2016-11-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
gcc/
* tree-ssa-math-opts.c (find_bswap_or_nop): Zero out bytes in cmpxchg
and cmpnop in two steps: first the ones not accessed in original gimple
expression in a endian independent way and then the ones not accessed
in the final result in an endian-specific way.
(bswap_replace): Stop doing big endian adjustment.
From-SVN: r242394
2016-11-14 Thomas Koenig <tkoenig@gcc.gnu.org>
* dump-parse-tree.c (show_code): Add prototype.
(gfc_debug_code): New function.
(show_code_node): Add space after SELECT TYPE.
From-SVN: r242393
PR bootstrap/78069
* common.opt: Add prefer-atomic as a new enum value for
-fprofile-update.
* coretypes.h: Likewise.
* doc/invoke.texi: Document the new option value.
* gcc.c: Replace atomic with prefer-atomic. Remove warning.
* tree-profile.c (tree_profiling): Select default value
of -fprofile-update when 'prefer-atomic' is selected.
PR bootstrap/78069
* gcc.dg/no_profile_instrument_function-attr-1.c: Update test
to match scanned pattern.
* gcc.dg/tree-ssa/ssa-lim-11.c: Likewise.
From-SVN: r242386
The second patch updates the Cortex-A57 scheduler now that we can differentiate
between shifts and bitfield inserts. The Cortex-A57 Software Optimization Guide
indicates that BFM operations use the integer multi-cycle pipeline, while ARM
UXTB/H instructions use the Integer 1 or Integer 0 pipelines, so swap the bfm
and extend reservations. This results in minor scheduling differences.
* config/arm/cortex-a57.md (cortex_a57_alu): Move extend here, bfm...
(cortex_a57_alu_shift): ...here.
From-SVN: r242385
SBFM and UBFM include all shifts on AArch64, which are simpler than bitfield
insert. Add a new bfx attribute for these instructions so that they can be
modelled more accurately in the future. There is no difference in code
generation.
* config/aarch64/aarch64.md (aarch64_ashl_sisd_or_int_<mode>3)
Use bfx attribute.
(aarch64_lshr_sisd_or_int_<mode>3): Likewise.
(aarch64_ashr_sisd_or_int_<mode>3): Likewise.
(<optab>si3_insn_uxtw): Likewise.
(<optab><mode>3_insn): Likewise.
(<ANY_EXTEND:optab><GPI:mode>_ashl<SHORT:mode>): Likewise.
(zero_extend<GPI:mode>_lshr<SHORT:mode>): Likewise.
(extend<GPI:mode>_ashr<SHORT:mode>): Likewise.
(<optab><mode>): Likewise.
(insv<mode>): Likewise.
(andim_ashift<mode>_bfiz): Likewise.
* config/aarch64/thunderx.md (thunderx_shift): Add bfx.
* config/arm/cortex-a53.md (cortex_a53_alu_shift): Likewise.
* config/arm/cortex-a57.md (cortex_a57_alu): Add bfx.
* config/arm/exynos-m1.md (exynos_m1_alu): Add bfx.
(exynos_m1_alu_p): Likewise.
* config/arm/types.md: Add bfx.
* config/arm/xgene1.md (xgene1_bfm): Add bfx.
From-SVN: r242384
The existing vector costs stop some beneficial vectorization. This is mostly
due to vector statement cost being set to 3 as well as vector loads having a
higher cost than scalar loads. This means that even when we vectorize 4x, it
is possible that the cost of a vectorized loop is similar to the scalar
version, and we fail to vectorize.
Using a cost of 3 for a vector operation suggests they are 3 times as
expensive as scalar operations. Since most vector operations have a
similar throughput as scalar operations, this is not correct.
Using slightly lower values for these heuristics now allows this loop
and many others to be vectorized. On a proprietary benchmark the gain
from vectorizing this loop is around 15-30% which shows vectorizing it is
indeed beneficial.
* config/aarch64/aarch64.c (cortexa57_vector_cost):
Change vec_stmt_cost, vec_align_load_cost and vec_unalign_load_cost.
From-SVN: r242383
Implement P0217R3 - C++17 structured bindings
* g++.dg/cpp1z/decomp1.C: New test.
* g++.dg/cpp1z/decomp2.C: New test.
* g++.dg/cpp1z/decomp3.C: New test.
* g++.dg/cpp1z/decomp4.C: New test.
* g++.dg/cpp1z/decomp5.C: New test.
* g++.dg/cpp1z/decomp6.C: New test.
* g++.dg/cpp1z/decomp7.C: New test.
* g++.dg/cpp1z/decomp8.C: New test.
* g++.dg/cpp1z/decomp9.C: New test.
* g++.dg/cpp1z/decomp10.C: New test.
