Improve stack adjustment by reusing a temporary move immediate from the epilog
if the register is still valid in the epilog. This generates smaller code for
leaf functions with a stack size of more then 4KB.
gcc/
* config/aarch64/aarch64.c (aarch64_add_constant_internal):
Add extra argument to allow emitting the move immediate.
Use add/sub with positive immediate.
(aarch64_add_constant): Add inline function.
(aarch64_add_sp): Likewise.
(aarch64_sub_sp): Likewise.
(aarch64_expand_prologue): Call aarch64_sub_sp.
(aarch64_expand_epilogue): Call aarch64_add_sp.
Decide when to leave out move.
(aarch64_output_mi_thunk): Call aarch64_add_constant.
testsuite/
* gcc.target/aarch64/test_frame_17.c: New test.
From-SVN: r241420
If the number of integer callee-saves is odd, any FP callee-saves use 8-byte
aligned LDP/STP. Since 16-byte alignment may be faster on some CPUs, align
the FP callee-saves to 16 bytes and use the alignment gap for the last FP
callee-save when possible.
gcc/
* config/aarch64/aarch64.c (aarch64_layout_frame):
Align FP callee-saves.
From-SVN: r241419
* include/experimental/bits/fs_ops.h
(exists(const path&, error_code&)): Clear error if status is known
(LWG 2725).
(status(const path&, error_code&)): Handle EOVERFLOW.
* testsuite/experimental/filesystem/operations/exists.cc: Test
overload taking an error_code.
From-SVN: r241417
2016-10-21 Eelis van der Weegen <eelis@eelis.net>
* include/bits/stl_algo.h (__gen_two_uniform_ints): Move logic out
of shuffle into new function.
(shuffle): Call __gen_two_uniform_ints.
(__sample<ForwardIterator, OutputIterator, Cat, Size, URBG>): Use
__gen_two_uniform_ints and perform two samples at a time.
From-SVN: r241414
* dwarf2out.c (ranges_table): Change into vec<dw_ranges, va_gc> *.
(ranges_by_label): Change into vec<dw_ranges_by_label, va_gc> *.
(ranges_table_allocated, ranges_table_in_use,
ranges_by_label_allocated, ranges_by_label_in_use,
RANGES_TABLE_INCREMENT): Removed.
(add_ranges_num): Use vec_safe_push into ranges_table.
(add_ranges_by_labels): Use vec_safe_push into ranges_by_label.
(output_ranges): Adjust for ranges_table and ranges_by_label
conversion from arrays to vec.
(add_high_low_attributes, dwarf2out_finish): Adjust for range_table
conversion from arrays to vec.
(dwarf2out_c_finalize): Don't clear ranges_table_allocated,
ranges_table_in_use, ranges_by_label_allocated and
ranges_by_label_in_use. Set ranges_by_label to NULL instead of 0.
From-SVN: r241409
gcc/ChangeLog:
* print-rtl-function.c (flag_compact): Move extern decl to...
* print-rtl.h (flag_compact): ...here.
* rtl-tests.c (selftests::assert_rtl_dump_eq): New function.
(ASSERT_RTL_DUMP_EQ): New macro.
(selftest::test_dumping_regs): New function.
(selftest::test_dumping_insns): New function.
(selftest::test_uncond_jump): Add uses of ASSERT_RTL_DUMP_EQ on
the insns.
(selftest::rtl_tests_c_tests): Call the new test functions.
From-SVN: r241405
2016-10-21 Paul Thomas <pault@gcc.gnu.org>
PR fortran/69566
* resolve.c (fixup_array_ref): New function.
(resolve_select_type): Gather up the rank and array reference,
if any, from the selector. Fix up the 'associate name' and the
'associate entities' as necessary.
* trans-expr.c (gfc_conv_class_to_class): If the symbol backend
decl is a FUNCTION_DECL, use the 'fake_result_decl' instead.
2016-10-21 Paul Thomas <pault@gcc.gnu.org>
PR fortran/69566
* gfortran.dg/select_type_37.f03: New test.
