Implement new serial algorithms from Parallelism TS (P0024R2)

These new (non-parallel) algorithms were added to C++17 along with the
parallel algorithms, but were missing from libstdc++.

* include/bits/algorithmfwd.h: Change title of doc group.
* include/bits/stl_algo.h (for_each_n): Add new C++17 algorithm from
P0024R2.
* include/bits/stl_numeric.h: Define doc group and add algos to it.
* include/std/numeric (__is_random_access_iter): New internal trait.
(reduce, transform_reduce, exclusive_scan, inclusive_scan)
(transform_exclusive_scan, transform_inclusive_scan): Likewise.
* testsuite/25_algorithms/for_each/for_each_n.cc: New test.
* testsuite/26_numerics/exclusive_scan/1.cc: New test.
* testsuite/26_numerics/inclusive_scan/1.cc: New test.
* testsuite/26_numerics/reduce/1.cc: New test.
* testsuite/26_numerics/transform_exclusive_scan/1.cc: New test.
* testsuite/26_numerics/transform_inclusive_scan/1.cc: New test.
* testsuite/26_numerics/transform_reduce/1.cc: New test.
* testsuite/util/testsuite_iterators.h (test_container::size()): New
member function.

From-SVN: r272459

xtensa: fix PR target/90922

Stack pointer adjustment code in prologue missed a case of no
callee-saved registers and a stack frame size bigger than 128 bytes.
Handle that case.

This fixes the following gcc tests with call0 ABI:
  gcc.c-torture/execute/stdarg-2.c
  gcc.dg/torture/pr55882.c
  gcc.dg/torture/pr57569.c

2019-06-18  Max Filippov  <jcmvbkbc@gmail.com>
gcc/
* config/xtensa/xtensa.c (xtensa_expand_prologue): Add stack
pointer adjustment for the case of no callee-saved registers and
stack frame bigger than 128 bytes.

From-SVN: r272455

[PR90921] Fortran OpenACC 'declare' directive's module handling causes duplicate data clauses

gcc/fortran/
PR fortran/90921
* trans-decl.c (finish_oacc_declare): Reset module_oacc_clauses
before scanning each namespace.
gcc/testsuite/
PR fortran/90921
* gfortran.dg/goacc/declare-3.f95: Update.

From-SVN: r272454

[PR85221] Set 'omp declare target', 'omp declare target link' attributes for Fortran OpenACC 'declare'd variables

gcc/fortran/
PR fortran/85221
* trans-decl.c (add_attributes_to_decl): Handle OpenACC 'declare'
directive.
gcc/testsuite/
PR fortran/85221
* gfortran.dg/goacc/declare-3.f95: New file.

From-SVN: r272453

[PR90859] Document status quo for "[OMP] Mappings for VLA different depending on 'target { c && { ! lp64 } }'"

gcc/testsuite/
PR middle-end/90859
* c-c++-common/goacc/firstprivate-mappings-1.c: Update.

From-SVN: r272452

Test cases to verify OpenACC 'firstprivate' mappings

gcc/testsuite/
* c-c++-common/goacc/firstprivate-mappings-1.c: New file.
* g++.dg/goacc/firstprivate-mappings-1.C: Likewise.
libgomp/
* testsuite/libgomp.oacc-c++/firstprivate-mappings-1.C: New file.
* testsuite/libgomp.oacc-c-c++-common/firstprivate-mappings-1.c:
Likewise.

From-SVN: r272451

Fix description of 'GOMP_MAP_FIRSTPRIVATE'

..., which got garbled in r230275.

include/
* gomp-constants.h (enum gomp_map_kind): Fix description of
'GOMP_MAP_FIRSTPRIVATE'.

From-SVN: r272450

Add missing results check in 'libgomp.fortran/allocatable3.f90'

libgomp/
* testsuite/libgomp.fortran/allocatable3.f90: Add missing results
check.

From-SVN: r272449

Add 'libgomp.oacc-fortran/allocatable-array-1.f90'

libgomp/
* testsuite/libgomp.oacc-fortran/allocatable-array-1.f90: New
file.

From-SVN: r272448

[PR90743] Fortran 'allocatable' with OpenACC data/OpenMP 'target' 'map' clauses

Test what OpenMP 5.0 has to say on this topic.  And, do the same for OpenACC.

libgomp/
PR fortran/90743
* oacc-parallel.c (GOACC_parallel_keyed): Handle NULL mapping
case.
* testsuite/libgomp.fortran/target-allocatable-1-1.f90: New file.
* testsuite/libgomp.fortran/target-allocatable-1-2.f90: Likewise.
* testsuite/libgomp.oacc-fortran/allocatable-1-1.f90: Likewise.
* testsuite/libgomp.oacc-fortran/allocatable-1-2.f90: Likewise.

From-SVN: r272447

[PR90861] Document status quo for OpenACC 'declare' not cleaning up for VLAs

gcc/testsuite/
PR testsuite/90861
* c-c++-common/goacc/declare-pr90861.c: New file.
libgomp/
PR testsuite/90861
* testsuite/libgomp.oacc-c-c++-common/declare-vla.c: Update.

From-SVN: r272446

[PR90868] Document status quo for duplicate OpenACC 'declare' directives for 'extern' variables

gcc/testsuite/
PR testsuite/90868
* c-c++-common/goacc/declare-1.c: Update.
* c-c++-common/goacc/declare-2.c: Likewise.

From-SVN: r272445

[PR90862] OpenACC 'declare' ICE when nested inside another construct

gcc/
PR middle-end/90862
* omp-low.c (check_omp_nesting_restrictions): Handle
GF_OMP_TARGET_KIND_OACC_DECLARE.
gcc/testsuite/
PR middle-end/90862
* c-c++-common/goacc/declare-1.c: Update.
* c-c++-common/goacc/declare-2.c: Likewise.
libgomp/
PR middle-end/90862
* testsuite/libgomp.oacc-c-c++-common/declare-1.c: Update.

From-SVN: r272444

re PR c++/84698 (ICE when using noexcept(noexcept()) declaration on global friend function of template class)

PR c++/84698
* g++.dg/cpp0x/noexcept42.C: New test.

From-SVN: r272443

re PR c++/71548 (Invalid declaration involving template template param causes crash)

PR c++/71548
* g++.dg/cpp0x/variadic177.C: New test.

From-SVN: r272438

i386.md (@cmp<mode>_1): Rename from cmp<mode>_1.

* config/i386/i386.md (@cmp<mode>_1): Rename from cmp<mode>_1.
(@add<mode>3_carry): Rename from add<mode>3_carry.
(@sub<mode>3_carry_ccc): Rename from sub<mode>3_carry_ccc.
(@sub<mode>3_carry_ccgz): Rename form sub<mode>3_carry_ccgz.
(@copysign<mode>3_const): Rename from copysign<mode>3_const.
(@copysign<mode>3_var): Rename from copysign<mode>3_var.
(@xorsign<mode>3_1): Rename from xorsign<mode>3_1.
(@x86_shift<mode>_adj_1): Rename from x86_shift<mode>_adj_1.
(@x86_shift<mode>_adj_2): Rename from x86_shift<mode>_adj_2.
(@x86_shift<mode>_adj_3): Rename from x86_shift<mode>_adj_3.
(cmpstrnsi): Use gen_cmp_1.
(lwp_slwpcb): Use gen_lwp_slwpcb_1.
(@lwp_slwpcb<mode>_1): Rename from lwp_slwpcb<mode>_1.
(@umonitor_<mode>): Rename from umonitor_<mode>.
* config/i386/i386-expand.c (ix86_expand_copysign):
Use gen_copysign3_const and gen_copysign3_var.
(ix86_expand_xorsign): Use gen_xorsign3_1.
(ix86_expand_branch): Use gen_sub3_carry_ccc,
gen_sub3_carry_ccgz and gen_cmp1.
(ix86_expand_int_addcc): Use gen_sub3_carry and gen_add3_carry.
(ix86_split_ashl): Use gen_x86_shift_adj_1 and gen_x86_shift_adj_2.
(ix86_split_ashr): Use gen_x86_shift_adj_1 and gen_x86_shift_adj_3.
(ix86_split_lshr): Ditto.
(ix86_expand_builtin) <case IX86_BUILTIN_UMONITOR>: Use gen_umonitor.

