implementations of vec_cmpne. Change the signature for
35095 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
35096 (representing vec_all_ne) to remove the previously described first
35097 argument of type RS6000_BTI_INTSI, as this was an artifact of
35098 reliance on the AltiVec predicate framework, which is no longer
35099 used in the implementation of these functions. Add
35100 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
35101 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
35102 since, unlike the AltiVec predicate framework implementation, we
35103 do not share function descriptors between vec_alle and vec_anyeq.
35104 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
35105 set of modes that receive special treatment even when
35106 TARGET_P9_VECTOR is true. The special treatment emits code that
35107 does not depend on Power9 instructions.
35108 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
35109 define_expand to not rely on AltiVec predicate framework.
35110 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35112 (vector_ne_v2di_p): Change this define_expand to not rely on
35113 AltiVec predicate framework.
35114 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
35116 (vector_ne_<mode>_p): Change this define_expand to not rely on
35117 AltiVec predicate framework.
35118 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35120 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
35121 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
35122 define_insn pattern.
35123 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
35124 define_insn pattern because the xvcmpne<VSs>. instruction is not
35126 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
35127 instruction is not supported.
35129 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35131 * config/nvptx/nvptx.c: Include intl.h.
35133 2017-03-01 Martin Jambor <mjambor@suse.cz>
35136 * ipa-prop.h (ipa_bits): Removed field known.
35137 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
35138 to pointers. Adjusted their comments to warn about their sharing.
35139 (ipcp_transformation_summary): Change bits to a vector of pointers.
35140 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
35141 (ipa_get_ipa_bits_for_value): Declare.
35142 * tree-vrp.h (value_range): Mark as GTY((for_user)).
35143 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
35144 (ipa_bits_hash_table): Likewise.
35145 (ipa_vr_ggc_hash_traits): Likewise.
35146 (ipa_vr_hash_table): Likewise.
35147 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
35148 being pointers and vr_known being removed.
35149 (ipa_set_jf_unknown): Likewise.
35150 (ipa_get_ipa_bits_for_value): New function.
35151 (ipa_set_jfunc_bits): Likewise.
35152 (ipa_get_value_range): New overloaded functions.
35153 (ipa_set_jfunc_vr): Likewise.
35154 (ipa_compute_jump_functions_for_edge): Use the above functions to
35155 construct bits and vr parts of jump functions.
35156 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
35157 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35159 (ipcp_grow_transformations_if_necessary): Also allocate
35160 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35162 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
35163 them. Fix too long lines.
35164 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
35165 vr_known being removed.
35166 (ipa_read_jump_function): Use new setter functions to construct bits
35167 and vr parts of jump functions or set them to NULL.
35168 (write_ipcp_transformation_info): Adjust for bits being pointers.
35169 (read_ipcp_transformation_info): Likewise.
35170 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
35172 Include gt-ipa-prop.h.
35173 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
35175 (ipcp_store_bits_results): Likewise.
35176 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
35177 Do not write to existing jump functions but use a temporary instead.
35179 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35182 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
35183 attempt to use its first operand as BIT_FIELD_REF base.
35185 2017-03-01 Richard Biener <rguenther@suse.de>
35187 PR middle-end/79721
35188 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
35189 interpolating formula in wrapping arithmetic.
35190 (chrec_apply): Convert chrec_evaluate return value to wanted type.
35192 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35194 PR tree-optimization/79734
35195 * tree-vect-generic.c (expand_vector_condition): Optimize
35196 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
35197 Handle VEC_COND_EXPR where comparison has different inner width from
35198 type's inner width.
35200 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
35202 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
35203 markup, and similar issues. Remove @opindex entries for things
35204 that aren't options. Add missing -mmpy-option entries.
35206 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35208 PR tree-optimization/79737
35209 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
35210 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
35211 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
35212 instead of byte_size. Formatting fix.
35213 (shift_bytes_in_array_right): Formatting fix.
35215 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
35218 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
35219 condition on optimize for the leaf function test.
35221 2017-02-28 Martin Liska <mliska@suse.cz>
35224 * read-rtl-function.c (function_reader::handle_unknown_directive):
35225 Bail out when one uses -flto.
35227 2017-02-28 Martin Liska <mliska@suse.cz>
35229 * common.opt: Replace space with tabular for options of <number>
35231 * config/i386/i386.opt: Show <number> value for
35232 -mlarge-data-threshold.
35233 * opts.c (print_filtered_help): Do not display number in hexadecimal
35236 2017-02-28 Martin Liska <mliska@suse.cz>
35238 * common.opt: Fix --help=option -Q for options which are of
35241 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
35243 * config/i386/i386.c (print_reg): Error out for values
35244 of 8-bit size in invalid integer register.
35246 2017-02-28 Martin Sebor <msebor@redhat.com>
35248 PR tree-optimization/79691
35249 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
35251 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35254 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
35255 gcc_unreachable with output_operand_lossage.
35257 2017-02-28 Richard Biener <rguenther@suse.de>
35259 PR tree-optimization/79740
35260 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
35262 (visit_nary_op): Insert the nary into the hashtable if we
35263 pattern-matched sth.
35264 * tree-ssa-pre.c (eliminate_insert): Robustify.
35266 2017-02-28 Richard Biener <rguenther@suse.de>
35268 PR middle-end/79731
35269 * fold-const.c (decode_field_reference): Reject out-of-bound
35272 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35274 * config/i386/i386.c: Include intl.h.
35275 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
35276 instead of just cond ? "..." : "...".
35277 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
35278 * coverage.c (read_counts_file): Likewise.
35279 * omp-offload.c: Include intl.h.
35280 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
35281 of just cond ? "..." : "...".
35282 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
35283 of just cond ? "..." : "...".
35285 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
35288 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
35290 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
35291 'tune for' CPU name.
35292 * config/arm/arm-cpu-data.h: Regenerated.
35294 2017-02-28 Richard Biener <rguenther@suse.de>
35296 PR tree-optimization/79732
35297 * tree-inline.c (expand_call_inline): Do not shadow var.
35299 2017-02-28 Richard Biener <rguenther@suse.de>
35301 PR tree-optimization/79723
35302 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
35303 address-space properly.
35305 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
35307 * doc/optinfo.texi (Optimization groups): Fix option used for
35309 * doc/invoke.texi (-fopt-info): Document "omp".
35310 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
35311 (OPTGROUP_ALL): Add OPTGROUP_OMP.
35312 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
35313 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
35314 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
35316 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
35318 * dumpfile.c (optgroup_options): Instead of "openmp", associate
35319 OPTGROUP_OMP with "omp".
35321 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
35324 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
35325 for arithmetic shift of unsigned V2DI.
35327 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
35329 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
35330 arc/linux.h headers.
35331 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
35332 (LINK_SPEC): Likewise.
35333 (ARC_TLS_EXTRA_START_SPEC): Likewise.
35334 (EXTRA_SPECS): Likewise.
35335 (STARTFILE_SPEC): Likewise.
35336 (ENDFILE_SPEC): Likewise.
35337 (LIB_SPEC): Likewise.
35338 (TARGET_SDATA_DEFAULT): Likewise.
35339 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
35340 (MULTILIB_DEFAULTS): Likewise.
35341 (DWARF2_UNWIND_INFO): Likewise.
35342 * config/arc/big.h: New file.
35343 * config/arc/elf.h: Likewise.
35344 * config/arc/linux.h: Likewise.
35345 * config/arc/t-uClibc: Remove.
35347 2017-02-27 Bin Cheng <bin.cheng@arm.com>
35349 PR tree-optimization/77536
35350 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
35351 (tree_transform_and_unroll_loop): Use above function to compute the
35352 estimated niter of unrolled loop and use it when scaling profile.
35353 Also use count info rather than frequency if it's non-zero.
35354 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
35355 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
35356 (vect_transform_loop): Call above function.
35358 2017-02-27 Richard Biener <rguenther@suse.de>
35360 PR tree-optimization/45397
35361 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
35362 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
35363 (visit_nary_op): Add pattern matching for CSEing sign-changed
35364 or truncated operations with wider ones.
35366 2017-02-27 Richard Biener <rguenther@suse.de>
35368 PR tree-optimization/79690
35369 * tree-vect-stmts.c (vectorizable_store): Use vector type
35370 built from the DR with address-space.
35372 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
35374 * doc/invoke.texi (Optimize Options): Refine the description
35375 of asan-use-after-return.
35377 2017-02-25 Alan Modra <amodra@gmail.com>
35379 PR rtl-optimization/79584
35380 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
35381 base, not ad->base_term, the reg within base. Remove assertion
35382 that ad->base == ad->base_term. Replace gen_int_mode using
35383 bogus mode with const0_rtx.
35385 2017-02-25 Jakub Jelinek <jakub@redhat.com>
35387 PR middle-end/79396
35388 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
35389 FMA_EXPR like tcc_binary or tcc_unary.
35391 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
35394 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
35396 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
35397 (output_loc_operands): Handle DW_OP_call_ref and
35398 DW_OP_GNU_variable_value.
35399 (struct variable_value_struct): New type.
35400 (struct variable_value_hasher): Likewise.
35401 (variable_value_hash): New variable.
35402 (string_types): Remove.
35403 (copy_loc_descr): New function.
35404 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
35405 (prepend_loc_descr_to_each): New function.
35406 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
35407 instead of add_loc_descr_to_each if the first argument is single
35408 location list and the second has multiple.
35409 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
35410 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
35411 when looking for variable value which doesn't have other location info.
35412 (loc_list_from_tree): Formatting fix.
35413 (gen_array_type_die): Simplify DW_AT_string_length handling.
35414 (adjust_string_types): Remove.
35415 (gen_subprogram_die): Don't call adjust_string_types nor test/set
35416 string_types. Call resolve_variable_values.
35417 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
35418 (resolve_addr_in_expr): Likewise. Add A argument.
35419 (copy_deref_exprloc): Remove deref argument. Adjust for the
35420 original expression being DW_OP_GNU_variable_value with optionally
35421 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
35422 optionally after it.
35423 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
35424 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
35425 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
35426 (variable_value_hasher::hash, variable_value_hasher::equal): New
35428 (resolve_variable_value_in_expr, resolve_variable_value,
35429 resolve_variable_values, note_variable_value_in_expr,
35430 note_variable_value): New functions.
35431 (dwarf2out_early_finish): Call note_variable_value on all toplevel
35434 2017-02-24 Jakub Jelinek <jakub@redhat.com>
35437 * opts.h (handle_generated_option): Add GENERATED_P argument.
35438 * opts-common.c (handle_option): Adjust function comment.
35439 (handle_generated_option): Add GENERATED_P argument, pass it to
35441 (control_warning_option): Pass false to handle_generated_option
35443 * opts.c (maybe_default_option): Pass true to handle_generated_option
35445 * optc-gen.awk: Likewise.
35447 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35449 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
35450 a REG, look at the REG it is a SUBREG of.
35451 (splitter for cmpeqsi_t): Ditto.
35453 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35455 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
35456 the special USEs with the pattern of the insn, not the insn itself.
35458 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
35461 * doc/invoke.texi: Document -mload-store-pairs.
35463 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35464 Sandra Loosemore <sandra@codesourcery.com>
35466 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
35467 argument isn't a CONST_INT.
35468 (nios2_alternate_compare_const): Assert op is a CONST_INT.
35469 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
35470 (nios2_validate_compare): Bypass alternate compare logic if *op2
35471 is not a CONST_INT.
35472 (ldstwm_operation_p): Return false if first_base is not a REG or
35473 if first_offset is not a CONST_INT.
35475 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35477 * config/cris/cris.md: Use correct operand in a define_peephole2.
35479 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35481 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
35483 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35485 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
35486 this_insn if it is an INSN or JUMP_INSN.
35487 (force_offsettable): Look at base, not at addr.
35488 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
35489 on things that aren't necessarily CONST_INTs.
35491 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
35493 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
35494 -mfpmath=sse is the default also for x86-32 targets with SSE2
35495 instruction set when @option{-ffast-math} is enabled
35497 2017-02-24 Jeff Law <law@redhat.com>
35499 PR rtl-optimizatoin/79286
35500 * ira.c (update_equiv_regs): Drop may_trap_p exception to
35503 2017-02-24 Richard Biener <rguenther@suse.de>
35505 PR tree-optimization/79389
35506 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
35509 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
35511 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
35512 function comment to reflect reality.
35513 (loop_exits_before_overflow): Fix typo in function description.
35515 2017-02-24 Richard Biener <rguenther@suse.de>
35517 PR tree-optimization/79389
35518 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
35519 properly that a threading opportunity exists. Detect conditional
35520 copy/constant propagation opportunities.
35522 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
35524 * config/visium/visium.md (type): Add trap.
35525 (b): New mode attribute.
35526 (*btst): Rename into...
35527 (*btst<mode>): ...this and adjust.
35528 (*cbranchsi4_btst_insn): Rename into...
35529 (*cbranch<mode>4_btst_insn): ...this and adjust.
35530 (trap): New define_insn.
35532 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35534 PR tree-optimization/79389
35535 * ifcvt.c (struct noce_if_info): Add rev_cond field.
35536 (noce_reversed_cond_code): New function.
35537 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
35538 reversed_comparison_code. Formatting fix.
35539 (noce_try_store_flag): Test rev_cond != NULL in addition to
35540 reversed_comparison_code.
35541 (noce_try_store_flag_constants): Likewise.
35542 (noce_try_store_flag_mask): Likewise.
35543 (noce_try_addcc): Use rev_cond if non-NULL instead of
35544 reversed_comparison_code.
35545 (noce_try_cmove_arith): Likewise. Formatting fixes.
35546 (noce_try_minmax, noce_try_abs): Clear rev_cond.
35547 (noce_find_if_block): Initialize rev_cond.
35548 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
35549 instead of false as last argument never attempt to reverse it
35552 2017-02-23 Bin Cheng <bin.cheng@arm.com>
35554 PR tree-optimization/79663
35555 * tree-predcom.c (combine_chains): Process refs in reverse order
35556 only for ZERO length chains, and add explaining comment.
35558 2017-02-23 Jeff Law <law@redhat.com>
35560 PR tree-optimization/79578
35561 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
35562 in call to operand_equal_p.
35564 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
35567 * config/i386/cpuid.h: Fix another undefined behavior.
35569 2017-02-23 Richard Biener <rguenther@suse.de>
35571 PR tree-optimization/79683
35572 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
35573 vector types for data-refs.
35575 2017-02-23 Martin Liska <mliska@suse.cz>
35577 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
35579 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35581 PR middle-end/79665
35582 * internal-fn.c (get_range_pos_neg): Moved to ...
35583 * tree.c (get_range_pos_neg): ... here. No longer static.
35584 * tree.h (get_range_pos_neg): New prototype.
35585 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
35586 are known to be in between 0 and signed maximum inclusive, try to
35587 expand both unsigned and signed divmod and use the cheaper one from
35590 2017-02-22 Jeff Law <law@redhat.com>
35592 PR tree-optimization/79578
35593 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
35594 to compare base operands.
35596 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
35599 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
35600 gpc_reg_operand instead of fpr_reg_operand.
35602 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
35604 * config/mips/mips.c (mips_return_in_memory): Force FP
35605 vector types to be returned in memory for o32 ABI.
35607 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35609 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
35610 instead of DW_TAG_member for static data member declarations and don't
35611 set no_linkage_name for static inline data members.
35612 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
35615 2017-02-22 Martin Liska <mliska@suse.cz>
35617 * doc/invoke.texi: Replace inequality signs with square brackets
35620 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
35623 * lra-constraints.c (simplify_operand_subreg): Handle
35624 WORD_REGISTER_OPERATIONS targets.
35626 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35629 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
35630 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
35631 elimination by swapping fld*.
35633 2017-02-22 Richard Biener <rguenther@suse.de>
35635 PR tree-optimization/79673
35636 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
35637 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
35638 irrelevant address-space qualifiers and avoiding a
35639 ADDR_SPACE_CONVERT_EXPR from fold_convert.
35641 2017-02-22 Richard Biener <rguenther@suse.de>
35643 PR tree-optimization/79666
35644 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
35645 to not symbolically negate if that may introduce undefined
35648 2017-02-22 Martin Liska <mliska@suse.cz>
35651 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
35652 * data-streamer-out.c (streamer_write_gcov_count_stream):
35654 * value-prof.c (stream_out_histogram_value): Make assert more
35655 precise based on type of counter.
35657 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
35660 * config/i386/i386.md (standard_x87sse_constant_load splitter):
35661 Use nonimmediate_operand instead of memory_operand for operand 1.
35662 (float-extend standard_x87sse_constant_load splitter): Ditto.
35664 2017-02-21 Jeff Law <law@redhat.com>
35666 PR tree-optimization/79621
35667 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
35668 blocks with edges to themselves.
35670 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35673 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
35674 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
35675 Use gimple_call_builtin_p.
35678 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
35679 on temporarily removed DEBUG_INSNs.
35681 PR tree-optimization/79649
35682 * tree-loop-distribution.c (classify_partition): Give up on
35683 non-generic address space loads/stores.
35685 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
35687 * doc/loop.texi (Loop manipulation): Remove nonexistent
35688 tree_ssa_loop_version from the documentation.
35689 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
35691 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35694 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
35695 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
35696 * config/rs6000/rs6000.c: Include except.h.
35697 (rs6000_expand_split_stack_prologue): Call
35698 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
35700 2017-02-21 Martin Jambor <mjambor@suse.cz>
35703 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
35704 have been analyzed.
35706 2017-02-21 Martin Jambor <mjambor@suse.cz>
35708 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
35709 for backward compatibility only.
35710 * doc/invoke.texi (Option Summary): Remove all references to
35711 -fipa-cp-alignment.
35713 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
35717 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35719 * lra-constraints.c (curr_insn_transform): Handle
35720 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35722 2017-02-21 Martin Liska <mliska@suse.cz>
35724 * config/i386/i386.opt: Replace -masm-dialect with -masm.
35726 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
35728 PR translation/79638
35729 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
35731 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
35734 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
35735 (arm_function_ok_for_sibcall): Return false for an indirect call by
35736 descriptor if all the argument registers are used.
35737 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
35738 alignment of the function.
35740 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35742 PR tree-optimization/61441
35743 * simplify-rtx.c (simplify_const_unary_operation): For
35744 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
35745 the sNaN unmodified.