Co-Authored-By: Jason Merrill <jason@redhat.com>
From-SVN: r242378
* call.c (build_new_method_call_1): Include template arguments in
error message.
(print_error_for_call_failure): Likewise.
(build_new_function_call): Pass them in.
* name-lookup.c (supplement_binding_1): Don't complain about a
conflict with an erroneous declaration.
* error.c (dump_decl): Fix printing of alias declaration.
* decl.c (make_typename_type): Call cxx_incomplete_type_error.
* parser.c (cp_parser_diagnose_invalid_type_name): Likewise.
* semantics.c (perform_koenig_lookup): Don't wrap an error in
TEMPLATE_ID_EXPR.
From-SVN: r242376
* doc/xml/manual/status_cxx2017.xml: Update status.
* doc/html/manual/status.html: Regenerate.
* include/bits/shared_ptr.h (shared_ptr(unique_ptr<_Yp, _Del>)): Add
extension constructor to maintain C++14 behaviour.
* include/bits/shared_ptr_base.h (__sp_array_delete): Add new struct.
(__shared_count(_Ptr, false_type), __shared_count(_Ptr, true_type)):
New constructors.
(__sp_compatible_with, __sp_is_constructible): Add specializations
for array support.
(__sp_is_constructible_arr, __sp_is_constructible_arrN): New helpers.
(__shared_ptr_access): New base class for observer member functions.
(__shared_ptr::element_type): Use remove_extent.
(__shared_ptr::_UniqCompatible): Add __sp_compatible_with check.
(__shared_ptr(_Yp*)): Use tag dispatching to call new __shared_count
constructor.
(__shared_ptr(unique_ptr<_Yp, _Del>)): Add extension constructor.
(__shared_ptr::operator*, __shared_ptr::operator->): Remove and
inherit from __shared_ptr_access base class.
(__shared_ptr::__has_esft_base): Return false for array types.
(__weak_ptr::element_type): Use remove_extent.
* include/experimental/bits/shared_ptr.h (__libfund_v1): Remove.
(__shared_ptr<__libfund_v1<_Tp>>): Remove specializations.
(__wak_ptr<__libfund_v1<_Tp>>): Likewise.
(experimental::__sp_compatible_v): Redefine using
__sp_compatible_with.
(experimental::__sp_is_constructible_v): Redefine using
__sp_is_constructible.
(get_deleter, operator<<): Change argument from __shared_ptr to
shared_ptr.
* testsuite/20_util/shared_ptr/cons/array.cc: New test.
* testsuite/20_util/shared_ptr/cons/unique_ptr_array.cc: Adjust for
new behaviour.
* testsuite/20_util/shared_ptr/observers/array.cc: Test observers for
arrays.
* testsuite/20_util/shared_ptr/observers/array_neg.cc: New test.
From-SVN: r242369
* gcc-interface/decl.c (gnat_to_gnu_entity): In assertion about known
Esize, protect with !is_type and change !Unknown_Esize to Known_Esize.
From-SVN: r242363
* gcc-interface/utils2.c (gnat_protect_expr): Also protect only the
address if the expression is the component of a dereference.
Do not use a reference type for the final temporary reference.
From-SVN: r242358
* c-ada-spec.c (print_ada_declaration): For typedef declarations, look
for nested types only if the type is a record or union and dump SLOC.
From-SVN: r242356
2016-11-13 Janus Weil <janus@gcc.gnu.org>
PR fortran/60952
* decl.c (match_procedure_in_type): Apply the FL_PROCEDURE attribute
to the target procedure.
2016-11-13 Janus Weil <janus@gcc.gnu.org>
PR fortran/60952
* gfortran.dg/typebound_proc_34.f90: New test.
From-SVN: r242352
The code generating traceback tables mistakenly does an early return
if !optional_tbtab, which causes it to miss the code generating the TOC
section. This only matters if the TOC will be empty since otherwise
the section is created elsewhere.
This patch fixes it.
PR target/77957
* config/rs6000/rs6000.c (rs6000_output_function_epilogue): Don't
return early if !optional_tbtab.
From-SVN: r242336
2016-11-12 Janus Weil <janus@gcc.gnu.org>
PR fortran/77501
* class.c (gfc_find_typebound_intrinsic_op): Remove an unnecessary
assert and nullification.
* decl.c (gfc_match_decl_type_spec): Use gfc_get_tbp_symtree,
fix indentation.
(gfc_match_generic): Remove an unnecessary assert.
Use gfc_get_tbp_symtree to avoid ICE.
2016-11-12 Janus Weil <janus@gcc.gnu.org>
PR fortran/77501
* gfortran.dg/typebound_generic_16.f90: New test.
From-SVN: r242335