From-SVN: r272432

* constexpr.c (cxx_eval_store_expression): Delay target evaluation.

From-SVN: r272431

Consolidate constexpr array handling.

* constexpr.c (eval_and_check_array_index): Split out from...
(cxx_eval_array_reference): ...here.
(cxx_eval_store_expression): Use it here, too.
(diag_array_subscript): Take location.  Strip location wrapper.

From-SVN: r272430

Handle constexpr conversion from and then to the same type.

* constexpr.c (cxx_eval_constant_expression): Handle conversion from
and then to the same type.

From-SVN: r272429

tree.c (build_constructor): Add MEM_STAT_DECL.

* tree.c (build_constructor): Add MEM_STAT_DECL.

gcc/cp/
* constexpr.c (unshare_constructor): Add MEM_STAT_DECL.

From-SVN: r272428

[AArch64] Add a new CC mode for SVE conditions

The SVE ACLE patches need to introduce a new CC_NZC mode for the
conditions that can be tested after a PTRUE.  In particular, LT needs
to map to "mi"/"first" and GE to "pl"/"nfrst", instead of the normal
CC mapping.

Another advantage of using a separate mode is that we can print the SVE
names of the conditions, which makes the output a bit easier to read.
It therefore seems like an independent improvement that can go in now.

The patch also avoids using (compare X (const_int 0)), because that gets
folded away when used with LTU and GEU ("cc"/"last" and "cs"/"nlast").
Just using an unspec should be OK.

The full set of conditions can't be tested without other SVE ACLE patches.

2019-06-18  Richard Sandiford  <richard.sandiford@arm.com>

gcc/
* config/aarch64/aarch64-modes.def (CC_NZC): New CC_MODE.
* config/aarch64/aarch64-sve.md (*<optab><mode>3_cc)
(ptest_ptrue<mode>, while_ult<GPI:mode><PRED_ALL:mode>)
(*while_ult<GPI:mode><PRED_ALL:mode>_cc, *cmp<cmp_op><mode>)
(*cmp<cmp_op><mode>_ptest, *cmp<cmp_op><mode>_cc)
(*pred_cmp<cmp_op><mode>_combine, *pred_cmp<cmp_op><mode>)
(vec_cmp<mode><vpred>, vec_cmpu<mode><vpred>, cbranch<mode>4):
Use CC_NZC instead of CC.
* config/aarch64/aarch64.md (condjump): Print a '.' in SVE conditions.
* config/aarch64/aarch64.c (aarch64_sve_condition_codes): New variable.
(aarch64_print_operand): Handle E_CC_NZCmode.
(aarch64_emit_sve_ptrue_op_cc): Use gen_set_clobber_cc_nzc instead
of gen_set_clobber_cc.

gcc/testsuite/
* gcc.target/aarch64/sve/struct_vect_18.c: Allow branches to
contain dots.
* gcc.target/aarch64/sve/struct_vect_19.c: Likewise.
* gcc.target/aarch64/sve/struct_vect_20.c: Likewise.
* gcc.target/aarch64/sve/struct_vect_21.c: Likewise.
* gcc.target/aarch64/sve/struct_vect_22.c: Likewise.
* gcc.target/aarch64/sve/struct_vect_23.c: Likewise.
* gcc.target/aarch64/sve/unroll-1.c: Likewise.
* gcc.target/aarch64/sve/while_1.c: Check for b.any.

From-SVN: r272427

[AArch64] Tabify aarch64-sve.md

2019-06-18  Richard Sandiford  <richard.sandiford@arm.com>

gcc/
* config/aarch64/aarch64-sve.md: Tabify file.

From-SVN: r272426

[AArch64] Factor out pfalse predicate creation

Following on from the previous ptrue patch.

2019-06-18  Richard Sandiford  <richard.sandiford@arm.com>

gcc/
* config/aarch64/aarch64-protos.h (aarch64_pfalse_reg): Declare.
* config/aarch64/aarch64.c (aarch64_pfalse_reg): New function.
* config/aarch64/aarch64-sve.md: Use it.

From-SVN: r272425

[AArch64] Factor out ptrue predicate creation

This is the first step to canonicalising predicate constants so that
they can be reused between modes.

2019-06-18  Richard Sandiford  <richard.sandiford@arm.com>

gcc/
* config/aarch64/aarch64-protos.h (aarch64_ptrue_reg): Declare.
* config/aarch64/aarch64.c (aarch64_ptrue_reg): New functions.
(aarch64_expand_sve_widened_duplicate, aarch64_expand_sve_mem_move)
(aarch64_maybe_expand_sve_subreg_move, aarch64_evpc_rev_local)
(aarch64_expand_sve_vec_cmp_int): Use it.
(aarch64_expand_sve_vec_cmp_float): Likewise.
* config/aarch64/aarch64-sve.md: Likewise throughout.

From-SVN: r272424

[AArch64] Simplify SVE IFN_COND patterns

This patch makes the binary IFN_COND patterns use the same approach
as the ternary patterns, with one pattern handling the cases in
which the "else" value isn't tied to one of the other inputs.

2019-06-18  Richard Sandiford  <richard.sandiford@arm.com>
    Kugan Vivekanandarajah  <kuganv@linaro.org>

gcc/
* config/aarch64/aarch64-sve.md (*cond_<optab><mode>_0): Delete.
(*cond_<optab><mode>_z): Fold into...
(*cond_<optab><mode>_any): ...here.  Also handle cases in which
operand 4 can be tied to operand 0 (either inherently or via RA).

From-SVN: r272423

pr81563.c (dg-final): Check that no registers are restored from %esp.

* gcc.target/i386/pr81563.c (dg-final): Check that no
registers are restored from %esp.

From-SVN: r272422

re PR debug/90900 (ICE in copy_rtx, at rtl.c:376)

2019-06-18  Richard Biener  <rguenther@suse.de>

PR debug/90900
* cfgexpand.c (expand_debug_expr): Treat NOTE_P DECL_RTL
as if optimized away.

* gcc.dg/gomp/pr90900.c: New testcase.

From-SVN: r272421

pr90866-2.c: Remove a pointless declaration to avoid compilation errors on arm-none-eabi.

gcc/testsuite/ChangeLog:

* gcc.dg/pr90866-2.c: Remove a pointless declaration
to avoid compilation errors on arm-none-eabi.

From-SVN: r272418

re PR ada/80590 (non-bootstrap build failure of Ada runtime)

PR ada/80590

* sem_ch5.adb (Analyze_Loop_Statement): Avoid exception propagation
during normal processing.

From-SVN: r272417

Avoid undefined behaviour in std::byte operators (LWG 2950)

* include/c_global/cstddef (std::byte): Perform arithmetic operations
in unsigned int to avoid promotion (LWG 2950).

From-SVN: r272415

aarch64: fix gcc.target/aarch64/pcs_attribute-2.c on non-gnu targets

Move the ifunc symbol tests into a separate file with dg-require-ifunc.
And added a base pcs ifunc symbol to the test for completeness.

gcc/testsuite/ChangeLog:

* gcc.target/aarch64/pcs_attribute-2.c: Remove ifunc usage.
* gcc.target/aarch64/pcs_attribute-3.c: New test.