From-SVN: r241403
While changing LABEL_REF_LABEL it might as well become an inline
function, so that its clearer what types are involved. Unfortunately
because it is still possible to use XEXP and related macros on a
LABEL_REF rtx you can still set the field to be a non insn rtx. The
other unfortunate thing is that the generators actually create LABEL_REF
rtx that refer to MATCH_x rtx, so there we actually need to use XEXP to
bypass the checking this patch adds.
gcc/ChangeLog:
2016-10-21 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
* rtl.h (label_ref_label): New function.
(set_label_ref_label): New function.
(LABEL_REF_LABEL): Delete.
* alias.c (rtx_equal_for_memref_p): Adjust.
* cfgbuild.c (make_edges): Likewise.
(purge_dead_tablejump_edges): Likewise.
* cfgexpand.c (convert_debug_memory_address): Likewise.
* cfgrtl.c (patch_jump_insn): Likewise.
* combine.c (distribute_notes): Likewise.
* cse.c (hash_rtx_cb): Likewise.
(exp_equiv_p): Likewise.
(fold_rtx): Likewise.
(check_for_label_ref): Likewise.
* cselib.c (rtx_equal_for_cselib_1): Likewise.
(cselib_hash_rtx): Likewise.
* emit-rtl.c (mark_label_nuses): Likewise.
* explow.c (convert_memory_address_addr_space_1): Likewise.
* final.c (output_asm_label): Likewise.
(output_addr_const): Likewise.
* gcse.c (add_label_notes): Likewise.
* genconfig.c (walk_insn_part): Likewise.
* genrecog.c (validate_pattern): Likewise.
* ifcvt.c (cond_exec_get_condition): Likewise.
(noce_emit_store_flag): Likewise.
(noce_get_alt_condition): Likewise.
(noce_get_condition): Likewise.
* jump.c (maybe_propagate_label_ref): Likewise.
(mark_jump_label_1): Likewise.
(redirect_exp_1): Likewise.
(rtx_renumbered_equal_p): Likewise.
* lra-constraints.c (operands_match_p): Likewise.
* print-rtl.c (print_value): Likewise.
* reload.c (find_reloads): Likewise.
* reload1.c (set_label_offsets): Likewise.
* reorg.c (get_branch_condition): Likewise.
* rtl-tests.c (test_uncond_jump): Likewise.
* rtl.c (rtx_equal_p_cb): Likewise.
(rtx_equal_p): Likewise.
* rtlanal.c (reg_mentioned_p): Likewise.
(rtx_referenced_p): Likewise.
(get_condition): Likewise.
* varasm.c (const_hash_1): Likewise.
(compare_constant): Likewise.
(const_rtx_hash_1): Likewise.
(output_constant_pool_1): Likewise.
From-SVN: r241401
Tweak find_valid_class_1 to consider a reg class if atleast one regno in
that class is ok.
Previously, even if no regno was in_hard_reg_set_p, the code goes ahead and
considers rclass as valid. bad was set only if a regno was in the reg
class *and* HARD_REGNO_MODE_OK was false - if both were false, bad wasn't
set and the reload got a wrong rclass. If that happened to be the best
one, this eventually lead to find_reg running out of registers to
spill, because the chosen rclass wouldn't have enough regs.
Also, it expected every regno in rclass to be valid for mode
i.e., if any regno fails HARD_REGNO_MODE_OK, it rejected the rclass. The
comments in the original commit for find_valid_class_1 say atleast one
regno is ok. This was updated to say "class which contains only
registers" when in_hard_reg_set_p was introduced in place of just
TEST_HARD_REG_BIT.
This commit fixes both of the above problems by not breaking out of the loop
on first unavailable regno. Instead, it computes the rclass size consisting
of all regnos in that class valid for the current mode.
