OpenACC 2.6 deep copy: Fortran front-end parts

        gcc/fortran/
        * gfortran.h (gfc_omp_map_op): Add OMP_MAP_ATTACH, OMP_MAP_DETACH.
        * openmp.c (gfc_match_omp_variable_list): Add allow_derived parameter.
        Parse derived-type member accesses if true.
        (omp_mask2): Add OMP_CLAUSE_ATTACH and OMP_CLAUSE_DETACH.
        (gfc_match_omp_map_clause): Add allow_derived parameter.  Pass to
        gfc_match_omp_variable_list.
        (gfc_match_omp_clauses): Support attach and detach.  Support derived
        types for appropriate OpenACC directives.
        (OACC_PARALLEL_CLAUSES, OACC_SERIAL_CLAUSES, OACC_KERNELS_CLAUSES,
        OACC_DATA_CLAUSES, OACC_ENTER_DATA_CLAUSES): Add OMP_CLAUSE_ATTACH.
        (OACC_EXIT_DATA_CLAUSES): Add OMP_CLAUSE_DETACH.
        (check_symbol_not_pointer): Don't disallow pointer objects of derived
        type.
        (resolve_oacc_data_clauses): Don't disallow allocatable derived types.
        (resolve_omp_clauses): Perform duplicate checking only for non-derived
        type component accesses (plain variables and arrays or array sections).
        Support component refs.
        * trans-expr.c (gfc_conv_component_ref,
        conv_parent_component_references): Make global.
        (gfc_maybe_dereference_var): New function, broken out of...
        (gfc_conv_variable): ...here.  Call above function.
        * trans-openmp.c (gfc_omp_privatize_by_reference): Support component
        refs.
        (gfc_trans_omp_array_section): New function, broken out of...
        (gfc_trans_omp_clauses): ...here.  Support component refs/derived
        types, attach and detach clauses.
        * trans.h (gfc_conv_component_ref, conv_parent_component_references,
        gfc_maybe_dereference_var): Add prototypes.

        gcc/testsuite/
        * gfortran.dg/goacc/derived-types.f90: New test.
        * gfortran.dg/goacc/derived-types-2.f90: New test.
        * gfortran.dg/goacc/derived-types-3.f90: New test.
        * gfortran.dg/goacc/data-clauses.f95: Adjust for expected errors.
        * gfortran.dg/goacc/enter-exit-data.f95: Likewise.

From-SVN: r279628

OpenACC 2.6 deep copy: C and C++ front-end parts

gcc/c-family/
* c-common.h (c_omp_map_clause_name): Add prototype.
* c-omp.c (c_omp_map_clause_name): New function.
* c-pragma.h (pragma_omp_clause): Add PRAGMA_OACC_CLAUSE_ATTACH and
PRAGMA_OACC_CLAUSE_DETACH.

gcc/c/
* c-parser.c (c_parser_omp_clause_name): Add parsing of attach and
detach clauses.
(c_parser_omp_variable_list): Add ALLOW_DEREF optional parameter.
Allow deref (->) in variable lists if true.
(c_parser_omp_var_list_parens): Add ALLOW_DEREF optional parameter.
Pass to c_parser_omp_variable_list.
(c_parser_oacc_data_clause): Support attach and detach clauses.  Update
call to c_parser_omp_variable_list.
(c_parser_oacc_all_clauses): Support attach and detach clauses.
(OACC_DATA_CLAUSE_MASK, OACC_ENTER_DATA_CLAUSE_MASK,
OACC_KERNELS_CLAUSE_MASK, OACC_PARALLEL_CLAUSE_MASK,
OACC_SERIAL_CLAUSE_MASK): Add PRAGMA_OACC_CLAUSE_ATTACH.
(OACC_EXIT_DATA_CLAUSE_MASK): Add PRAGMA_OACC_CLAUSE_DETACH.
* c-typeck.c (handle_omp_array_sections_1): Reject subarrays for attach
and detach.  Support deref.
(handle_omp_array_sections): Use GOMP_MAP_ATTACH_DETACH instead of
GOMP_MAP_ALWAYS_POINTER for OpenACC.
(c_oacc_check_attachments): New function.
(c_finish_omp_clauses): Check attach/detach arguments for being
pointers using above.  Support deref.

gcc/cp/
* parser.c (cp_parser_omp_clause_name): Support attach and detach
clauses.
(cp_parser_omp_var_list_no_open): Add ALLOW_DEREF optional parameter.
Parse deref if true.
(cp_parser_omp_var_list): Add ALLOW_DEREF optional parameter.  Pass to
cp_parser_omp_var_list_no_open.
(cp_parser_oacc_data_clause): Support attach and detach clauses.
Update call to cp_parser_omp_var_list_no_open.
(cp_parser_oacc_all_clauses): Support attach and detach.
(OACC_DATA_CLAUSE_MASK, OACC_ENTER_DATA_CLAUSE_MASK,
OACC_KERNELS_CLAUSE_MASK, OACC_PARALLEL_CLAUSE_MASK,
OACC_SERIAL_CLAUSE_MASK): Add PRAGMA_OACC_CLAUSE_ATTACH.
(OACC_EXIT_DATA_CLAUSE_MASK): Add PRAGMA_OACC_CLAUSE_DETACH.
* semantics.c (handle_omp_array_sections_1): Reject subarrays for
attach and detach.
(handle_omp_array_sections): Use GOMP_MAP_ATTACH_DETACH instead of
GOMP_MAP_ALWAYS_POINTER for OpenACC.
(cp_oacc_check_attachments): New function.
(finish_omp_clauses): Use above function.  Allow structure fields and
class members to appear in OpenACC data clauses.  Support
GOMP_MAP_ATTACH_DETACH.  Support deref.

gcc/testsuite/
* c-c++-common/goacc/deep-copy-arrayofstruct.c: New test.
* c-c++-common/goacc/mdc-1.c: New test.
* c-c++-common/goacc/mdc-2.c: New test.
* gcc.dg/goacc/mdc.C: New test.

Co-Authored-By: Cesar Philippidis <cesar@codesourcery.com>
From-SVN: r279627

OpenACC 2.6 deep copy: middle-end parts

gcc/
* gimplify.c (gimplify_omp_var_data): Add GOVD_MAP_HAS_ATTACHMENTS.
(insert_struct_comp_map): Support derived-type member mappings
for arrays with descriptors which use GOMP_MAP_TO_PSET.  Support
GOMP_MAP_ATTACH_DETACH.
(gimplify_scan_omp_clauses): Tidy up OACC_ENTER_DATA/OACC_EXIT_DATA
mappings.  Handle attach/detach clauses and component references.
(gimplify_adjust_omp_clauses_1): Skip adjustments for explicit
attach/detach clauses.
(gimplify_omp_target_update): Handle struct mappings and finalize for
detach operations.
* omp-low.c (lower_omp_target): Support GOMP_MAP_ATTACH,
GOMP_MAP_DETACH, GOMP_MAP_FORCE_DETACH.
* tree-pretty-print.c (dump_omp_clause): Likewise, plus
GOMP_MAP_ATTACH_DETACH.

include/
* gomp-constants.h (gomp_map_kind): Add GOMP_MAP_ATTACH_DETACH.

From-SVN: r279626

OpenACC 2.6 deep copy: libgomp parts

include/
* gomp-constants.h (GOMP_MAP_FLAG_SPECIAL_4, GOMP_MAP_DEEP_COPY):
Define.
(gomp_map_kind): Add GOMP_MAP_ATTACH, GOMP_MAP_DETACH,
GOMP_MAP_FORCE_DETACH.

libgomp/
* libgomp.h (struct target_var_desc): Add do_detach flag.
* oacc-init.c (acc_shutdown_1): Free aux block if present.
* oacc-mem.c (find_group_last): Add SIZES parameter. Support
struct components.  Tidy up and add some new checks.
(goacc_enter_data_internal): Update call to find_group_last.
(goacc_exit_data_internal): Support detach operations and
GOMP_MAP_STRUCT.
(GOACC_enter_exit_data): Handle initial GOMP_MAP_STRUCT or
GOMP_MAP_FORCE_PRESENT in finalization detection code.  Handle
attach/detach in enter/exit data detection code.
* target.c (gomp_map_vars_existing): Initialise do_detach field of
tgt_var_desc.
(gomp_map_vars_internal): Support attach.
(gomp_unmap_vars_internal): Support detach.

From-SVN: r279625

OpenACC 2.6 deep copy: attach/detach API routines

libgomp/
* libgomp.h (struct splay_tree_aux): Add attach_count field.
(gomp_attach_pointer, gomp_detach_pointer): Add prototypes.
* libgomp.map (OACC_2.6): New section. Add acc_attach,
acc_attach_async, acc_detach, acc_detach_async, acc_detach_finalize,
acc_detach_finalize_async.
* oacc-mem.c (acc_attach_async, acc_attach, goacc_detach_internal,
acc_detach, acc_detach_async, acc_detach_finalize,
acc_detach_finalize_async): New functions.
* openacc.h (acc_attach, acc_attach_async, acc_detach,
(acc_detach_async, acc_detach_finalize, acc_detach_finalize_async): Add
prototypes.
* target.c (gomp_attach_pointer, gomp_detach_pointer): New functions.
(gomp_remove_var_internal): Free attachment counts if present.
* testsuite/libgomp.oacc-c-c++-common/deep-copy-3.c: New test.
* testsuite/libgomp.oacc-c-c++-common/deep-copy-5.c: New test.

