HLT crash in run 388037: segmentation violation in PixelTrackProducerFromSoAAlpaka<pixelTopology::HIonPhase1>::produce
#46656
Comments
|
cms-bot internal usage |
|
A new Issue was created by @mmusich. @Dr15Jones, @antoniovilela, @makortel, @mandrenguyen, @rappoccio, @sextonkennedy, @smuzaffar can you please review it and eventually sign/assign? Thanks. cms-bot commands are listed here |
|
type tracking |
|
@cms-sw/tracking-pog-l2 FYI |
|
so apparently the crash happens here: Interestingly in the reproducer log, few lines above the crash we have also: |
|
assign RecoTracker/PixelTrackFitting |
|
New categories assigned: reconstruction @jfernan2,@mandrenguyen you have been requested to review this Pull request/Issue and eventually sign? Thanks |
cmsbuild
added
reconstruction-pending
pending-signatures
and removed
pending-assignment
labels
|
assign heterogeneous, hlt |
|
New categories assigned: heterogeneous,hlt @fwyzard,@makortel,@Martin-Grunewald,@mmusich you have been requested to review this Pull request/Issue and eventually sign? Thanks |
|
this diff --git a/RecoTracker/PixelTrackFitting/plugins/storeTracks.h b/RecoTracker/PixelTrackFitting/plugins/storeTracks.h
index fb9f169c3e5..ace9c7e9f56 100644
--- a/RecoTracker/PixelTrackFitting/plugins/storeTracks.h
+++ b/RecoTracker/PixelTrackFitting/plugins/storeTracks.h
@@ -32,9 +32,11 @@ void storeTracks(Ev& ev, const TWH& tracksWithHits, const TrackerTopology& ttopo
const auto& hits = tracksWithHits[i].second;
for (unsigned int k = 0; k < hits.size(); k++) {
- auto* hit = hits[k]->clone(); // need to clone (at least if from SoA)
- track->appendHitPattern(*hit, ttopo);
- recHits->push_back(hit);
+ if (hits[k]) {
+ auto* hit = hits[k]->clone(); // need to clone (at least if from SoA)
+ track->appendHitPattern(*hit, ttopo);
+ recHits->push_back(hit);
+ }
}
tracks->push_back(*track);
delete track;avoids the crash. Probably there is a better way starting from upstream. |
|
Two new pieces of evidence:
[1] diff --git a/RecoLocalTracker/SiPixelRecHits/plugins/SiPixelRecHitFromSoAAlpaka.cc b/RecoLocalTracker/SiPixelRecHits/plugins/SiPixelRecHitFromSoAAlpaka.cc
index a76ff6af49a..cc129f8a437 100644
--- a/RecoLocalTracker/SiPixelRecHits/plugins/SiPixelRecHitFromSoAAlpaka.cc
+++ b/RecoLocalTracker/SiPixelRecHits/plugins/SiPixelRecHitFromSoAAlpaka.cc
@@ -127,7 +127,7 @@ void SiPixelRecHitFromSoAAlpaka<TrackerTraits>::produce(edm::StreamID streamID,
gind,
maxHitsInModule);
- nhits = std::min(nhits, maxHitsInModule);
+ //nhits = std::min(nhits, maxHitsInModule);
LogDebug("SiPixelRecHitFromSoAAlpaka") << "in det " << gind << "conv " << nhits << " hits from " << dsv.size()
<< " legacy clusters" << ' ' << lc << ',' << fc; |
|
@cms-sw/tracking-pog-l2 @AdrianoDee Do you have any feedback ? |
|
@missirol having a look |
|
Ok the problem is that this should actually beTrackerTraits::maxHitsInModule that takes into account the fact that for HIon we may have more than 1024 hits per module. Proposed fixes:
The problem ends up appearing in |
|
Thanks @AdrianoDee |
|
I verified that cherry-picking AdrianoDee@9b8e10f on top of |
|
+1 |
|
I think this issue was closed too hastily (let's at least wait for deployment :) ) - I can't reopen it though. |
In run 388037 (PbPb collisions, HLT release
CMSSW_14_1_4_patch3), we got the following segmentation violation:The log file from the HLT node can be found at https://cernbox.cern.ch/s/pnmiGV9LkISCWqU (25MB -- too large to be posted on gitHub).
The issue is reproducible with the following script (run on
lxplus8-gpuinCMSSW_14_1_4_patch3):@cms-sw/hlt-l2 @cms-sw/heterogeneous-l2 FYI
The text was updated successfully, but these errors were encountered: