Open VinInn opened 3 years ago
A new Issue was created by @VinInn Vincenzo Innocente.
@Dr15Jones, @perrotta, @dpiparo, @makortel, @smuzaffar, @qliphy can you please review it and eventually sign/assign? Thanks.
cms-bot commands are listed here
assign core,hlt,reconstruction
New categories assigned: core,hlt,reconstruction
@Dr15Jones,@smuzaffar,@slava77,@Martin-Grunewald,@makortel,@jpata,@missirol you have been requested to review this Pull request/Issue and eventually sign? Thanks
I'm not sure if ASAN can impact (e.g. the order of) floating point computations, or if this should be interpreted as a symptom of some code reading from incorrect memory locations (that are still technically valid so ASAN doesn't catch those).
Maybe we should check offline reco: just to make sure is not a HLT only feature....
I run reco on 200 events. Reco itself as much less errors than HLT (only few events affected): but is different.
[innocent@patatrack02 asan]$ grep MSG-w asanReco.log | grep -v MemoryCheck
%MSG-w XrdAdaptorInternal: file_open 04-Oct-2021 15:14:41 CEST pre-events
%MSG-w XrdAdaptorInternal: file_open 04-Oct-2021 15:14:41 CEST pre-events
%MSG-w XrdAdaptor: file_open 04-Oct-2021 15:14:43 CEST pre-events
%MSG-w BasicTrajectoryState: TrackProducer:jetCoreRegionalStepTracks 04-Oct-2021 15:30:16 CEST Run: 1 Event: 1662
%MSG-w BasicTrajectoryState: TrackProducer:jetCoreRegionalStepTracks 04-Oct-2021 15:30:16 CEST Run: 1 Event: 1662
%MSG-w BasicTrajectoryState: TrackProducer:jetCoreRegionalStepTracks 04-Oct-2021 15:30:16 CEST Run: 1 Event: 1662
%MSG-w BasicTrajectoryState: TrackProducer:jetCoreRegionalStepTracks 04-Oct-2021 15:30:16 CEST Run: 1 Event: 1662
%MSG-w BasicTrajectoryState: TrackProducer:jetCoreRegionalStepTracks 04-Oct-2021 15:30:16 CEST Run: 1 Event: 1662
%MSG-w BasicTrajectoryState: TrackProducer:jetCoreRegionalStepTracks 04-Oct-2021 15:30:16 CEST Run: 1 Event: 1662
%MSG-w BasicTrajectoryState: TrackProducer:jetCoreRegionalStepTracks 04-Oct-2021 15:30:16 CEST Run: 1 Event: 1662
%MSG-w BasicTrajectoryState: TrackProducer:jetCoreRegionalStepTracks 04-Oct-2021 15:30:16 CEST Run: 1 Event: 1662
%MSG-w BasicTrajectoryState: TrackProducer:jetCoreRegionalStepTracks 04-Oct-2021 15:30:16 CEST Run: 1 Event: 1662
%MSG-w BasicTrajectoryState: TrackProducer:jetCoreRegionalStepTracks 04-Oct-2021 15:30:16 CEST Run: 1 Event: 1662
%MSG-w TrackNaN: TrackProducer:jetCoreRegionalStepTracks 04-Oct-2021 15:30:16 CEST Run: 1 Event: 1662
%MSG-w OutOfBounds: TrackProducer:initialStepTracksPreSplitting 04-Oct-2021 15:32:13 CEST Run: 1 Event: 1670
%MSG-w OutOfBounds: TrackProducer:initialStepTracks 04-Oct-2021 15:32:34 CEST Run: 1 Event: 1670
[innocent@patatrack02 asan]$ grep MSG-w oriReco.log | grep -v MemoryCheck
%MSG-w XrdAdaptorInternal: file_open 04-Oct-2021 16:14:53 CEST pre-events
%MSG-w XrdAdaptorInternal: file_open 04-Oct-2021 16:14:53 CEST pre-events
%MSG-w XrdAdaptor: file_open 04-Oct-2021 16:14:54 CEST pre-events
%MSG-w OutOfBounds: TrackProducer:initialStepTracksPreSplitting 04-Oct-2021 16:19:35 CEST Run: 1 Event: 1670
%MSG-w OutOfBounds: TrackProducer:initialStepTracks 04-Oct-2021 16:19:36 CEST Run: 1 Event: 1670
%MSG-w OutOfBounds: TrackProducer:mixedTripletStepTracks 04-Oct-2021 16:20:39 CEST Run: 1 Event: 1691
%MSG-w PFTrackTransformer: PFElecTkProducer:pfTrackElec 04-Oct-2021 16:24:15 CEST Run: 1 Event: 1761
I have compared running a hlt menu on 1000 events in /cvmfs/cms-ib.cern.ch/week0/slc7_amd64_gcc10/cms/cmssw/CMSSW_12_1_ASAN_X_2021-09-29-2300 vs /cvmfs/cms-ib.cern.ch/week0/slc7_amd64_gcc10/cms/cmssw/CMSSW_12_1_X_2021-09-29-2300
and results differs
for instance error found are different
ASAN
Normal build
and hlt results as well
I do not think this was expected