Co-Authored-By: Thomas Schwinge <thomas@codesourcery.com>
From-SVN: r279624

Factor out duplicate code in gimplify_scan_omp_clauses

gcc/
* gimplify.c (insert_struct_comp_map, extract_base_bit_offset): New.
(gimplify_scan_omp_clauses): Outline duplicated code into calls to
above two functions.

From-SVN: r279623

Use gomp_map_val for OpenACC host-to-device address translation

libgomp/
* libgomp.h (gomp_map_val): Add prototype.
* oacc-parallel.c (GOACC_parallel_keyed): Use gomp_map_val instead of
open-coding device-address calculation.
* target.c (gomp_map_val): Make global. Use OFFSET_POINTER in
non-present case.

Co-Authored-By: Cesar Philippidis <cesar@codesourcery.com>
From-SVN: r279622

OpenACC reference count overhaul

libgomp/
* libgomp.h (struct splay_tree_key_s): Substitute dynamic_refcount
field for virtual_refcount.
(enum gomp_map_vars_kind): Add GOMP_MAP_VARS_OPENACC_ENTER_DATA.
(gomp_free_memmap): Remove prototype.
* oacc-init.c (acc_shutdown_1): Iteratively call gomp_remove_var
instead of calling gomp_free_memmap.
* oacc-mem.c (acc_map_data): Use virtual_refcount instead of
dynamic_refcount.
(acc_unmap_data): Open code instead of forcing target_mem_desc's
to_free field to NULL then calling gomp_unmap_vars.  Handle
REFCOUNT_INFINITY on target blocks.
(goacc_enter_data): Rename to...
(goacc_enter_datum): ...this.  Remove MAPNUM parameter and special
handling for mapping groups.  Use virtual_refcount instead of
dynamic_refcount.  Use GOMP_MAP_VARS_OPENACC_ENTER_DATA for
map_map_vars_async call.  Re-do lookup for target pointer return value.
(acc_create, acc_create_async, acc_copyin, acc_copyin_async): Call
renamed goacc_enter_datum function.
(goacc_exit_data): Rename to...
(goacc_exit_datum): ...this.  Update for virtual_refcount semantics.
(acc_delete, acc_delete_async, acc_delete_finalize,
acc_delete_finalize_async, acc_copyout, acc_copyout_async,
acc_copyout_finalize, acc_copyout_finalize_async): Call renamed
goacc_exit_datum function.
(gomp_acc_remove_pointer, find_pointer): Remove functions.
(find_group_last, goacc_enter_data_internal, goacc_exit_data_internal):
New functions.
(GOACC_enter_exit_data): Use goacc_enter_data_internal and
goacc_exit_data_internal helper functions.
* target.c (gomp_map_vars_internal): Handle
GOMP_MAP_VARS_OPENACC_ENTER_DATA.  Update for virtual_refcount
semantics.
(gomp_unmap_vars_internal): Update for virtual_refcount semantics.
(gomp_load_image_to_device, omp_target_associate_ptr): Zero-initialise
virtual_refcount field instead of dynamic_refcount.
(gomp_free_memmap): Remove function.
* testsuite/libgomp.oacc-c-c++-common/unmap-infinity-1.c: New test.
* testsuite/libgomp.c-c++-common/unmap-infinity-2.c: New test.
* testsuite/libgomp.oacc-c-c++-common/pr92843-1.c: Add XFAIL.

From-SVN: r279621

Use aux struct in libgomp for infrequently-used/API-specific data

libgomp/
* libgomp.h (struct splay_tree_aux): New.
(struct splay_tree_key_s): Replace link_key field with aux pointer.
* target.c (gomp_map_vars_internal): Adjust for link_key being moved
to aux struct.
(gomp_remove_var_internal): Free aux block if present.
(gomp_load_image_to_device): Zero-initialise aux field instead of
link_key field.
(omp_target_associate_pointer): Zero-initialise aux field.

Co-Authored-By: Thomas Schwinge <thomas@codesourcery.com>
From-SVN: r279620

Daily bump.

From-SVN: r279619

re PR target/92905 (Spills float-int union to memory)

2019-12-19  Vladimir Makarov  <vmakarov@redhat.com>

PR target/92905
* lra-constraints.c (process_alt_operands): Check offmemok when
processing preferred_reload_class.

2019-12-19  Vladimir Makarov  <vmakarov@redhat.com>

PR target/92905
* gcc.target/i386/pr92905.c: New test.

From-SVN: r279596

Update g++.dg/ext/sve-sizeless-2.C after r279471

2019-12-19  Richard Sandiford  <richard.sandiford@arm.com>

gcc/testsuite/
* g++.dg/ext/sve-sizeless-2.C: Don't expect an error for
alias templates.

From-SVN: r279588

Allow constants in amdgcn extends and truncates

2019-12-19  Andrew Stubbs  <ams@codesourcery.com>

gcc/
* config/gcn/gcn-valu.md
(<convop><VEC_ALL1REG_INT_ALT:mode><VEC_ALL1REG_INT_MODE:mode>2<exec>):
Change input predcate to gcn_alu_operand.
(extend<VEC_ALL1REG_INT_ALT:mode><VEC_ALL1REG_INT_MODE:mode>2<exec>):
Likewise.
(truncv64di<mode>2): Likewise.
(truncv64di<mode>2_exec): Likewise.
(<convop><mode>v64di2): Likewise.
(<convop><mode>v64di2_exec): Likewise.

From-SVN: r279587

Use V64SI for all amdgcn add-with-carry insns

2019-12-19  Andrew Stubbs  <ams@codesourcery.com>

gcc/
* config/gcn/gcn-valu.md (*plus_carry_dpp_shr_<mode>): Rename to ...
(*plus_carry_dpp_shr_v64si): ... this, and replace all
VEC_1REG_INT_MODE with V64SI.

From-SVN: r279584

Prevent conversion of character data in array constructors.

Fix for PR fortran/92896 [10 Regression] [DEC] ICE in reduce_unary, at
fortran/arith.c:1283.

This was caused by an unintended side affect of "Allow CHARACTER literals
in assignments and data statements" (revision 277975). If the conversion
occurs in a array constructor it is rejected.

From-SVN: r279583

hash-map-tests.c: add a selftest involving int_hash

gcc/ChangeLog:
* hash-map-tests.c (selftest::test_map_of_int_to_strings): New
selftest.
(selftest::hash_map_tests_c_tests): Call it.

From-SVN: r279582

gimple const-correctness fixes

This patch converts various "gimple *" to "const gimple *" and similar
fixes for gimple subclasses, adding is_a_helper for gimple subclasses
to support the const form of as_a, and adding a few "const" overloads
of accessors.

This is enough to make pp_gimple_stmt_1's stmt const.

