* omp-low.c (lower_rec_input_clauses): For lastprivate conditional
references, lookup in in hash map MEM_REF operand instead of the
MEM_REF itself.
(lower_omp_1): When looking for lastprivate conditional assignments,
handle MEM_REFs with REFERENCE_TYPE operands.
* testsuite/libgomp.c++/lastprivate-conditional-1.C: New test.
* testsuite/libgomp.c++/lastprivate-conditional-2.C: New test.
From-SVN: r271948
* gimplify.c (gimplify_scan_omp_clauses): Don't sorry_at on lastprivate
conditional on combined for simd.
* omp-low.c (struct omp_context): Add combined_into_simd_safelen0
member.
(lower_rec_input_clauses): For gimple_omp_for_combined_into_p max_vf 1
constructs, don't remove lastprivate_conditional_map, but instead set
ctx->combined_into_simd_safelen0 and adjust hash_map, so that it points
to parent construct temporaries.
(lower_lastprivate_clauses): Handle ctx->combined_into_simd_safelen0
like !ctx->lastprivate_conditional_map.
(lower_omp_1) <case GIMPLE_ASSIGN>: If up->combined_into_simd_safelen0,
use up->outer context instead of up.
* omp-expand.c (expand_omp_for_generic): Perform cond_var bump even if
gimple_omp_for_combined_p.
(expand_omp_for_static_nochunk): Likewise.
(expand_omp_for_static_chunk): Add forgotten cond_var bump that was
probably moved over into expand_omp_for_generic rather than being copied
there.
gcc/cp/
* cp-tree.h (CP_OMP_CLAUSE_INFO): Allow for any clauses up to _condvar_
instead of only up to linear.
gcc/testsuite/
* c-c++-common/gomp/lastprivate-conditional-2.c (foo): Don't expect
a sorry_at on any of the clauses.
libgomp/
* testsuite/libgomp.c-c++-common/lastprivate-conditional-7.c: New test.
* testsuite/libgomp.c-c++-common/lastprivate-conditional-8.c: New test.
* testsuite/libgomp.c-c++-common/lastprivate-conditional-9.c: New test.
* testsuite/libgomp.c-c++-common/lastprivate-conditional-10.c: New test.
From-SVN: r271907
* gimplify.c (struct gimplify_omp_ctx): Add clauses member.
(gimplify_scan_omp_clauses): Initialize ctx->clauses.
(gimplify_adjust_omp_clauses_1): Transform lastprivate conditional
explicit clause on combined parallel into implicit shared clause.
(gimplify_adjust_omp_clauses): Move lastprivate conditional clause
and firstprivate if the decl has one too from combined parallel to
the worksharing construct.
gcc/testsuite/
* c-c++-common/gomp/lastprivate-conditional-2.c (foo): Don't expect
sorry on lastprivate conditional on parallel for.
* c-c++-common/gomp/lastprivate-conditional-3.c (foo): Add tests for
lastprivate conditional warnings on parallel for constructs.
* c-c++-common/gomp/lastprivate-conditional-4.c: New test.
libgomp/
* testsuite/libgomp.c-c++-common/lastprivate_conditional_4.c: Rename
to ...
* testsuite/libgomp.c-c++-common/lastprivate-conditional-4.c: ... this.
* testsuite/libgomp.c-c++-common/lastprivate-conditional-5.c: New test.
* testsuite/libgomp.c-c++-common/lastprivate-conditional-6.c: New test.
From-SVN: r271733
* omp-low.c (lower_omp_1) <case GIMPLE_ASSIGN>: Look through ordered,
critical, taskgroup and section regions when looking for a region
with non-NULL lastprivate_conditional_map.
* testsuite/libgomp.c-c++-common/lastprivate-conditional-3.c: New test.
From-SVN: r271672
PR libgomp/90641
* work.c (gomp_init_work_share): Instead of aligning final ordered
value to multiples of long long alignment, align to that the
first part (ordered team ids) and if inline_ordered_team_ids
is not on a long long alignment boundary within the structure,
use __alignof__ (long long) - 1 pad size always.
* loop.c (GOMP_loop_start): Fix *mem computation if
inline_ordered_team_ids is not aligned on long long alignment boundary
within the structure.
* loop-ull.c (GOMP_loop_ull_start): Likewise.
* sections.c (GOMP_sections2_start): Likewise.