From-SVN: r272414

[nvptx] Use define_insn parametrization

Parametrize some define_insn to simplify code in define_expands generating
those insns.

Build and reg-tested on x86_64 with nvptx accelerator.

2019-06-18  Tom de Vries  <tdevries@suse.de>

* config/nvptx/nvptx-protos.h (gen_set_softstack_insn): Remove.
* config/nvptx/nvptx.c (gen_set_softstack_insn): Remove.
* config/nvptx/nvptx.md (define_insn "set_softstack_<mode>"): Rename to ...
(define_insn "@set_softstack_<mode>"): ... this.
(define_insn "omp_simt_enter_<mode>"): Rename to ...
(define_insn "@omp_simt_enter_<mode>"): ... this.
(define_insn "omp_simt_exit_<mode>"): Rename to ...
(define_insn "@omp_simt_exit_<mode>"): ... this.

From-SVN: r272413

[nvptx] Fix __main missing prototype warning in crt0.c

Atm we see:
...
libgcc/config/nvptx/crt0.c:36:1: warning: no previous prototype for \
  ‘__main’ [-Wmissing-prototypes]
...

Fix this by adding the prototype.

Build and reg-tested on nvptx.
Build and reg-tested on x86_64 with nvptx accelerator.

2019-06-18  Tom de Vries  <tdevries@suse.de>

* config/nvptx/crt0.c (__main): Declare.

From-SVN: r272412

Restore correct iv step for fully-masked loops

r272233 introduced a large number of execution failures on SVE.
The patch hard-coded an IV step of VF, but for SLP groups it needs
to be VF * group size.

Also, iv_precision had type widest_int but only needs to be unsigned int.

2019-06-18  Richard Sandiford  <richard.sandiford@arm.com>

gcc/
* tree-vect-loop-manip.c (vect_set_loop_masks_directly): Remove
vf parameter.  Restore the previous iv step of nscalars_step,
but give it iv_type rather than compare_type.  Tweak code order
to match the comments.
(vect_set_loop_condition_masked): Update accordingly.
* tree-vect-loop.c (vect_verify_full_masking): Use "unsigned int"
for iv_precision.  Tweak comment formatting.

From-SVN: r272411

Darwin - strip trailing whitespace from a file, NFC.

2019-06-18  Iain Sandoe  <iain@sandoe.co.uk>

        * config/darwin.c: Strip trailing whitespace.

From-SVN: r272410

[Darwin] The need for FDE symbols is dependent on linker used, not OS rev.

For very old toolchains, the compiler generated extra symbols that mark the
start of each FDE.  We no longer need this (since xcode 3 era).

Since we have detection of the linker version, we can use that directly to
determine if support is needed.

2019-06-18  Iain Sandoe  <iain@sandoe.co.uk>

* config/darwin.c (darwin_emit_unwind_label): New default to false.
(darwin_override_options): Set darwin_emit_unwind_label as needed.

From-SVN: r272409

Fix IPA-CP topological sorting

2019-06-18  Martin Jambor  <mjambor@suse.cz>

PR ipa/90889
* ipa-cp.c (ignore_edge_p): Do not ignore edges when only the
caller does not have flag_ipa_cp set.

From-SVN: r272408

[Vectorizer] Support masking fold left reductions

This patch adds support in the vectorizer for masking fold left reductions.
This avoids the need to insert a conditional assignement with some identity
value.

From-SVN: r272407

libsanitizer - remove reference to __cxa_rethrow_primary_exception.

For some Darwin versions the absence of the rethrow_primary_exception
symbol causes almost all sanitizer tests to fail.
The symbol is not present in libstdc++ and, therefore is is correct to
remove the reference to it for all platforms.  We do this by adding a
new guard "ASAN_HAS_CXA_RETHROW_PRIMARY_EXCEPTION".

libsanitizer/

2019-06-18  Iain Sandoe  <iain@sandoe.co.uk>

PR libsanitizer/87880
* asan/asan_interceptors.h:
(ASAN_INTERCEPT___CXA_RETHROW_PRIMARY_EXCEPTION): New.
* asan/Makefile.am (DEFS): Add
ASAN_HAS_CXA_RETHROW_PRIMARY_EXCEPTION, defined to 0.
* asan/Makefile.in: Regenerated.
* asan/libtool-version: Bump version.

From-SVN: r272406

Add predict_doloop_p target hook

    Add one target hook predict_doloop_p, it return true if we can predict it
    is possible to use a low-overhead loop, it can help ivopts to make some
    better decisions.

    PR middle-end/80791
    * target.def (predict_doloop_p): New hook.
    * targhooks.h (default_predict_doloop_p): New declaration.
    * targhooks.c (default_predict_doloop_p): New function.
    * doc/tm.texi.in (TARGET_PREDICT_DOLOOP_P): New hook.
    * doc/tm.texi: Regenerate.
    * config/rs6000/rs6000.c (rs6000_predict_doloop_p): New function.
    (TARGET_PREDICT_DOLOOP_P): New macro.
    * tree-ssa-loop-ivopts.c (generic_predict_doloop_p): New function.

From-SVN: r272405

Daily bump.

From-SVN: r272404

omp-low.c (struct omp_context): Add scan_inclusive field.

* omp-low.c (struct omp_context): Add scan_inclusive field.
(scan_omp_1_stmt) <case GIMPLE_OMP_SCAN>: Set ctx->scan_inclusive
if inclusive scan.
(struct omplow_simd_context): Add lastlane member.
(lower_rec_simd_input_clauses): Add rvar argument, handle inscan
reductions.  Build 2 or 3 argument .GOMP_SIMD_LANE calls rather than
1 or 2 argument.
(lower_rec_input_clauses): Handle inscan reductions in simd contexts.
(lower_lastprivate_clauses): Set TREE_THIS_NOTRAP on the ARRAY_REF.
(lower_omp_scan): New function.
(lower_omp_1) <case GIMPLE_OMP_SCAN>: Use lower_omp_scan.
* tree-ssa-dce.c (eliminate_unnecessary_stmts): For IFN_GOMP_SIMD_LANE
check 3rd argument if present rather than 2nd.
* tree-vectorizer.h (struct _loop_vec_info): Add scan_map member.
(struct _stmt_vec_info): Change simd_lane_access_p from bool into
2-bit bitfield.
* tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
scan_map.  For IFN_GOMP_SIMD_LANE check 3rd argument if present rather
than 2nd.
(_loop_vec_info::~_loop_vec_info): Delete scan_map.
* tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Allow two
different STMT_VINFO_SIMD_LANE_ACCESS_P refs if they have the same
init.
(vect_find_stmt_data_reference): Encode in ->aux the 2nd
IFN_GOMP_SIMD_LANE argument.
(vect_analyze_data_refs): Set STMT_VINFO_SIMD_LANE_ACCESS_P from the
encoded ->aux value.
* tree-vect-stmts.c: Include attribs.h.
(vectorizable_call): Adjust comment about IFN_GOMP_SIMD_LANE.
(scan_operand_equal_p, check_scan_store, vectorizable_scan_store): New
functions.
(vectorizable_load): For STMT_VINFO_SIMD_LANE_ACCESS_P tests use != 0.
(vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P > 1.
cp/
* semantics.c (finish_omp_clauses): For OMP_CLAUSE_REDUCTION_INSCAN
set need_copy_assignment.
testsuite/
* gcc.dg/vect/vect-simd-8.c: New test.
* gcc.dg/vect/vect-simd-9.c: New test.
* g++.dg/vect/simd-2.cc: New test.
* g++.dg/gomp/scan-1.C: New test.

From-SVN: r272399

re PR target/62055 (missed optimization: recognize fnabs (FP negative absolute value) (x86-64))

PR target/62055
* config/i386/i386.md (*nabstf2_1): New insn pattern.
(*nabs<mode>2_1): Ditto.
(nabs sse-reg splitter): New splitter.
* config/i386/sse.md (*nabs<mode>2): New insn_and_split pattern.

testsuite/ChangeLog:

PR target/62055
* gcc.target/i386/fnabs.c: New test.

From-SVN: r272396

PR c++/83820 - excessive attribute arguments not detected.

* parser.c (cp_parser_std_attribute): Detect excessive arguments.

* g++.dg/cpp0x/gen-attrs-67.C: New test.

From-SVN: r272395

Fix AIX test failure due to replacement operator delete

On AIX the sized delete defined in the library will call the non-sized
delete defined in the library, not the replacement version defined in
the test file. By also replacing sized delete we make the test pass
everywhere.

* testsuite/20_util/allocator/1.cc: Add sized delete, which fixes a
failure on AIX.

From-SVN: r272391

re PR bootstrap/90873 (-Wmaybe-uninitialized warning in gcc/tree-ssa-forwprop.c breaks 32-bit bootstrap)

PR bootstrap/90873.
* tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Fix
TMR index check.

From-SVN: r272390

[nvptx] Fix missing mode warnings in nvptx.md, omp part

Fix these warnings:
...
gcc/config/nvptx/nvptx.md:1074:1: warning: operand 0 missing mode?
gcc/config/nvptx/nvptx.md:1240:1: warning: operand 0 missing mode?
gcc/config/nvptx/nvptx.md:1240:1: warning: operand 1 missing mode?
gcc/config/nvptx/nvptx.md:1240:1: warning: operand 2 missing mode?
gcc/config/nvptx/nvptx.md:1268:1: warning: operand 0 missing mode?
...

Build and reg-tested on x86_64 with nvptx accelerator.

2019-06-17  Tom de Vries  <tdevries@suse.de>

* config/nvptx/nvptx-protos.h (gen_set_softstack_insn): Declare.
* config/nvptx/nvptx.c (gen_set_softstack_insn): New function.
* config/nvptx/nvptx.md (define_insn "set_softstack_insn"): Rename to
...
(define_insn "set_softstack_<mode>"): ... this.  Use P iterator on
match_operand 0.
(define_insn "omp_simt_enter_insn"): Rename to ...
(define_insn "omp_simt_enter_<mode>"): ... this.  Use P iterator on
match_operand 0, 1 and 2, as well as the unspec_volatile result.
(define_expand "omp_simt_enter): Use gen_omp_simt_enter_di and
gen_omp_simt_enter_si.
(define_expand "omp_simt_exit"): New.
(define_insn "omp_simt_exit"): Rename to ...
(define_insn "omp_simt_exit_<mode>"): ... this.  Use P iterator on
match_operand 0.

From-SVN: r272388

[netbsd][aarch64] add netbsd/aarch64 target

Initial support for AArch64 running NetBSD.

Matthew Green <mrg@eterna.com.au>
Maya Rashish <coypu@sdf.org>

gcc:
* config.gcc (aarch64*-*-netbsd*): New target.
* config/aarch64/aarch64-netbsd.h: New file.
* config/aarch64/t-aarch64-netbsd: Likewise.

libgcc:
* config.host (aarch64*-*-netbsd*): New case.

Co-Authored-By: Maya Rashish <coypu@sdf.org>
From-SVN: r272387

Add 'noexcept' to std::lerp

* include/c_global/cmath (__lerp, lerp): Add noexcept (LWG 3201).

From-SVN: r272386

PR libstdc++/90281 Fix string conversions for filesystem::path

Fix several bugs in the encoding conversions for filesystem::path that
prevent conversion of Unicode characters outside the Basic Multilingual
Plane, and prevent returning basic_string specializations with
alternative allocator types.

The std::codecvt_utf8 class template is not suitable for UTF-16
conversions because it uses UCS-2 instead. For conversions between UTF-8
and UTF-16 either std::codecvt<C, char, mbstate> or
codecvt_utf8_utf16<C> must be used.

The __str_codecvt_in and __str_codecvt_out utilities do not
return false on a partial conversion (e.g. for invalid or incomplete
Unicode input). Add new helpers that treat partial conversions as
errors, and use them for all filesystem::path conversions.

PR libstdc++/90281 Fix string conversions for filesystem::path
* include/bits/fs_path.h (u8path) [_GLIBCXX_FILESYSTEM_IS_WINDOWS]:
Use codecvt_utf8_utf16 instead of codecvt_utf8. Use
__str_codecvt_in_all to fail for partial conversions and throw on
error.
[!_GLIBCXX_FILESYSTEM_IS_WINDOWS && _GLIBCXX_USE_CHAR8_T]
(path::_Cvt<char8_t>): Add explicit specialization.
[_GLIBCXX_FILESYSTEM_IS_WINDOWS] (path::_Cvt::_S_wconvert): Remove
overloads.
[_GLIBCXX_FILESYSTEM_IS_WINDOWS] (path::_Cvt::_S_convert): Use
if-constexpr instead of dispatching to _S_wconvert. Use codecvt
instead of codecvt_utf8. Use __str_codecvt_in_all and
__str_codecvt_out_all.
[!_GLIBCXX_FILESYSTEM_IS_WINDOWS] (path::_Cvt::_S_convert): Use
codecvt instead of codecvt_utf8. Use __str_codecvt_out_all.
(path::_S_str_convert) [_GLIBCXX_FILESYSTEM_IS_WINDOWS]: Use
codecvt_utf8_utf16 instead of codecvt_utf8. Construct return values
with allocator. Use __str_codecvt_out_all. Fallthrough to POSIX code
after converting to UTF-8.
(path::_S_str_convert): Use codecvt instead of codecvt_utf8. Use
__str_codecvt_in_all.
(path::string): Fix initialization of string types with different
allocators.
(path::u8string) [_GLIBCXX_FILESYSTEM_IS_WINDOWS]: Use
codecvt_utf8_utf16 instead of codecvt_utf8. Use __str_codecvt_out_all.
* include/bits/locale_conv.h (__do_str_codecvt): Reorder static and
runtime conditions.
(__str_codecvt_out_all, __str_codecvt_in_all): New functions that
return false for partial conversions.
* include/experimental/bits/fs_path.h (u8path):
[_GLIBCXX_FILESYSTEM_IS_WINDOWS]: Implement correctly for mingw.
[_GLIBCXX_FILESYSTEM_IS_WINDOWS] (path::_Cvt::_S_wconvert): Add
missing handling for char8_t. Use codecvt and codecvt_utf8_utf16
instead of codecvt_utf8. Use __str_codecvt_in_all and
__str_codecvt_out_all.
[!_GLIBCXX_FILESYSTEM_IS_WINDOWS] (path::_Cvt::_S_convert): Use
codecvt instead of codecvt_utf8. Use __str_codecvt_out_all.
(path::string) [_GLIBCXX_FILESYSTEM_IS_WINDOWS]: Use
codecvt_utf8_utf16 instead of codecvt_utf8. Construct return values
with allocator. Use __str_codecvt_out_all and __str_codecvt_in_all.
(path::string) [!_GLIBCXX_FILESYSTEM_IS_WINDOWS]: Use
__str_codecvt_in_all.
(path::u8string) [_GLIBCXX_FILESYSTEM_IS_WINDOWS]: Use
codecvt_utf8_utf16 instead of codecvt_utf8. Use __str_codecvt_out_all.
* src/c++17/fs_path.cc (path::_S_convert_loc): Use
__str_codecvt_in_all.
* src/filesystem/path.cc (path::_S_convert_loc): Likewise.
* testsuite/27_io/filesystem/path/construct/90281.cc: New test.
* testsuite/27_io/filesystem/path/factory/u8path.cc: New test.
* testsuite/27_io/filesystem/path/native/string.cc: Test with empty
strings and with Unicode characters outside the basic multilingual
plane.
* testsuite/27_io/filesystem/path/native/alloc.cc: New test.
* testsuite/experimental/filesystem/path/construct/90281.cc: New test.
* testsuite/experimental/filesystem/path/factory/u8path.cc: New test.
* testsuite/experimental/filesystem/path/native/alloc.cc: New test.
* testsuite/experimental/filesystem/path/native/string.cc: Test with
empty strings and with Unicode characters outside the basic
multilingual plane.