gcc/ChangeLog:
* gimple-predict.h (gimple_predict_predictor): Make "gs" param
const.
(gimple_predict_outcome): Likewise.
* gimple-pretty-print.c (do_niy): Likewise.
(dump_unary_rhs): Likewise.
(dump_binary_rhs): Likewise.
(dump_ternary_rhs): Likewise.
(dump_gimple_assign): Likewise.
(dump_gimple_return): Likewise.
(dump_gimple_call_args): Likewise.
(pp_points_to_solution): Make "pt" param const.
(dump_gimple_call): Make "gs" param const.
(dump_gimple_switch): Likewise.
(dump_gimple_cond): Likewise.
(dump_gimple_label): Likewise.
(dump_gimple_goto): Likewise.
(dump_gimple_bind): Likewise.
(dump_gimple_try): Likewise.
(dump_gimple_catch): Likewise.
(dump_gimple_eh_filter): Likewise.
(dump_gimple_eh_must_not_throw): Likewise.
(dump_gimple_eh_else): Likewise.
(dump_gimple_resx): Likewise.
(dump_gimple_eh_dispatch): Likewise.
(dump_gimple_debug): Likewise.
(dump_gimple_omp_for): Likewise.
(dump_gimple_omp_continue): Likewise.
(dump_gimple_omp_single): Likewise.
(dump_gimple_omp_taskgroup): Likewise.
(dump_gimple_omp_target): Likewise.
(dump_gimple_omp_teams): Likewise.
(dump_gimple_omp_sections): Likewise.
(dump_gimple_omp_block): Likewise.
(dump_gimple_omp_critical): Likewise.
(dump_gimple_omp_ordered): Likewise.
(dump_gimple_omp_scan): Likewise.
(dump_gimple_omp_return): Likewise.
(dump_gimple_transaction): Likewise.
(dump_gimple_asm): Likewise.
(dump_gimple_phi): Make "phi" param const.
(dump_gimple_omp_parallel): Make "gs" param const.
(dump_gimple_omp_task): Likewise.
(dump_gimple_omp_atomic_load): Likewise.
(dump_gimple_omp_atomic_store): Likewise.
(dump_gimple_mem_ops): Likewise.
(pp_gimple_stmt_1): Likewise.  Add "const" to the various as_a <>
casts throughout.
* gimple-pretty-print.h (gimple_stmt_1): Make gimple * param const.
* gimple.h (is_a_helper <const gdebug *>::test): New.
(is_a_helper <const ggoto *>::test): New.
(is_a_helper <const glabel *>::test): New.
(is_a_helper <const geh_else *>::test): New.
(is_a_helper <const geh_mnt *>::test): New.
(is_a_helper <const gswitch *>::test): New.
(is_a_helper <const gtry *>::test): New.
(is_a_helper <const greturn *>::test): New.
(gimple_call_tail_p): Make param const.
(gimple_call_return_slot_opt_p): Likewise.
(gimple_call_va_arg_pack_p): Likewise.
(gimple_call_use_set): Add const overload.
(gimple_call_clobber_set): Likewise.
(gimple_has_lhs): Make param const.
(gimple_bind_body): Likewise.
(gimple_catch_handler): Likewise.
(gimple_eh_filter_failure): Likewise.
(gimple_eh_must_not_throw_fndecl): Likewise.
(gimple_eh_else_n_body): Likewise.
(gimple_eh_else_e_body): Likewise.
(gimple_try_eval): Likewise.
(gimple_try_cleanup): Likewise.
(gimple_phi_arg): Add const overload.
(gimple_phi_arg_def): Make param const.
(gimple_phi_arg_edge): Likewise.
(gimple_phi_arg_location): Likewise.
(gimple_phi_arg_has_location): Likewise.
(gimple_debug_bind_get_var): Likewise.
(gimple_debug_bind_get_value): Likewise.
(gimple_debug_source_bind_get_var): Likewise.
(gimple_debug_source_bind_get_value): Likewise.
(gimple_omp_body): Likewise.
(gimple_omp_for_collapse): Likewise.
(gimple_omp_for_pre_body): Likewise.
(gimple_transaction_body): Likewise.
* tree-eh.c (lookup_stmt_eh_lp_fn): Make param "t" const.
(lookup_stmt_eh_lp): Likewise.
* tree-eh.h (lookup_stmt_eh_lp_fn): Make param const.
(lookup_stmt_eh_lp): Likewise.
* tree-ssa-alias.h (pt_solution_empty_p): Make param const.
* tree-ssa-structalias.c (pt_solution_empty_p): Likewise.

From-SVN: r279581

[AArch64] Fix handling of npatterns>1 constants for partial SVE modes

For partial SVE vectors of element X, we want to treat duplicates
of single X elements in the same way as for full vectors of X.
But if a constant instead contains a repeating pattern of X elements,
the transition from one value to the next must happen at container
boundaries rather than element boundaries.  E.g. a VNx4HI should
in that case contain the same number of constants as a VNx4SI.

Fixing this means that we need a reinterpret from the container-based
mode to the partial mode; e.g. in the above example we need a
reinterpret from VNx4SI to VNx4HI.  We can't use subregs for that
because they're forbidden by aarch64_can_change_class_mode; we should
handle them in the same way as for big-endian instead.

2019-12-19  Richard Sandiford  <richard.sandiford@arm.com>

gcc/
* config/aarch64/aarch64.c (aarch64_simd_valid_immediate): When
handling partial SVE vectors, use the container mode rather than
the element mode if the constant isn't a single-element duplicate.
* config/aarch64/aarch64-sve.md (@aarch64_sve_reinterpret<mode>):
Check targetm.can_change_mode_class instead of BYTES_BIG_ENDIAN.

gcc/testsuite/
* gcc.target/aarch64/sve/mixed_size_9.c: New test.

From-SVN: r279580

comment tweak

From-SVN: r279579

PR c++/52320 - EH cleanups for partially constructed arrays.

This testcase wasn't fixed by the 66139 patch; split_nonconstant_init_1 was
failing to add a cleanup for an array member of a class (e.g. e1) that will
run if initializing a later member (e.g. e2) throws.

* typeck2.c (split_nonconstant_init_1): Add nested parm.
Add cleanup for whole array if true.

From-SVN: r279577

PR c++/66139 - EH cleanups for partially constructed aggregates.

There were several overlapping PRs about failure to clean up fully
constructed subobjects when an exception is thrown during aggregate
initialization of a temporary.  I fixed this for non-temporaries in the
context of 57510, but that fix didn't handle temporaries.  So this patch
does split_nonconstant_init at gimplification time, which is much smaller
than alternatives I tried.

PR c++/57510
* cp-gimplify.c (cp_gimplify_init_expr): Use split_nonconstant_init.
* typeck2.c (split_nonconstant_init): Handle non-variable dest.
(split_nonconstant_init_1): Clear TREE_SIDE_EFFECTS.
* tree.c (is_local_temp): New.

From-SVN: r279576

Add changelog for commit r279573 (added myself to write-after-approval)

2019-12-19  Stam Markianos-Wright  <stam.markianos-wright@arm.com>

* MAINTAINERS (write_after_approval): Add myself.

From-SVN: r279575

Implement sub-dword add/sub on amdgcn

2019-12-19  Andrew Stubbs  <ams@codesourcery.com>

gcc/
* config/gcn/gcn-valu.md (addv64si3<exec_clobber>): Rename to ...
(add<mode>3<exec_clobber>): ... this, and use VEC_ALL1REG_INT_MODE.
(addv64si3_dup<exec_clobber>): Rename to ...
(add<mode>3_dup<exec_clobber>): ... this, and use VEC_ALL1REG_INT_MODE.
(subv64si3<exec_clobber>): Rename to ...
(sub<mode>3<exec_clobber>): ... this, and use VEC_ALL1REG_INT_MODE.

From-SVN: r279574

Add myself to MAINTAINERS

From-SVN: r279573

[AArch64] Reject invalid subregs involving partial SVE modes

When adding partial SVE modes, I'd remembered to handle reloads
in a similar way to full big-endian SVE vectors, but forgot the
just-as-important mode-change rules.

2019-12-19  Richard Sandiford  <richard.sandiford@arm.com>

gcc/
* config/aarch64/aarch64.c (aarch64_can_change_mode_class):
Don't allow changes between partial SVE modes and other kinds
of mode.  Don't allow changes between two partial SVE modes
if they have different container or element sizes.

gcc/testsuite/
* gcc.target/aarch64/sve/mixed_size_8.c: New test.

From-SVN: r279572

[AArch64] Handle arguments and return types with partial SVE modes

Partial SVE modes can be picked up and used by the vector_size(N)
attribute.[*] This means that we need to cope with arguments and return
values with partial SVE modes, which previously triggered asserts like:

  /* Generic vectors that map to SVE modes with -msve-vector-bits=N are
     passed by reference, not by value.  */
  gcc_assert (!aarch64_sve_mode_p (mode));

The ABI for these types is fixed from pre-SVE days, and must in any case
be the same for all -msve-vector-bits=N values.  All we need to do is
ensure that the vectors are passed and returned in the traditional way.

[*] Advanced SIMD always wins for 64-bit and 128-bit vectors though.

2019-12-19  Richard Sandiford  <richard.sandiford@arm.com>

gcc/
* config/aarch64/aarch64.c (aarch64_function_value_1): New function,
split out from...
(aarch64_function_value): ...here.  Handle partial SVE modes by
pretending that they have the associated/traditional integer mode,
then wrap the result in the real mode.
(aarch64_layout_arg): Take an orig_mode argument and pass it to
aarch64_function_arg_alignment.  Handle partial SVE modes analogously
to aarch64_function_value.
(aarch64_function_arg): Update call accordingly.
(aarch64_function_arg_advance): Likewise.

gcc/testsuite/
* gcc.target/aarch64/sve/pcs/gnu_vectors_3.c: New test.

From-SVN: r279571

Revert accidental commit

From-SVN: r279570

Don't mangle attributes that have a space in their name

The SVE port needs to maintain a different type identity for
GNU vectors and "SVE vectors" even during LTO, since the types
use different ABIs.  The easiest way of doing that seemed to be
to use type attributes.  However, these type attributes shouldn't
be user-facing; they're just a convenient way of representing the
types internally in GCC.

There are already several internal-only attributes, such as "fn spec"
and "omp declare simd".  They're distinguished from normal user-facing
attributes by having a space in their name, which means that it isn't
possible to write them directly in C or C++.

Taking the same approach mostly works well for SVE.  The only snag
I've hit so far is that the new attribute needs to (and only exists to)
affect type identity.  This means that it would normally get included
in mangled names, to distinguish it from types without the attribute.