From-SVN: r271671
PR libgomp/90585
* plugin/plugin-hsa.c: Include gstdint.h. Include inttypes.h only if
HAVE_INTTYPES_H is defined.
(print_uint64_t): New typedef.
(PRIu64): Define if HAVE_INTTYPES_H is not defined.
(print_kernel_dispatch, run_kernel): Use PRIu64 macro instead of
"lu", cast uint64_t HSA_DEBUG and fprintf arguments to print_uint64_t.
(release_kernel_dispatch): Likewise. Cast shadow->debug to uintptr_t
before casting to void *.
* plugin/plugin-nvptx.c: Include gstdint.h instead of stdint.h.
* oacc-mem.c: Don't include config.h nor stdint.h.
* target.c: Don't include config.h.
* oacc-cuda.c: Likewise.
* oacc-host.c: Don't include stdint.h.
From-SVN: r271597
... to avoid compilation overhead, and to keep simple '-foffload=[...]'
handling in test cases.
libgomp/
* testsuite/libgomp.oacc-c++/c++.exp: Specify
"-foffload=$offload_target".
* testsuite/libgomp.oacc-c/c.exp: Likewise.
* testsuite/libgomp.oacc-fortran/fortran.exp: Likewise.
* testsuite/lib/libgomp.exp
(check_effective_target_openacc_nvidia_accel_configured): Remove,
as (conceptually) merged into
check_effective_target_openacc_nvidia_accel_selected. Adjust all
users.
From-SVN: r269109
PR middle-end/89002
* gimplify.c (gimplify_omp_for): When adding OMP_CLAUSE_*_GIMPLE_SEQ
for lastprivate/linear IV, push gimplify context around gimplify_assign
and, if it needed any temporaries, pop it into a gimple bind around the
sequence.
* testsuite/libgomp.c/pr89002.c: New test.
From-SVN: r268346
PR middle-end/89002
* gimplify.c (gimplify_omp_for): When adding OMP_CLAUSE_*_GIMPLE_SEQ
for lastprivate/linear IV, push gimplify context around gimplify_assign
and, if it needed any temporaries, pop it into a gimple bind around the
sequence.
* testsuite/libgomp.c/pr89002.c: New test.
From-SVN: r268345
I wrote a test-case:
...
int
main (void)
{
for (unsigned i = 0; i < 128; ++i)
{
acc_init (acc_device_nvidia);
acc_shutdown (acc_device_nvidia);
}
return 0;
}
...
and ran it under valgrind. The only leak location reported with a frequency
of 128, was the allocation of ptx_devices in nvptx_init.
Fix this by freeing ptx_devices in GOMP_OFFLOAD_fini_device, once
instantiated_devices drops to 0.
2019-01-24 Tom de Vries <tdevries@suse.de>
* plugin/plugin-nvptx.c (GOMP_OFFLOAD_fini_device): Free ptx_devices
once instantiated_devices drops to 0.
From-SVN: r268237
Consider test-case:
...
int
main (void)
{
#pragma acc parallel async
;
#pragma acc parallel async
;
#pragma acc wait
return 0;
}
...
This fails with:
...
libgomp: cuMemAlloc error: invalid argument
Segmentation fault (core dumped)
...
The cuMemAlloc error is due to the fact that we're try to allocate 0 bytes.
Fix this by preventing calling map_push with size zero argument in nvptx_exec.
This also has the consequence that for the abort-1.c test-case, we end up
calling cuMemFree during map_fini for the struct cuda_map allocated in
map_init, which fails because an abort happened. Fix this by calling
cuMemFree with CUDA_CALL_NOCHECK in cuda_map_destroy.
2019-01-23 Tom de Vries <tdevries@suse.de>
PR target/PR88946
* plugin/plugin-nvptx.c (cuda_map_destroy): Use CUDA_CALL_NOCHECK for
cuMemFree.
(nvptx_exec): Don't call map_push if mapnum == 0.
* testsuite/libgomp.oacc-c-c++-common/pr88946.c: New test.
From-SVN: r268178
There are currently two situations where this assert triggers:
...
libgomp/plugin/plugin-nvptx.c: map_fini: Assertion `!s->map->active' failed.
...
First, in abort-1.c, a parallel region triggering an abort:
...
int
main (void)
{
#pragma acc parallel
abort ();
return 0;
}
...