From-SVN: r272385

[PR c++/90754] name lookup ICE

https://gcc.gnu.org/ml/gcc-patches/2019-06/msg00952.html
PR c++/90754
* name-lookup.c (lookup_type_scope_1): Calll qualify_lookup before
checking context.

PR c++/90754
* g++.dg/lookup/pr90754.C: New.

From-SVN: r272384

tree-ssa-alias.c (aliasing_component_refs_p): Consider only the access path from base to first VIEW_CONVERT_EXPR or BIT_FIELD_REF.

* tree-ssa-alias.c (aliasing_component_refs_p): Consider only
the access path from base to first VIEW_CONVERT_EXPR or
BIT_FIELD_REF.

From-SVN: r272383

Improve PR64242 testcase

Clear the input array to avoid the testcase accidentally
passing with an incorrect frame pointer.

Committed as obvious.

testsuite/
PR middle-end/64242
* gcc.c-torture/execute/pr64242.c: Improve test.

From-SVN: r272382

Simplify node ownership in _Hashtable members

Introduce an RAII type to manage nodes in unordered containers while
they are being inserted. If the caller always owns a node until it is
inserted, then the insertion functions don't need to deallocate on
failure. This allows a FIXME in the node re-insertion API to be removed.

Also change extract(const key_type&) to not call extract(const_iterator)
anymore.  This avoids looping through the bucket nodes again to find the
node before the one being extracted.

2019-06-17  François Dumont  <fdumont@gcc.gnu.org>
    Jonathan Wakely  <jwakely@redhat.com>

* include/bits/hashtable.h (struct _Hashtable::_Scoped_node): New type.
(_Hashtable::_M_insert_unique_node): Add key_type parameter. Don't
deallocate node if insertion fails.
(_Hashtable::_M_insert_multi_node): Likewise.
(_Hashtable::_M_reinsert_node): Pass additional key argument.
(_Hashtable::_M_reinsert_node_multi): Likewise. Remove FIXME.
(_Hashtable::_M_extract_node(size_t, __node_base*)): New function.
(_Hashtable::extract(const_iterator)): Use _M_extract_node.
(_Hashtable::extract(const _Key&)): Likewise.
(_Hashtable::_M_merge_unique): Pass additional key argument.
(_Hashtable::_M_emplace<Args>(true_type, Args&&...)): Likewise. Use
_Scoped_node.
(_Hashtable::_M_insert): Likewise.
* include/bits/hashtable_policy.h (_Map_base::operator[]): Likewise.
(_Hashtable_alloc): Add comments to functions with misleading names.

Co-Authored-By: Jonathan Wakely <jwakely@redhat.com>
From-SVN: r272381

tree-ssa-alias.c (nonoverlapping_component_refs_p): Also truncate access path on BIT_FIELD_REFs.

* tree-ssa-alias.c (nonoverlapping_component_refs_p): Also truncate
access path on BIT_FIELD_REFs.

From-SVN: r272380

Remove dead code (PR ipa/90874).

2019-06-17  Martin Liska  <mliska@suse.cz>

PR ipa/90874
* ipa-utils.h (odr_type_p): Remove dead code.

From-SVN: r272378

Remove support for alternative Solaris 11.4 ld -V output

* configure.ac (ld_vers) <*-*-solaris2*>: Remove support for
alternative Solaris 11.4 format.
* configure: Regenerate.

From-SVN: r272377

Fix tests that fail without PCH

The recent change to stop transitively including <string> broke some
tests, but only when the library is configured without PCH, because
otherwise the <string> header still gets included via the precompiled
<bits/stdc++.h> header.

* testsuite/20_util/bad_function_call/what.cc: Include <string> header
for std::string.
* testsuite/20_util/shared_ptr/cons/weak_ptr_expired.cc: Likewise.
* testsuite/20_util/tuple/cons/allocator_with_any.cc: Include <memory>
header for std::allocator.
* testsuite/23_containers/array/tuple_interface/tuple_element.cc: Add
using-declaration for std::size_t.
* testsuite/23_containers/array/tuple_interface/tuple_size.cc:
Likewise.
* testsuite/23_containers/deque/cons/55977.cc: Include <istream> for
std::istream.
* testsuite/23_containers/vector/cons/55977.cc: Likewise.
* testsuite/experimental/map/erasure.cc: Include <string> for
std::string.
* testsuite/experimental/unordered_map/erasure.cc: Likewise.

From-SVN: r272376

re PR ada/80590 (non-bootstrap build failure of Ada runtime)

PR ada/80590

* exp_ch9.adb (Expand_N_Delay_Relative_Statement): Swap the two
conditions to avoid a unnecessary exception propagation in the default
case.

From-SVN: r272372

Daily bump.

From-SVN: r272370

re PR d/90603 (ICE in functionParameters, at d/dmd/expression.c:1553)

PR d/90603
d/dmd: Merge upstream dmd 792f0fdf2

Fixes segmentation fault in functionParameters, and other related
semantic bugs in forward or recursively referenced declarations.

Reviewed-on: https://github.com/dlang/dmd/pull/10046

From-SVN: r272366

[nvptx] Fix some missing mode warnings in nvptx.md

Fix these warnings:
...
gcc/config/nvptx/nvptx.md:748:1: warning: operand 0 missing mode?
gcc/config/nvptx/nvptx.md:757:1: warning: operand 1 missing mode?
gcc/config/nvptx/nvptx.md:1514:1: warning: source missing a mode?
...

Build and reg-tested on x86_64 with nvptx accelerator.

2019-06-17  Tom de Vries  <tdevries@suse.de>

* config/nvptx/nvptx.md (define_insn "call_insn"): Rename to ...
(define_insn "call_insn_<mode>"): ... this.  Use P iterator on
match_operand 0.
(define_insn "call_value_insn"): Rename to ...
(define_insn "call_value_insn_<mode>"): this.  Use P iterator on
match_operand 0.
(define_insn "nvptx_red_partition"): Set unspec_volatile result mode to
DI.

From-SVN: r272365

re PR middle-end/64242 (Longjmp expansion incorrect)

PR middle-end/64242
* config/pa/pa.md (nonlocal_goto): Restore frame pointer last.  Add
frame clobbers and schedule block.
(builtin_longjmp): Likewise.

