On 9/22/20 6:38 PM, Tom de Vries wrote:
> [ was: Re: [Patch] [middle-end & nvptx] gcc/tracer.c: Don't split BB
> with SIMT LANE [PR95654] ]
>
> On 9/16/20 8:20 PM, Alexander Monakov wrote:
>>
>>
>> On Wed, 16 Sep 2020, Tom de Vries wrote:
>>
>>> [ cc-ing author omp support for nvptx. ]
>>
>> The issue looks familiar. I recognized it back in 2017 (and LLVM people
>> recognized it too for their GPU targets). In an attempt to get agreement
>> to fix the issue "properly" for GCC I found a similar issue that affects
>> all targets, not just offloading, and filed it as PR 80053.
>>
>> (yes, there are no addressable labels involved in offloading, but
>> nevertheless
>> the nature of the middle-end issue is related)
>
> Hi Alexander,
>
> thanks for looking into this.
>
> Seeing that the attempt to fix things properly is stalled, for now I'm
> proposing a point-fix, similar to the original patch proposed by Tobias.
>
> Richi, Jakub, OK for trunk?
>
I've had to modify this patch in two ways:
- the original test-case stopped failing, though not the
minimized one, so I added that one as a test-case
- only testing for ENTER_ALLOC and EXIT, and not explicitly for VOTE_ANY
in ignore_bb_p also stopped working, so I've added that now.
Re-tested and committed.
Thanks,
- Tom
[omp, ftracer] Don't duplicate blocks in SIMT region
When running the libgomp testsuite on x86_64-linux with nvptx accelerator on
the test-case included in this patch, we run into:
...
FAIL: libgomp.fortran/pr95654.f90 -O3 -fomit-frame-pointer -funroll-loops \
-fpeel-loops -ftracer -finline-functions execution test
...
The test-case is a minimal version of this FAIL:
...
FAIL: libgomp.fortran/pr66199-5.f90 -O3 -fomit-frame-pointer -funroll-loops \
-fpeel-loops -ftracer -finline-functions execution test
...
but that one has stopped failing at commit c2ebf4f10de "openmp: Add support
for non-rect simd and improve collapsed simd support".
The problem is that ftracer duplicates a block containing GOMP_SIMT_VOTE_ANY.
That is, before ftracer we have (dropping the GOMP_SIMT_ prefix):
...
bb4(ENTER_ALLOC)
*----------+
| \
| \
| v
| *
v bb8
*<------------*
bb5(VOTE_ANY)
*-------------+
| |
| |
| |
| |
| v
| *
v bb7(XCHG_IDX)
*<------------*
bb6(EXIT)
...
The XCHG_IDX internal-fn does inter-SIMT-lane communication, which for nvptx
maps onto shfl, an operator which has the requirement that the warp executing
the operator is convergent. The warp diverges at bb4, and
reconverges at bb5, and does not diverge by going to bb7, so the shfl is
indeed executed by a convergent warp.
After ftracer, we have:
...
bb4(ENTER_ALLOC)
*----------+
| \
| \
| \
| \
v v
* *
bb5(VOTE_ANY) bb8(VOTE_ANY)
* *
|\ /|
| \ +--------+ |
| \/ |
| /\ |
| / +----------v
|/ *
v bb7(XCHG_IDX)
*<--------------*
bb6(EXIT)
...
The warp diverges again at bb5, but does not reconverge again before bb6, so
the shfl is executed by a divergent warp, which causes the FAIL.
Fix this by making ftracer ignore blocks containing ENTER_ALLOC, VOTE_ANY and
EXIT, effectively treating the SIMT region conservatively.
An argument can be made that the test needs to be added in a more
generic place, like gimple_can_duplicate_bb_p or some such, and that ftracer
then needs to use the generic test. But that's a discussion with a much
broader scope, so I'm leaving that for another patch.
Bootstrapped and reg-tested on x86_64-linux.
Build on x86_64-linux with nvptx accelerator, tested with libgomp.
gcc/ChangeLog:
PR fortran/95654
* tracer.c (ignore_bb_p): Ignore GOMP_SIMT_ENTER_ALLOC,
GOMP_SIMT_VOTE_ANY and GOMP_SIMT_EXIT.
libgomp/ChangeLog:
2020-10-05 Tom de Vries <tdevr...@suse.de>
PR fortran/95654
* testsuite/libgomp.fortran/pr95654.f90: New test.
---
gcc/tracer.c | 18 ++++++++++++++++++
libgomp/testsuite/libgomp.fortran/pr95654.f90 | 11 +++++++++++
2 files changed, 29 insertions(+)
diff --git a/gcc/tracer.c b/gcc/tracer.c
index 82ede722534..5e51752d89f 100644
--- a/gcc/tracer.c
+++ b/gcc/tracer.c
@@ -108,6 +108,24 @@ ignore_bb_p (const_basic_block bb)
return true;
}
+ for (gimple_stmt_iterator gsi = gsi_start_bb (CONST_CAST_BB (bb));
+ !gsi_end_p (gsi); gsi_next (&gsi))
+ {
+ gimple *g = gsi_stmt (gsi);
+
+ /* An IFN_GOMP_SIMT_ENTER_ALLOC/IFN_GOMP_SIMT_EXIT call must be
+ duplicated as part of its group, or not at all.
+ The IFN_GOMP_SIMT_VOTE_ANY is currently part of such a group,
+ so the same holds there, but it could be argued that the
+ IFN_GOMP_SIMT_VOTE_ANY could be generated after that group,
+ in which case it could be duplicated. */
+ if (is_gimple_call (g)
+ && (gimple_call_internal_p (g, IFN_GOMP_SIMT_ENTER_ALLOC)
+ || gimple_call_internal_p (g, IFN_GOMP_SIMT_EXIT)
+ || gimple_call_internal_p (g, IFN_GOMP_SIMT_VOTE_ANY)))
+ return true;
+ }
+
return false;
}
diff --git a/libgomp/testsuite/libgomp.fortran/pr95654.f90 b/libgomp/testsuite/libgomp.fortran/pr95654.f90
new file mode 100644
index 00000000000..2dddd3df4fe
--- /dev/null
+++ b/libgomp/testsuite/libgomp.fortran/pr95654.f90
@@ -0,0 +1,11 @@
+! { dg-do run }
+program main
+ implicit none
+ integer :: d1
+ !$omp target map(from: d1)
+ !$omp teams distribute parallel do simd default(none) lastprivate(d1) num_teams (2) num_threads (1)
+ do d1 = 0, 31
+ end do
+ !$omp end target
+ if (d1 /= 32) stop 3
+end program main