If that computed size is zero, the rclass would be skipped, as it won't
beat best_size. Otherwise, the computed size is used to choose the best
rclass, instead of the static size from reg_class_size.
gcc/
2016-10-21 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
PR target/71627
* reload.c (find_valid_class_1): Allow regclass if atleast one
regno in regclass is ok. Compute and use rclass size based on
actually available regnos for mode in rclass.
gcc/testsuite/
2016-10-21 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
PR target/71627
* gcc.target/avr/pr71627.c: New test
From-SVN: r241400
PR rtl-optimization/78038
* ree.c (get_defs): Return NULL if a defining insn for REG cannot
be deduced to set REG through the RTL structure.
(make_defs_and_copies_lists): Return false on a failing get_defs call.
* gcc.target/aarch64/pr78038.c: New test.
From-SVN: r241395
2016-10-21 Richard Biener <rguenther@suse.de>
PR tree-optimization/78051
* tree-vrp.c (evrp_dom_walker::before_dom_children): Update stmt
and mark replaced if folding did something.
* g++.dg/torture/pr78051.C: New testcase.
From-SVN: r241394
2016-10-20 Jerry DeLisle <jvdelisle@gcc.gnu.org>
* io/transfer.c (finalize_transfer): Free format data in child
procedures. (st_read_done, st_write_done): Don't free format
hash table.
From-SVN: r241392
* config/i386/i386.c (ix86_fold_builtin): Handle IX86_BUILTIN_INFQ
and IX86_BUILTIN_HUGE_VALQ here ...
(ix86_expand_builtin): ... not here.
From-SVN: r241387
I started to copy the Go 1.7 interface code, but the gc and gccgo
representations of interfaces are too different. So instead I rewrote
the gccgo interface code from C to Go. The code is largely the same as
it was, but the names are more like those used in the gc runtime.
I also copied over the string comparison functions, and tweaked the
compiler to use eqstring when comparing strings for equality.
Reviewed-on: https://go-review.googlesource.com/31591
From-SVN: r241384
PR lto/78049
* lto-streamer-in.c (fixup_call_stmt_edges_1): Replace value
comparison with STMT_UID_NOT_IN_RANGE.
(fixup_call_stmt_edges): Do not fixup edges of a thunk in
LTRANS.
From-SVN: r241380
* compare-elim.c (conforming_compare): Accept UNSPECs.
(find_comparison_dom_walker::before_dom_children): Deal with
instructions both using and killing the flags register.
(equivalent_reg_at_start): New function extracted from...
(try_eliminate_compare): ...here. Use it and add support for
registers and UNSPECs as second operand of the compare.
* config/visium/visium-modes.def (CCV): New.
* config/visium/predicates.md (visium_v_comparison_operator): New.
(visium_branch_operator): Deal with CCV mode.
* config/visium/visium.c (visium_select_cc_mode): Likewise.
(output_cbranch): Likewise.
* config/visium/visium.md (UNSPEC_{ADD,SUB,NEG}V): New constants.
(uaddv<mode>4): New expander.
(addv<mode>4): Likewise.
(add<mode>3_insn_set_carry): New instruction.
(add<mode>3_insn_set_overflow): Likewise.
(addsi3_insn_set_overflow): Likewise.
(usubv<mode>4): New expander.
(subv<mode>4): Likewise.
(sub<mode>3_insn_set_carry): New instruction.
(sub<mode>3_insn_set_overflow): Likewise.
(subsi3_insn_set_overflow): Likewise.
(unegv<mode>3): New expander.
(negv<mode>3): Likewise.
(neg<mode>2_insn_set_overflow): New instruction.
(addv_tst<mode>): Likewise.
(subv_tst<mode>): Likewise.
(negv_tst<mode>): Likewise.
(cbranch<mode>4_addv_insn): New splitter and instruction.
(cbranch<mode>4_subv_insn): Likewise.
(cbranch<mode>4_negv_insn): Likewise.
From-SVN: r241379
2016-10-20 Richard Biener <rguenther@suse.de>
* tree-ssa-alias.c (ptrs_compare_unequal): Remove code duplication.
Handle decls possibly not bound.
* tree-ssa-structalias.c (get_constraint_for_ssa_var): Add
nothing_id for decls that might not be bound if we are interested
for the address.