35747 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
35749 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
35750 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
35751 instead of SYSTEM_HEADER_DIR.
35753 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
35754 Martin Liška <mliska@suse.cz>
35756 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
35757 Fix typos and grammar, use active voice, and clarify.
35759 2017-02-20 Marek Polacek <polacek@redhat.com>
35761 PR middle-end/79537
35762 * gimplify.c (gimplify_expr): Handle unused *&&L;.
35765 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
35767 2017-02-20 Jakub Jelinek <jakub@redhat.com>
35770 * config/i386/i386.c (ix86_expand_builtin): Handle
35771 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
35772 ix86_builtins_isa[fcode].isa as a requirement of those
35773 flags and any other flag in the bitmask.
35774 (ix86_init_mmx_sse_builtins): Use 0 instead of
35775 ~OPTION_MASK_ISA_64BIT as mask.
35776 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
35777 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
35778 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
35779 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
35781 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35784 * lra-constraints.c (split_reg): Check requested split mode
35785 is supported by the register.
35787 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35789 * lra-constraints.c (simplify_operand_subreg): Remove early
35792 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35795 * lra-constraints.c (curr_insn_transform): Tighten condition
35796 for converting SUBREG reloads from OP_OUT to OP_INOUT.
35798 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35801 * lra-constraints.c (curr_insn_transform): Handle
35802 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35804 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
35807 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
35809 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
35811 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
35814 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
35817 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35819 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
35822 2017-02-18 Jakub Jelinek <jakub@redhat.com>
35824 * final.c (last_columnnum, override_columnnum): New variables.
35825 (final_start_function): Set last_columnnum, pass it to begin_prologue
35826 hook and pass 0 to dwarf2out_begin_prologue.
35827 (final_scan_insn): Update override_columnnum. Pass last_columnnum
35828 to source_line debug hook.
35829 (notice_source_line): Compute last_columnnum and for debug_column_info
35830 return true on column changes.
35831 * debug.h (struct gcc_debug_hooks): Add column argument to
35832 source_line and begin_prologue hooks.
35833 (debug_nothing_int_charstar_int_bool): Remove prototype.
35834 (debug_nothing_int_int_charstar,
35835 debug_nothing_int_int_charstar_int_bool): New prototypes.
35836 (dwarf2out_begin_prologue): Add column argument.
35837 * debug.c (do_nothing_debug_hooks): Adjust source_line and
35838 begin_prologue hooks.
35839 (debug_nothing_int_charstar_int_bool): Remove.
35840 (debug_nothing_int_int_charstar,
35841 debug_nothing_int_int_charstar_int_bool): New functions.
35842 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
35843 through to dwarf2out_source_line.
35844 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
35845 (dwarf2out_source_line): Add column argument, emit it if requested.
35846 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
35848 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35849 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35850 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
35851 through to dwarf2out_begin_prologue.
35852 (vmsdbgout_source_line): Add column argument, pass it through to
35853 dwarf2out_source_line.
35854 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
35855 dbxout_source_line caller.
35856 (dbxout_source_line): Add column argument.
35858 * common.opt (gno-column-info, gcolumn-info): New options.
35859 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
35860 (check_die): Also test for multiple DW_AT_decl_column attributes.
35861 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
35862 DW_AT_decl_column if requested.
35863 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
35865 (gen_variable_die): Likewise.
35866 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
35867 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
35870 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
35871 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
35872 (ix86_handle_option): Handle OPT_m3dnowa.
35873 * doc/invoke.texi (-m3dnowa): Document.
35874 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
35875 -m3dnowa instead of -m3dnow -march=athlon.
35878 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
35879 instead of gcc_assert for K, r and R code checks. Formatting fixes.
35881 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35884 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
35885 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
35886 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
35887 generator for vsx_xxpermdi_<mode>_be.
35888 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
35889 force big-endian semantics.
35890 (vsx_xxpermdi_<mode>_be): New define_expand with same
35891 implementation as previous version of vsx_xxpermdi_<mode>.
35893 2017-02-17 Jakub Jelinek <jakub@redhat.com>
35895 PR tree-optimization/79327
35896 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
35897 variable, its initialization and use.
35899 2017-02-17 Julia Koval <julia.koval@intel.com>
35901 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
35902 (OPTION_MASK_ISA_PKU_UNSET): New.
35903 (ix86_handle_option): Handle -mrdpid.
35904 * config/i386/cpuid.h (bit_RDPID): New.
35905 * config/i386/driver-i386.c (host_detect_local_cpu):
35906 Detect RDPID feature.
35907 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
35908 * config/i386/i386-c.c (ix86_target_macros_internal):
35910 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
35911 (ix86_valid_target_attribute_inner_p): Add "rdpid".
35912 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
35913 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
35914 * config/i386/i386.md (define_insn "rdpid"): New.
35915 * config/i386/i386.opt Add -mrdpid.
35916 * config/i386/immintrin.h (_rdpid_u32): New.
35918 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
35920 PR rtl-optimization/79541
35921 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
35922 instead of transforming it into USE.
35924 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
35926 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
35927 If HONOR_SNANS (SFmode) force the input to a register.
35928 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
35929 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
35930 an frsp or similar insn.
35932 2017-02-17 Martin Liska <mliska@suse.cz>
35934 PR rtl-optimization/79577
35935 * params.def (selsched-max-sched-times): Increase minimum to 1.
35937 2017-02-17 Martin Liska <mliska@suse.cz>
35939 PR rtl-optimization/79574
35940 * gcse.c (want_to_gcse_p): Prevent integer overflow.
35942 2017-02-17 Martin Liska <mliska@suse.cz>
35944 PR tree-optimization/79529
35945 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
35946 ssa_defined_default_def_p to handle cases which are implicitly
35948 * tree-ssa.c (ssa_defined_default_def_p): New function.
35949 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
35950 which are implicitly defined.
35951 * tree-ssa.h (ssa_defined_default_def_p): Declare.
35953 2017-02-17 Richard Biener <rguenther@suse.de>
35955 PR middle-end/79576
35956 * params.def (max-ssa-name-query-depth): Limit to 10.
35958 2017-02-17 Richard Biener <rguenther@suse.de>
35960 PR tree-optimization/79552
35961 * tree-ssa-structalias.c (visit_loadstore): Properly verify
35964 2017-02-17 Richard Biener <rguenther@suse.de>
35967 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
35969 2017-02-17 Marek Polacek <polacek@redhat.com>
35971 PR middle-end/79536
35972 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
35973 (fold_negate_expr): New wrapper.
35975 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
35977 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
35978 Correct terminology and de-emphasize pre-standard behavior.
35980 2017-02-16 Alan Modra <amodra@gmail.com>
35982 PR rtl-optimization/79286
35983 * ira.c (def_dominates_uses): New function.
35984 (update_equiv_regs): Don't create an equivalence for insns that
35985 may trap where the register def does not dominate the use.
35987 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
35989 PR rtl-optimization/78127
35990 * lra.c (lra): Call lra_eliminate before finish the loop after
35993 2017-02-16 Richard Biener <rguenther@suse.de>
35995 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
35997 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
35998 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
35999 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36000 (isl_val_int_from_wi): New function.
36001 (extract_affine_gmp): Rename to ...
36002 (extract_affine_wi): ... this, take a widest_int.
36003 (extract_affine_int): Just wrap extract_affine_wi.
36004 (add_param_constraints): Use isl_val_int_from_wi.
36005 (add_loop_constraints): Likewise, and extract_affine_wi.
36007 2017-02-15 Jeff Law <law@redhat.com>
36009 PR middle-end/79521
36010 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36011 ira_init_register_move_cost_if_necessary.
36013 2017-02-15 Martin Sebor <msebor@redhat.com>
36015 PR middle-end/32003
36016 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36017 removed in a prior commit.
36019 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36021 PR tree-optimization/79347
36022 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36023 counters during peeling.
36025 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36027 * Makefile.in (site.exp): Remove "set ISLVER".
36029 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36032 * real.c (real_from_integer): Call real_convert even for decimal.
36034 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36037 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36039 2017-02-14 Andrew Pinski <apinski@cavium.com>
36041 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
36042 cores and change the partno/implementer to be correct.
36043 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
36044 the 'B" as the implementer.
36045 * config/aarch64/aarch64-tune.md: Regenerate.
36047 2017-02-14 Carl Love <cel@us.ibm.com>
36049 * config/rs6000/rs6000.c: Add case statement entry to make the
36050 xvcvuxdsp built-in argument unsigned.
36051 * config/rs6000/vsx.md: Fix the source and return operand types so they
36052 match the instruction definitions from the ISA document. Fix typo
36053 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
36056 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
36059 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
36060 member early_clobber_alts.
36061 * lra-lives.c (reg_early_clobber_p): New.
36062 (process_bb_lives): Use it.
36063 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
36064 (debug_operand_data): Initialize early_clobber_alts.
36065 (setup_operand_alternative): Set up early_clobber_alts.
36066 (collect_non_operand_hard_regs): Ditto. Pass early clobber
36067 alternatives to new_insn_reg.
36068 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
36070 (lra_update_insn_regno_info): Pass the new arg.
36072 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36074 PR middle-end/79505
36075 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
36076 (new_oacc_loop_raw): Don't clear already cleared fields.
36079 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
36080 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
36081 _mm512_prefetch_i64gather_ps): New inline functions and macros.
36083 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
36086 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
36088 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
36091 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
36092 the extra instruction to the right place to store 128-bit constant
36095 2017-02-14 Martin Sebor <msebor@redhat.com>
36097 PR middle-end/79448
36098 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
36099 warning for strings of unknown length.
36101 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
36103 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
36105 2017-02-14 Jeff Law <law@redhat.com>
36108 * ira-costs.c (scan_one_insn): Initialize register move costs
36109 for pseudos seen in USE/CLOBBER insns.
36111 PR tree-optimization/79095
36112 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
36113 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
36114 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
36115 if the operands are known to be not equal, then the resulting range
36117 (intersect_ranges): If the new range is ~[0,0] and the old range is
36118 wide, then prefer ~[0,0].
36119 * tree-vrp.c (overflow_comparison_p_1): New function.
36120 (overflow_comparison_p): New function.
36121 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
36122 if NAME is used in an overflow test.
36123 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
36124 overflow check that can be expressed as an equality test, then adjust
36125 ops to be that equality test.
36127 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36129 * config/s390/s390-builtin-types.def: Remove flags argument.
36130 * config/s390/s390.c (s390_init_builtins): Likewise.
36132 2017-02-14 Martin Liska <mliska@suse.cz>
36134 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
36135 vector. Fix trailing white spaces.
36137 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
36139 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
36142 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36144 PR rtl-optimization/68664
36145 * config/arm/arm.c (arm_sched_can_speculate_insn):
36146 New function. Declare prototype.
36147 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36149 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36151 PR rtl-optimization/68664
36152 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
36154 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36156 2017-02-14 Amit Pawar <amit.pawar@amd.com>
36158 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
36159 max skip bytes for function, loop and jump.
36161 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36163 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
36164 ABS_EXPR for gimple dump.
36166 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36169 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
36171 PR tree-optimization/79408
36172 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
36173 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
36174 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
36175 also if rhs1 is INTEGER_CST.
36177 2017-02-14 Richard Biener <rguenther@suse.de>
36179 PR middle-end/79432
36180 * tree-into-ssa.c (insert_phi_nodes): When the function can
36181 have abnormal edges rewrite SSA names with broken use-def
36182 dominance out of SSA and register them for PHI insertion.
36184 2017-02-13 Martin Sebor <msebor@redhat.com>
36186 PR middle-end/79496
36187 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
36188 clearing info.nowrite flag when snprintf size argument is a range.
36190 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36192 * cprop.c (cprop_jump): Add missing space in string literal.
36193 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
36194 (get_constraint_for_component_ref): Likewise.
36195 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
36196 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
36197 * lra-constraints.c (process_alt_operands): Likewise.
36198 * ipa-inline.c (inline_small_functions): Likewise.
36199 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
36200 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
36201 * trans-mem.c (diagnose_tm_1_op): Likewise.
36202 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
36203 (grid_parallel_clauses_gridifiable): Likewise.
36205 * config/nvptx/mkoffload.c (process): Add space in between
36208 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
36209 "MOD4_SSE_REGS" and "ALL_REGS".
36211 * spellcheck.c (test_data): Add , in between "foo" and "food".
36213 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36216 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
36217 boundary crossing check and subsequent code generation agree.
36219 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36221 * config/aarch64/aarch64.c (has_memory_op): Delete.
36222 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
36225 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36227 PR rtl-optimization/79388
36228 PR rtl-optimization/79450
36229 * combine.c (distribute_notes): When removing TEM_INSN for which
36230 corresponding dest has last value recorded, invalidate that last
36233 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36235 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
36236 of explicit '@'. Add missing assembly comment marker on branch costs
36239 2017-02-13 Nathan Sidwell <nathan@acm.org>
36241 * gengtype-lex.l (<in_struct>): Add '/'.
36243 2017-02-13 Martin Liska <mliska@suse.cz>
36246 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
36248 2017-02-13 Richard Biener <rguenther@suse.de>
36250 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
36252 * configure: Re-generate.
36253 * config.in: Likewise.
36254 * graphite-dependences.c: Simplify as if
36255 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
36256 * graphite-isl-ast-to-gimple.c: Likewise.
36257 * graphite-optimize-isl.c: Likewise.
36258 * graphite-poly.c: Likewise.
36259 * graphite-sese-to-poly.c: Likewise.
36260 * graphite.h: Likewise.
36261 * toplev.c: Include isl/version.h and use isl_version () for
36262 printing the ISL version.
36263 * doc/install.texi: Update ISL requirement.
36265 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36267 * doc/standards.texi (Standards): Update reference to
36270 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36272 * doc/extend.texi (Named Address Spaces): sourceware.org now
36274 * doc/install.texi (Binaries): Ditto.
36277 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36279 * doc/cpp.texi: Replace "stringify"/"stringification" with C
36280 standard terminology "stringize"/"stringizing" throughout.
36281 * doc/cppinternals.texi: Likewise.
36283 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36285 * doc/extend.texi: Fix some spelling mistakes and typos.
36286 * doc/invoke.texi: Likewise.
36288 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36291 * params.def (inline-min-speedup) Change from 10 to 8.
36293 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36295 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
36298 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36301 * ipa-inline-analysis.c (get_minimal_bb): New function.
36302 (record_modified): Use it.
36303 (remap_edge_change_prob): Handle also ancestor functions.
36305 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
36307 * doc/contrib.texi (Contributors): Remove broken link into
36308 the Mauve CVS repository.
36310 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36312 PR middle-end/79454
36313 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
36314 result computation whenever lhs doesn't have vector mode, not
36315 just when it has BLKmode.
36317 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36319 * doc/makefile.texi (profiledbootstrap): Refer to the
36320 installation instructions only in textual form.
36322 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36325 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
36327 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36329 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
36330 (Specific): Update mingw-w64 reference.
36332 (Specific): Remove broken link to Renesas RX processor.
36334 2017-02-10 Richard Biener <rguenther@suse.de>
36336 * toplev.c (process_options): Do not mention obsolete graphite
36337 options when printing sorry message about missing graphite support.
36338 Mention -floop-nest-optimize.
36340 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
36342 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
36343 (vtst_p16): Likewise.
36344 (vtstq_p8): Likewise.
36345 (vtstq_p16): Likewise.
36347 (vtstq_p64): Likewise.
36348 * config/arm/arm_neon.h (vgetq_lane_p64): New.
36349 (vset_lane_p64): New.
36350 (vsetq_lane_p64): New.
36352 2017-02-10 Jakub Jelinek <jakub@redhat.com>
36354 PR tree-optimization/79411
36355 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
36356 stmt operands are SSA_NAMEs used in abnormal phis.
36357 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
36360 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36363 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
36366 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36368 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
36370 2017-02-09 Jakub Jelinek <jakub@redhat.com>
36372 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
36376 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
36377 not arbitrary TREE_CONSTANT.
36380 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
36381 "omp declare target link" attribute unless is_global_var.
36382 * omp-offload.c (find_link_var_op): Likewise.
36384 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
36385 Chung-Lin Tang <cltang@codesourcery.com>
36387 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
36389 (gimplify_adjust_omp_clauses): Don't delete TILE.
36390 (gimplify_omp_for): Deal with TILE.
36391 * internal-fn.c (expand_GOACC_TILE): New function.
36392 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
36394 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
36395 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
36397 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
36398 avoid DIV for outermost collapse var.
36399 (expand_oacc_for): Insert tile element loop as needed. Adjust.
36400 Remove out of date comments, fix whitespace.
36401 * omp-general.c (omp_extract_for_data): Deal with tiling.
36402 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
36403 adjust OLF_DIM_BASE value.
36404 (struct omp_for_data): Add tiling field.
36405 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
36406 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
36407 for auto loops. Remove default auto determining, moved to
36408 oacc_loop_fixed_partitions.
36409 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
36410 stmts, add e_mask field.
36411 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
36412 (oacc_thread_numbers): Use oacc_dim_call.
36413 (oacc_xform_tile): New.
36414 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
36415 (finish_oacc_loop): Adjust for ifns vector.
36416 (oacc_loop_discover_walk): Append loop abstraction sites to list,
36417 add case for GOACC_TILE fns.
36418 (oacc_loop_xform_loop): Delete.
36419 (oacc_loop_process): Iterate over call list directly, and add
36420 handling for GOACC_TILE fns.
36421 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
36423 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
36424 vector partitioning to outer loops. Assign 2 partitions to loops
36425 when available. Add TILE handling.
36426 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
36427 (execite_oacc_device_lower): Process GOACC_TILE fns,
36428 ignore unknown specs.
36429 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
36430 * tree.c (omp_clause_num_ops): Adjust TILE ops.
36431 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
36433 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
36435 * configure.ac (ACX_BUGURL): Update.
36436 * configure: Regenerate.
36438 2017-02-09 Richard Biener <rguenther@suse.de>
36440 PR tree-optimization/69823
36441 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
36442 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
36444 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36446 * config/arc/arc-c.def: Add __NPS400__ definition.
36447 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
36448 (TARGET_NPS400): Define.
36450 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36452 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
36454 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
36455 pointer, arch_info.
36456 (arc_cpu_types): Fill the arch_info field with a pointer into the
36457 arc_arch_types table.