From-SVN: r272361

MSP430: Implement 64-bit shifts in assembly code

gcc/ChangeLog:

2019-06-16  Jozef Lawrynowicz  <jozef.l@mittosystems.com>

* config/msp430/msp430.c (msp430_expand_helper): Setup arguments which
describe how to perform MSPABI compliant 64-bit shift.
* config/msp430/msp430.md (ashldi3): New define_expand.
(ashrdi3): New define_expand.
(lshrdi3): New define_expand.

libgcc/ChangeLog:

2019-06-16  Jozef Lawrynowicz  <jozef.l@mittosystems.com>

* config/msp430/slli.S (__mspabi_sllll): New library function for
performing a logical left shift of a 64-bit value.
* config/msp430/srai.S (__mspabi_srall): New library function for
performing a arithmetic right shift of a 64-bit value.
* config/msp430/srll.S (__mspabi_srlll): New library function for
performing a logical right shift of a 64-bit value.

gcc/testsuite/ChangeLog:

2019-06-16  Jozef Lawrynowicz  <jozef.l@mittosystems.com>

* gcc.target/msp430/mspabi_sllll.c: New test.
* gcc.target/msp430/mspabi_srall.c: New test.
* gcc.target/msp430/mspabi_srlll.c: New test.
* gcc.c-torture/execute/shiftdi-2.c: New test.

From-SVN: r272360

sourcebuild.texi: Document new effective target keyword longlong64.

gcc/ChangeLog

2019-06-16  Jozef Lawrynowicz  <jozef.l@mittosystems.com>

* doc/sourcebuild.texi: Document new effective target keyword
longlong64.

gcc/testsuite/ChangeLog

2019-06-16  Jozef Lawrynowicz  <jozef.l@mittosystems.com>

* lib/target-supports.exp: Add check_effective_target_longlong64.

From-SVN: r272359

tree-ssa-alias.c (indirect_ref_may_alias_decl_p, [...]): Revert accidental commits.

* tree-ssa-alias.c (indirect_ref_may_alias_decl_p,
indirect_refs_may_alias_p): Revert accidental commits.

From-SVN: r272358

alias-access-path-4.c: New testcase.

* gcc.dg/tree-ssa/alias-access-path-4.c: New testcase.
* gcc.dg/tree-ssa/alias-access-path-5.c: New testcase.

* tree-ssa-alias.c (aliasing_component_refs_p): Watch for arrays
at the end of structures.

From-SVN: r272357

[Darwin] The need for picsym stubs is dependent on linker used, not OS rev.

For very old toolchains, the compiler generated pic symbol stubs that
provide the necessary indirections.  We no longer need this (since xcode
3 era) and it's more efficient for the linker to make one stub when it
knows a symbol is needed that for us to emit them speculatively in every
object.

Our current codegen is making the assumption that a specific OS version
uses a specific linker version - and therefore the presence of support
could be based on the target OS rev.  Of course, that's way too simplistic
(most likely bogus for cross-toolchains) and we want to make things explict.

Since we have detection of the linker version, we can use that directly
(A config test for support for stub-less linking might also be feasible
but much more involved).

Finally, should the user wish to generate code that caters for export to
use in an environment with an older toolchain, the generation of stubs
can be forced from the command line.

In addition to the points above, branch islanding and a long branch opt
for PowerPC Darwin has become conflated with the emission of these stubs
(which is confusing, and will be rectified in a following patch).

2019-06-16  Iain Sandoe  <iain@sandoe.co.uk>

* config/darwin.c (machopic_indirect_call_target): Use renamed
darwin_picsymbol_stubs to decide on output.
(darwin_override_options): Handle darwin_picsymbol_stubs.
* config/darwin.h (MIN_LD64_OMIT_STUBS): New.
(LD64_VERSION): Revise default.
* config/darwin.opt: (mpic-symbol-stubs): New option.
(darwin_picsymbol_stubs): New variable.
* config/i386/darwin.h (TARGET_MACHO_BRANCH_ISLANDS):
rename to TARGET_MACHO_PICSYM_STUBS.
* config/i386/i386.c (output_pic_addr_const): Likewise.
* config/i386/i386.h Likewise.
* config/rs6000/darwin.h: Likewise.
* config/rs6000/rs6000.c (rs6000_call_darwin_1): Use renamed
darwin_picsymbol_stubs.

From-SVN: r272356

[Darwin, opts] Add RejectNegative for cases with explict inverses.

These cases specify the negative option specifically, so we should not
accept a regular -no-xxxx for them.

2019-06-16  Iain Sandoe  <iain@sandoe.co.uk>

* config/darwin.opt (prebind, noprebind, seglinkedit,
noseglinkedit): Add RejectNegative.

From-SVN: r272355

tree-ssa-alias.c (nonoverlapping_component_refs_p): Fix pasto in my previous patch.

* tree-ssa-alias.c (nonoverlapping_component_refs_p): Fix pasto
in my previous patch.

From-SVN: r272354

dump_parse_tree (debug): Add verison for formal arglist.

2019-06-16  Thomas Koenig  <tkoenig@gcc.gnu.org>

    * dump_parse_tree (debug): Add verison for formal arglist.
    Do not crash when a gfc_expr is NULL.

From-SVN: r272353

re PR d/90863 (ICE in StatementSemanticVisitor::visit, at d/dmd/statementsem.c:1992)

PR d/90863
d/dmd: Merge upstream dmd 6e44734cc

Fixes segmentation fault in StatementSemanticVisitor::visit.

Reviewed-on: https://github.com/dlang/dmd/pull/10033

From-SVN: r272352

re PR d/90559 (Out of memory because of negative length)

PR d/90559
d/dmd: Merge upstream dmd 7afcc60c3

Partially fixes out of memory because of negative length.

Reviewed-on: https://github.com/dlang/dmd/pull/10025

gcc/d/ChangeLog:

2019-06-16  Iain Buclaw  <ibuclaw@gdcproject.org>

PR d/90559
* d-target.cc (Target::_init): Reduce max static data size to INT_MAX.

From-SVN: r272351

d/dmd: Merge upstream dmd f8e38c001

Fixes bug where foreach(int) doesn't work on BigEndian targets by
deprecating the use of index types smaller than a size_t/ptrdiff_t.

Reviewed-on: https://github.com/dlang/dmd/pull/10009

From-SVN: r272350

d/dmd: Merge upstream dmd 974650488

Adds static function VarDeclaration::create to the dmd C++ interface.

Reviewed-on: https://github.com/dlang/dmd/pull/10008

From-SVN: r272349

re PR d/90560 (ICE in visit, at d/dmd/dcast.c:1872)

PR d/90560
d/dmd: Merge upstream dmd c6887d9bb

Fixes segmentation fault in castTo::CastTo::visit.

Reviewed-on: https://github.com/dlang/dmd/pull/10007

From-SVN: r272348

re PR d/90762 (ICE in resolvePropertiesX, at d/dmd/expression.c:251)

PR d/90762
d/dmd: Merge upstream dmd b0cd59177

Fixes segmentation fault in resolvePropertiesX.

Reviewed-on: https://github.com/dlang/dmd/pull/10006

From-SVN: r272347

re PR d/90761 (ICE in visit, at d/dmd/dcast.c:883)

PR d/90761
d/dmd: Merge upstream dmd d912f4e49

Fixes segmentation fault in implicitConvTo::ImplicitConvTo::visit.

Reviewed-on: https://github.com/dlang/dmd/pull/10005

From-SVN: r272346

re PR d/90651 (ICE in FuncDeclaration::semantic3, at d/dmd/func.c:1524)

PR d/90651
d/dmd: Merge upstream dmd 0f6cbbcad

Fixes segmentation fault in FuncDeclaration::semantic3.

Reviewed-on: https://github.com/dlang/dmd/pull/10003

gcc/d/ChangeLog:

2019-06-16  Iain Buclaw  <ibuclaw@gdcproject.org>

* typeinfo.cc (object_module): New variable.
(make_frontend_typeinfo): Update signature.  Set temporary on
generated TypeInfo classes.
(create_tinfo_types): Set object_module.  Move generation of front-end
typeinfo into ...
(create_frontend_tinfo_types): ... New function.
(layout_typeinfo): Call create_frontend_tinfo_types.
(layout_classinfo): Likewise.
(layout_cpp_typeinfo): Likewise.
(create_typeinfo): Likewise.