However, the SVE ABI specifies a separate mangling for SVE vector types,
rather than using an attribute mangling + a normal vector mangling.
So we need some way of suppressing the attribute mangling for this case.

There are currently no other target-independent or target-specific
internal-only attributes that affect type identity, so this patch goes
for the simplest fix of skipping mangling for attributes whose names
contain a space.  Other options I thought about were:

(1) Also make sure that targetm.mangled_type returns nonnull.

(2) Check directly for the target-specific name.

(3) Add a new target hook.

(4) Add new information to attribute_spec.  This would be very invasive
    at this stage, but maybe we should consider replacing all the boolean
    fields with flags?  That should make the tables slightly easier to
    read and would make adding new flags much simpler in future.

2019-12-19  Richard Sandiford  <richard.sandiford@arm.com>

gcc/cp/
* mangle.c (write_CV_qualifiers_for_type): Don't mangle attributes
that contain a space.

From-SVN: r279569

re PR fortran/92977 (ICE in gfc_trans_omp_atomic, at fortran/trans-openmp.c:3526)

PR fortran/92977
* frontend-passes.c (call_external_blas): Use || instead of |.

From-SVN: r279567

Fix symver attribute with LTO

* cgraph.c (cgraph_node_cannot_be_local_p_1): Prevent targets of
symver attributes to be localized.
* ipa-visibility.c (cgraph_externally_visible_p,
varpool_node::externally_visible_p): Likewise.
* symtab.c (symtab_node::verify_base): Check visibility of symbol
versions.

* lto-common.c (read_cgraph_and_symbols): Work around binutils
PR25424

Co-Authored-By: Xi Ruoyao <xry111@mengyan1223.wang>
From-SVN: r279566

ipa: duplicate ipa_size_summary for cloned nodes

* ipa-fnsummary.h (ipa_size_summary): Remove copy consturctor.
(ipa_size_summary_t): Add duplicate method; move to heap.
* ipa-fnsumary.c (ipa_fn_summary_alloc): Fix allocation.

From-SVN: r279563

Handle aggregate pass-through for self-recursive call (PR ipa/92794)

2019-12-19  Feng Xue  <fxue@os.amperecomputing.com>

        PR ipa/92794
        * ipa-cp.c (self_recursive_agg_pass_through_p): New function.
        (intersect_with_plats): Use error_mark_node as place holder
        when aggregate jump function is simple pass-through for
        self-recursive call.
        (intersect_with_agg_replacements): Likewise.
        (intersect_aggregates_with_edge): Likewise.
        (find_aggregate_values_for_callers_subset): Likewise.

2019-12-19  Feng Xue  <fxue@os.amperecomputing.com>

        PR ipa/92794
        * gcc.dg/ipa/92794.c: New test.

From-SVN: r279561

Daily bump.

From-SVN: r279560

PR c++/91165 follow-on tweak

I talked in the PR about possibly stripping the location from the args in
the hash table, since if we use the cache the locations would be wrong, but
didn't actually do anything about that.  Then I noticed that there's already
unshare_expr_without_location...

* constexpr.c (cxx_eval_call_expression): Use
unshare_expr_without_location.

From-SVN: r279557

Add diagnostic_metadata and CWE support

This patch adds support for associating a diagnostic message with an
optional diagnostic_metadata object, so that plugins can add extra data
to their diagnostics (e.g. mapping a diagnostic to a taxonomy or coding
standard such as from CERT or MISRA).

Currently this only supports associating a CWE identifier with a
diagnostic (which is what I'm using for the warnings in the analyzer
patch kit), but adding a diagnostic_metadata class allows for future
growth in this area without an explosion of further "warning_at"
overloads for all of the different kinds of custom data that a plugin
might want to add.

This version of the patch renames the overly-general
-fdiagnostics-show-metadata to -fdiagnostics-show-cwe and adds test
coverage for it via a plugin.

It also adds a note to the documentation that no GCC diagnostics
currently use this; it's a feature for plugins (and, at some point,
I hope, the analyzer).

gcc/ChangeLog:
* common.opt (fdiagnostics-show-cwe): Add.
* diagnostic-core.h (class diagnostic_metadata): New forward decl.
(warning_at): Add overload taking a const diagnostic_metadata &.
(emit_diagnostic_valist): Add overload taking a
const diagnostic_metadata *.
* diagnostic-format-json.cc: Include "diagnostic-metadata.h".
(json_from_metadata): New function.
(json_end_diagnostic): Call it to add "metadata" child for
diagnostics with metadata.
(diagnostic_output_format_init): Clear context->show_cwe.
* diagnostic-metadata.h: New file.
* diagnostic.c: Include "diagnostic-metadata.h".
(diagnostic_impl): Add const diagnostic_metadata * param.
(diagnostic_n_impl): Likewise.
(diagnostic_initialize): Initialize context->show_cwe.
(diagnostic_set_info_translated): Initialize diagnostic->metadata.
(get_cwe_url): New function.
(print_any_cwe): New function.
(diagnostic_report_diagnostic): Call print_any_cwe if the
diagnostic has non-NULL metadata.
(emit_diagnostic): Pass NULL as the metadata in the call to
diagnostic_impl.
(emit_diagnostic_valist): Likewise.
(emit_diagnostic_valist): New overload taking a
const diagnostic_metadata *.
(inform): Pass NULL as the metadata in the call to
diagnostic_impl.
(inform_n): Likewise for diagnostic_n_impl.
(warning): Likewise.
(warning_at): Likewise.  Add overload that takes a
const diagnostic_metadata &.
(warning_n): Pass NULL as the metadata in the call to
diagnostic_n_impl.
(pedwarn): Likewise for diagnostic_impl.
(permerror): Likewise.
(error): Likewise.
(error_n): Likewise.
(error_at): Likewise.
(sorry): Likewise.
(sorry_at): Likewise.
(fatal_error): Likewise.
(internal_error): Likewise.
(internal_error_no_backtrace): Likewise.
* diagnostic.h (diagnostic_info::metadata): New field.
(diagnostic_context::show_cwe): New field.
* doc/invoke.texi (-fno-diagnostics-show-cwe): New option.
* opts.c (common_handle_option): Handle OPT_fdiagnostics_show_cwe.
* toplev.c (general_init): Initialize global_dc->show_cwe.

gcc/testsuite/ChangeLog:
* gcc.dg/plugin/diagnostic-test-metadata.c: New test.
* gcc.dg/plugin/diagnostic_plugin_test_metadata.c: New test plugin.
* gcc.dg/plugin/plugin.exp (plugin_test_list): Add them.

From-SVN: r279556

re PR fortran/92977 (ICE in gfc_trans_omp_atomic, at fortran/trans-openmp.c:3526)

PR fortran/92977
* frontend-passes.c (in_omp_atomic): New variable.
(cfe_expr_0, matmul_to_var_expr, matmul_temp_args,
inline_matmul_assign, call_external_blas): Don't optimize in
EXEC_OMP_ATOMIC.
(optimize_namespace): Clear in_omp_atomic.
(gfc_code_walker): Set in_omp_atomic for EXEC_OMP_ATOMIC, save/restore
it around.

* gfortran.dg/gomp/pr92977.f90: New test.

From-SVN: r279554

re PR middle-end/86416 ([OpenMP] Offloading - better lto1 error message if mode not supported on offloading target)

PR middle-end/86416
* testsuite/libgomp.c/pr86416-1.c (main): Use L suffixes rather than
q or none.
* testsuite/libgomp.c/pr86416-2.c (main): Use Q suffixes rather than
L or none.

From-SVN: r279552

Add OpenACC 2.6's no_create

The clause makes any device code use the local memory address for each
of the variables specified unless the given variable is already present
on the current device.

2019-12-19  Julian Brown  <julian@codesourcery.com>
            Maciej W. Rozycki  <macro@codesourcery.com>
            Tobias Burnus  <tobias@codesourcery.com>
            Thomas Schwinge  <thomas@codesourcery.com>

        gcc/
        * omp-low.c (lower_omp_target): Support GOMP_MAP_NO_ALLOC.
        * tree-pretty-print.c (dump_omp_clause): Likewise.

        gcc/c-family/
        * c-pragma.h (pragma_omp_clause): Add
        PRAGMA_OACC_CLAUSE_NO_CREATE.

        gcc/c/
        * c-parser.c (c_parser_omp_clause_name): Support no_create.
        (c_parser_oacc_data_clause): Likewise.
        (c_parser_oacc_all_clauses): Likewise.
        (OACC_DATA_CLAUSE_MASK, OACC_KERNELS_CLAUSE_MASK)
        (OACC_PARALLEL_CLAUSE_MASK, OACC_SERIAL_CLAUSE_MASK): Add
        PRAGMA_OACC_CLAUSE_NO_CREATE.
        * c-typeck.c (handle_omp_array_sections): Support
        GOMP_MAP_NO_ALLOC.