36458 (arc_selected_cpu): Declare.
36459 * config/arc/arc.c (arc_selected_cpu): Make global.
36460 (arc_selected_arch): Delete.
36461 (arc_base_cpu): Delete.
36462 (arc_override_options): Remove references to deleted variables,
36463 update access to arch information.
36464 (ARC_OPT): Update access to arch information.
36465 (ARC_OPTX): Likewise.
36466 * config/arc/arc.h (arc_base_cpu): Remove declaration.
36467 (TARGET_ARC600): Update access to arch information.
36468 (TARGET_ARC601): Likewise.
36469 (TARGET_ARC700): Likewise.
36470 (TARGET_EM): Likewise.
36471 (TARGET_HS): Likewise.
36472 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
36475 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
36478 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
36479 condition/operands for integer GE/LE/GEU/LEU operations.
36481 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
36483 PR translation/79397
36484 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
36487 2017-02-08 Martin Jambor <mjambor@suse.cz>
36490 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
36491 whether allocation happened.
36492 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
36493 nothing was allocated.
36495 2017-02-08 Jakub Jelinek <jakub@redhat.com>
36497 PR tree-optimization/79408
36498 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
36499 constant, but SSA_NAME with a known integer range, use the minimum
36500 of that range instead of op1 to determine if modulo can be replaced
36501 with its first operand.
36503 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36505 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
36507 2017-02-08 Richard Biener <rguenther@suse.de>
36509 PR tree-optimization/71824
36510 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36511 Check all loops contained in the merged region.
36513 2017-02-07 Andrew Pinski <apinski@cavium.com>
36515 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
36517 2017-02-07 Andrew Pinski <apinski@cavium.com>
36519 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
36520 (thunderxt88): Likewise.
36521 (thunderxt81): Disable LSE and change v8.1 to v8.
36522 (thunderxt83): Likewise.
36524 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36525 Richard Biener <rguenther@suse.de>
36527 PR middle-end/79399
36528 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
36529 type from int to size_t.
36530 * ira-costs.c (struct_costs_size): Change type from int to size_t.
36532 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36534 PR rtl-optimization/79386
36535 * cprop.c (bypass_conditional_jumps): Initialize
36536 bypass_last_basic_block already before splitting bbs after
36537 unconditional traps...
36538 (bypass_conditional_jumps): ... rather than here.
36541 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
36542 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
36543 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
36544 fix -masm=intel patterns.
36546 2017-02-07 Richard Biener <rguenther@suse.de>
36548 PR tree-optimization/79256
36549 PR middle-end/79278
36550 * builtins.c (get_object_alignment_2): Use min_align_of_type
36551 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
36552 and ADJUST_FIELD_ALIGN.
36554 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
36556 * doc/tm.texi: Regenerate.
36557 * stor-layout.c (layout_decl): Adjust.
36558 (update_alignment_for_field): Likewise.
36559 (place_field): Likewise.
36560 (min_align_of_type): Likewise.
36561 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
36562 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
36563 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
36564 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
36565 * config/frv/frv.c (frv_adjust_field_align): Likewise.
36566 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
36567 * config/i386/i386.c (x86_field_alignment): Likewise.
36568 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
36569 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
36570 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
36571 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
36572 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
36573 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
36577 2017-01-30 Richard Biener <rguenther@suse.de>
36579 PR tree-optimization/79256
36580 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36581 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36584 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
36586 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
36587 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
36588 builtins to SImode and emit a zero-extend, if necessary.
36590 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36592 * docs/invoke.texi (RISC-V Options): Alphabetize.
36594 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36596 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
36599 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36601 * config/riscv/riscv.c: New file.
36602 * common/config/riscv/riscv-common.c: Likewise.
36603 * config.gcc: Likewise.
36604 * config/riscv/constraints.md: Likewise.
36605 * config/riscv/elf.h: Likewise.
36606 * config/riscv/generic.md: Likewise.
36607 * config/riscv/linux.h: Likewise.
36608 * config/riscv/multilib-generator: Likewise.
36609 * config/riscv/peephole.md: Likewise.
36610 * config/riscv/pic.md: Likewise.
36611 * config/riscv/predicates.md: Likewise.
36612 * config/riscv/riscv-builtins.c: Likewise.
36613 * config/riscv/riscv-c.c: Likewise.
36614 * config/riscv/riscv-ftypes.def: Likewise.
36615 * config/riscv/riscv-modes.def: Likewise.
36616 * config/riscv/riscv-opts.h: Likewise.
36617 * config/riscv/riscv-protos.h: Likewise.
36618 * config/riscv/riscv.h: Likewise.
36619 * config/riscv/riscv.md: Likewise.
36620 * config/riscv/riscv.opt: Likewise.
36621 * config/riscv/sync.md: Likewise.
36622 * config/riscv/t-elf-multilib: Likewise.
36623 * config/riscv/t-linux: Likewise.
36624 * config/riscv/t-linux-multilib: Likewise.
36625 * config/riscv/t-riscv: Likewise.
36626 * configure.ac: Likewise.
36627 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
36628 Waterman as RISC-V maintainers.
36629 * doc/install.texi: Add RISC-V entries.
36630 * doc/invoke.texi: Add RISC-V options section.
36631 * doc/md.texi: Add RISC-V constraints section.
36632 * configure: Regenerated.
36634 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
36637 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
36638 false values to be constant vectors with all 0 or all 1 bits set.
36639 (vcondu<mode><mode>): Likewise.
36640 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
36642 (fpmask_comparison_operator): Update comment.
36643 (vecint_comparison_operator): New predicate.
36644 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
36645 vector conditionals when the true and false values are constant
36646 vectors with all 0 bits or all 1 bits set.
36648 2017-02-06 Martin Sebor <msebor@redhat.com>
36650 PR tree-optimization/79376
36651 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
36653 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
36655 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
36656 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
36657 to simplify split condition.
36659 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36661 * omp-expand.c (oxpand_omp_atomic_fetch_op,
36662 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
36665 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
36667 PR rtl-optimization/68664
36668 * target.def (can_speculate_insn): New hook.
36669 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
36670 * doc/tm.texi: Regenerate.
36671 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
36672 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
36673 (rs6000_sched_can_speculate_insn): New function.
36675 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36677 PR tree-optimization/79284
36678 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
36679 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
36680 vectorizable_mask_load_store, vectorizable_operation,
36681 vect_is_simple_cond, get_same_sized_vectype): Use it instead
36682 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
36683 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
36684 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
36685 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
36686 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
36687 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
36688 is_gimple_assign (stmt). Replace another such test with
36689 is_gimple_assign (stmt).
36691 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36694 * config/avr/avr.c (rtl-iter.h): Include it.
36695 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
36696 (avr_legitimate_combined_insn): ...and implementation.
36698 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
36700 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
36701 * config/s390/s390.c (s390_const_operand_ok)
36702 (s390_canonicalize_comparison, s390_extract_part)
36703 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
36704 (s390_contiguous_bitmask_p, s390_rtx_costs)
36705 (legitimize_pic_address): Likewise.
36706 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
36707 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
36708 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
36709 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
36710 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
36712 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36714 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
36715 REGNO($0) == REGNO($1).
36717 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36719 * config/s390/linux.h(SIZE_TYPE): Add comment.
36721 2017-02-06 Julian Brown <julian@codesourcery.com>
36722 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36723 Virendra Pathak <virendra.pathak@broadcom.com>
36725 * config/aarch64/aarch64-cores.def: Change the scheduler
36727 * config/aarch64/aarch64.md: Include thunderx2t99.md.
36728 * config/aarch64/thunderx2t99.md: New file.
36730 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36732 * doc/standards.texi (Go Language): Update link to language
36735 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36737 * tree-eh.c (lower_resx): Sanitize profile.
36738 (cleanup_empty_eh_move_lp): Likewise.
36740 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36743 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
36745 * cfgloopmanip.h (loop_version): Update prototype.
36746 * modulo-sched.c (sms_schedule): Update call of loop_version.
36747 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
36748 * tree-parloops.c (gen_parallel_loop): Likewise.
36749 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
36750 * tree-ssa-loop-split.c (split_loop): Likewise.
36751 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
36752 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
36754 2017-02-05 Martin Liska <mliska@suse.cz>
36757 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
36759 (print_operand_address): Initialize a struct to zero.
36761 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36763 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
36764 garbage collector only in textual form.
36766 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36768 * doc/extend.texi (x86 specific memory model extensions for
36769 transactional memory): Simplify a phrase.
36771 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
36774 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
36775 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
36776 (atomic_storedi_1): Likewise.
36778 2017-02-04 Jakub Jelinek <jakub@redhat.com>
36780 PR tree-optimization/79338
36781 * tree-parloops.c (gather_scalar_reductions): Don't call
36782 vect_analyze_loop_form for loop->inner before destroying loop's
36785 2017-02-03 Martin Sebor <msebor@redhat.com>
36787 PR tree-optimization/79327
36788 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
36789 when precision has resulted in leading zeros.
36790 (format_integer): Adjust the likely counter to assume an unknown
36791 argument that may be zero is non-zero.
36793 2017-02-03 Jason Merrill <jason@redhat.com>
36796 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
36797 avoid copying non-taken branch.
36799 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36801 PR tree-optimization/79340
36802 * tree-vect-loop.c (vectorizable_reduction): Release
36803 vec_defs elements after safe_splicing them into other vectors.
36806 PR tree-optimization/79327
36807 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
36808 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
36810 (format_integer): Use wide_int_to_tree instead of build_int_cst
36811 + to_?hwi. If argmin is NULL, just set argmin and argmax to
36812 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
36813 of shortest and longest sequence.
36815 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
36817 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
36818 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
36820 2017-02-03 Walter Lee <walt@tilera.com>
36823 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
36824 after initial stackframe link reg save.
36825 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
36827 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36830 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
36831 wu for stxssp alternative.
36833 2017-02-03 Martin Sebor <msebor@redhat.com>
36835 PR tree-optimization/79352
36836 * gimple-fold.c (get_range_strlen): Add argument.
36837 (get_range_strlen): Change return type to bool.
36838 (get_maxval_strlen): Pass in a dummy argument.
36839 * gimple-fold.h (get_range_strlen): Change return type to bool.
36840 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
36841 * tree.h (array_at_struct_end_p): Add argument.
36842 * tree.c (array_at_struct_end_p): Handle it.
36844 2017-02-03 Martin Liska <mliska@suse.cz>
36847 * multiple_target.c (create_dispatcher_calls): Redirect edge
36848 from a caller of a dispatcher.
36849 (expand_target_clones): Make the clones local.
36850 (ipa_target_clone): Do both target clones and resolvers.
36851 (ipa_dispatcher_calls): Remove the pass.
36852 (pass_dispatcher_calls::gate): Likewise.
36853 (make_pass_dispatcher_calls): Likewise.
36854 * passes.def (pass_target_clone): Put as very first IPA early
36857 2017-02-03 Martin Liska <mliska@suse.cz>
36859 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
36860 in case of a function with ifunc attribute.
36862 2017-02-03 Martin Liska <mliska@suse.cz>
36864 * cgraph.c (cgraph_node::dump): Dump function version info.
36865 * symtab.c (symtab_node::dump_base): Add missing new line.
36867 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36869 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
36870 (ifcombine_ifandif): Use it.
36872 2017-02-03 Martin Liska <mliska@suse.cz>
36874 * doc/invoke.texi: Document default value for
36875 use-after-scope-direct-emission-threshold.
36877 2017-02-03 Martin Liska <mliska@suse.cz>
36879 PR tree-optimization/79339
36880 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
36881 (format_floating): Likewise.
36883 2017-02-03 Martin Liska <mliska@suse.cz>
36886 * ipa-prop.c (ipa_node_params_t::insert): Remove current
36888 (ipa_node_params_t::remove): Likewise.
36889 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
36890 initialization from removed ipa_node_params_t::insert.
36891 (ipa_node_params::~ipa_node_params): Move from removed
36892 ipa_node_params_t::release.
36893 * symbol-summary.h (symbol_summary::m_released): New member.
36894 Do not release a summary twice. Do not allow to call finalizer
36895 for types of a summary that live in GGC memory.
36897 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36899 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
36902 2017-02-02 Martin Sebor <msebor@redhat.com>
36904 PR middle-end/79275
36905 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
36906 (format_string): Tighten up the range of output for non-constant
36907 strings and correct the expected range for wide non-constant strings.
36909 2017-02-02 Martin Sebor <msebor@redhat.com>
36911 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
36913 PR middle-end/32003
36914 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
36916 (-fdump-tree-@var): Add to index and document how to come up
36917 with pass-specific option and dump file names.
36918 (-fdump-passes): Clarify where to look for output.
36920 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36922 PR middle-end/77445
36923 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
36924 statistics of the analyzed path; allow threading for speed when
36925 any of BBs along the path are optimized for speed.
36927 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
36929 PR middle-end/78468
36930 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
36931 settings of the virtual registers.
36934 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
36936 * explow.c (get_dynamic_stack_size): Take known alignment of stack
36937 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
36940 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36942 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
36943 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
36945 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36947 * config/s390/s390.md: Add missing comments with the expanded
36949 * config/s390/vector.md: Likewise.
36950 * config/s390/vx-builtins.md: Likewise.
36952 2017-02-02 Jakub Jelinek <jakub@redhat.com>
36955 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
36956 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
36957 conditions on a single line.
36959 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36961 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36962 __S390_VX__ to __VX__.
36964 2017-02-01 Andrew Pinski <apinski@cavium.com>
36966 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
36967 stmt_info to record_stmt_cost.
36968 (vect_get_known_peeling_cost): Pass stmt_info if known to
36970 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
36971 cpu_vector_cost field into
36972 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
36973 field into vec_int_stmt_cost and vec_fp_stmt_cost.
36974 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
36975 splitting of scalar_stmt_cost and vec_stmt_cost.
36976 (thunderx_vector_cost): Likewise.
36977 (cortexa57_vector_cost): LIkewise.
36978 (exynosm1_vector_cost): Likewise.
36979 (xgene1_vector_cost): Likewise.
36980 (thunderx2t99_vector_cost): Improve after the splitting of the two
36982 (aarch64_builtin_vectorization_cost): Update for the splitting of
36983 scalar_stmt_cost and vec_stmt_cost.
36985 2017-02-01 Torvald Riegel <triegel@redhat.com>
36986 Richard Henderson <rth@redhat.com>
36988 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
36989 conditional on existance of a fast atomic load.
36990 * optabs-query.c (can_atomic_load_p): New function.
36991 * optabs-query.h (can_atomic_load_p): Declare it.
36992 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
36993 no fast atomic load is available for the particular size of access.
36994 (expand_atomic_compare_and_swap): Likewise.
36995 (expand_atomic_load): Likewise.
36996 (expand_atomic_store): Likewise.
36997 (expand_atomic_fetch_op): Likewise.
36998 * testsuite/lib/target-supports.exp
36999 (check_effective_target_sync_int_128): Remove x86 because it provides
37000 no fast atomic load.
37001 (check_effective_target_sync_int_128_runtime): Likewise.
37003 2017-02-01 Richard Biener <rguenther@suse.de>
37005 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37006 (graphite_transform_loops): Provide opt-info for optimized nests.
37007 * tree-parloop.c (parallelize_loops): Provide opt-info for
37008 parallelized loops.
37010 2017-02-01 Richard Biener <rguenther@suse.de>
37012 PR middle-end/79315
37013 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37014 was not set before.
37016 2017-02-01 Richard Biener <rguenther@suse.de>
37018 PR tree-optimization/71824
37019 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37020 Verify the loops are valid in the merged SESE region.
37021 (scop_detection::can_represent_loop_1): Check analyzing the
37022 evolution of the number of iterations in the region succeeds.
37024 2017-01-31 Ian Lance Taylor <iant@golang.org>
37026 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37027 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37029 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37031 PR preprocessor/79210
37032 * input.c (get_substring_ranges_for_loc): Replace line_width
37033 assertion with error-handling.
37035 2017-01-31 Richard Biener <rguenther@suse.de>
37037 PR tree-optimization/77318
37038 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37039 (create_pw_aff_from_tree): Take loop parameter.
37040 (add_condition_to_pbb): Pass loop of the condition to
37041 create_pw_aff_from_tree.
37043 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37045 * config/s390/s390.c (s390_asan_shadow_offset): New function.
37046 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
37048 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
37052 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
37054 (convert_int_to_float128): Likewise.
37055 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
37056 (convert_int_to_float128): Likewise.
37057 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
37058 (UNSPEC_IEEE128_CONVERT): Likewise.
37059 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
37060 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
37061 Use local variables for IBM extended format.
37062 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
37063 (fix_trunc<mode>si2_fprs): Likewise.
37064 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
37065 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
37066 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
37067 to know that we can now have integers of all sizes in vector
37069 (fix<uns>_<mode>di2_hw): Likewise.
37070 (float<uns>_<mode>si2_hw): Likewise.
37071 (fix_<mode>si2_hw): Likewise.
37072 (fixuns_<mode>si2_hw): Likewise.
37073 (float<uns>_<mode>di2_hw): Likewise.
37074 (float_<mode>di2_hw): Likewise.
37075 (float_<mode>si2_hw): Likewise.
37076 (floatuns_<mode>di2_hw): Likewise.
37077 (floatuns_<mode>si2_hw): Likewise.
37078 (xscvqp<su>wz_<mode>): Delete, no longer used.
37079 (xscvqp<su>dz_<mode>): Likewise.
37080 (xscv<su>dqp_<mode>): Likewise.
37081 (ieee128_mfvsrd_64bit): Likewise.
37082 (ieee128_mfvsrd_32bit): Likewise.
37083 (ieee128_mfvsrwz): Likewise.
37084 (ieee128_mtvsrw): Likewise.
37085 (ieee128_mtvsrd_64bit): Likewise.
37086 (ieee128_mtvsrd_32bit): Likewise.
37088 2017-01-31 Martin Liska <mliska@suse.cz>
37091 * ipa-prop.c (ipa_free_all_node_params): Call release method
37092 instead of ~sumbol_summary to not to trigger double times
37095 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
37097 PR tree-optimization/71691
37098 * bitmap.h (class auto_bitmap): New.
37099 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
37100 is_maybe_undefined instead of ssa_undefined_value_p.
37102 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37104 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37105 __S390_ARCH_LEVEL__ to __ARCH__.