(get_constraint_for_component_ref): Deal with that.
From-SVN: r241378
* include/backward/auto_ptr.h (__shared_ptr(auto_ptr&&))
(shared_ptr(auto_ptr&&)): Adjust template parameter lists.
* include/bits/shared_ptr.h (__sp_compatible_with)
(__sp_is_constructible): New helper traits for shared_ptr.
(shared_ptr::_Convertible): Replace with _Constructible.
(shared_ptr::_Constructible, shared_ptr::_Assignable): Forward checks
to base class.
(shared_ptr::shared_ptr, shared_ptr::operator=): Constrain template
with _Constructible and _Assignable.
(shared_ptr::shared_ptr(shared_ptr<_Tp1>, _Tp*)): Use element_type
instead of _Tp.
(operator<): Likewise.
(operator>): Define in terms of operator<.
(static_pointer_cast, const_pointer_cast, dynamic_pointer_cast): Use
element_type instead of _Tp.
(reinterpret_pointer_cast): Define for C++17.
(weak_ptr::_Convertible): Replace with _Constructible.
(weak_ptr::_Constructible, weak_ptr::_Assignable): Forward checks
to base class.
(weak_ptr::weak_ptr, weak_ptr::operator=): Constrain templates
with _Constructible and _Assignable.
* include/bits/shared_ptr_base.h (__shared_ptr::_Convertible): Replace
with _Compatible.
(__shared_ptr::_SafeConv): New constraint for incoming raw pointers.
(__shared_ptr::_Compatible): New constraint for converting from
other types of shared_ptr and weak_ptr.
(__shared_ptr::_Assignable): Define in terms of _Compatible.
(__shared_ptr::_UniqCompatible, __shared_ptr::_UniqAssignable): New
constraints for converting from unique_ptr.
(__shared_ptr::__shared_ptr, __shared_ptr::operator=): Constrain
template with _SaveConf, _Compatible and _Assignable. Remove
__glibcxx_function_requires concept checks. Add static assertion for
deleter expression being well-formed.
(__shared_ptr::__shared_ptr(__shared_ptr<_Tp1>, _Tp*))
(__shared_ptr::operator*, __shared_ptr::operator->)
(__shared_ptr::get, __shared_ptr::_M_ptr): Use element_type instead
of _Tp.
(operator<): Likewise.
(operator>): Define in terms of operator<.
(static_pointer_cast, const_pointer_cast, dynamic_pointer_cast): Use
element_type instead of _Tp.
(reinterpret_pointer_cast): Define for C++17.
(weak_ptr::_Convertible): Replace with _Compatible.
(weak_ptr::_Compatible, weak_ptr::_Assignable): New constraints for
conversions from other types of weak_ptr and shared_ptr.
(__weak_ptr::__weak_ptr, __weak_ptr::operator=): Constrain templates
with _Constructible and _Assignable.
(__weak_ptr::_M_ptr): Use element_type instead of _Tp.
* testsuite/20_util/shared_ptr/assign/auto_ptr_neg.cc: Adjust
dg-error pattern.
* testsuite/20_util/shared_ptr/cons/auto_ptr.cc: Test conversions.
* testsuite/20_util/shared_ptr/cons/unique_ptr.cc: Likewise.
* testsuite/20_util/shared_ptr/cons/void_neg.cc: Likewise.
* testsuite/20_util/shared_ptr/casts/reinterpret.cc: New test.
From-SVN: r241373
not in the template parameters.
* include/std/optional (operator=(_Up&&)): Move SFINAE
from template parameters to the return type.
(operator=(const optional<_Up>&)): Likewise.
(operator=(optional<_Up>&&)): Likewise.
From-SVN: r241372
gcc/fortran/ChangeLog:
2016-10-20 Andre Vehreschild <vehre@gcc.gnu.org>
* class.c (gfc_build_class_symbol): Set the kind of _len to
gfc_charlen_int_kind to catch changes of the charlen kind.
From-SVN: r241367