The abort is detected in nvptx_exec as the CUDA_ERROR_ILLEGAL_INSTRUCTION
return status of the cuStreamSynchronize call after kernel launch, which is
then handled by calling non-returning function GOMP_PLUGIN_fatal.
Consequently, the map_pop in nvptx_exec that in case of cuStreamSynchronize
success would remove or inactive the element added by the map_push earlier in
nvptx_exec, does not trigger. With the element no longer active, but still
marked active and a member of s->map, we run into the assert during
GOMP_OFFLOAD_fini_device, which is triggered from atexit handler
gomp_target_fini (which is triggered by the GOMP_PLUGIN_fatal mentioned above
calling exit).
Second, in pr88941.c, an async parallel region without wait:
...
int
main (void)
{
#pragma acc parallel async
;
/* no #pragma acc wait */
return 0;
}
...
Because nvptx_exec is handling an async region, it does not call map_pop for
the element added by map_push, but schedules an kernel execution completion
event to call map_pop. Again, we run into the assert during
GOMP_OFFLOAD_fini_device, which is triggered from atexit handler
gomp_target_fini, but the exit in this case is triggered by returning from main.
So either the kernel is still running, or the kernel has completed but the
corresponding event that is supposed to call map_pop is stuck in the event
queue, waiting for an event_gc.
Fix this by removing the assert, and skipping the freeing of device memory if
the map is still marked active (though in the async case, this is more a
workaround than an fix).
2019-01-23 Tom de Vries <tdevries@suse.de>
PR target/88941
PR target/88939
* plugin/plugin-nvptx.c (cuda_map_destroy): Handle map->active case.
(map_fini): Remove "assert (!s->map->active)".
* testsuite/libgomp.oacc-c-c++-common/pr88941.c: New test.
From-SVN: r268177
The map field of a struct ptx_stream is a FIFO. The FIFO is implemented as a
single linked list, with pop-from-the-front semantics.
The function map_pop pops an element, either by:
- deallocating the element, if there is more than one element
- or marking the element inactive, if there's only one element
The responsibility of map_push is to push an element to the back, as well as
selecting the element to push, by:
- allocating an element, or
- reusing the element at the front if inactive and big enough, or
- dropping the element at the front if inactive and not big enough, and
allocating one that's big enough
The current implemention gets at least the first and most basic scenario wrong:
> map = cuda_map_create (size);
We create an element, and assign it to map.
> for (t = s->map; t->next != NULL; t = t->next)
> ;
We determine the last element in the fifo.
> t->next = map;
We append the new element.
> s->map = map;
But here, we throw away the rest of the FIFO, and declare the FIFO to be just
the new element.
This problem causes the test-case asyncwait-1.c to fail intermittently on some
systems. The pr87835.c test-case added here is a a minimized and modified
version of asyncwait-1.c (avoiding the kernel construct) that is more likely to
fail.
Fix this by rewriting map_pop more robustly, by:
- seperating the function in two phases: select element, push element
- when reusing or dropping an element, making sure that the element is cleanly
popped from the queue
- rewriting the push element part in such a way that it can handle all cases
without needing if statements, such that each line is exercised for each of
the three cases.
2019-01-23 Tom de Vries <tdevries@suse.de>
PR target/87835
* plugin/plugin-nvptx.c (map_push): Fix adding of allocated element.
* testsuite/libgomp.oacc-c-c++-common/pr87835.c: New test.
From-SVN: r268176
Add some test-cases that set vector length using -fopenacc-dim.
2019-01-12 Tom de Vries <tdevries@suse.de>
* testsuite/libgomp.oacc-c-c++-common/pr85486-2.c: New test.
* testsuite/libgomp.oacc-c-c++-common/vector-length-128-2.c: New test.
* testsuite/libgomp.oacc-c-c++-common/vector-length-128-5.c: New test.
* testsuite/libgomp.oacc-fortran/gemm-2.f90: New test.
From-SVN: r267897
Add some test-cases using vector_length 64.
2019-01-12 Tom de Vries <tdevries@suse.de>
* testsuite/libgomp.oacc-c-c++-common/vector-length-64-1.c: New test.
* testsuite/libgomp.oacc-c-c++-common/vector-length-64-2.c: New test.
* testsuite/libgomp.oacc-c-c++-common/vector-length-64-3.c: New test.