37107 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37109 PR tree-optimization/79267
37110 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
37111 if should_remove_lhs_p is true.
37113 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
37116 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
37117 (add_alignment_attribute): New.
37118 (base_type_die): Add alignment attribute.
37119 (subrange_type_die): Likewise.
37120 (modified_type_die): Likewise.
37121 (gen_array_type_die): Likewise.
37122 (gen_descr_array_type_die: Likewise.
37123 (gen_enumeration_type_die): Likewise.
37124 (gen_subprogram_die): Likewise.
37125 (gen_variable_die): Likewise.
37126 (gen_field_die): Likewise.
37127 (gen_ptr_to_mbr_type_die): Likewise.
37128 (gen_struct_or_union_type_die): Likewise.
37129 (gen_subroutine_type_die): Likewise.
37130 (gen_typedef_die): Likewise.
37131 (base_type_cmp): Compare alignment attribute.
37133 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37136 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
37137 (setb_unsigned) New pattern for setb with CCUNS.
37138 * config/rs6000/rs6000.c (expand_block_compare): Use a different
37139 subfc./subfe sequence to avoid overflow problems. Generate a
37140 shorter sequence with cmpld/setb for power9.
37141 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
37142 for generating subfc. instruction.
37143 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
37144 now uses this instruction.
37146 2017-01-30 Ian Lance Taylor <iant@google.com>
37149 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
37150 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
37152 2017-01-30 Martin Sebor <msebor@redhat.com>
37154 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
37155 Move constant to the right of a relational operator.
37156 (get_mpfr_format_length, format_character, format_string): Ditto.
37157 (should_warn_p, maybe_warn): Same.
37159 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
37161 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
37164 * asan.c (get_translation_unit_decl): Remove function.
37165 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
37167 2017-01-30 Martin Liska <mliska@suse.cz>
37169 PR gcov-profile/79259
37170 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
37171 -fprofile-generate.
37173 2017-01-30 Martin Liska <mliska@suse.cz>
37176 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
37177 Initialize variables with NULL value.
37179 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
37182 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
37184 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
37186 2017-01-30 Richard Biener <rguenther@suse.de>
37188 PR tree-optimization/79276
37189 * tree-vrp.c (process_assert_insertions): Properly adjust common
37190 when removing a duplicate.
37192 2017-01-30 Richard Biener <rguenther@suse.de>
37194 PR tree-optimization/79256
37195 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37196 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37198 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
37200 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
37203 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
37204 ("*r<noxa>sbg_<mode>_sll_bitmask")
37205 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
37206 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
37207 Use contiguous_bitmask_nowrap_operand.
37209 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37212 * config/rs6000/altivec.h (vec_xl): Revise #define.
37213 (vec_xst): Likewise.
37215 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
37217 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
37219 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
37221 PR rtl-optimization/79194
37222 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
37223 traps before call to bypass_conditional_jumps.
37225 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37227 PR tree-optimization/71374
37228 * lra-constraints.c (check_conflict_input_operands): New.
37229 (match_reload): Use it.
37231 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37234 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
37235 account to calculate conflict_set.
37237 2017-01-27 Bin Cheng <bin.cheng@arm.com>
37239 PR rtl-optimization/78559
37240 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
37241 other_insn in combine.
37243 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
37245 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
37246 uint16_type_node for BT_UINT16.
37248 2017-01-27 David Malcolm <dmalcolm@redhat.com>
37250 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
37251 "RTL Tests" to menu.
37252 (GIMPLE Tests): New node.
37253 (RTL Tests): New node.
37255 2017-01-27 Richard Biener <rguenther@suse.de>
37257 PR tree-optimization/79245
37258 * tree-loop-distribution.c (distribute_loop): Apply cost
37259 modeling also to detected patterns.
37261 2017-01-27 Richard Biener <rguenther@suse.de>
37263 PR tree-optimization/71433
37264 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
37265 (compare_assert_loc): New function.
37266 (process_assert_insertions): Sort and optimize assert locations
37267 to remove duplicates and push down identical assertions on
37268 edges to their destination block.
37270 2017-01-27 Richard Biener <rguenther@suse.de>
37272 PR tree-optimization/79244
37273 * tree-vrp.c (remove_range_assertions): Forcefully propagate
37274 out SSA names even if abnormal.
37276 2017-01-27 Jakub Jelinek <jakub@redhat.com>
37278 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
37279 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
37280 instead of MPFR_RNDN.
37282 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
37285 * arm.c (arm_option_override): Don't call build_target_option_node
37286 until after doing all option overrides.
37287 (arm_valid_target_attribute_tree): Likewise.
37289 2017-01-27 Martin Liska <mliska@suse.cz>
37291 * doc/invoke.texi (-fprofile-arcs): Document profiling support
37292 for {cd}tors and C++ {cd}tors.
37294 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
37296 * config/s390/s390.md ("*setmem_long_and")
37297 ("*setmem_long_and_31z"): Use zero_extend instead of and.
37299 2017-01-26 Martin Sebor <msebor@redhat.com>
37301 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
37304 2017-01-26 Martin Sebor <msebor@redhat.com>
37306 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
37307 HAVE_DFmode before using XFmode or DFmode.
37308 (parse_directive): Avoid using the z length modifier to avoid
37309 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
37311 PR middle-end/78703
37312 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
37313 to accept adjustment as an array.
37314 (get_int_range): New function.
37315 (struct directive): Make width and prec arrays.
37316 (directive::set_width, directive::set_precision): Call get_int_range.
37317 (format_integer, format_floating): Handle width and precision ranges.
37318 (format_string, parse_directive): Same.
37320 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37323 * dwarf2out.c (generate_skeleton_bottom_up): For children with
37324 comdat_type_p set, just clone them, but keep the children in the
37328 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
37329 which have direct callers with -fvar-tracking-assignments enabled
37331 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
37332 inside of type units.
37334 2017-01-26 Martin Sebor <msebor@redhat.com>
37336 PR middle-end/78703
37337 * gimple-ssa-sprintf.c (struct result_range): Add likely and
37339 (struct format_result): Replace number_chars, number_chars_min,
37340 and number_chars_max with a single member of struct result_range.
37342 (format_result::operator+=): Adjust.
37343 (struct fmtresult): Remove bounded. Handle likely and unlikely
37345 (fmtresult::adjust_for_width_or_precision): New function.
37346 (fmtresult:type_max_digits): New function.
37347 (bytes_remaining): Handle likely and unlikely counters.
37348 (min_bytes_remaining): Remove.
37349 (format_percent): Simplify.
37350 (format_integer, format_floating): Set likely and unlikely counters.
37351 (get_string_length, format_character, format_string): Same.
37352 (format_plain, should_warn_p): New function.
37353 (maybe_warn): Call should_warn_p. Update diagnostic messages
37354 and handle those for all directives, including plain strings.
37355 (format_directive): Handle likely and unlikely counters.
37356 Remove unnecessary quoting from diagnostics. Add an informational
37358 (add_bytes): Remove.
37359 (pass_sprintf_length::compute_format_length): Simplify.
37360 (try_substitute_return_value): Handle likely and unlikely counters.
37362 2017-01-26 Carl Love <cel@us.ibm.com>
37364 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
37365 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
37367 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
37370 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
37371 endianess for subregs into account.
37372 * lra-constraints.c (lra_constraints): Do risky transformations
37373 always on the first iteration.
37374 * lra-lives.c (check_pseudos_live_through_calls): Add arg
37375 last_call_used_reg_set.
37376 (process_bb_lives): Define and use last_call_used_reg_set.
37377 * lra.c (lra): Always continue after lra_constraints on the first
37380 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
37382 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
37384 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
37386 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37388 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
37389 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
37390 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
37391 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
37392 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
37393 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
37394 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
37395 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
37396 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
37398 2017-01-26 Marek Polacek <polacek@redhat.com>
37401 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
37402 for the third operand.
37404 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37406 PR middle-end/79236
37407 * omp-low.c (struct omp_context): Add simt_stmt field.
37408 (scan_omp_for): Return omp_context *.
37409 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
37410 context to the _simt_ SIMD stmt.
37411 (lower_omp_for): For combined SIMD with sibling _simt_
37412 SIMD, make sure to use the same decls in _looptemp_
37413 clauses as in the sibling.
37415 2017-01-26 David Sherwood <david.sherwood@arm.com>
37417 PR middle-end/79212
37418 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
37421 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37424 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
37425 emit fld b; fld a; if possible.
37427 * brig-builtins.def: Update copyright years.
37428 * config/arm/arm_acle_builtins.def: Update copyright years.
37430 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
37433 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
37434 constraint instead of o for the stxsd instruction.
37436 2017-01-25 Carl Love <cel@us.ibm.com>
37438 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
37439 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
37441 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
37443 * doc/invoke.texi (C++ Dialect Options): Fix typo.
37445 2017-01-25 Richard Biener <rguenther@suse.de>
37447 PR tree-optimization/69264
37448 * target.def (vector_alignment_reachable): Improve documentation.
37449 * doc/tm.texi: Regenerate.
37450 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
37452 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
37453 earlier changes with respect to TYPE_USER_ALIGN.
37454 (vector_alignment_reachable_p): Likewise. Improve dumping.
37456 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37459 * config/arm/arm.md (xordi3): Force constant operand into a register
37462 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37464 * doc/invoke.texi (-fstore-merging): Correct default optimization
37465 levels at which it is enabled.
37466 (-O): Move -fstore-merging from list to...
37469 2017-01-25 Richard Biener <rguenther@suse.de>
37472 * omp-expand.c: Include debug.h.
37473 (expand_omp_taskreg): Make sure to generate early debug before
37474 outlining anything from a function.
37475 (expand_omp_target): Likewise.
37476 (grid_expand_target_grid_body): Likewise.
37478 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
37481 * asan.c (get_translation_unit_decl): New function.
37482 (asan_add_global): Extract modules file name from globals
37483 TRANSLATION_UNIT_DECL name.
37485 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
37488 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
37489 for long calls with APCS frame and VFP.
37491 2017-01-24 David Malcolm <dmalcolm@redhat.com>
37493 * cfg.c (original_copy_tables_initialized_p): New function.
37494 * cfg.h (original_copy_tables_initialized_p): New decl.
37495 * cfgrtl.c (relink_block_chain): Guard the call to
37496 free_original_copy_tables with a call to
37497 original_copy_tables_initialized_p.
37498 * cgraph.h (symtab_node::native_rtl_p): New decl.
37499 * cgraphunit.c (symtab_node::native_rtl_p): New function.
37500 (symtab_node::needed_p): Don't assert for early assembly output
37501 for __RTL functions.
37502 (cgraph_node::finalize_function): Set "force_output" for __RTL
37504 (cgraph_node::analyze): Bail out early for __RTL functions.
37505 (analyze_functions): Update assertion to support __RTL functions.
37506 (cgraph_node::expand): Bail out early for __RTL functions.
37507 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
37509 * function.h (struct function): Update comment for field
37511 * gimple-expr.c: Include "tree-pass.h".
37512 (gimple_has_body_p): Return false for __RTL functions.
37513 * Makefile.in (OBJS): Add run-rtl-passes.o.
37514 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
37516 (gcc::pass_manager::get_clean_slate): New accessor.
37517 * passes.c: Include "insn-addr.h".
37518 (should_skip_pass_p): Add logging. Update logic for running
37519 "expand" to be compatible with both __GIMPLE and __RTL. Guard
37520 property-provider override so it is only done for gimple passes.
37522 (skip_pass): New function.
37523 (execute_one_pass): Call skip_pass when skipping passes.
37524 * read-md.c (md_reader::read_char): Support filtering
37525 the input to a subset of line numbers.
37526 (md_reader::md_reader): Initialize fields
37527 m_first_line and m_last_line.
37528 (md_reader::read_file_fragment): New function.
37529 * read-md.h (md_reader::read_file_fragment): New decl.
37530 (md_reader::m_first_line): New field.
37531 (md_reader::m_last_line): New field.
37532 * read-rtl-function.c (function_reader::create_function): Only
37533 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
37534 curr_properties. Set DECL_INITIAL to a dummy block.
37535 (read_rtl_function_body_from_file_range): New function.
37536 * read-rtl-function.h (read_rtl_function_body_from_file_range):
37538 * run-rtl-passes.c: New file.
37539 * run-rtl-passes.h: New file.
37541 2017-01-24 Jeff Law <law@redhat.com>
37543 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
37546 2017-01-24 Bin Cheng <bin.cheng@arm.com>
37548 PR tree-optimization/79159
37549 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
37550 (record_nonwrapping_iv): Improve boundary using above function if no
37551 value range information.
37553 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
37554 Martin Jambor <mjambor@suse.cz>
37556 * brig-builtins.def: New file.
37557 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
37558 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
37559 (DEF_HSAIL_SAT_BUILTIN): Likewise.
37560 (DEF_HSAIL_INTR_BUILTIN): Likewise.
37561 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
37562 * builtin-types.def (BT_INT8): New.
37563 (BT_INT16): Likewise.
37564 (BT_UINT8): Likewise.
37565 (BT_UINT16): Likewise.
37566 (BT_FN_ULONG): Likewise.
37567 (BT_FN_UINT_INT): Likewise.
37568 (BT_FN_UINT_ULONG): Likewise.
37569 (BT_FN_UINT_LONG): Likewise.
37570 (BT_FN_UINT_PTR): Likewise.
37571 (BT_FN_ULONG_PTR): Likewise.
37572 (BT_FN_INT8_FLOAT): Likewise.
37573 (BT_FN_INT16_FLOAT): Likewise.
37574 (BT_FN_UINT32_FLOAT): Likewise.
37575 (BT_FN_UINT16_FLOAT): Likewise.
37576 (BT_FN_UINT8_FLOAT): Likewise.
37577 (BT_FN_UINT64_FLOAT): Likewise.
37578 (BT_FN_UINT16_UINT32): Likewise.
37579 (BT_FN_UINT32_UINT16): Likewise.
37580 (BT_FN_UINT16_UINT16_UINT16): Likewise.
37581 (BT_FN_INT_PTR_INT): Likewise.
37582 (BT_FN_UINT_PTR_UINT): Likewise.
37583 (BT_FN_LONG_PTR_LONG): Likewise.
37584 (BT_FN_ULONG_PTR_ULONG): Likewise.
37585 (BT_FN_VOID_UINT64_UINT64): Likewise.
37586 (BT_FN_UINT8_UINT8_UINT8): Likewise.
37587 (BT_FN_INT8_INT8_INT8): Likewise.
37588 (BT_FN_INT16_INT16_INT16): Likewise.
37589 (BT_FN_INT_INT_INT): Likewise.
37590 (BT_FN_UINT_FLOAT_UINT): Likewise.
37591 (BT_FN_FLOAT_UINT_UINT): Likewise.
37592 (BT_FN_ULONG_UINT_UINT): Likewise.
37593 (BT_FN_ULONG_UINT_PTR): Likewise.
37594 (BT_FN_ULONG_ULONG_ULONG): Likewise.
37595 (BT_FN_UINT_UINT_UINT): Likewise.
37596 (BT_FN_VOID_UINT_PTR): Likewise.
37597 (BT_FN_UINT_UINT_PTR: Likewise.
37598 (BT_FN_UINT32_UINT64_PTR): Likewise.
37599 (BT_FN_INT_INT_UINT_UINT): Likewise.
37600 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
37601 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
37602 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
37603 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
37604 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
37605 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
37606 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
37607 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
37608 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
37609 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
37610 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
37611 * doc/frontends.texi: List BRIG FE.
37612 * doc/install.texi (Testing): Add BRIG tesring requirements.
37613 * doc/invoke.texi (Overall Options): Mention BRIG.
37614 * doc/standards.texi (Standards): Doucment BRIG HSA version.
37616 2017-01-24 Richard Biener <rguenther@suse.de>
37618 PR translation/79208
37619 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
37621 2017-01-24 Martin Jambor <mjambor@suse.cz>
37624 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
37625 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
37626 and known_contexts.
37628 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
37630 PR middle-end/79123
37631 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
37632 casts from signed to unsigned really don't have a range.
37634 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
37636 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
37637 GMP_RNDx for compatiblity.
37639 2017-01-24 Martin Liska <mliska@suse.cz>
37642 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
37643 that would prevent us to call alloca with -1 as argument.
37645 2017-01-24 Jakub Jelinek <jakub@redhat.com>
37647 * dwarf2out.c (output_compilation_unit_header, output_file_names):
37648 Avoid -Wformat-security warning.
37650 2017-01-23 Andrew Pinski <apinski@cavium.com>
37652 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
37655 2017-01-23 Martin Sebor <msebor@redhat.com>
37657 PR middle-end/78703
37658 * gimple-ssa-sprintf.c (warn_level): New global.
37659 (format_integer): Use it here and throughout the rest of the file.
37660 Use the same switch to compute sign as base.
37661 (maybe_warn): New function.
37662 (format_directive): Factor out warnings into maybe_warn.
37663 Add debugging output. Use warn_level.
37664 (add_bytes): Use warn_level.
37665 (pass_sprintf_length::compute_format_length): Add debugging output.
37666 (try_substitute_return_value): Same.
37667 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
37669 PR middle-end/78703
37670 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
37671 (struct fmtresult, format_integer, format_floating): Adjust.
37672 (fmtresult::fmtresult): Set max correctly in two argument ctor.
37673 (get_string_length, format_string,format_directive): Same.
37674 (pass_sprintf_length::compute_format_length): Same.
37675 (try_substitute_return_value): Simplify slightly.
37677 PR middle-end/78703
37678 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
37679 (fmtresult::operator+=): Outlined.
37680 (struct fmtresult): Add ctors.
37681 (struct conversion_spec): Rename...
37682 (struct directive): ...to this. Add and remove data members.
37683 (directive::set_width, directive::set_precision): New functions.
37684 (format_percent): Use fmtresult ctor.
37685 (get_width_and_precision): Remove.
37686 (format_integer): Make naming changes. Avoid computing width and
37688 (format_floating): Same. Adjust indentation.
37689 (format_character, format_none): New functions.
37690 (format_string): Moved character handling to format_character.
37691 (format_directive): Remove arguments, change return type.
37692 (parse_directive): New function.
37693 (pass_sprintf_length::compute_format_length): Move directive
37694 parsing to parse_directive.
