Initialise the errors_ data member of SiPixelDigiErrorsSoAFromCUDA

[fwyzard]

PR description:

Initialise the errors_ data member of the SiPixelDigiErrorsSoAFromCUDA producer.

Prevents reading from uninitialised memory when the first event of a job has zero error words, and fixes a crash observed at HLT during data taking.

PR validation:

Without this fix, it is possible to reproduce the error observed online running over the "error stream" events that triggered it.
With this fix, no crash is observed after running 30+ times on the same input.

[fwyzard]

type bug-fix

[fwyzard]

urgent

[fwyzard]

enable gpu

[fwyzard]

please test

[cmsbuild]

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-38476/30692

• This PR adds an extra 12KB to repository

• There are other open Pull requests which might conflict with changes you have proposed:

  • File EventFilter/SiPixelRawToDigi/plugins/SiPixelDigiErrorsSoAFromCUDA.cc modified in PR(s): New Heterogeneous Memory Pool #37952

[cmsbuild]

A new Pull Request was created by @fwyzard (Andrea Bocci) for master.

It involves the following packages:

• EventFilter/SiPixelRawToDigi (reconstruction)

@jpata, @clacaputo can you please review it and eventually sign? Thanks.
@OzAmram, @VinInn, @Martin-Grunewald, @missirol, @dkotlins, @ferencek, @tvami this is something you requested to watch as well.
@perrotta, @dpiparo, @qliphy you are the release manager for this.

cms-bot commands are listed here

[cmsbuild]

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-3206f6/25711/summary.html
COMMIT: 532c257
CMSSW: CMSSW_12_5_X_2022-06-22-1100/el8_amd64_gcc10
Additional Tests: GPU
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/38476/25711/install.sh to create a dev area with all the needed externals and cmssw changes.

GPU Comparison Summary

Summary:

• No significant changes to the logs found
• Reco comparison results: 0 differences found in the comparisons
• DQMHistoTests: Total files compared: 4
• DQMHistoTests: Total histograms compared: 19864
• DQMHistoTests: Total failures: 9
• DQMHistoTests: Total nulls: 0
• DQMHistoTests: Total successes: 19855
• DQMHistoTests: Total skipped: 0
• DQMHistoTests: Total Missing objects: 0
• DQMHistoSizes: Histogram memory added: 0.0 KiB( 3 files compared)
• Checked 12 log files, 9 edm output root files, 4 DQM output files
• TriggerResults: no differences found

Comparison Summary

Summary:

• No significant changes to the logs found
• Reco comparison results: 6 differences found in the comparisons
• DQMHistoTests: Total files compared: 50
• DQMHistoTests: Total histograms compared: 3659099
• DQMHistoTests: Total failures: 13
• DQMHistoTests: Total nulls: 1
• DQMHistoTests: Total successes: 3659063
• DQMHistoTests: Total skipped: 22
• DQMHistoTests: Total Missing objects: 0
• DQMHistoSizes: Histogram memory added: 0.004 KiB( 49 files compared)
• DQMHistoSizes: changed ( 312.0 ): 0.004 KiB MessageLogger/Warnings
• Checked 208 log files, 45 edm output root files, 50 DQM output files
• TriggerResults: no differences found

[VinInn]

and why valgrind, ubsan, asan etc did not catch it?

[fwyzard]

I think, because the problem only happens if the first event has gpuDigiErrors.nErrorWords() == 0.
If the first event has some errors, the SimpleVector is initialised when reading the data from the gpu, and for all events after the SimpleVector is reset inside produce() after putting the produce in the event.

[clacaputo]

+reconstruction

[cmsbuild]

This pull request is fully signed and it will be integrated in one of the next master IBs (tests are also fine). This pull request will now be reviewed by the release team before it's merged. @perrotta, @dpiparo, @qliphy (and backports should be raised in the release meeting by the corresponding L2)

[perrotta]

+1