From-SVN: r272345

re PR d/90650 (ICE in fold_convert_loc, at fold-const.c:2552)

PR d/90650
d/dmd: Merge upstream dmd ab03e2918

Fixes internal compiler error in fold_convert_loc.

Reviewed-on: https://github.com/dlang/dmd/pull/9996

gcc/testsuite/ChangeLog:

2019-06-16  Iain Buclaw  <ibuclaw@gdcproject.org>

PR d/90650
* gdc.dg/pr90650a.d: New test.
* gdc.dg/pr90650b.d: New test.

From-SVN: r272344

re PR d/90604 (ICE in sizemask, at d/dmd/mtype.c:2542)

PR d/90604
d/dmd: Merge upstream dmd f30c5dc79

Fixes internal compiler error in Type::sizemask.

Reviewed-on: https://github.com/dlang/dmd/pull/9998

From-SVN: r272343

re PR d/90602 (ICE: null field)

PR d/90602
d/dmd: Merge upstream dmd 420cce2a6

Fixes internal compiler error during CTFE.

Reviewed-on: https://github.com/dlang/dmd/pull/9997

From-SVN: r272342

re PR d/90661 (ICE in AlignDeclaration::syntaxCopy, at d/dmd/attrib.c:670)

PR d/90661
d/dmd: Merge upstream dmd c74e624c9

Fixes segmentation fault in AlignDeclaration::syntaxCopy.

Reviewed-on: https://github.com/dlang/dmd/pull/10001

From-SVN: r272341

re PR d/90651 (ICE in FuncDeclaration::semantic3, at d/dmd/func.c:1524)

PR d/90651
d/dmd: Merge upstream dmd 78dc31152

Fixes bug where the object module was not always implicitly imported.

Reviewed-on: https://github.com/dlang/dmd/pull/9999

From-SVN: r272340

re PR d/90660 (ICE in TypeQualified::resolveHelper, at d/dmd/mtype.c:6738)

PR d/90660
d/dmd: Merge upstream dmd bbc5ea66a

Fixes segmentation fault in TypeQualified::resolveHelper.

Reviewed-on: https://github.com/dlang/dmd/pull/10000

From-SVN: r272339

[openacc, parloops] Fix SIGSEGV in oacc_entry_exit_ok_1

When compiling the test-case with r268755, we run into a SIGSEGV in
oacc_entry_exit_ok_1 when trying to dereference a NULL red:
...
                      struct reduction_info *red;
                      red = reduction_phi (reduction_list, use_stmt);
                      tree val = PHI_RESULT (red->keep_res);
...

Fix this by handling ref == NULL.

Bootstrapped and reg-tested on x86_64.
Build and reg-tested on x86_64 with nvptx accelerator.

2019-06-16  Tom de Vries  <tdevries@suse.de>

PR tree-optimization/89376
* tree-parloops.c (oacc_entry_exit_ok_1): Handle red == NULL.

* testsuite/libgomp.oacc-c-c++-common/pr89376.c: New test.

From-SVN: r272338

Daily bump.

From-SVN: r272337

decl.c (gfc_match_derived_decl): Dummy argument cannot be a derived type.

2019-06-15  Steven G. Kargl  <kargl@gcc.gnu.org>

* decl.c (gfc_match_derived_decl): Dummy argument cannot be a derived
 type.

2019-06-15  Steven G. Kargl  <kargl@gcc.gnu.org>

* gfortran.dg/dummy_derived_typed.f90: New test.

From-SVN: r272333

Sync top-level change from gdb

This patch syncs a change to the top-level configury from gdb.  It
should not affect the gcc build at all.  Tested by rebuilding.

ChangeLog
2019-06-15  Tom Tromey  <tom@tromey.com>

* configure.ac (host_libs): Add gnulib.
* configure: Rebuild.
* Makefile.def (host_modules, dependencies): Add gnulib.
* Makefile.in: Rebuild.

From-SVN: r272332

invoke.texi (Spec Files): Update location of the Fortran spec file.

* doc/invoke.texi (Spec Files): Update location of the
Fortran spec file.

From-SVN: r272331

extend.texi (Common Function Attributes): Clarify no_sanitize.

* doc/extend.texi (Common Function Attributes): Clarify
no_sanitize.  Fix grammar.

From-SVN: r272330

alias-access-path-2.c: New testcase.

* gcc.dg/tree-ssa/alias-access-path-2.c: New testcase.

* tree-ssa-alias.c (alias_stats): Add
nonoverlapping_component_refs_p_may_alias,
nonoverlapping_component_refs_p_no_alias,
nonoverlapping_component_refs_of_decl_p_may_alias,
nonoverlapping_component_refs_of_decl_p_no_alias.
(dump_alias_stats): Dump them.
(nonoverlapping_component_refs_of_decl_p): Add stats.
(nonoverlapping_component_refs_p): Add stats; do not stop on first
ARRAY_REF.

From-SVN: r272329

ieee_4.f90: Un-xfail on i?86-*-freebsd.

2019-06-15  Steven G. Kargl  <kargl@gcc.gnu.org>

* gfortran.dg/ieee/ieee_4.f90: Un-xfail on i?86-*-freebsd.

From-SVN: r272328

[objective-c/c++, testsuite, 3/3] Workaround for PR90709.

Use the CF-CFString.h, F-NS*.h proxy headers where needed in the objective-c++
testsuite.  Make minor adjustments to tests as required, providing that those
do not alter the test intent.

2019-06-15  Iain Sandoe  <iain@sandoe.co.uk>

PR objc/90709
* obj-c++.dg/proto-lossage-7.mm: Use proxy headers.
* obj-c++.dg/strings/const-cfstring-2.mm: Likewise.
* obj-c++.dg/strings/const-cfstring-5.mm: Likewise
* obj-c++.dg/strings/const-str-12.mm: Likewise.
* obj-c++.dg/syntax-error-1.mm: Likewise.
* obj-c++.dg/torture/strings/const-cfstring-1.mm: Likewise.
* obj-c++.dg/torture/strings/const-str-10.mm: Likewise.
* obj-c++.dg/torture/strings/const-str-11.mm: Likewise.
* obj-c++.dg/torture/strings/const-str-9.mm: Likewise.
* obj-c++.dg/cxx-ivars-3.mm: Skip on later Darwin, where the 10.4 API
in no longer supported, also on m64 where there's no meaning to it.
* obj-c++.dg/isa-field-1.mm: Suppress unwanted warning, add comment why.
* obj-c++.dg/objc-gc-3.mm: Skip for Darwin > 16, the API use is an error
there.
* obj-c++.dg/qual-types-1.mm: Prune a spurious l64 warning.
* obj-c++.dg/stubify-1.mm: Tidy up after better compiler warnings.
* obj-c++.dg/stubify-2.mm: Likewise.
* obj-c++.dg/try-catch-1.mm: Likewise.
* obj-c++.dg/try-catch-3.mm: Likewise.

From-SVN: r272327

[objective-c/c++, testsuite, 2/3] Workaround for PR90709.

Use the CF-CFString.h, F-NS*.h proxy headers where needed in the objective-c
testsuite.  Make minor adjustments to tests as required, providing that those
do not alter the test intent.