37696 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37698 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
37699 (assign_assembler_name_if_needed): ... this.
37700 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
37701 (assign_assembler_name_if_needed): ... this.
37702 (free_lang_data_in_cgraph): Adjust callers.
37703 * cgraphunit.c (cgraph_node::analyze): Likewise.
37704 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
37707 2017-01-23 Richard Biener <rguenther@suse.de>
37709 PR tree-optimization/79088
37710 PR tree-optimization/79188
37711 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
37712 resetting loop bounds after last path deletion. Reset loop
37713 bounds of the target loop, make code match the comments.
37714 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
37715 Make sure loops need no fixups.
37717 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
37719 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
37720 exponent support with double type for first argument.
37721 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
37722 type returned by __builtin_vec_extract_sig,
37723 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
37724 functions from "vector int" to "vector unsigned int" or from
37725 "vector long long int" to "vector unsigned long long int".
37726 Changed type returned by __builtin_vec_extract_exp,
37727 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
37728 functions from "vector int" to "vector unsigned int" or from
37729 "vector long long int" to "vector unsigned long long int".
37730 Changed return type of __builtin_vec_test_data_class,
37731 __builtin_vec_test_data_class_sp, and
37732 __builtin_vec_test_data_class_dp from "vector int" to
37733 "vector bool int" or from "vector long long int" to "vector bool
37734 long long int" and changed second argument type from "unsigned
37735 int" to "int". Added new overloaded function forms "vector float
37736 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
37737 "vector float __builtin_vec_insert_exp_sp (vector float, vector
37738 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
37739 double, vector unsigned long long int)" and "vector double
37740 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
37741 long int)". Changed return type of
37742 __builtin_scalar_test_data_class and
37743 __builtin_scalar_test_data_class_sp and
37744 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
37745 int" and changed second argument from "unsigned int" to "int".
37746 Changed type returned by __builtin_scalar_test_neg,
37747 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
37748 from "int" to "bool int". Added new overloaded function form
37749 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
37750 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
37751 exponent double-precision with floating point first argument.
37752 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
37753 documentation of scalar_test_data_class, scalar_test_neg,
37754 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
37755 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
37756 vec_test_data_class built-in functions to reflect refinements in
37757 their type signatures.
37759 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
37761 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
37763 (aarch64_elf_asm_destructor): Likewise.
37765 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
37767 PR rtl-optimization/78634
37768 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
37769 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
37770 * ifcvt.c (noce_try_cmove): Add missing cost check.
37772 PR rtl-optimization/71724
37773 * combine.c (if_then_else_cond): Look for situations where it is
37774 beneficial to undo the work of one of the recursive calls.
37776 2017-01-23 Bin Cheng <bin.cheng@arm.com>
37778 PR tree-optimization/70754
37779 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
37780 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
37781 combined stmt before it if not NULL.
37782 (combine_chains): Process refs reversely and compute dominance point
37785 2017-01-23 Martin Liska <mliska@suse.cz>
37787 PR tree-optimization/79196
37788 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
37789 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
37791 (strlen_optimize_stmt): Call the renamed function.
37793 2017-01-23 Michael Matz <matz@suse.de>
37795 PR tree-optimization/78384
37796 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
37798 2017-01-23 Richard Biener <rguenther@suse.de>
37800 PR tree-optimization/79186
37801 * tree-vrp.c (register_new_assert_for): Make sure we've seen
37802 both incoming edges before moving an assert.
37804 2017-01-23 Martin Jambor <mjambor@suse.cz>
37806 * ipa-prop.c (load_from_param_1): Removed.
37807 (load_from_unmodified_param): Bits from load_from_param_1 put back
37809 (load_from_param): Removed.
37810 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
37811 with stmt. Reverted back to use of load_from_unmodified_param.
37813 2017-01-23 Martin Jambor <mjambor@suse.cz>
37816 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
37817 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
37818 field a pointer to garbage collected vector, mark lattices and
37819 ipcp_orig_node with GTY((skip)).
37820 (ipa_get_param_count): Adjust to descriptors being a pointer.
37821 (ipa_get_param): Likewise.
37822 (ipa_get_type): Likewise.
37823 (ipa_get_param_move_cost): Likewise.
37824 (ipa_set_param_used): Likewise.
37825 (ipa_get_controlled_uses): Likewise.
37826 (ipa_set_controlled_uses): Likewise.
37827 (ipa_is_param_used): Likewise.
37828 (ipa_node_params_t): Move into garbage collector. New methods insert
37830 (ipa_node_params_sum): Annotate wth GTY(()).
37831 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
37833 (ipa_load_from_parm_agg): Adjust declaration.
37834 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
37835 * ipa-profile.c (ipa_profile): Likewise.
37836 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
37837 (ipa_populate_param_decls): Make descriptors parameter garbage
37839 (ipa_dump_param): Adjust to descriptors being a pointer.
37840 (ipa_alloc_node_params): Likewise.
37841 (ipa_initialize_node_params): Likewise.
37842 (load_from_param_1): Make descriptors parameter garbage collected.
37843 (load_from_unmodified_param): Likewise.
37844 (load_from_param): Likewise.
37845 (ipa_load_from_parm_agg): Likewise.
37846 (ipa_node_params::~ipa_node_params): Removed.
37847 (ipa_free_all_node_params): Remove call to delete operator.
37848 (ipa_node_params_t::insert): New.
37849 (ipa_node_params_t::remove): Likewise.
37850 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
37851 copy known_csts and known_contexts vectors.
37852 (ipa_read_node_info): Adjust to descriptors being a pointer.
37853 (ipcp_modif_dom_walker): Make m_descriptors field garbage
37855 (ipcp_transform_function): Make descriptors variable garbage
37858 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
37860 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
37861 * config/i386/avx512dqintrin.h: Ditto.
37862 * config/i386/avx512fintrin.h: Ditto.
37863 * config/i386/i386.c: Handle new builtins.
37864 * config/i386/i386-builtin.def: Add new builtins.
37865 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
37866 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
37868 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37869 Martin Liska <mliska@suse.cz>
37871 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
37872 * asan.c (asan_expand_poison_ifn): Support stores and use
37873 appropriate ASAN report function.
37874 * internal-fn.c (expand_ASAN_POISON_USE): New function.
37875 * internal-fn.def (ASAN_POISON_USE): Declare.
37876 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
37877 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
37878 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
37879 ASAN_POISON calls w/o LHS.
37880 * tree-ssa.c (execute_update_addresses_taken): Create clobber
37881 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
37882 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
37883 * gimplify.c (asan_poison_variables): Add attribute
37884 use_after_scope_memory to variables that really needs to live
37886 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
37887 having the attribute.
37889 2017-01-23 Martin Liska <mliska@suse.cz>
37891 * asan.c (create_asan_shadow_var): New function.
37892 (asan_expand_poison_ifn): Likewise.
37893 * asan.h (asan_expand_poison_ifn): New declaration.
37894 * internal-fn.c (expand_ASAN_POISON): Likewise.
37895 * internal-fn.def (ASAN_POISON): New builtin.
37896 * sanopt.c (pass_sanopt::execute): Expand
37897 asan_expand_poison_ifn.
37898 * tree-inline.c (copy_decl_for_dup_finish): Make function
37900 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
37901 * tree-ssa.c (is_asan_mark_p): New function.
37902 (execute_update_addresses_taken): Rewrite local variables
37903 (identified just by use-after-scope as addressable) into SSA.
37905 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37907 * doc/install.texi (Specific): opensource.apple.com uses https
37908 now. Remove trailing slash.
37910 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37912 * README.Portability: Remove note on an Irix compatibility issue.
37914 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
37916 * gcov.c (INCLUDE_ALGORITHM): Define.
37917 (INCLUDE_VECTOR): Define.
37918 No longer include <vector> and <algorithm> directly.
37920 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
37922 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
37924 * doc/invoke.texi (Code Gen Options): Ditto.
37926 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
37929 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
37931 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
37933 rtl-optimization/79125
37934 * cprop.c (local_cprop_pass): Handle cases where we make an
37935 unconditional trap.
37937 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
37941 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
37942 read from, for big endian.
37944 2017-01-20 Jiong Wang <jiong.wang@arm.com>
37946 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
37947 register pauth builtins for LP64 only.
37949 2017-01-20 Marek Polacek <polacek@redhat.com>
37952 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
37955 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37957 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
37959 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
37960 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
37961 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
37963 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37966 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
37967 in big-endian mode when they are not a single duplicated value.
37969 2017-01-20 Richard Biener <rguenther@suse.de>
37971 * BASE-VER: Bump to 7.0.1.
37973 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37975 * omp-low.c (omplow_simd_context): New struct. Use it...
37976 (lower_rec_simd_input_clauses): ...here and...
37977 (lower_rec_input_clauses): ...here to hold common data. Adjust all
37978 references to idx, lane, max_vf, is_simt.
37980 2017-01-20 Graham Markall <graham.markall@embecosm.com>
37982 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
37985 2017-01-20 Martin Jambor <mjambor@suse.cz>
37987 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
37988 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
37990 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
37991 (GTFILES): Rename hsa.c to hsa-common.c.
37992 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
37993 * hsa-dump.c: Likewise.
37994 * hsa-gen.c: Likewise.
37995 * hsa-regalloc.c: Likewise.
37996 * ipa-hsa.c: Likewise.
37997 * omp-expand.c: Likewise.
37998 * omp-low.c: Likewise.
37999 * toplev.c: Likewise.
38001 2017-01-20 Marek Polacek <polacek@redhat.com>
38004 * doc/invoke.texi: Document -Wduplicated-branches.
38005 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38006 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38007 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38008 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38009 return 0 only when not OEP_LEXICOGRAPHIC.
38010 (fold_build_cleanup_point_expr): Use the expression
38011 location when building CLEANUP_POINT_EXPR.
38012 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38013 * tree.c (add_expr): Handle error_mark_node.
38015 2017-01-20 Martin Liska <mliska@suse.cz>
38018 * tree-profile.c (init_ic_make_global_vars): Do not call
38020 (gimple_init_gcov_profiler): Likewise.
38022 2017-01-20 Martin Liska <mliska@suse.cz>
38025 * cgraph.h (maybe_create_reference): Remove argument and
38027 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38029 * ipa-cp.c (create_specialized_node): Likewise.
38030 * symtab.c (symtab_node::maybe_create_reference): Handle
38031 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38033 2017-01-20 Martin Liska <mliska@suse.cz>
38035 * read-rtl-function.c (function_reader::create_function): Use
38036 build_decl instread of build_decl_stat.
38038 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38040 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38041 * config/i386/avx512dqintrin.h: Ditto.
38042 * config/i386/avx512fintrin.h: Ditto.
38043 * config/i386/i386-builtin-types.def: Add new types.
38044 * config/i386/i386.c: Handle new types.
38045 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
38046 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
38047 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
38048 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
38049 (__builtin_ia32_kshiftridi): New.
38050 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
38052 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
38056 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
38057 define to rs6000_init_stack_protect_guard.
38058 (rs6000_init_stack_protect_guard): New function.
38060 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38061 Yunqiang Su <yunqiang.su@imgtec.com>
38063 * config.gcc (supported_defaults): Add madd4.
38064 (with_madd4): Add validation.
38065 (all_defaults): Add madd4.
38066 * config/mips/mips.opt (mmadd4): New option.
38067 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38069 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38071 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
38072 (ISA_HAS_FUSED_MADD4): Likewise.
38073 * doc/invoke.texi (-mmadd4): Document the new option.
38074 * doc/install.texi (--with-madd4): Document the new option.
38076 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38078 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
38079 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
38080 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
38081 (aarch64_init_pauth_hint_builtins): New.
38082 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
38083 (aarch64_expand_builtin): Expand new builtins.
38085 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38087 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
38088 * combine-stack-adj.c (no_unhandled_cfa): Handle
38089 REG_CFA_TOGGLE_RA_MANGLE.
38090 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
38091 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
38092 info for return address signing.
38093 (aarch64_expand_epilogue): Likewise.
38095 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38097 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
38098 * config/aarch64/aarch64-protos.h
38099 (aarch64_return_address_signing_enabled): New declaration.
38100 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
38102 (aarch64_expand_prologue): Sign return address before it's pushed onto
38104 (aarch64_expand_epilogue): Authenticate return address fetched from
38106 (aarch64_override_options): Sanity check for ILP32 and ISA level.
38107 (aarch64_attributes): New function attributes for "sign-return-address".
38108 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
38109 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
38110 ("*do_return"): Generate combined instructions according to key index.
38111 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
38112 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
38114 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
38115 * config/aarch64/aarch64.opt (msign-return-address=): New.
38116 * doc/extend.texi (AArch64 Function Attributes): Documents
38117 "sign-return-address=".
38118 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
38120 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38122 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
38123 overall option summary.
38125 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38127 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
38128 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
38129 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
38130 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
38132 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
38134 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
38135 -mpower9-minmax by default for -mcpu=power9.
38136 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
38137 128-bit floating point.
38139 2017-01-20 Alan Modra <amodra@gmail.com>
38141 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
38142 optimizing for size.
38144 2017-01-20 Alan Modra <amodra@gmail.com>
38147 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
38148 for strcmp and strncmp from corresponding builtin decl.
38150 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38152 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
38153 instead of i386/rtems-64.h.
38154 * config/i386/rtems-64.h: Remove.
38156 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38160 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
38162 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
38164 2017-01-19 Tamar Christina <tamar.christina@arm.com>
38166 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
38167 Change int to HOST_WIDE_INT.
38168 * config/aarch64/aarch64-protos.h
38169 (aarch64_simd_gen_const_vector_dup): Likewise.
38170 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
38172 2017-01-19 David Malcolm <dmalcolm@redhat.com>
38174 * langhooks-def.h (lhd_type_for_size): New decl.
38175 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
38176 * langhooks.c (lhd_type_for_size): New function, taken from
38179 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
38181 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
38182 define_bypass for CR latency.
38183 (power9-cracked-alu): Update bypass latency and remove power9-branch.
38184 (power9-alu2): Add define_bypass for CR latency.
38186 (power9-mul): Update insn latency.
38187 (power9-mul-compare): Update insn latency, bypass latency and remove
38190 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38192 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
38194 * config/aarch64/aarch64.md
38195 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
38196 aarch64_nopcrelative_literal_loads.
38197 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
38199 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
38201 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
38202 TARGET_LOONGSON_3A.
38203 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
38205 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
38208 * config.gcc (supported_defaults): Add lxc1-sxc1.
38209 (with_lxc1_sxc1): Add validation.
38210 (all_defaults): Add lxc1-sxc1.
38211 * config/mips/mips.opt (mlxc1-sxc1): New option.
38212 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38214 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38215 __mips_no_lxc1_sxc1.
38216 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
38217 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
38218 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
38220 2017-01-19 Richard Biener <rguenther@suse.de>
38222 PR tree-optimization/72488
38223 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
38224 sure to restore SSA info.
38225 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
38227 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
38229 PR rtl-optimization/79121
38230 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
38231 of the inner type when shifting an extended value.
38233 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38236 * symtab.c (symtab_node::equal_address_to): Fix comparing of
38237 interposable aliases.
38239 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
38242 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
38243 Use the evmergelohi instruction.
38244 (mov_si<mode>_e500_subreg4_2_le): Likewise.
38245 (mov_sitf_e500_subreg8_2_be): Likewise.
38246 (mov_sitf_e500_subreg12_2_le): Likewise.
38247 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
38248 (mov_si<mode>_e500_subreg4_2_be): Likewise.
38249 (mov_sitf_e500_subreg8_2_le): Likewise.
38250 (mov_sitf_e500_subreg12_2_be): Likewise.
38252 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38254 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
38255 attribute from vecsimple to vecperm.
38256 (altivec_vbpermq2): Likewise.
38258 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38261 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
38263 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38264 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
38265 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
38266 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
38267 case where N arg is SIZE_MAX.
38268 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
38269 (cmpstrsi): Add pattern.
38271 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
38273 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
38274 __builtin_vec_revb builtins.
38275 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
38276 built-in functions to support generation of the ISA 3.0 XXBR<x>
38277 vector byte reverse instructions.
38278 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
38279 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
38280 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
38281 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
38282 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
38283 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
38284 (P9V_BUILTIN_VEC_REVB): Likewise.
38285 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
38286 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
38287 (p9_xxbrq_v16qi): Likewise.
38288 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
38289 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
38290 (p9_xxbrh_v8hi): Likewise.
38291 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
38292 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
38293 vec_revb built-in functions.
38295 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
38297 PR rtl-optimization/78952
38298 * config/i386/i386.md (any_extract): New code iterator.
38299 (*insvqi_2): Use any_extract for source operand.
38300 (*insvqi_3): Use any_shiftrt for source operand.
38302 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
38304 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
38306 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
38308 2017-01-18 Matthias Klose <doko@ubuntu.com>
38310 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
38312 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38314 * config/rs6000/altivec.h (vec_bperm): Change #define.
38315 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
38316 (altivec_vbpermq2): New define_insn.
38317 (altivec_vbpermd): Likewise.
38318 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
38319 function interface.
38320 (VBPERMD): Likewise.
38321 (VBPERM): New polymorphic function interface.
38322 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
38323 Add entries for P9V_BUILTIN_VEC_VBPERM.
38324 * doc/extend.texi: Add interfaces for vec_bperm.
38326 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38328 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
38329 first letter of error messages.
38330 (s390_resolve_overloaded_builtin): Likewise.
38331 * config/s390/s390.c (s390_expand_builtin): Likewise.
38332 (s390_invalid_arg_for_unprototyped_fn): Likewise.
38333 (s390_valid_target_attribute_inner_p): Likewise.
38334 * config/s390/s390.md ("tabort"): Likewise.
38336 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
38338 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
38339 (ISA_AVOID_DIV_HILO): New macro.
38340 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
38341 (ISA_HAS_DDIV): Likewise.
38343 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38345 * doc/invoke.texi (fabi-version): Correct number of occurrences.
38347 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38349 * doc/invoke.texi (fabi-version): Spelling fix.
38351 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38354 * doc/invoke.texi (fabi-version): Mention mangling fix for
38357 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38360 * doc/invoke.texi (fabi-version): Document discriminator mangling.
38362 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
38365 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
38366 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
38368 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
38370 (stack_protect_test): Ditto.