        gcc/cp/
        * parser.c (cp_parser_omp_clause_name): Support no_create.
        (cp_parser_oacc_data_clause): Likewise.
        (cp_parser_oacc_all_clauses): Likewise.
        (OACC_DATA_CLAUSE_MASK, OACC_KERNELS_CLAUSE_MASK)
        (OACC_PARALLEL_CLAUSE_MASK): Add PRAGMA_OACC_CLAUSE_NO_CREATE.
        * semantics.c (handle_omp_array_sections): Support no_create.

        gcc/fortran/
        * gfortran.h (gfc_omp_map_op): Add OMP_MAP_NO_ALLOC.
        * openmp.c (omp_mask2): Add OMP_CLAUSE_NO_CREATE.
        (gfc_match_omp_clauses): Support no_create.
        (OACC_PARALLEL_CLAUSES, OACC_KERNELS_CLAUSES)
        (OACC_DATA_CLAUSES): Add OMP_CLAUSE_NO_CREATE.
        * trans-openmp.c (gfc_trans_omp_clauses_1): Support
        OMP_MAP_NO_ALLOC.

        gcc/testsuite/
        * gfortran.dg/goacc/common-block-1.f90: Add no_create-clause tests.
        * gfortran.dg/goacc/common-block-1.f90: Likewise.
        * gfortran.dg/goacc/data-clauses.f95: Likewise.
        * gfortran.dg/goacc/data-tree.f95: Likewise.
        * gfortran.dg/goacc/kernels-tree.f95: Likewise.
        * gfortran.dg/goacc/parallel-tree.f95: Likewise.

        include/
        * gomp-constants.h (gomp_map_kind): Support GOMP_MAP_NO_ALLOC.

        libgomp/
        * target.c (gomp_map_vars_async): Support GOMP_MAP_NO_ALLOC.
        * testsuite/libgomp.oacc-c-c++-common/no_create-1.c: New test.
        * testsuite/libgomp.oacc-c-c++-common/no_create-2.c: New test.
        * testsuite/libgomp.oacc-c-c++-common/no_create-3.c: New test.
        * testsuite/libgomp.oacc-c-c++-common/no_create-4.c: New test.
        * testsuite/libgomp.oacc-c-c++-common/no_create-5.c: New test.
        * testsuite/libgomp.oacc-fortran/no_create-1.f90: New test.
        * testsuite/libgomp.oacc-fortran/no_create-2.f90: New test.
        * testsuite/libgomp.oacc-fortran/no_create-3.F90: New test.

Reviewed-by: Thomas Schwinge <thomas@codesourcery.com>
Co-Authored-By: Maciej W. Rozycki <macro@codesourcery.com>
Co-Authored-By: Thomas Schwinge <thomas@codesourcery.com>
Co-Authored-By: Tobias Burnus <tobias@codesourcery.com>
From-SVN: r279551

* ira.c (ira): Use simple LRA algorithm when not optimizing.

From-SVN: r279550

typeck.c (cxx_sizeof_or_alignof_type): Add location_t parameter and use it throughout.

/gcc/cp
2019-12-18  Paolo Carlini  <paolo.carlini@oracle.com>

* typeck.c (cxx_sizeof_or_alignof_type): Add location_t parameter
and use it throughout.
(cxx_sizeof_expr): Likewise.
(cxx_alignof_expr): Likewise.
(cxx_sizeof_or_alignof_expr): Likewise.
(cxx_alignas_expr): Update call.
* decl.c (fold_sizeof_expr): Likewise.
* pt.c (tsubst_copy): Likewise.
(tsubst_copy_and_build): Likewise.
* except.c (build_throw): Add location_t parameter and use it.
(expand_end_catch_block): Update call.
* parser.c (cp_parser_unary_expression): Update
cxx_sizeof_or_alignof_type and cxx_sizeof_or_alignof_expr calls,
pass the compound location.
(cp_parser_throw_expression): Likewise pass the combined location
to build_throw.
* cp-tree.h: Update declarations.

* semantics.c (finish_handler_parms): Use DECL_SOURCE_LOCATION.
* decl2.c (check_classfn): Likewise.

* except.c (is_admissible_throw_operand_or_catch_parameter):
Exploit cp_expr_loc_or_input_loc in one place.

* except.c (create_try_catch_expr): Remove, unused.

/libcc1
2019-12-18  Paolo Carlini  <paolo.carlini@oracle.com>

* libcp1plugin.cc (plugin_build_unary_expr): Update build_throw
and cxx_sizeof_or_alignof_expr calls.
(plugin_build_unary_type_expr): Likewise for
cxx_sizeof_or_alignof_type.

/gcc/testsuite
2019-12-18  Paolo Carlini  <paolo.carlini@oracle.com>

* g++.dg/diagnostic/alignof2.C: New.
* g++.dg/diagnostic/alignof3.C: Likewise.
* g++.dg/diagnostic/incomplete-type-1.C: Likewise.
* g++.dg/warn/Wcatch-value-3b.C: Likewise.
* g++.dg/cpp0x/alignof3.C: Check location(s) too.
* g++.dg/cpp1z/decomp-bitfield1.C: Likewise.
* g++.dg/cpp1z/has-unique-obj-representations2.C: Likewise.
* g++.dg/expr/sizeof3.C: Likewise.
* g++.dg/ext/flexary6.C: Likewise.
* g++.dg/ext/vla4.C: Likewise.
* g++.dg/template/sizeof11.C: Likewise.
* g++.dg/warn/Wcatch-value-1.C: Likewise.
* g++.dg/warn/Wcatch-value-2.C: Likewise.
* g++.dg/warn/Wcatch-value-3.C: Likewise.
* g++.old-deja/g++.brendan/sizeof1.C: Likewise.
* g++.old-deja/g++.brendan/sizeof3.C: Likewise.
* g++.old-deja/g++.brendan/sizeof4.C: Likewise.
* g++.old-deja/g++.eh/ctor1.C: Likewise.
* g++.old-deja/g++.jason/ambig1.C: Likewise.
* g++.old-deja/g++.other/sizeof4.C: Likewise.

From-SVN: r279543

Fix POWER dfp test case target tests.

PR bootstrap/92661
* gcc.target/powerpc/pr92661.c: New test.
* gcc.target/powerpc/dfp-dd.c: Add dg-require-effective-target hard_dfp.
Remove unneeded powerpc_fprs test.
* gcc.target/powerpc/dfp-td.c: Likewise.
* gcc.target/powerpc/dfp-dd-2.c: Add dg-require-effective-target dfp.
* gcc.target/powerpc/dfp-td-2.c: Likewise.
* gcc.target/powerpc/dfp-td-3.c: Likewise.
* gcc.target/powerpc/dfp/dfp.exp: Remove rs6000-*-* and
powerpc*-*-darwin* target tests.  Add check_effective_target_dfp test.
* gcc.target/powerpc/dfp/dtstsfi-0.c: Remove unneeded target test.
Remove unneeded dg-skip-if.
* gcc.target/powerpc/dfp/dtstsfi-1.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-10.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-11.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-12.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-13.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-14.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-15.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-16.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-17.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-18.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-19.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-2.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-20.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-21.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-22.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-23.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-24.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-25.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-26.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-27.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-28.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-29.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-3.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-30.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-31.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-32.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-33.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-34.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-35.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-36.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-37.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-38.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-39.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-4.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-40.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-41.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-42.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-43.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-44.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-45.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-46.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-47.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-48.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-49.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-5.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-50.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-51.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-52.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-53.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-54.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-55.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-56.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-57.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-58.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-59.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-6.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-60.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-61.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-62.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-63.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-64.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-65.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-66.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-67.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-68.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-69.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-7.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-70.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-71.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-72.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-73.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-74.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-75.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-76.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-77.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-78.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-79.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-8.c: Likewise.
* gcc.target/powerpc/dfp/dtstsfi-9.c: Likewise.

From-SVN: r279542

Drop unused member from cpp_string_location_reader (PR preprocessor/92982)

libcpp/ChangeLog:
PR preprocessor/92982
* charset.c
(cpp_string_location_reader::cpp_string_location_reader): Delete
initialization of m_line_table.
* include/cpplib.h (cpp_string_location_reader::m_line_table):
Delete unused member.

From-SVN: r279541

[OpenACC] Refactor 'goacc_enter_data' so that it can be called from 'goacc_insert_pointer', "present" case, and simplify

No functional changes.

libgomp/
* oacc-mem.c (goacc_enter_data): Refactor, so that it can be
called...
(goacc_insert_pointer): ... from here, "present" case.
(goacc_insert_pointer): Inline function into...
(GOACC_enter_exit_data): ... here, and simplify.

From-SVN: r279540

[OpenACC] Refactor 'goacc_enter_data' so that it can be called from 'goacc_insert_pointer', "not present" case

No functional changes.

libgomp/
* oacc-mem.c (goacc_enter_data): Refactor, so that it can be
called...
(goacc_insert_pointer): ... from here, "not present" case.

From-SVN: r279539

[OpenACC] Refactor 'goacc_remove_pointer' interface

No functional changes.

libgomp/
* oacc-mem.c (goacc_remove_pointer): Refactor interface.  Adjust
all users.