From-SVN: r267895
Add test-cases for "[nvptx] Force vl32 if calling vector-partitionable
routines".
2019-01-12 Tom de Vries <tdevries@suse.de>
PR target/85486
* testsuite/libgomp.oacc-c-c++-common/pr85486-3.c: New test.
* testsuite/libgomp.oacc-c-c++-common/pr85486.c: New test.
From-SVN: r267894
Add test-cases for PR85381.
2019-01-12 Tom de Vries <tdevries@suse.de>
PR target/85381
* testsuite/libgomp.oacc-c-c++-common/pr85381-5.c: New test.
* testsuite/libgomp.oacc-c-c++-common/pr85381.c: New test.
From-SVN: r267893
Add various reduction test-cases with vector length 128.
2019-01-12 Tom de Vries <tdevries@suse.de>
* testsuite/libgomp.oacc-c-c++-common/vred2d-128.c: New test.
* testsuite/libgomp.oacc-fortran/gemm.f90: New test.
* testsuite/libgomp.oacc-c-c++-common/vector-length-128-10.c: New test.
From-SVN: r267892
Add various test-cases with vector length 128.
2019-01-12 Tom de Vries <tdevries@suse.de>
* testsuite/libgomp.oacc-c-c++-common/vector-length-128-7.c: New test.
* testsuite/libgomp.oacc-c-c++-common/vector-length-128-4.c: New test.
* testsuite/libgomp.oacc-c-c++-common/vector-length-128-6.c: New test.
From-SVN: r267891
Update message in nvptx libgomp plugin about insufficient resources to launch
kernel, to accommodate for the fact the vector_length can now be variable.
2019-01-12 Tom de Vries <tdevries@suse.de>
* plugin/plugin-nvptx.c (nvptx_exec): Update insufficient hardware
resources diagnostic.
From-SVN: r267890
Allow vector_length clauses to accept values larger than warp size. Note that
this does not enable setting vector_length to values larger than warp size using
-fopenacc-dim.
2019-01-12 Tom de Vries <tdevries@suse.de>
* config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Take larger vector
lengths into account.
* testsuite/libgomp.oacc-c-c++-common/vector-length-128-1.c: Expect
vector length to be 128.
* testsuite/libgomp.oacc-c-c++-common/parallel-dims.c: Expect vector
length 2097152 to be reduced to 1024 instead of 32.
From-SVN: r267889
When using a compiler build with:
...
+#define PTX_DEFAULT_VECTOR_LENGTH PTX_CTA_SIZE
...
consider a test-case:
...
int
main (void)
{
#pragma acc parallel vector_length (64)
#pragma acc loop worker
for (unsigned int i = 0; i < 32; i++)
#pragma acc loop vector
for (unsigned int j = 0; j < 64; j++)
;
return 0;
}
...
If num_workers is 16, either because:
- we add a "num_workers (16)" clause on the parallel directive, or
- we set "GOMP_OPENACC_DIM=:16:", or
- the libgomp plugin chooses 16 num_workers
we run into an illegal instruction at runtime, because a bar.sync instruction
tries to use a barrier 16. The instruction is illegal, because ptx supports
only 16 barriers per CTA, and the valid range is 0..15.
The problem is that with a warp-multiple vector length, we use a code generation
scheme with a per-worker barrier. And because barrier zero is reserved for
per-cta barrier, only the remaining 15 barriers can be used as per-worker
barrier, and consequently we can't use num_workers larger than 15.
This problem occurs only for vector_length 64. For vector_length 32, we use a
different code generation scheme, and for vector_length >= 96, the maximum
num_workers is not big enough not to trigger this problem.
Also, this problem only occurs for num_workers 16. As explained above,
num_workers 15 is safe to use, and 16 is already the maximum num_workers for
vector_length 64.
This patch fixes the problem in both the compiler (handling "num_workers (16)")
and in the libgomp nvptx plugin (with and without "GOMP_OPENACC_DIM=:16:").
2019-01-11 Tom de Vries <tdevries@suse.de>
* config/nvptx/nvptx.c (PTX_CTA_NUM_BARRIERS, PTX_PER_CTA_BARRIER)
(PTX_NUM_PER_CTA_BARRIER, PTX_FIRST_PER_WORKER_BARRIER)
(PTX_NUM_PER_WORKER_BARRIERS): Define.
(nvptx_apply_dim_limits): Prevent vector_length 64 and
num_workers 16.