38371 * config/rs6000/rs6000.opt (mstack-protector-guard=,
38372 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
38374 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
38375 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
38376 -mstack-protector-guard-offset=.
38377 (RS/6000 and PowerPC Options): Ditto.
38379 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38381 * config/i386/i386.h (MASK_CLASS_P): New define.
38382 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
38383 there are no registers from different register sets also when
38384 mask registers are used. Update function comment.
38385 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
38386 to (*k/*r) and (*k/*km) alternatives.
38388 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
38390 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
38391 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
38392 (EH_RETURN_HANDLER_RTX): New define.
38393 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
38394 Force frame pointer in EH return functions.
38395 (aarch64_expand_epilogue): Add barrier for eh_return.
38396 (aarch64_final_eh_return_addr): Remove.
38397 (aarch64_eh_return_handler_rtx): New function.
38398 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
38400 (aarch64_eh_return_handler_rtx): New prototype.
38402 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38404 * config/rs6000/altivec.h (vec_rlmi): New #define.
38405 (vec_vrlnm): Likewise.
38406 (vec_rlnm): Likewise.
38407 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
38408 (UNSPEC_VRLNM): Likewise.
38409 (VIlong): New mode iterator.
38410 (altivec_vrl<VI_char>mi): New define_insn.
38411 (altivec_vrl<VI_char>nm): Likewise.
38412 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
38414 (VRLDNM): Likewise.
38415 (RLNM): New polymorphic function entry.
38416 (VRLWMI): New monomorphic function entry.
38417 (VRLDMI): Likewise.
38418 (RLMI): New polymorphic function entry.
38419 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
38420 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
38421 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
38424 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38427 * dwarf2out.c (field_byte_offset): Restore the
38428 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
38429 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
38430 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
38433 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38436 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
38438 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38441 * dwarf2out.c (add_data_member_location_attribute): For constant
38442 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
38443 instead of DW_AT_data_member_location, DW_AT_bit_offset and
38444 DW_AT_byte_size attributes.
38446 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38448 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
38449 after forcing to constant memory when the code model is medium.
38451 2017-01-17 Julia Koval <julia.koval@intel.com>
38454 * config/i386/avx512fintrin.h
38455 (_mm512_i32gather_ps): Change __addr type to void const*.
38456 (_mm512_mask_i32gather_ps): Ditto.
38457 (_mm512_i32gather_pd): Ditto.
38458 (_mm512_mask_i32gather_pd): Ditto.
38459 (_mm512_i64gather_ps): Ditto.
38460 (_mm512_mask_i64gather_ps): Ditto.
38461 (_mm512_i64gather_pd): Ditto.
38462 (_mm512_mask_i64gather_pd): Ditto.
38463 (_mm512_i32gather_epi32): Ditto.
38464 (_mm512_mask_i32gather_epi32): Ditto.
38465 (_mm512_i32gather_epi64): Ditto.
38466 (_mm512_mask_i32gather_epi64): Ditto.
38467 (_mm512_i64gather_epi32): Ditto.
38468 (_mm512_mask_i64gather_epi32): Ditto.
38469 (_mm512_i64gather_epi64): Ditto.
38470 (_mm512_mask_i64gather_epi64): Ditto.
38471 (_mm512_i32scatter_ps): Change __addr type to void*.
38472 (_mm512_mask_i32scatter_ps): Ditto.
38473 (_mm512_i32scatter_pd): Ditto.
38474 (_mm512_mask_i32scatter_pd): Ditto.
38475 (_mm512_i64scatter_ps): Ditto.
38476 (_mm512_mask_i64scatter_ps): Ditto.
38477 (_mm512_i64scatter_pd): Ditto.
38478 (_mm512_mask_i64scatter_pd): Ditto.
38479 (_mm512_i32scatter_epi32): Ditto.
38480 (_mm512_mask_i32scatter_epi32): Ditto.
38481 (_mm512_i32scatter_epi64): Ditto.
38482 (_mm512_mask_i32scatter_epi64): Ditto.
38483 (_mm512_i64scatter_epi32): Ditto.
38484 (_mm512_mask_i64scatter_epi32): Ditto.
38485 (_mm512_i64scatter_epi64): Ditto.
38486 (_mm512_mask_i64scatter_epi64): Ditto.
38487 * config/i386/avx512pfintrin.h
38488 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
38489 (_mm512_mask_prefetch_i32gather_ps): Ditto.
38490 (_mm512_mask_prefetch_i64gather_pd): Ditto.
38491 (_mm512_mask_prefetch_i64gather_ps): Ditto.
38492 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
38493 (_mm512_prefetch_i32scatter_ps): Ditto.
38494 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
38495 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
38496 (_mm512_prefetch_i64scatter_pd): Ditto.
38497 (_mm512_prefetch_i64scatter_ps): Ditto.
38498 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
38499 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
38500 * config/i386/avx512vlintrin.h
38501 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
38502 (_mm_mmask_i32gather_ps): Ditto.
38503 (_mm256_mmask_i32gather_pd): Ditto.
38504 (_mm_mmask_i32gather_pd): Ditto.
38505 (_mm256_mmask_i64gather_ps): Ditto.
38506 (_mm_mmask_i64gather_ps): Ditto.
38507 (_mm256_mmask_i64gather_pd): Ditto.
38508 (_mm_mmask_i64gather_pd): Ditto.
38509 (_mm256_mmask_i32gather_epi32): Ditto.
38510 (_mm_mmask_i32gather_epi32): Ditto.
38511 (_mm256_mmask_i32gather_epi64): Ditto.
38512 (_mm_mmask_i32gather_epi64): Ditto.
38513 (_mm256_mmask_i64gather_epi32): Ditto.
38514 (_mm_mmask_i64gather_epi32): Ditto.
38515 (_mm256_mmask_i64gather_epi64): Ditto.
38516 (_mm_mmask_i64gather_epi64): Ditto.
38517 (_mm256_i32scatter_ps): Change __addr type to void*.
38518 (_mm256_mask_i32scatter_ps): Ditto.
38519 (_mm_i32scatter_ps): Ditto.
38520 (_mm_mask_i32scatter_ps): Ditto.
38521 (_mm256_i32scatter_pd): Ditto.
38522 (_mm256_mask_i32scatter_pd): Ditto.
38523 (_mm_i32scatter_pd): Ditto.
38524 (_mm_mask_i32scatter_pd): Ditto.
38525 (_mm256_i64scatter_ps): Ditto.
38526 (_mm256_mask_i64scatter_ps): Ditto.
38527 (_mm_i64scatter_ps): Ditto.
38528 (_mm_mask_i64scatter_ps): Ditto.
38529 (_mm256_i64scatter_pd): Ditto.
38530 (_mm256_mask_i64scatter_pd): Ditto.
38531 (_mm_i64scatter_pd): Ditto.
38532 (_mm_mask_i64scatter_pd): Ditto.
38533 (_mm256_i32scatter_epi32): Ditto.
38534 (_mm256_mask_i32scatter_epi32): Ditto.
38535 (_mm_i32scatter_epi32): Ditto.
38536 (_mm_mask_i32scatter_epi32): Ditto.
38537 (_mm256_i32scatter_epi64): Ditto.
38538 (_mm256_mask_i32scatter_epi64): Ditto.
38539 (_mm_i32scatter_epi64): Ditto.
38540 (_mm_mask_i32scatter_epi64): Ditto.
38541 (_mm256_i64scatter_epi32): Ditto.
38542 (_mm256_mask_i64scatter_epi32): Ditto.
38543 (_mm_i64scatter_epi32): Ditto.
38544 (_mm_mask_i64scatter_epi32): Ditto.
38545 (_mm256_i64scatter_epi64): Ditto.
38546 (_mm256_mask_i64scatter_epi64): Ditto.
38547 (_mm_i64scatter_epi64): Ditto.
38548 (_mm_mask_i64scatter_epi64): Ditto.
38549 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
38550 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
38551 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
38552 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
38553 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
38554 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
38555 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
38556 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
38557 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
38558 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
38559 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
38560 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
38561 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
38562 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
38563 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
38564 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
38565 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
38566 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
38567 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
38568 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
38569 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
38570 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
38571 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
38572 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
38573 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
38574 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
38575 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
38576 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
38577 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
38578 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
38579 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
38580 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
38581 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
38582 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
38583 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
38584 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
38585 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
38586 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
38587 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
38588 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
38589 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
38590 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
38591 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
38592 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
38593 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
38594 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
38595 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
38596 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
38597 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
38598 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
38599 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
38600 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
38601 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
38602 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
38603 definitions accordingly.
38605 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
38606 Kuan-Lin Chen <kuanlinchentw@gmail.com>
38609 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
38612 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
38615 * ira-conflicts.c (ira_build_conflicts): Update total conflict
38616 hard regs for inner regno.
38618 2017-01-17 Martin Liska <mliska@suse.cz>
38621 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
38622 assumption and add comment.
38624 2017-01-17 Nathan Sidwell <nathan@acm.org>
38626 * ipa-visibility.c (localize_node): New function, broken out of ...
38627 (function_and_variable_visibility): ... here. Call it.
38629 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38631 PR middle-end/77445
38632 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
38633 correctly set frequency of oudgoing edge.
38634 (duplicate_thread_path): Fix profile updating.
38636 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38639 * configure.ac: Add GCC_BASE_VER.
38640 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
38641 version from BASE-VER file.
38642 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
38643 (gcc.o): Depend on $(BASEVER).
38644 * common.opt (dumpfullversion): New option.
38645 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
38646 * doc/invoke.texi: Document -dumpfullversion.
38647 * doc/install.texi: Document --with-gcc-major-version-only.
38648 * configure: Regenerated.
38650 2017-01-17 Richard Biener <rguenther@suse.de>
38652 PR tree-optimization/71433
38653 * tree-vrp.c (register_new_assert_for): Merge same asserts
38654 on all incoming edges.
38655 (process_assert_insertions_for): Handle insertions at the
38658 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
38660 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
38661 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
38663 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
38666 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
38669 2017-01-17 Alan Modra <amodra@gmail.com>
38672 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
38673 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
38674 symbolic stack limit when pic.
38676 2017-01-16 Martin Sebor <msebor@redhat.com>
38678 PR tree-optimization/78608
38679 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
38681 2017-01-16 Jeff Law <law@redhat.com>
38684 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
38685 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
38686 for several include directories that may be relative to sysroot.
38687 * config/i386/x-mingw32 (gplus_includedir): Define.
38688 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
38689 (native_system_includedir): Likewise.
38690 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
38691 override if TARGET_SYSTEM_ROOT is defined.
38692 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
38694 PR tree-optimization/79090
38695 PR tree-optimization/33562
38696 PR tree-optimization/61912
38697 PR tree-optimization/77485
38698 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
38699 and computed trims into the dump file.
38701 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38703 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
38705 2017-01-16 Jakub Jelinek <jakub@redhat.com>
38708 * gimplify.c (gimplify_init_constructor): If want_value and
38709 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
38713 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
38714 sequence. Formatting fixes.
38715 (doloop_optimize): Formatting fixes.
38718 * gcc.c (debug_level_greater_than_spec_func): New function.
38719 (static_spec_functions): Add debug-level-gt spec function.
38720 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
38722 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38723 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38724 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
38725 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
38726 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
38727 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
38729 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
38731 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
38732 QImode fixups to general and mask registers only.
38734 2017-01-16 Carl Love <cel@us.ibm.com>
38736 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
38737 for built-in functions
38738 vector signed char vec_nabs (vector signed char)
38739 vector signed short vec_nabs (vector signed short)
38740 vector signed int vec_nabs (vector signed int)
38741 vector signed long long vec_nabs (vector signed long long)
38742 vector float vec_nabs (vector float)
38743 vector double vec_nabs (vector double)
38744 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
38746 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
38747 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
38748 * doc/extend.texi: Update the documentation file for the new built-in
38751 2017-01-16 Martin Sebor <msebor@redhat.com>
38753 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
38756 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38758 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
38759 UNSPEC_VSX__XXSPLTD to require special splat handling.
38761 2017-01-16 David Malcolm <dmalcolm@redhat.com>
38764 * system.h: Poison strndup.
38766 2017-01-16 Alan Modra <amodra@gmail.com>
38769 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
38772 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
38774 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
38776 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
38778 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
38779 call recog here. Assert that INSN_CODE (insn) is non-negative.
38781 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38784 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
38786 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
38787 in the currently scheduled RTL fragment.
38789 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38791 PR rtl-optimization/78751
38792 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
38795 2017-01-14 Jeff Law <law@redhat.com>
38797 PR tree-optimization/79090
38798 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
38799 variable length stores.
38800 (compute_trims): Delete dead assignment to *trim_tail.
38801 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
38804 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
38806 PR rtl-optimization/78626
38807 PR rtl-optimization/78727
38808 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
38809 of a block, and split such blocks after everything else is finished.
38811 2017-01-14 Alan Modra <amodra@gmail.com>
38814 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
38815 target legitimate_combined_insn.
38816 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
38817 (rs6000_legitimate_combined_insn): New function.
38818 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
38820 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
38821 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
38822 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
38824 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
38826 * doc/frontends.texi (G++ and GCC): Remove references to Java.
38828 2017-01-13 Jeff Law <law@redhat.com>
38830 PR tree-optimization/33562
38831 PR tree-optimization/61912
38832 PR tree-optimization/77485
38833 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
38835 (delete_dead_assignment): Likewise.
38836 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
38837 statement to delete_dead_call and delete_dead_assignment.
38839 2017-01-13 David Malcolm <dmalcolm@redhat.com>
38842 * substring-locations.c (format_warning_va): Strengthen case 1 so
38843 that both endpoints of the substring must be within the format
38844 range for just the substring to be printed.
38846 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
38848 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
38849 * config/i386/i386.c (ix86_target_string): Add missing options
38850 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
38851 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
38852 flags_other and ix86_target_other to flags2_other. Display unknown
38854 (ix86_valid_target_attribute_inner_p): Add missing options and
38855 reorder options by implied ISAs, as in ix86_target_string.
38857 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38859 * hash-table.h (hash_table::too_empty_p): New function.
38860 (hash_table::expand): Use it.
38861 (hash_table::traverse): Likewise.
38862 (hash_table::empty_slot): Use sizeof (value_type) instead of
38863 sizeof (PTR) to convert bytes to elements. Shrink the table
38864 if the current size is excessive for the current number of
38867 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38869 * ira-costs.c (record_reg_classes): Break from the inner loop
38870 early once alt_fail is known to be true. Update outer loop
38871 handling accordingly.
38873 2017-01-13 Jeff Law <law@redhat.com>
38875 * tree-ssa-dse.c (decrement_count): New function.
38876 (increment_start_addr, maybe_trim_memstar_call): Likewise.
38877 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
38878 when we know the partially dead statement is a mem* function.
38880 PR tree-optimization/61912
38881 PR tree-optimization/77485
38882 * tree-ssa-dse.c: Include expr.h.
38883 (maybe_trim_constructor_store): New function.
38884 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
38886 PR tree-optimization/33562
38887 PR tree-optimization/61912
38888 PR tree-optimization/77485
38889 * doc/invoke.texi: Document new dse-max-object-size param.
38890 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
38891 * tree-ssa-dse.c: Include params.h.
38892 (dse_store_status): New enum.
38893 (initialize_ao_ref_for_dse): New, partially extracted from
38895 (valid_ao_ref_for_dse, normalize_ref): New.
38896 (setup_live_bytes_from_ref, compute_trims): Likewise.
38897 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
38898 (maybe_trim_partially_dead_store): Likewise.
38899 (maybe_trim_complex_store): Likewise.
38900 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
38901 Track what bytes live from the original store. Return tri-state
38902 for dead, partially dead or live.
38903 (dse_dom_walker): Add constructor, destructor and new private members.
38904 (delete_dead_call, delete_dead_assignment): New extracted from
38906 (dse_optimize_stmt): Make a member of dse_dom_walker.
38907 Use initialize_ao_ref_for_dse.
38909 PR tree-optimization/33562
38910 PR tree-optimization/61912
38911 PR tree-optimization/77485
38912 * sbitmap.h (bitmap_count_bits): Prototype.
38913 (bitmap_clear_range, bitmap_set_range): Likewise.
38914 * sbitmap.c (bitmap_clear_range): New function.
38915 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
38917 2017-01-13 Martin Liska <mliska@suse.cz>
38920 * function.c (set_cfun): Add new argument force.
38921 * function.h (set_cfun): Likewise.
38922 * ipa-inline-transform.c (inline_call): Use the function when
38923 strict alising from is dropped for function we inline to.
38925 2017-01-13 Richard Biener <rguenther@suse.de>
38927 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
38928 for dumping GIMPLE INTEGER_CSTs.
38930 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
38932 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
38933 to 201112L since C++17.
38935 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
38938 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
38939 if -fsanitize=kernel-address is present.
38941 2017-01-13 Richard Biener <rguenther@suse.de>
38943 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
38944 as _Literal ( type ) number in case usual suffixes do not
38945 preserve all information.
38947 2017-01-13 Richard Biener <rguenther@suse.de>
38949 PR tree-optimization/77283
38950 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
38951 and ssa-iterators.h.
38952 (is_feasible_trace): Implement a cost model based on joiner
38955 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
38958 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
38959 char or short to __float128/_Float128 directly.
38961 2017-01-12 Martin Sebor <msebor@redhat.com>
38963 to -Wformat-overflow.
38964 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
38965 (min_bytes_remaining): Same.
38966 (get_string_length): Same.
38967 (format_string): Same.
38968 (format_directive): Same.
38970 (pass_sprintf_length::handle_gimple_call): Same.
38972 2017-01-12 Jakub Jelinek <jakub@redhat.com>
38974 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
38975 info.nowrite calls with no lhs that can't throw. Return bool
38976 whether gsi_remove has been called or not.
38977 (pass_sprintf_length::handle_gimple_call): Return bool whether
38978 try_substitute_return_value called gsi_remove. Formatting fix.
38979 (pass_sprintf_length::execute): Don't use gsi_remove if
38980 handle_gimple_call returned true.
38983 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
38984 be removed due to side-effects, don't remove following barrier nor
38985 turn the successor edge into fallthru edge.
38987 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38990 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
38991 element-reversing loads and stores as not swappable.
38993 2017-01-12 Nathan Sidwell <nathan@acm.org>
38994 Nicolai Stange <nicstange@gmail.com>
38996 * combine.c (try_combine): Don't ignore result of overlap checking
38997 loop. Combine overlap & asm check into single loop.