From-SVN: r279538

[OpenACC] Refactor 'GOACC_enter_exit_data' to call 'goacc_enter_data', 'goacc_exit_data'

No functional changes.

libgomp/
* oacc-mem.c (GOACC_enter_exit_data): Refactor code to call
'goacc_enter_data', 'goacc_exit_data'.

From-SVN: r279537

[OpenACC] Refactor 'delete_copyout' into 'goacc_exit_data'

Change 'FLAG_COPYOUT', 'FLAG_FINALIZE' into the usual map kind.

No functional changes.

libgomp/
* oacc-mem.c (delete_copyout): Refactor into...
(goacc_exit_data): ... this.  Adjust all users.

From-SVN: r279536

[OpenACC] Refactor 'present_create_copy' into 'goacc_enter_data'

Every caller passes in 'FLAG_PRESENT', 'FLAG_CREATE'.  Change the remaining
'FLAG_COPY' into the usual map kind.

No functional changes.

libgomp/
* oacc-mem.c (present_create_copy): Refactor into...
(goacc_enter_data): ... this.  Adjust all users.

From-SVN: r279535

Assert in 'libgomp/target.c:gomp_unmap_vars_internal' that we're not unmapping 'tgt' while it's still in use

libgomp/
* target.c (gomp_unmap_vars_internal): Add a safeguard to
'gomp_remove_var'.

From-SVN: r279534

[OpenACC] In 'libgomp/target.c:gomp_to_device_kind_p', handle 'GOMP_MAP_FORCE_FROM' like 'GOMP_MAP_FROM'

Fix oversight from r254194 "Coalesce host to device transfers in libgomp".

libgomp/
* target.c (gomp_to_device_kind_p): Handle 'GOMP_MAP_FORCE_FROM'
like 'GOMP_MAP_FROM'.

From-SVN: r279533

[PR92726, PR92970, PR92984] [OpenACC] Clarify 'acc_delete' etc. for 'NULL'-in, non-present data, or size zero

PR92970 "OpenACC 2.5: 'acc_delete' etc. on non-present data is a no-op" is an
actual bug fix, and the other ones are fall-out, currently undefined behavior.

libgomp/
PR libgomp/92726
PR libgomp/92970
PR libgomp/92984
* oacc-mem.c (delete_copyout): No-op behavior if 'lookup_host'
fails.
(GOACC_enter_exit_data): Simplify accordingly.
* testsuite/libgomp.oacc-c-c++-common/pr92970-1.c: New file,
subsuming...
* testsuite/libgomp.oacc-c-c++-common/lib-17.c: ... this file...
* testsuite/libgomp.oacc-c-c++-common/lib-18.c: ..., and this
file.
* testsuite/libgomp.oacc-c-c++-common/pr92984-1.c: New file,
subsuming...
* testsuite/libgomp.oacc-c-c++-common/lib-21.c: ... this file...
* testsuite/libgomp.oacc-c-c++-common/lib-29.c: ..., and this
file.
* testsuite/libgomp.oacc-c-c++-common/pr92726-1.c: New file,
subsuming...
* testsuite/libgomp.oacc-c-c++-common/lib-28.c: ... this file.

From-SVN: r279532

[OpenACC] Elaborate/simplify 'exit data' 'finalize' handling

No functional changes.

gcc/
* gimplify.c (gimplify_omp_target_update): Elaborate 'exit data'
'finalize' handling.
gcc/testsuite/
* c-c++-common/goacc/finalize-1.c: Extend.
* gfortran.dg/goacc/finalize-1.f: Likewise.
libgomp/
* oacc-mem.c (GOACC_enter_exit_data): Simplify 'exit data'
'finalize' handling.

From-SVN: r279531

[PR92848] [OpenACC] Use 'GOMP_MAP_VARS_ENTER_DATA' for dynamic data lifetimes

libgomp/
PR libgomp/92848
* oacc-mem.c (acc_map_data, present_create_copy)
(goacc_insert_pointer): Use 'GOMP_MAP_VARS_ENTER_DATA'.
(acc_unmap_data, delete_copyout, goacc_remove_pointer): Adjust.
* testsuite/libgomp.oacc-c-c++-common/lib-50.c: Remove.
* testsuite/libgomp.oacc-c-c++-common/pr92848-1-d-a.c: New file
* testsuite/libgomp.oacc-c-c++-common/pr92848-1-d-p.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/pr92848-1-r-a.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/pr92848-1-r-p.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/subset-subarray-mappings-1-r-p.c:
Remove "XFAIL"s.

From-SVN: r279530

Make 'libgomp/target.c:gomp_unmap_tgt' 'static' again

This got changed to 'attribute_hidden' in r271128, but it's not actually used
outside of 'libgomp/target.c'.

libgomp/
* target.c (gomp_unmap_tgt): Make it 'static'.
* libgomp.h (gomp_unmap_tgt): Remove.

From-SVN: r279529

PR 86416 – improve lto1 diagnostic if a mode does not exist

        PR middle-end/86416
        *  Makefile.in (CFLAGS-lto-streamer-in.o): Pass target_noncanonical on.
        * lto-streamer-in.c (lto_input_mode_table): Improve unsupported-mode
        diagnostic.

        PR middle-end/86416
        * testsuite/libgomp.c/pr86416-1.c: New.
        * testsuite/libgomp.c/pr86416-2.c: New.

From-SVN: r279528

re PR fortran/70853 (ICE on pointing to null, in gfc_add_block_to_block, at fortran/trans.c:1599)

2019-12-18  Harald Anlauf  <anlauf@gmx.de>

PR fortran/70853
* trans-expr.c (gfc_trans_pointer_assignment): Reject bounds
remapping if pointer target is NULL().

PR fortran/70853
* gfortran.dg/pr70853.f90: New test.

From-SVN: r279527

[AArch64] Fixup core tunings

Several tuning settings in cores.def are not consistent.
Set the tuning for Cortex-A76AE and Cortex-A77 to neoversen1 so
it is the same as for Cortex-A76 and Neoverse N1.
Set the tuning for Neoverse E1 to cortexa73 so it's the same as for
Cortex-A65. Set the scheduler for Cortex-A65 and Cortex-A65AE to
cortexa53.

    gcc/
* config/aarch64/aarch64-cores.def:
("cortex-a76ae"): Use neoversen1 tuning.
("cortex-a77"): Likewise.
("cortex-a65"): Use cortexa53 scheduler.
("cortex-a65ae"): Likewise.
("neoverse-e1"): Use cortexa73 tuning.

From-SVN: r279526

IPA-CP: Remove bogus static keyword (PR 92971)

2019-12-18  Martin Jambor  <mjambor@suse.cz>

PR ipa/92971
* ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Fix
          definition of values, release memory on exit.

testsuite/
* gcc.dg/ipa/ipcp-agg-12.c: New test.

From-SVN: r279525

* config/avr/avr-mcus.def: Typo.

From-SVN: r279524

ipa-param-manipulation.h (get_original_index): Declare.

* ipa-param-manipulation.h (get_original_index): Declare.
* ipa-param-manipulation.c (ipa_param_adjustments::get_original_index):
New member function.
* ipa-prop.c (ipcp_get_parm_bits): New function.
* ipa-prop.h (ipcp_get_parm_bits): Declare.
* tree-ssa-ccp.c: Include cgraph.h, alloc-pool.h, symbol-summary.h,
ipa-utils.h and ipa-prop.h
(get_default_value): Use ipcp_get_parm_bits.

* gcc.dg/ipa/ipa-bit-cp.c: New testcase.
* gcc.dg/ipa/ipa-bit-cp-1.c: New testcase.
* gcc.dg/ipa/ipa-bit-cp-2.c: New testcase.

Co-Authored-By: Martin Jambor <mjambor@suse.cz>
From-SVN: r279523

PR c++/12333 - X::~X() with implicit this->.

this->X::~X() is handled by finish_class_member_access_expr and its
lookup_destructor subroutine; let's use it in cp_parser_lookup_name for the
case where this-> is implicit.

I tried replacing the other destructor code here with just the call to
lookup_destructor, but that regressed handling of naming the destructor
outside a non-static member function.

* parser.c (cp_parser_lookup_name): Use lookup_destructor.
* typeck.c (lookup_destructor): No longer static.

From-SVN: r279522

Fix vect/pr65947-8.c testcase for amdgcn.

2019-12-18  Andrew Stubbs  <ams@codesourcery.com>

gcc/testsuite/
* gcc.dg/vect/pr65947-8.c: Change pass conditions for amdgcn.

From-SVN: r279521

re PR lto/92972 (gcc/lto-wrapper.c:443: identical branches ?)

PR lto/92972
* lto-wrapper.c (merge_and_complain): Use just "-fno-pie" instead of
big ? "-fno-pie" : "-fno-pie".  Formatting fixes.  Fix comment typo.

From-SVN: r279520

trans.c (Pragma_to_gnu): Push a diagnostics state for pragma Warnings (Off) before turning off all the...