2019-06-15  Iain Sandoe  <iain@sandoe.co.uk>

PR objc/90709
* objc.dg/encode-7-next-64bit.m: Use proxy headers.
* objc.dg/image-info.m: Likewise.
* objc.dg/method-6.m: Likewise.
* objc.dg/no-extra-load.m: Likewise.
* objc.dg/objc-foreach-4.m: Likewise.
* objc.dg/objc-foreach-5.m: Likewise.
* objc.dg/proto-lossage-7.m: Likewise.
* objc.dg/strings/const-cfstring-2.m: Likewise.
* objc.dg/strings/const-cfstring-5.m: Likewise.
* objc.dg/strings/const-str-12b.m: Likewise.
* objc.dg/symtab-1.m: Likewise.
* objc.dg/torture/strings/const-cfstring-1.m: Likewise.
* objc.dg/torture/strings/const-str-10.m: Likewise.
* objc.dg/torture/strings/const-str-11.m: Likewise.
* objc.dg/torture/strings/const-str-9.m: Likewise.
* objc.dg/zero-link-1.m: Likewise.
* objc.dg/zero-link-2.m: Likewise.
* objc.dg/zero-link-3.m: Likewise.
* objc.dg/isa-field-1.m: Suppress unwanted warning, add comment why.
* objc.dg/headers.m: XFAIL for Darwin14-19.
* objc.dg/objc-gc-4.m: Skip for Darwin > 16, the API use is an error
there.

From-SVN: r272326

[objective-c/c++, testsuite, 1/3] Workaround for PR90709.

Since we cannot parse the current NeXT headers, because of PR90709 and its
dependents, we have a large amount of testsuite noise for Darwin platforms.
In order to restore the usefulness of the testsuite, we are going add headers
without the modern syntax elements that trigger the bug, and use these for
test runs on newer Darwin.

The headers are imported from GNUStep, with some local modifications to make
sure that __BLOCKS__ is honoured as a gate for Apple-style blocks closures.

CF-CFString.h, F-NS*.h are proxy headers that use the installed CoreFoundation
or Foundation headers on systems <= Darwin12 and the GNUStep headers for newer.

2019-06-15  Iain Sandoe  <iain@sandoe.co.uk>

PR objc/90709
* objc-obj-c++-shared/CF-CFString.h: New.
* objc-obj-c++-shared/F-NSArray.h: New.
* objc-obj-c++-shared/F-NSAutoreleasePool.h: New.
* objc-obj-c++-shared/F-NSObject.h: New.
* objc-obj-c++-shared/F-NSString.h: New.
* objc-obj-c++-shared/F-NSValue.h: New.
* objc-obj-c++-shared/GNUStep/CoreFoundation/CFArray.h: New.
* objc-obj-c++-shared/GNUStep/CoreFoundation/CFAvailability.h: New.
* objc-obj-c++-shared/GNUStep/CoreFoundation/CFBase.h: New.
* objc-obj-c++-shared/GNUStep/CoreFoundation/CFCharacterSet.h: New.
* objc-obj-c++-shared/GNUStep/CoreFoundation/CFData.h: New.
* objc-obj-c++-shared/GNUStep/CoreFoundation/CFDictionary.h: New.
* objc-obj-c++-shared/GNUStep/CoreFoundation/CFLocale.h: New.
* objc-obj-c++-shared/GNUStep/CoreFoundation/CFString.h: New.
* objc-obj-c++-shared/GNUStep/Foundation/NSArray.h: New.
* objc-obj-c++-shared/GNUStep/Foundation/NSAutoreleasePool.h: New.
* objc-obj-c++-shared/GNUStep/Foundation/NSDate.h: New.
* objc-obj-c++-shared/GNUStep/Foundation/NSEnumerator.h: New.
* objc-obj-c++-shared/GNUStep/Foundation/NSGeometry.h: New.
* objc-obj-c++-shared/GNUStep/Foundation/NSObjCRuntime.h: New.
* objc-obj-c++-shared/GNUStep/Foundation/NSObject.h: New.
* objc-obj-c++-shared/GNUStep/Foundation/NSRange.h: New.
* objc-obj-c++-shared/GNUStep/Foundation/NSString.h: New.
* objc-obj-c++-shared/GNUStep/Foundation/NSValue.h: New.
* objc-obj-c++-shared/GNUStep/Foundation/NSZone.h: New.
* objc-obj-c++-shared/GNUStep/GNUstepBase/GNUstep.h: New.
* objc-obj-c++-shared/GNUStep/GNUstepBase/GSBlocks.h: New.
* objc-obj-c++-shared/GNUStep/GNUstepBase/GSConfig.h: New.
* objc-obj-c++-shared/GNUStep/GNUstepBase/GSObjCRuntime.h: New.
* objc-obj-c++-shared/GNUStep/GNUstepBase/GSVersionMacros.h: New.
* objc-obj-c++-shared/GNUStep/GNUstepBase/NSArray+GNUstepBase.h: New.
* objc-obj-c++-shared/GNUStep/GNUstepBase/NSMutableString+GNUstepBase.h:
New.
* objc-obj-c++-shared/GNUStep/GNUstepBase/NSNumber+GNUstepBase.h: New.
* objc-obj-c++-shared/GNUStep/GNUstepBase/NSObject+GNUstepBase.h: New.
* objc-obj-c++-shared/GNUStep/GNUstepBase/NSString+GNUstepBase.h: New.

From-SVN: r272325

[nvptx, libgomp] Update pr85381-{2,4}.c test-cases

After the fix for "PR tree-optimization/89713 - Assume loop with an exit is
finite" ( r272234 ) empty oacc loops are removed before expand.

Update pr85381-{2,4}.c accordingly.

2019-06-15  Tom de Vries  <tdevries@suse.de>

PR tree-optimization/89713
* testsuite/libgomp.oacc-c-c++-common/pr85381-2.c: Expect no bar.sync.
* testsuite/libgomp.oacc-c-c++-common/pr85381-4.c: Same.

From-SVN: r272324

i386.md (and<mode>3): Generate zero-extends for TARGET_ZERO_EXTEND_WITH_AND &&...

* config/i386/i386.md (and<mode>3): Generate zero-extends for
TARGET_ZERO_EXTEND_WITH_AND && optimize_function_for_speed_p (cfun))
only.
(*anddi3_doubleword): Split before reload. Merge with
anddi->zext pre-reload splitter.
(*andndi3_doubleword): Split before reload.
(*<code>di3_doubleword): Ditto.
(*one_cmpldi2_doubleword): Ditto.

From-SVN: r272323

re PR middle-end/90779 (Fortran array initialization in offload regions)

PR middle-end/90779
* gimplify.c: Include omp-offload.h and context.h.
(gimplify_bind_expr): Add "omp declare target" attributes
to static block scope variables inside of target region or target
functions.

* c-c++-common/goacc/routine-5.c (func2): Don't expect error for
static block scope variable in #pragma acc routine.

* testsuite/libgomp.c/pr90779.c: New test.
* testsuite/libgomp.fortran/pr90779.f90: New test.

From-SVN: r272322

[openacc] Disable pass_thread_jumps for IFN_UNIQUE

If we compile the openacc testcase with -fopenacc -O2, we run into a SIGSEGV
or assert.  The root cause for this is that pass_thread_jumps breaks the
invariant that OACC_FORK and OACC_JOIN mark the start and end of a
single-entry-single-exit region.

Fix this by bailing out when encountering an IFN_UNIQUE in
thread_jumps::profitable_jump_thread_path.

Bootstrapped and reg-tested on x86_64.
Build and reg-tested libgomp on x86_64 with nvptx accelerator.

2019-06-15  Tom de Vries  <tdevries@suse.de>

PR tree-optimization/90009
* tree-ssa-threadbackward.c (thread_jumps::profitable_jump_thread_path):
Return NULL if bb contains IFN_UNIQUE.

* testsuite/libgomp.oacc-c-c++-common/pr90009.c: New test.

From-SVN: r272321