38999 2017-01-12 Richard Biener <rguenther@suse.de>
39001 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39002 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39004 2017-01-12 Richard Biener <rguenther@suse.de>
39006 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39007 and TS_TARGET_OPTION directly derive from TS_BASE.
39008 * tree-core.h (tree_optimization_option): Derive from tree_base.
39009 (tree_target_option): Likewise.
39011 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39013 * config/i386/i386.c (memory_address_length): Increase len
39014 only when rip_relative_addr_p returns false.
39016 2017-01-11 Julia Koval <julia.koval@intel.com>
39018 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39019 (OPTION_MASK_ISA_SGX_SET): New.
39020 (ix86_handle_option): Handle OPT_msgx.
39021 * config.gcc: Added sgxintrin.h.
39022 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39023 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39024 * config/i386/i386.c (ix86_target_string): Add -msgx.
39026 (ix86_option_override_internal): Handle new options.
39027 (ix86_valid_target_attribute_inner_p): Add sgx.
39028 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39029 * config/i386/i386.opt: Add msgx.
39030 * config/i386/sgxintrin.h: New file.
39031 * config/i386/x86intrin.h: Add sgxintrin.h.
39033 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39036 * fold-const.c (maybe_nonzero_address): Return 1 for function
39038 (tree_single_nonzero_warnv_p): Don't handle function local objects
39042 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
39045 2017-01-11 David Malcolm <dmalcolm@redhat.com>
39048 * opts.c: Include "spellcheck.h"
39049 (struct string_fragment): New struct.
39050 (struct edit_distance_traits<const string_fragment &>): New
39052 (get_closest_sanitizer_option): New function.
39053 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
39055 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39057 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
39059 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
39060 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
39061 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
39062 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
39063 for initial die_offset if dwarf_split_debug_info.
39064 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
39065 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
39067 (output_skeleton_debug_sections): Formatting fix. Use
39068 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
39069 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
39071 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
39073 * config/arm/cortex-a53.md: Add bypasses for
39074 cortex_a53_r2f_cvt.
39075 (cortex_a53_r2f): Only use for transfers.
39076 (cortex_a53_f2r): Likewise.
39077 (cortex_a53_r2f_cvt): Add reservation for conversions.
39078 (cortex_a53_f2r_cvt): Likewise.
39080 2017-01-11 Tamar Christina <tamar.christina@arm.com>
39082 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
39083 to all inlined functions, change static to extern.
39085 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
39088 * config/arm/arm.c (legitimize_pic_address): Handle reference to
39090 (arm_assemble_integer): Likewise.
39092 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
39094 * config.gcc: Use new awk script to check CPU, FPU and architecture
39095 parameters for --with-... options.
39096 * config/arm/parsecpu.awk: New file
39097 * config/arm/arm-cpus.in: New file.
39098 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
39100 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
39102 * config/arm/t-arm: Update dependency rules.
39103 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
39104 of processing .def files.
39105 * config/arm/genopt.sh: Deleted.
39106 * config/arm/gentune.sh: Deleted.
39107 * config/arm/arm-cores.def: Deleted.
39108 * config/arm/arm-arches.def: Deleted.
39109 * config/arm/arm-fpus.def: Deleted.
39110 * config/arm/arm-tune.md: Regenerated.
39111 * config/arm/arm-tables.opt: Regenerated.
39112 * config/arm/arm-cpu.h: New generated file.
39113 * config/arm/arm-cpu-data.h: New generated file.
39114 * config/arm/arm-cpu-cdata.h: New generated file.
39116 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
39119 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
39121 (input_varpool_node): Unpack dynamically_initialized bit.
39123 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
39125 PR rtl-optimization/79032
39126 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
39127 the alignment of the adjusted memory reference against that of MODE,
39128 instead of the alignment of the original memory reference.
39130 2017-01-11 Martin Jambor <mjambor@suse.cz>
39132 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
39134 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
39135 decorated functions.
39137 2017-01-11 Richard Biener <rguenther@suse.de>
39139 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
39140 set range/nonnull info for PHI results. Do not set it on
39141 stmts marked for removal.
39143 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
39145 * expr.c (store_field): In the bitfield case, fetch the return value
39146 from the registers before applying a single big-endian adjustment.
39147 Always do a final load for a BLKmode value not larger than a word.
39149 2017-01-10 David Malcolm <dmalcolm@redhat.com>
39152 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39153 that we correctly handle column numbers greater than
39154 LINE_MAP_MAX_COLUMN_NUMBER.
39156 2017-01-10 Martin Sebor <msebor@redhat.com>
39158 PR middle-end/78245
39159 * gimple-ssa-sprintf.c (get_destination_size): Call
39160 {init,fini}object_sizes.
39161 * tree-object-size.c (addr_object_size): Adjust.
39162 (pass_through_call): Adjust.
39163 (pass_object_sizes::execute): Adjust.
39164 * tree-object-size.h (fini_object_sizes): Declare.
39166 2017-01-10 Martin Sebor <msebor@redhat.com>
39168 PR tree-optimization/78775
39169 * builtins.c (get_size_range): Move...
39170 * calls.c: ...to here.
39171 (alloc_max_size): Accept zero argument.
39172 (operand_signed_p): Remove.
39173 (maybe_warn_alloc_args_overflow): Call get_size_range.
39174 * calls.h (get_size_range): Declare.
39176 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
39178 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
39179 from TI's devices.csv file as of September 2016.
39180 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
39182 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
39184 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
39185 * doc/invoke.texi: Likewise.
39186 * doc/md.texi: Likewise.
39187 * doc/objc.texi: Likewise.
39189 2017-01-10 Joshua Conner <joshconner@google.com>
39191 * config/arm/fuchsia-elf.h: New file.
39192 * config/fuchsia.h: New file.
39193 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
39194 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
39196 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
39198 2016-01-10 Richard Biener <rguenther@suse.de>
39200 PR tree-optimization/79034
39201 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
39202 Propagate out degenerate PHIs in the joiner.
39204 2017-01-10 Martin Liska <mliska@suse.cz>
39206 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
39207 (sort_congruence_classes_by_decl_uid): Likewise.
39208 (sort_congruence_class_groups_by_decl_uid): Likewise.
39209 (sem_item_optimizer::merge_classes): Sort class, groups in these
39210 classes and members in the groups by DECL_UID of declarations.
39211 This would make merge operations stable.
39213 2017-01-10 Martin Liska <mliska@suse.cz>
39215 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
39216 usage of m_classes_vec.
39217 (sem_item_optimizer::~sem_item_optimizer): Likewise.
39218 (sem_item_optimizer::get_group_by_hash): Likewise.
39219 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
39220 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
39221 (sem_item_optimizer::verify_classes): Likewise.
39222 (sem_item_optimizer::process_cong_reduction): Likewise.
39223 (sem_item_optimizer::dump_cong_classes): Likewise.
39224 (sem_item_optimizer::merge_classes): Likewise.
39225 * ipa-icf.h (congruence_class_hash): Rename from
39226 congruence_class_group_hash. Remove declaration of m_classes_vec.
39228 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
39230 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
39231 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
39232 * config.gcc: Add avx512vpopcntdqintrin.h.
39233 * config/i386/avx512vpopcntdqintrin.h: New.
39234 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
39235 * config/i386/i386-builtin-types.def: Add new types.
39236 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
39237 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
39238 __builtin_ia32_vpopcountq_v8di_mask): New.
39239 * config/i386/i386-c.c (ix86_target_macros_internal): Define
39240 __AVX512VPOPCNTDQ__.
39241 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
39242 (PTA_AVX512VPOPCNTDQ): Define.
39243 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
39244 TARGET_AVX512VPOPCNTDQ_P): Define.
39245 * config/i386/i386.opt: Add mavx512vpopcntdq.
39246 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
39247 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
39249 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39251 PR middle-end/77484
39252 * predict.def (PRED_CALL): Set to 67.
39254 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
39256 * expr.c (store_field): In the bitfield case, if the value comes from
39257 a function call and is of an aggregate type returned in registers, do
39258 not modify the field mode; extract the value in all cases if the mode
39259 is BLKmode and the size is not larger than a word.
39261 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
39264 * config/i386/cpuid.h: Fix undefined behavior.
39266 2017-01-04 Jeff Law <law@redhat.com>
39268 PR tree-optimization/79007
39269 PR tree-optimization/67955
39270 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
39271 conservative for pt.null when flag_non_call_exceptions is on.
39273 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39275 PR translation/79019
39276 PR translation/79020
39277 * params.def (PARAM_INLINE_MIN_SPEEDUP,
39278 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
39279 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
39281 * config/avr/avr.opt (maccumulate-args): Likewise.
39282 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
39283 * common.opt (freport-bug): Likewise.
39284 * cif-code.def (CIF_FINAL_ERROR): Likewise.
39285 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
39286 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
39287 translatable string.
39288 * config/i386/i386.c (function_value_32): Likewise.
39289 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
39290 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
39292 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
39293 * common/config/msp430/msp430-common.c (msp430_handle_option):
39295 * symtab.c (symtab_node::verify_base): Likewise.
39296 * opts.c (set_debug_level): Likewise.
39297 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
39298 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
39299 missing whitespace to translatable strings.
39300 * config/avr/avr.md (bswapsi2): Fix typo in comment.
39301 * config/sh/superh.h: Likewise.
39302 * config/i386/xopintrin.h: Likewise.
39303 * config/i386/znver1.md: Likewise.
39304 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
39305 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
39306 * double-int.h (struct double_int): Likewise.
39307 * double-int.c (div_and_round_double): Likewise.
39308 * wide-int.cc: Likewise.
39309 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
39310 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
39311 * cfgcleanup.c (crossjumps_occured): Renamed to ...
39312 (crossjumps_occurred): ... this.
39313 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
39316 PR tree-optimization/78899
39317 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
39318 returning bool return struct loop *, NULL for failure and the new
39320 (versionable_outer_loop_p): Don't version outer loop if it has
39321 dont_vectorized bit set.
39322 (tree_if_conversion): When versioning outer loop, ensure
39323 tree_if_conversion is performed also on the inner loop of the
39324 non-vectorizable outer loop copy.
39325 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
39326 LOOP_VECTORIZED in inner loop of the scalar outer loop and
39327 prevent vectorization of it.
39328 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
39329 the outer loop vectorization of the non-scalar version is attempted
39330 before vectorization of the inner loop in scalar version. If
39331 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
39332 vectorization of its inner loop.
39333 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
39334 has 2 inner loops, rename also on edges from bb whose single pred
39335 is outer_loop->header. Fix typo in function comment.
39337 2017-01-09 Martin Sebor <msebor@redhat.com>
39340 * asan.c (asan_emit_stack_protection): Increase local buffer size
39341 to avoid snprintf truncation warning.
39343 2017-01-09 Andrew Pinski <apinski@cavium.com>
39345 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
39346 to reference thunderx2t99 for the tuning structure
39347 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
39349 (thunderx2t99_extra_costs): This.
39350 * config/aarch64/aarch64-tune.md: Regenerate.
39351 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
39352 (vulcan_addrcost_table): This.
39353 (vulcan_regmove_cost): Rename to ...
39354 (thunderx2t99_regmove_cost): This.
39355 (vulcan_vector_cost): Rename to ...
39356 (thunderx2t99_vector_cost): this.
39357 (vulcan_branch_cost): Rename to ...
39358 (thunderx2t99_branch_cost): This.
39359 (vulcan_tunings): Rename to ...
39360 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
39361 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
39363 2017-01-09 Martin Jambor <mjambor@suse.cz>
39367 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
39368 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
39369 (propagate_vr_accross_jump_function): Use the above function for all
39370 value range computations for pass-through jump functions and type
39371 converasion from explicit value range values.
39372 (ipcp_propagate_stage): Do not attempt to deduce types of formal
39373 parameters from TYPE_ARG_TYPES.
39374 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
39375 (ipa_write_node_info): Stream type of the actual argument.
39376 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
39378 2017-01-09 Martin Liska <mliska@suse.cz>
39381 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
39382 (lookup_compiler): Do not show error message with have_E.
39384 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39386 PR tree-optimization/78938
39387 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
39388 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
39389 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
39390 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
39393 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39395 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
39398 2017-01-09 Richard Biener <rguenther@suse.de>
39400 PR tree-optimization/78997
39401 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
39402 name condition properly.
39404 2017-01-09 Richard Biener <rguenther@suse.de>
39407 * dwarf2out.c (is_cxx): New overload with context.
39408 (is_naming_typedef_decl): Use it.
39410 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39412 * invoke.texi (Option Summary): Correct spacing in option lists
39413 and add line breaks to fix over-long lines.
39415 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39417 PR middle-end/17660
39419 * extend.texi (Common Variable Attributes): Add xref to GCC
39420 Internals manual to explain mode attribute keywords.
39422 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39425 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
39426 and Preprocessor Options.
39427 (Options for Linking): Document -pthread here....
39428 (RS/6000 and PowerPC Options): ...not here.
39429 (Solaris 2 Options): ...or here.
39430 * doc/cppopts.texi: Document -pthread.
39432 2017-01-08 Martin Sebor <msebor@redhat.com>
39434 PR middle-end/77708
39435 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
39436 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
39437 New member functions.
39438 (format_directive): Used them.
39440 (pass_sprintf_length::handle_gimple_call): Same.
39441 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
39442 to avoid truncation for any argument.
39443 (extract_affine_mul): Same.
39444 * tree.c (get_file_function_name): Same.
39446 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39448 PR middle-end/77484
39449 * predict.def (PRED_INDIR_CALL): Set to 86.
39451 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39453 PR preprocessor/54124
39454 * doc/cppopts.texi: Reformat -d subtable to list the full name
39455 of the options. Add cross-reference to the docs for the general
39456 compiler -d options.
39457 * doc/invoke.texi (Developer Options): Add cross-reference to the
39458 preprocessor-specific -d option documentation.
39460 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39462 PR preprocessor/13498
39463 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
39464 redudant material, and reflect new command-line options.
39465 (System Headers): Likewise.
39467 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39469 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
39470 -isystem, and -idirafter. Copy-edit.
39471 * doc/cppopts.texi: Copy-edit. Remove contradiction about
39472 default for -ftrack-macro-expansion. Delete obsolete and
39473 badly-formatted implementation details about -fdebug-cpp output.
39474 * doc/cppwarnopts.texi: Copy-edit.
39476 2017-01-07 David Malcolm <dmalcolm@redhat.com>
39479 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39480 that the transition from a max line width >= 1<<10 to narrower
39481 lines works correctly.
39483 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
39485 * doc/options.texi (PerFunction): New.
39486 * opt-functions.awk (switch_flags): Map both Optimization and
39487 PerFunction to CL_OPTIMIZATION.
39488 * opth-gen.awk: Test for PerFunction flag along with
39490 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
39491 it only when the latter is present. Skip those that don't in
39492 the hash function generator.
39493 * common.opt (fvar-tracking): Mark as PerFunction instead of
39495 (fvar-tracking-assignments): Likewise.
39496 (fvar-tracking-assignments-toggle): Likewise.
39497 (fvar-tracking-uninit): Likewise.
39499 2017-01-07 Jakub Jelinek <jakub@redhat.com>
39501 PR translation/79018
39502 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
39505 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
39508 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
39509 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
39510 TARGET_LONG_JUMP_TABLE_OFFSETS.
39511 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
39512 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
39513 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
39514 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
39515 * config/m68k/m68k.md (tablejump expander): Likewise.
39516 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
39517 TARGET_LONG_JUMP_TABLE_OFFSETS.
39518 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
39519 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
39521 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
39522 David Holsgrove <david.holsgrove@xilinx.com>
39524 * common/config/microblaze/microblaze-common.c
39525 (TARGET_EXCEPT_UNWIND_INFO): Remove.
39526 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
39528 * config/microblaze/microblaze.c (microblaze_must_save_register)
39529 (microblaze_expand_epilogue, microblaze_return_addr): Handle
39531 (microblaze_eh_return): New function.
39532 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
39533 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
39534 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
39535 * config/microblaze/microblaze.md (eh_return): New pattern.
39537 2017-01-06 Jakub Jelinek <jakub@redhat.com>
39539 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
39540 GCC_DIAGNOSTIC_STRINGIFY): Define.
39542 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
39544 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39546 * config/arm/arm.md (<mcrr>): New.
39548 * config/arm/arm.c (arm_arch5te): New.
39549 (arm_option_override): Set arm_arch5te.
39550 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
39552 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
39553 (arm_mcrr_qualifiers): ... this. New.
39554 (MRRC_QUALIFIERS): Define to...
39555 (arm_mrrc_qualifiers): ... this. New.
39556 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
39558 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
39559 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
39560 (MRRCI, mrrc, MRRC): New.
39561 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
39562 VUNSPEC_MRRC2): New.
39564 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39566 * config/arm/arm.md (<mcr>): New.
39568 * config/arm/arm.c (arm_coproc_builtin_available): Add
39569 support for mcr, mrc, mcr2 and mrc2.
39570 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
39571 (arm_mcr_qualifiers): ... this. New.
39572 (MRC_QUALIFIERS): Define to ...
39573 (arm_mrc_qualifiers): ... this. New.
39574 (MCR_QUALIFIERS): Define to ...
39575 (arm_mcr_qualifiers): ... this. New.
39576 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
39578 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
39579 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
39580 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
39581 VUNSPEC_MRC2): New.
39583 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39585 * config/arm/arm.md (*ldc): New.
39589 * config/arm/arm.c (arm_coproc_builtin_available): Add
39590 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
39591 (arm_coproc_ldc_stc_legitimate_address): New.
39592 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
39593 'qualifier_const_pointer'.
39594 (LDC_QUALIFIERS): Define to...
39595 (arm_ldc_qualifiers): ... this. New.
39596 (STC_QUALIFIERS): Define to...
39597 (arm_stc_qualifiers): ... this. New.
39598 * config/arm/arm-protos.h
39599 (arm_coproc_ldc_stc_legitimate_address): New.
39600 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
39601 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
39602 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
39603 stc2, stcl, stc2l): New.
39604 * config/arm/constraints.md (Uz): New.
39605 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
39606 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
39607 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
39608 VUNSPEC_STC2L): New.
39610 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39612 * config/arm/arm.md (<cdp>): New.