* gcc-interface/trans.c (Pragma_to_gnu) <Pragma_Warnings>: Push a
diagnostics state for pragma Warnings (Off) before turning off all
the warnings and only pop it for pragma Warnings (On).

From-SVN: r279519

[Ada] Missing accessibility check on access discriminants

2019-12-18  Justin Squirek  <squirek@adacore.com>

gcc/ada/

* sem_ch6.adb (Analyze_Function_Return): Modify handling of
extended return statements to check accessibility of access
discriminants.
(Check_Aggregate_Accessibility): Removed.
(Check_Return_Obj_Accessibility): Added to centralize checking
of return aggregates and subtype indications in the case of an
extended return statement.

From-SVN: r279518

[Ada] Fix uninitialized out parameter in s-regpat.adb

2019-12-18  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* libgnat/s-regpat.adb (Parse_Literal, Parse_Piece): Ensure
Expr_Flags is always fully initialized.

From-SVN: r279517

[Ada] Atomic aspect on formal generic params now supported in Ada 202x mode

2019-12-18  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* libgnat/s-atopar.ads, libgnat/s-atopex.ads (Atomic_Type): Can
now be marked Atomic. This requires marking the unit Ada 202x.

From-SVN: r279516

[Ada] Simplify Big_Integer and Big_Real interface

2019-12-18  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* libgnat/a-nbnbin.ads, libgnat/a-nbnbin.adb,
libgnat/a-nbnbre.ads, libgnat/a-nbnbre.adb: Replace
Optional_Big_* types by a simple check and exception raise in
Get_Bignum.
(Set_Bignum): Arg should be 'out' and not 'in out'.
(Invalid_Big_Integer, No_Big_Real): Removed.
(Is_Valid): Now convention Intrinsic.

From-SVN: r279515

[Ada] Fix three-letter typos like "sss" in comments and docs

2019-12-18  Piotr Trojanek  <trojanek@adacore.com>

gcc/ada/

* doc/gnat_rm/implementation_defined_pragmas.rst,
doc/gnat_rm/obsolescent_features.rst,
doc/gnat_ugn/gnat_and_program_execution.rst, exp_attr.adb,
exp_ch9.adb, init.c, libgnat/s-valrea.adb, par-ch6.adb,
sem_attr.adb, sem_ch4.adb, sem_util.ads: Fix trivial typos.
* gnat_rm.texi, gnat_ugn.texi: Regenerate.

From-SVN: r279514

[Ada] Missing accessibility actuals on calls to interface conversion functions

2019-12-18  Gary Dismukes  <dismukes@adacore.com>

gcc/ada/

* sem_res.adb (Resolve_Type_Conversion): Add handling for access
types with designated operand and target types that are
referenced in places that have a limited view of an interface
type by retrieving the nonlimited view when it exists.  Add ???
comments related to missing limited_with_clause handling for
Target (in the non-access case).

From-SVN: r279513

[Ada] AI12-0282: shared variable control aspects on formal types

2019-12-18  Ed Schonberg  <schonberg@adacore.com>

gcc/ada/

* par-ch12.adb (P_Formal_Derived_Type_Definition): In Ada_2020
the keyword WITH can indicate the start of aspect specifications
and not a private type extension.
* sem_ch12.adb (Analyze_Formal_Type): Indicate that it is a
first subtype.
(Instantiate_Type): New procedure
Check_Shared_Variable_Control_Aspects to verify matching rules
between formal and actual types. Note that an array type with
aspect Atomic_Components is considered compatible with an array
type whose component type is Atomic, even though the array types
do not carry the same aspect.
* sem_ch13.adb (Analyze_One_Aspect): Allow shared variable
control aspects to appear on formal types.
(Rep_Item_Too_Early): Exclude aspects on formal types.
* sem_prag.adb (Mark_Type): Handle properly pragmas that come
from aspects on formal types.
(Analyze_Pragma, case Atomic_Components): Handle formal types.

From-SVN: r279512

[Ada] Minor housekeeping work in Create_Standard

2019-12-18  Eric Botcazou  <ebotcazou@adacore.com>

gcc/ada/

* cstand.adb (Create_Standard): Remove duplicate line and
adjust.

From-SVN: r279511

[Ada] Reserving switch d_K for known problem isssues detection

2019-12-18  Javier Miranda  <miranda@adacore.com>

gcc/ada/

* debug.adb: Document -gnatd_K as a reserved switch for the
detection of known problem issues of previous releases.

From-SVN: r279510

[Ada] Reject aspect specifications on number constants

2019-12-18  Ghjuvan Lacambre  <lacambre@adacore.com>

gcc/ada/

* par-ch13.adb: Check if declarations allow aspect
specifications.

From-SVN: r279509

[Ada] Einfo: fix typo in comment

2019-12-18  Piotr Trojanek  <trojanek@adacore.com>

gcc/ada/

* einfo.ads (Is_Ghost_Entity): Fix typo in comment.

From-SVN: r279508

[Ada] Do not propagate Object_Size onto Size for composite types

2019-12-18  Eric Botcazou  <ebotcazou@adacore.com>

gcc/ada/

* layout.adb (Layout_Type): In the case of composite types, do
not copy the Esize onto the RM_Size if the latter is not set.
* gcc-interface/decl.c (gnat_to_gnu_entity) <E_Record_Type>:
Also cap the alignment if an Object_Size clause has been
specified.  Pass VAR_DECL in the call to validate_size for the
Esize of a type.
(validate_size): Be prepared to give an error on an Object_Size
clause.

From-SVN: r279507

[Ada] Document the introduction of the Object_Size attribute in Ada 2020

2019-12-18  Eric Botcazou  <ebotcazou@adacore.com>

gcc/ada/

* einfo.ads (Handling of Type'Size Value): Add references to the
introduction of Object_Size in Ada 2020.
* sem_eval.adb (Subtypes_Statically_Match): Likewise.

From-SVN: r279506

[Ada] Wrong error on hidden must-override primitive

2019-12-18  Bob Duff  <duff@adacore.com>

gcc/ada/

* sem_ch3.adb (Derive_Subprogram): Do not set the
Requires_Overriding flag in the above-mentioned case.

From-SVN: r279505

[Ada] Bad "already use-visible" warning re: use in private part

2019-12-18  Bob Duff  <duff@adacore.com>

gcc/ada/

* sem_ch8.adb (Note_Redundant_Use): It was already checking for
a use clause in the visible part of the child. Add an additional
check for a use clause in the context clause of the child.

From-SVN: r279504

Daily bump.

From-SVN: r279483

PR c++/61339 - add warning for mismatch between struct and class

gcc/c-family/ChangeLog:

PR c++/61339
* c.opt (-Wmismatched-tags, -Wredundant-tags): New options.

gcc/cp/ChangeLog:

PR c++/61339
* parser.c (cp_parser_maybe_warn_enum_key): New function.
(class_decl_loc_t): New class.
(cp_parser_elaborated_type_specifier): Call
cp_parser_maybe_warn_enum_key.
(cp_parser_class_head): Call cp_parser_check_class_key.
(cp_parser_check_class_key): Add arguments.  Call class_decl_loc_t::add.
(c_parse_file): Call class_decl_loc_t::diag_mismatched_tags.

gcc/testsuite/ChangeLog:

PR c++/61339
* g++.dg/warn/Wmismatched-tags.C: New test.
* g++.dg/warn/Wredundant-tags.C: New test.
* g++.dg/pch/Wmismatched-tags.C: New test.
* g++.dg/pch/Wmismatched-tags.Hs: New test header.

gcc/ChangeLog:

PR c++/61339
* doc/invoke.texi (-Wmismatched-tags, -Wredundant-tags): Document
new C++ options.

From-SVN: r279480

Generate PADDI to add large constants if -mcpu=future.

2019-12-12  Michael Meissner  <meissner@linux.ibm.com>

* config/rs6000/predicates.md (add_operand): Allow eI constants.
* config/rs6000/rs6000.md (add<mode>3): Add alternative to
generate PADDI for 34-bit constants if -mcpu=future.

From-SVN: r279476

Use PLI to load up 32-bit SImode constants if -mcpu=future.

2019-12-17  Michael Meissner  <meissner@linux.ibm.com>

* config/rs6000/rs6000.md (movsi_internal1): Add alternative to
use PLI to load up 32-bit constants if -mcpu=future.

From-SVN: r279475

Use PLI to load up large constants if -mcpu=future.

2019-12-17  Michael Meissner  <meissner@linux.ibm.com>

* config/rs6000/rs6000.c (num_insns_constant_gpr): Return 1 if the
constant can be loaded with PLI if -mcpu=future.
* config/rs6000/rs6000.md (movdi_internal64): Add alternative to
use PLI to load up 34-bit constants if -mcpu=future.

From-SVN: r279474

PR c++/79592 - missing explanation of invalid constexpr.