* plugin/plugin-nvptx.c (nvptx_exec): Prevent vector_length 64 and
num_workers 16.
From-SVN: r267838
Before the commit "[libgomp, testsuite, openacc] Don't use const int for
dimensions", the "const int" construct was used to set launch dimensions in
reductions-[1-5].c. In the case of -xc -O0, the const int is implemented as a
variable by the C front-end. Consequently, the nvptx back-end generated
warnings that vector_length was overridden to be hard-coded, rather than left to
be set at runtime. The test-cases silenced these warnings by switching off all
warnings in the accelerator compiler using "-foffload=-w".
Given that no warnings occur anymore, remove the "-foffload=-w" setting.
2019-01-11 Tom de Vries <tdevries@suse.de>
* testsuite/libgomp.oacc-c-c++-common/reduction-1.c: Remove
-foffload=-w.
* testsuite/libgomp.oacc-c-c++-common/reduction-2.c: Same.
* testsuite/libgomp.oacc-c-c++-common/reduction-3.c: Same.
* testsuite/libgomp.oacc-c-c++-common/reduction-4.c: Same.
* testsuite/libgomp.oacc-c-c++-common/reduction-5.c: Same.
From-SVN: r267836
Add a test-case that tests the "insufficient resources" fatal in the nvptx
libgomp plugin.
2019-01-11 Tom de Vries <tdevries@suse.de>
* testsuite/libgomp.oacc-c-c++-common/insufficient-resources.c: New
test.
From-SVN: r267835
libgomp/
* config/rtems/affinity-fmt.c: New file. Include affinity-fmt.c,
undefining HAVE_GETPID and HAVE_GETHOSTNAME, and mapping fwrite to
write.
From-SVN: r267751
Const int is handled differently at -O0 for -xc and -xc++, which can cause noise
in testsuite/libgomp.oacc-c-c++-common test-cases (which are both run for c and
c++) if const int is used for launch dimensions.
Fix this by using #defines instead.
2019-01-09 Tom de Vries <tdevries@suse.de>
PR target/88756
* testsuite/libgomp.oacc-c-c++-common/reduction-1.c (ng, nw, vl): Use
#define instead of "const int".
* testsuite/libgomp.oacc-c-c++-common/reduction-2.c (ng, nw, vl): Same.
* testsuite/libgomp.oacc-c-c++-common/reduction-3.c (ng, nw, vl): Same.
* testsuite/libgomp.oacc-c-c++-common/reduction-4.c (ng, nw, vl): Same.
* testsuite/libgomp.oacc-c-c++-common/reduction-5.c (ng, nw, vl): Same.
From-SVN: r267747
When using a compiler build with:
...
+#define PTX_DEFAULT_VECTOR_LENGTH PTX_CTA_SIZE
+#define PTX_MAX_VECTOR_LENGTH PTX_CTA_SIZE
...
and running the libgomp testsuite, we run into an execution failure in
parallel-loop-1.c, due to a cuda launch failure:
...
nvptx_exec: kernel f6_none_none$_omp_fn$0: launch gangs=480, workers=0, \
vectors=1024
libgomp: cuLaunchKernel error: invalid argument
...
because workers == 0.
The workers variable is set to 0 here in nvptx_exec:
...
workers = blocks / actual_vectors;
...
because actual_vectors is 1024, and blocks is 768:
...
cuOccupancyMaxPotentialBlockSize: grid = 10, block = 768
...
Fix this by ensuring that workers is at least one.
2019-01-09 Tom de Vries <tdevries@suse.de>
* plugin/plugin-nvptx.c (nvptx_exec): Make sure to launch with at least
one worker.
From-SVN: r267746
The vector-length-128-3.c test-case uses GOMP_OPENACC_DIM=-:-:128, but '-' is
not yet supported on trunk. Use GOMP_OPENACC_DIM=::128 instead.
2019-01-07 Tom de Vries <tdevries@suse.de>
* testsuite/libgomp.oacc-c-c++-common/vector-length-128-3.c: Fix
GOMP_OPENACC_DIM argument.
From-SVN: r267624
Feng Xue
Jakub Jelinek
Rainer Orth
Thomas Schwinge
Chung-Lin Tang
Kevin Buettner
Jakub Jelinek
Richard Biener
Tom de Vries
Nathan Sidwell
Sebastian Huber