39613 * config/arm/arm.c (neon_const_bounds): Rename this ...
39614 (arm_const_bounds): ... this.
39615 (arm_coproc_builtin_available): New.
39616 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
39617 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
39618 (CDP_QUALIFIERS): Define to...
39619 (arm_cdp_qualifiers): ... this. New.
39621 (arm_expand_builtin_args): Add case for 6 arguments.
39622 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
39623 (arm_const_bounds): ... this.
39624 (arm_coproc_builtin_available): New.
39625 * config/arm/arm_acle.h (__arm_cdp): New.
39627 * config/arm/arm_acle_builtins.def (cdp): New.
39629 * config/arm/iterators.md (CDPI,CDP,cdp): New.
39630 * config/arm/neon.md: Rename all 'neon_const_bounds' to
39631 'arm_const_bounds'.
39632 * config/arm/types.md (coproc): New.
39633 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
39634 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
39635 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
39636 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
39638 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39640 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
39641 (UBINOP_QUALIFIERS): New.
39643 (acle_builtin_data): New. Change comment.
39644 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
39645 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
39646 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
39647 arm_acle_builtins.def.
39648 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
39649 (arm_init_acle_builtins): New.
39650 (CRC32_BUILTIN): Remove.
39651 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
39652 crc32cb, crc32ch and crc32cw.
39653 (arm_init_crc32_builtins): Remove.
39654 (arm_init_builtins): Use arm_init_acle_builtins rather
39655 than arm_init_crc32_builtins.
39656 (arm_expand_acle_builtin): New.
39657 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
39658 * config/arm/arm_acle_builtins.def: New.
39660 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39662 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
39663 (arm_builtin_datum): ... this.
39664 (arm_init_neon_builtin): Rename to ...
39665 (arm_init_builtin): ... this. Add a new parameters PREFIX
39666 and USE_SIG_IN_NAME.
39667 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
39668 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
39669 'arm_builtin_datum'.
39670 (arm_init_vfp_builtins): Likewise.
39671 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
39672 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
39673 (arm_expand_neon_args): Rename to ...
39674 (arm_expand_builtin_args): ... this. Rename builtin_arg
39675 enum values and differentiate between ARG_BUILTIN_MEMORY
39676 and ARG_BUILTIN_NEON_MEMORY.
39677 (arm_expand_neon_builtin_1): Rename to ...
39678 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
39679 values, arm_expand_builtin_args and add bool parameter NEON.
39680 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
39681 (arm_expand_vfp_builtin): Likewise.
39682 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
39684 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39686 PR middle-end/77484
39687 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
39688 * predict.c (tree_estimate_probability_bb): Reverse direction of
39689 polymorphic call predictor.
39691 2017-01-06 David Malcolm <dmalcolm@redhat.com>
39693 * passes.c (execute_one_pass): Split out pass-skipping logic into...
39694 (determine_pass_name_match): ...this new function and...
39695 (should_skip_pass_p): ...this new function.
39697 2017-01-06 Nathan Sidwell <nathan@acm.org>
39699 * ipa-visibility.c (function_and_variable_visibility): Reformat
39700 comments and long lines. Remove extrneous if.
39701 * symtab.c (symtab_node::make_decl_local): Fix code format.
39702 (symtab_node::set_section_for_node): Fix comment typo.
39704 2017-01-06 Martin Liska <mliska@suse.cz>
39707 * lra-constraints.c: Rename invariant to lra_invariant.
39708 * predict.c (set_even_probabilities): Initialize e to NULL.
39710 2017-01-05 Martin Sebor <msebor@redhat.com>
39712 PR tree-optimization/78910
39713 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
39714 (format_integer): Correct off-by-one error in the handling
39715 of precision with negative numbers in signed conversions..
39717 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
39719 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
39721 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39723 PR tree-optimization/71016
39724 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
39725 factor_out_conditional_conversion. Formatting fix.
39726 (factor_out_conditional_conversion): Add cond_stmt argument.
39727 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
39728 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
39731 2017-01-05 David Malcolm <dmalcolm@redhat.com>
39733 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
39734 read-rtl-function.o, and selftest-rtl.o.
39735 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
39736 (selftest::aarch64_test_loading_full_dump): New function.
39737 (selftest::aarch64_run_selftests): New function.
39738 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
39739 selftest::aarch64_run_selftests.
39740 * config/i386/i386.c
39741 (selftest::ix86_test_loading_dump_fragment_1): New function.
39742 (selftest::ix86_test_loading_call_insn): New function.
39743 (selftest::ix86_test_loading_full_dump): New function.
39744 (selftest::ix86_test_loading_unspec): New function.
39745 (selftest::ix86_run_selftests): Call the new functions.
39746 * emit-rtl.c (maybe_set_max_label_num): New function.
39747 * emit-rtl.h (maybe_set_max_label_num): New decl.
39748 * function.c (instantiate_decls): Guard call to
39749 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
39750 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
39752 * gensupport.c (gen_reader::gen_reader): Pass "false"
39753 for new "compact" param of rtx_reader.
39754 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
39755 rather than an empty string for NULL strings.
39756 * read-md.c: Potentially include config.h rather than bconfig.h.
39757 Wrap include of errors.h with #ifdef GENERATOR_FILE.
39758 (have_error): New global, copied from errors.c.
39759 (md_reader::read_name): Rename to...
39760 (md_reader::read_name_1): ...this, adding "out_loc" param,
39761 and converting "missing name or number" to returning false, rather
39763 (md_reader::read_name): Reimplement in terms of read_name_1.
39764 (md_reader::read_name_or_nil): New function.
39765 (md_reader::read_string): Handle "(nil)" by returning NULL.
39766 (md_reader::md_reader): Add new param "compact".
39767 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
39768 (md_reader::read_file): New method.
39769 * read-md.h (md_reader::md_reader): Add new param "compact".
39770 (md_reader::read_file): New method.
39771 (md_reader::is_compact): New accessor.
39772 (md_reader::read_name): Convert return type from void to file_location.
39773 (md_reader::read_name_or_nil): New decl.
39774 (md_reader::read_name_1): New decl.
39775 (md_reader::m_compact): New field.
39776 (noop_reader::noop_reader): Pass "false" for new "compact" param
39778 (rtx_reader::rtx_reader): Add new "compact" param.
39779 (rtx_reader::read_rtx_operand): Make virtual and convert return
39780 type from void to rtx.
39781 (rtx_reader::read_until): New decl.
39782 (rtx_reader::handle_any_trailing_information): New virtual function.
39783 (rtx_reader::postprocess): New virtual function.
39784 (rtx_reader::finalize_string): New virtual function.
39785 (rtx_reader::m_in_call_function_usage): New field.
39786 (rtx_reader::m_reuse_rtx_by_id): New field.
39787 * read-rtl-function.c: New file.
39788 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
39789 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
39790 (selftest::verify_three_block_rtl_cfg): New decl.
39791 * read-rtl-function.h: New file.
39792 * read-rtl.c: Potentially include config.h rather than bconfig.h.
39793 For host, include function.h, memmodel.h, and emit-rtl.h.
39794 (one_time_initialization): New function.
39795 (struct compact_insn_name): New struct.
39796 (compact_insn_names): New array.
39797 (find_code): Handle insn codes in compact dumps.
39798 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
39799 (bind_subst_iter_and_attr): Likewise.
39800 (add_condition_to_string): Likewise.
39801 (add_condition_to_rtx): Likewise.
39802 (apply_attribute_uses): Likewise.
39803 (add_current_iterators): Likewise.
39804 (apply_iterators): Likewise.
39805 (initialize_iterators): Guard usage of apply_subst_iterator with
39806 #ifdef GENERATOR_FILE.
39807 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
39808 (md_reader::read_mapping): Likewise.
39809 (add_define_attr_for_define_subst): Likewise.
39810 (add_define_subst_attr): Likewise.
39811 (read_subst_mapping): Likewise.
39812 (check_code_iterator): Likewise.
39813 (rtx_reader::read_rtx): Likewise. Move one-time initialization
39815 (one_time_initialization): New function.
39816 (rtx_reader::read_until): New method.
39817 (read_flags): New function.
39818 (parse_reg_note_name): New function.
39819 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
39820 Handle reuse_rtx ids.
39821 Wrap iterator lookup within #ifdef GENERATOR_FILE.
39822 Add parsing support for RTL dumps, mirroring the special-cases in
39823 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
39824 values, and calling handle_any_trailing_information.
39825 (rtx_reader::read_rtx_operand): Convert return type from void
39826 to rtx, returning return_rtx. Handle case 'e'. Call
39827 finalize_string on XSTR and XTMPL fields.
39828 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
39829 "(nil)" values were omitted. Call the postprocess vfunc on the
39831 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
39832 class ctor. Initialize m_in_call_function_usage. Call
39833 one_time_initialization.
39834 * rtl-tests.c (selftest::test_uncond_jump): Call
39835 set_new_first_and_last_insn.
39836 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
39837 * selftest-rtl.c: New file.
39838 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
39839 (selftest::get_insn_by_uid): New decl.
39840 * selftest-run-tests.c (selftest::run_tests): Call
39841 read_rtl_function_c_tests.
39842 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
39843 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
39846 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
39848 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
39849 operands in a special way. Assert that pos+len <= mode precision.
39851 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39853 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
39854 3 argument Alias with unlimited for the negative form.
39855 (fno-vect-cost-model): Removed.
39857 2017-01-05 Martin Liska <mliska@suse.cz>
39859 * hsa-gen.c (gen_hsa_divmod): New function.
39860 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
39862 2017-01-05 Martin Liska <mliska@suse.cz>
39865 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
39868 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39870 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
39871 small constant length operands.
39873 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39875 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
39876 between loop iterations.
39878 2017-01-05 Martin Liska <mliska@suse.cz>
39881 * gimplify.c (gimplify_decl_expr): Compare to
39882 asan_poisoned_variables instread of checking flags.
39883 (gimplify_target_expr): Likewise.
39884 (gimplify_expr): Likewise.
39885 (gimplify_function_tree): Conditionally initialize
39886 asan_poisoned_variables.
39888 2017-01-04 Jeff Law <law@redhat.com>
39890 PR tree-optimizatin/78812
39891 * rtl.h (contains_mem_rtx_p): Prototype.
39892 * ifcvt.c (containts_mem_rtx_p): Move from here to...
39893 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
39894 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
39895 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
39896 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
39898 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39900 * input.c (assert_char_at_range): Default-initialize actual_range.
39902 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39904 * df-scan.c (df_ref_create_structure): Make regno unsigned,
39905 to match the caller.
39907 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39909 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
39910 insns after final jump in test to emit dummy move.
39912 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39914 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
39915 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
39917 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39919 * multiple_target.c (create_dispatcher_calls): Init e_next.
39920 * tree-ssa-loop-split.c (split_loop): Init border.
39921 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
39924 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
39929 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
39930 (altivec_register_operand): Do not return true if the operand
39931 contains a SUBREG mixing SImode and SFmode.
39932 (vsx_register_operand): Likewise.
39933 (vsx_reg_sfsubreg_ok): New predicate.
39934 (vfloat_operand): Do not return true if the operand contains a
39935 SUBREG mixing SImode and SFmode.
39936 (vint_operand): Likewise.
39937 (vlogical_operand): Likewise.
39938 (gpc_reg_operand): Likewise.
39939 (int_reg_operand): Likewise.
39940 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
39941 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
39942 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
39944 (rs6000_emit_move_si_sf_subreg): New helper function.
39945 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
39946 fixup SUBREGs involving SImode and SFmode.
39947 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
39948 numbers for the new peephole2 optimization.
39949 (peephole2 for SFmode unions): New peephole2 to optimize cases in
39950 the GLIBC math library that do AND/IOR/XOR operations on single
39951 precision floating point.
39952 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
39953 target macros to say whether we need to avoid SUBREGs mixing
39955 (TARGET_ALLOW_SF_SUBREG): Likewise.
39956 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
39957 (UNSPEC_SI_FROM_SF): Likewise.
39958 (iorxor): Change spacing.
39959 (and_ior_xor): New iterator for AND, IOR, and XOR.
39960 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
39961 (movdi_from_sf_zero_ext): Likewise.
39962 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
39963 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
39964 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
39965 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
39966 (fms<mode>4): Likewise.
39967 (fnma<mode>4): Likewise.
39968 (fnms<mode>4): Likewise.
39969 (nfma<mode>4): Likewise.
39970 (nfms<mode>4): Likewise.
39972 2017-01-04 Marek Polacek <polacek@redhat.com>
39975 * doc/invoke.texi: Document -Wpointer-compare.
39977 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39979 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
39982 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
39983 descriptions for -gdwarf-5 and emit them as uleb128 instead of
39986 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
39989 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
39990 documentation of the powerpc_popcntb_ok attribute.
39991 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
39992 code to issue warning messages if a requested CPU configuration is
39993 not supported by the binary (assembler and loader) toolchain.
39994 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
39995 made to define a built-in function that has been disabled.
39996 (paired_init_builtins): Add assertion to prevent ICE if attempt is
39997 made to define a built-in function that has been disabled.
39998 (altivec_init_builtins): Add comment explaining why definition
39999 of the DST built-in functions is not preceded by an assertion
40000 check. Add assertions to prevent ICE if attempts are made to
40001 define an altivec predicate or an abs* built-in function that has
40003 (htm_init_builtins): Add comment explaining why definition of the
40004 htm built-in functions is not preceded by an assertion check.
40006 2017-01-04 Jeff Law <law@redhat.com>
40008 PR tree-optimizatin/67955
40009 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40010 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40011 the points-to solution does not include pt_null. Use DECL_PT_UID
40014 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40016 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40017 Use gen_int_mode instead of gen_lopwart for const_int operands.
40019 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40021 PR tree-optimization/71563
40022 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40023 out of range value - has low bits known to be zero.
40025 2017-01-04 Alan Modra <amodra@gmail.com>
40027 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40028 * configure: Regenerate.
40029 * config.in: Regenerate.
40031 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40034 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40035 a substring of the message, but strcmp with the whole message. Ifdef
40036 ENABLE_NLS, translate the message first using dgettext.
40038 2017-01-03 Jeff Law <law@redhat.com>
40040 PR tree-optimizatin/78856
40041 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
40042 (mark_threaded_blocks): Remove code to truncate thread paths that
40043 cross multiple loop headers. Instead invalidate the cached loop
40044 iteration information and handle case of a thread path walking
40045 into an irreducible region.
40047 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
40050 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
40051 assertions. Add support for doing the signbit if the IEEE 128-bit
40052 floating point value is in a GPR.
40053 * config/rs6000/rs6000.md (Fsignbit): Delete.
40054 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
40055 Update the length attribute if the value is in a GPR.
40056 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
40057 the sign or zero extension instruction, since the value is always 0/1.
40058 (signbit<mode>2_dm2): Delete using <Fsignbit>.
40061 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
40062 extracting SImode to a GPR register so that we can generate a
40063 store, limit the vector to be in a traditional Altivec register
40064 for the vextuwrx instruction.
40066 2017-01-03 Ian Lance Taylor <iant@google.com>
40068 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
40070 2017-01-03 Martin Sebor <msebor@redhat.com>
40072 PR tree-optimization/78696
40073 * gimple-ssa-sprintf.c (format_floating): Correct handling of
40074 precision. Use MPFR for %f for greater fidelity. Correct handling
40076 (pass_sprintf_length::compute_format_length): Set width and precision
40077 specified by asrerisk to void_node for vararg functions.
40078 (try_substitute_return_value): Adjust dump output.
40080 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
40082 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
40084 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
40086 * doc/invoke.texi (SPARC options): Document -mlra as the default.
40087 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
40088 -mlra/-mno-lra was passed to the compiler.
40090 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
40092 PR rtl-optimization/65618
40093 * emit-rtl.c (try_split): Move initialization of "before" and
40094 "after" to just before the call to emit_insn_after_setloc.
40096 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
40098 * doc/md.texi (Standard Names): Remove reference to Java frontend.
40100 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
40102 * dwarf2out.c (gen_enumeration_type_die): When
40103 -gno-strict-dwarf, add a DW_AT_encoding attribute.
40105 2017-01-03 Jakub Jelinek <jakub@redhat.com>
40107 PR tree-optimization/78965
40108 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
40109 Change first argument from const call_info & to call_info &. For %n
40110 set info.nowrite to false.
40112 PR middle-end/78901
40113 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
40114 possibly throwing calls.
40116 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
40117 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
40118 and fns handling, rather than in a separate case SSA_NAME.
40120 2017-01-02 Jeff Law <law@redhat.com>
40122 * config/darwin-driver.c (darwin_driver_init): Const-correctness
40123 fixes for first_period and second_period variables.
40125 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
40128 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
40129 (*insvqi_1): New insn pattern.
40130 (*insvqi_1_mem_rex64): Ditto.
40131 (*insvqi_2): Ditto.
40132 (*insvqi_3): Rename from *insvqi.
40134 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
40136 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
40138 * doc/cfg.texi (Edges): Remove reference to Java.
40139 (Maintaining the CFG): Ditto.
40141 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40143 PR middle-end/77674
40144 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
40145 transparent aliases.
40147 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40149 PR middle-end/77484
40150 * predict.def (PRED_CALL): Update hitrate.
40151 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
40152 * predict.c (tree_estimate_probability_bb): Split CALL predictor
40153 into direct/indirect/polymorphic variants.
40155 2017-01-01 Jakub Jelinek <jakub@redhat.com>
40157 Update copyright years.
40159 * gcc.c (process_command): Update copyright notice dates.
40160 * gcov-dump.c (print_version): Ditto.
40161 * gcov.c (print_version): Ditto.
40162 * gcov-tool.c (print_version): Ditto.
40163 * gengtype.c (create_file): Ditto.
40164 * doc/cpp.texi: Bump @copying's copyright year.
40165 * doc/cppinternals.texi: Ditto.
40166 * doc/gcc.texi: Ditto.
40167 * doc/gccint.texi: Ditto.
40168 * doc/gcov.texi: Ditto.
40169 * doc/install.texi: Ditto.
40170 * doc/invoke.texi: Ditto.
40172 Copyright (C) 2017 Free Software Foundation, Inc.
40174 Copying and distribution of this file, with or without modification,
40175 are permitted in any medium without royalty provided the copyright
40176 notice and this notice are preserved.