We changed months back to use the pre-generic form for constexpr evaluation,
but explain_invalid_constexpr_fn was still using DECL_SAVED_TREE.  This
mostly works, but misses some issues due to folding.  So with this patch we
save the pre-generic form of constexpr functions even when we know they
can't produce a constant result.

* constexpr.c (register_constexpr_fundef): Do store the body of a
template instantiation that is not potentially constant.
(explain_invalid_constexpr_fn): Look it up.
(cxx_eval_call_expression): Check fundef->result.

From-SVN: r279473

PR c++/92576 - redeclaration of variable template.

The variable templates patch way back when forgot to add handling here.  The
simplest answer seems to be recursing to the underlying declaration.

* decl.c (redeclaration_error_message): Recurse for variable
templates.

From-SVN: r279472

* name-lookup.c (get_std_name_hint): Add std::byte.

From-SVN: r279471

re PR c++/59655 (incorrect diagnostic on templatized function with lambda parameter)

PR c++/59655
* pt.c (push_tinst_level_loc): If limit_bad_template_recursion,
set TREE_NO_WARNING on tldcl.
* decl2.c (no_linkage_error): Treat templates with TREE_NO_WARNING
as defined during error recovery.

* g++.dg/cpp0x/diag3.C: New test.

From-SVN: r279470

re PR target/92841 (Optimize -fstack-protector-strong code generation a bit)

PR target/92841
* config/i386/i386.md (@stack_protect_set_1_<mode>,
@stack_protect_test_1_<mode>): Use output_asm_insn.
(*stack_protect_set_2_<mode>, *stack_protect_set_3): New define_insns
and corresponding define_peephole2s.

* gcc.target/i386/pr92841.c: New test.

From-SVN: r279468

Revert "Fix vector testcases for amdgcn."

Apologies everyone. :-(

From-SVN: r279466

Fix vector testcases for amdgcn.

2019-12-17  Andrew Stubbs  <ams@codesourcery.com>

gcc/testsuite/
* gcc.dg/vect/pr65947-8.c: Change pass conditions for amdgcn.
* gcc.dg/vect/vect-multitypes-11.c: Ensure that main isn't vectorized.
* gcc.dg/vect/vect-multitypes-12.c: Likewise.

From-SVN: r279465

symtab.c (symtab_node::get_partitioning_class): Aliases of external symbols are external.

* symtab.c (symtab_node::get_partitioning_class): Aliases of external
symbols are external.

From-SVN: r279464

[ARM] Add support for -mpure-code in thumb-1 (v6m)

This patch extends support for -mpure-code to all thumb-1 processors,
by removing the need for MOVT.

Symbol addresses are built using upper8_15, upper0_7, lower8_15 and
lower0_7 relocations, and constants are built using sequences of
movs/adds and lsls instructions.

The extension of the *thumb1_movhf pattern uses always the same size
(6) although it can emit a shorter sequence when possible. This is
similar to what *arm32_movhf already does.

CASE_VECTOR_PC_RELATIVE is now false with -mpure-code, to avoid
generating invalid assembly code with differences from symbols from
two different sections (the difference cannot be computed by the
assembler).

Tests pr45701-[12].c needed a small adjustment to avoid matching
upper8_15 when looking for the r8 register.

Test no-literal-pool.c is augmented with __fp16, so it now uses
-mfp16-format=ieee.

Test thumb1-Os-mult.c generates an inline code sequence with
-mpure-code and computes the multiplication by using a sequence of
add/shift rather than using the multiply instruction, so we skip it in
presence of -mpure-code.

With -mcpu=cortex-m0, the pure-code/no-literal-pool.c fails because
code like:
static char *p = "Hello World";
char *
testchar ()
{
  return p + 4;
}

generates 2 indirections (I removed non-essential directives/code)
          .section        .rodata
  .LC0:
  .ascii  "Hello World\000"
  .data
  p:
  .word   .LC0
  .section        .rodata
  .LC2:
  .word   p
  .section .text,"0x20000006",%progbits
  testchar:
  push    {r7, lr}
  add     r7, sp, #0
  movs    r3, #:upper8_15:#.LC2
  lsls    r3, #8
  adds    r3, #:upper0_7:#.LC2
  lsls    r3, #8
  adds    r3, #:lower8_15:#.LC2
  lsls    r3, #8
  adds    r3, #:lower0_7:#.LC2
  ldr     r3, [r3]
  ldr     r3, [r3]
  adds    r3, r3, #4
  movs    r0, r3
  mov     sp, r7
  @ sp needed
  pop     {r7, pc}

By contrast, when using -mcpu=cortex-m4, the code looks like:
        .section        .rodata
.LC0:
.ascii  "Hello World\000"
.data
p:
.word   .LC0
testchar:
push    {r7}
add     r7, sp, #0
movw    r3, #:lower16:p
movt    r3, #:upper16:p
ldr     r3, [r3]
adds    r3, r3, #4
mov     r0, r3
mov     sp, r7
pop     {r7}
bx      lr

I haven't found yet how to make code for cortex-m0 apply upper/lower
relocations to "p" instead of .LC2. The current code looks functional,
but could be improved.

2019-10-18  Christophe Lyon  <christophe.lyon@linaro.org>

gcc/
* config/arm/arm-protos.h (thumb1_gen_const_int): Add new prototype.
* config/arm/arm.c (arm_option_check_internal): Remove restriction
on MOVT for -mpure-code.
(thumb1_gen_const_int): New function.
(thumb1_legitimate_address_p): Support -mpure-code.
(thumb1_rtx_costs): Likewise.
(thumb1_size_rtx_costs): Likewise.
(arm_thumb1_mi_thunk): Likewise.
* config/arm/arm.h (CASE_VECTOR_PC_RELATIVE): Likewise.
* config/arm/thumb1.md (thumb1_movsi_symbol_ref): New.
(*thumb1_movhf): Support -mpure-code.

gcc/testsuite/
* gcc.target/arm/pr45701-1.c: Adjust for -mpure-code.
* gcc.target/arm/pr45701-2.c: Likewise.
* gcc.target/arm/pure-code/no-literal-pool.c: Add tests for
__fp16.
* gcc.target/arm/pure-code/pure-code.exp: Remove thumb2 and movt
conditions.
* gcc.target/arm/thumb1-Os-mult.c: Skip if -mpure-code is used.

From-SVN: r279463

Add myself to write after approval.

2019-12-17  Mihail Ionescu  <mihail.ionescu@arm.com>

* MAINTAINERS (write_after_approval): Add myself.

From-SVN: r279461

Add pointer to PR92772

2019-12-17  Andrew Stubbs  <ams@codesourcery.com>

* tree-vect-loop.c (vect_create_epilog_for_reduction): Mention pr92772
in the comments.

From-SVN: r279460

Add extract_last for amdgcn

2019-12-17  Andrew Stubbs  <ams@codesourcery.com>

gcc/
* config/gcn/gcn-valu.md (extract_last_<mode>): New expander.
(fold_extract_last_<mode>): New expander.

gcc/testsuite/
* lib/target-supports.exp
(check_effective_target_vect_fold_extract_last): Add amdgcn.

From-SVN: r279459

Add clz and ctz for amdgcn

2019-12-17  Andrew Stubbs  <ams@codesourcery.com>

gcc/
* config/gcn/gcn.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
(CTZ_DEFINED_VALUE_AT_ZERO): Define.
* config/gcn/gcn.md (s_mnemonic): Add clz and ctz.
(expander): Likewise.
(countzeros): New code iterator.
(<expander>si2): New insn pattern.
(<expander>di2): New insn pattern.

From-SVN: r279458

libgomp/openacc.f90 – clean-up public/private attributes

        * config/accel/openacc.f90 (module openacc_kinds): Use 'PUBLIC' to mark
        all symbols as public except for the 'use …, only' imported symbol,
        which is private.
        (module openacc): Default to 'PRIVATE' to exclude openacc_internal; mark
        all symbols from module openacc_kinds as PUBLIC
        * openacc.f90: Add comment with crossref to that file and openmp_lib.h;
        fix comment typo.
        * openacc_lib.h (acc_device_gcn): Add this PARAMETER.

From-SVN: r279456

re PR target/92962 (Documentation: x86 Options - znver2 missing RDPID and WBNOINVD)

PR target/92962
* common/config/i386/i386-common.c (processor_alias_table): Formatting
fixes.
* doc/invoke.texi (bdver3, bdver4, znver1): Add missing closing paren.
(znver2): Likewise.  Add RDPID and WBNOINVD, remove spurious comma
before CLWB.

From-SVN: r279455

Add abs pattern to handle {si,di} mode abs to avoid pmax/cmove conversion.

2019-12-17  Hongyu Wang  <hongyu.wang@intel.com>

gcc/
PR target/92651
* config/i386/i386.h (TARGET_EXPAND_ABS): New macro.
* config/i386/x86-tune.def (X86_TUNE_EXPAND_ABS): New.
* config/i386/i386.md (abs<SWI48x>2): New define_expand.

gcc/testsuite
* gcc.target/i386/pr92651.c: New testcase.

From-SVN: r279452