Skip to content

Conversation

@smuzaffar
Copy link
Contributor

This fixes #42059 ASAN runtime issue. This looks like a GCC bug as reported in #40902 (comment) . Local tests shows that failing ASAN workflow runs after this change

@smuzaffar
Copy link
Contributor Author

test parameters:
workflow = 20034.0

@smuzaffar
Copy link
Contributor Author

please test

@cmsbuild
Copy link
Contributor

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-42077/36056

  • This PR adds an extra 16KB to repository

@cmsbuild
Copy link
Contributor

A new Pull Request was created by @smuzaffar (Malik Shahzad Muzaffar) for master.

It involves the following packages:

  • L1Trigger/Phase2L1Taus (l1)

@epalencia, @aloeliger can you please review it and eventually sign? Thanks.
@Martin-Grunewald, @missirol this is something you requested to watch as well.
@perrotta, @dpiparo, @rappoccio you are the release manager for this.

cms-bot commands are listed here

@smuzaffar
Copy link
Contributor Author

please test for CMSSW_13_2_ASAN_X

@VinInn
Copy link
Contributor

VinInn commented Jun 24, 2023

The use of regex for such a simple text replacement looks an overkill to me.

@cmsbuild
Copy link
Contributor

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-3a1d9e/33371/summary.html
COMMIT: 07f077b
CMSSW: CMSSW_13_2_X_2023-06-23-2300/el8_amd64_gcc11
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/42077/33371/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

  • You potentially removed 10 lines from the logs
  • Reco comparison results: 13 differences found in the comparisons
  • DQMHistoTests: Total files compared: 49
  • DQMHistoTests: Total histograms compared: 3295604
  • DQMHistoTests: Total failures: 9
  • DQMHistoTests: Total nulls: 0
  • DQMHistoTests: Total successes: 3295573
  • DQMHistoTests: Total skipped: 22
  • DQMHistoTests: Total Missing objects: 0
  • DQMHistoSizes: Histogram memory added: 0.0 KiB( 48 files compared)
  • Checked 212 log files, 163 edm output root files, 49 DQM output files
  • TriggerResults: no differences found

@aloeliger
Copy link
Contributor

+l1

@cmsbuild
Copy link
Contributor

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, @rappoccio (and backports should be raised in the release meeting by the corresponding L2)

@cmsbuild
Copy link
Contributor

-1

Failed Tests: RelVals
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-3a1d9e/33372/summary.html
COMMIT: 07f077b
CMSSW: CMSSW_13_2_ASAN_X_2023-06-23-2300/el8_amd64_gcc11
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/42077/33372/install.sh to create a dev area with all the needed externals and cmssw changes.

RelVals

  • 20034.020034.0_TTbar_14TeV+2026D86/step2_TTbar_14TeV+2026D86.log
  • 24834.024834.0_TTbar_14TeV+2026D98/step2_TTbar_14TeV+2026D98.log
  • 23234.023234.0_TTbar_14TeV+2026D94/step2_TTbar_14TeV+2026D94.log
Expand to see more relval errors ...

@smuzaffar
Copy link
Contributor Author

hold

looks like this did not fix the ASAN runtime errors

@cmsbuild
Copy link
Contributor

Pull request has been put on hold by @smuzaffar
They need to issue an unhold command to remove the hold state or L1 can unhold it for all

@smuzaffar
Copy link
Contributor Author

The use of regex for such a simple text replacement looks an overkill to me.

I agree, may be we just use std::string replace here.

@makortel
Copy link
Contributor

hold

looks like this did not fix the ASAN runtime errors

The failures are different from #42059

==3070823==ERROR: AddressSanitizer: heap-buffer-overflow on address 0x60300b5678dc at pc 0x14db50c035a4 bp 0x7fff7a046290 sp 0x7fff7a046288
WRITE of size 2 at 0x60300b5678dc thread T0
    #0 0x14db50c035a3 in TauNNIdHW::compute(l1t::PFCandidate const&, std::vector<l1t::PFCandidate, std::allocator<l1t::PFCandidate> >&) (/cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02790/el8_amd64_gcc11/cms/cmssw/CMSSW_13_2_ASAN_X_2023-06-23-2300/lib/el8_amd64_gcc11/libL1TriggerPhase2L1ParticleFlow.so+0x45c5a3)
...

that would be better to be followed up in a separate issue.

@dan131riley
Copy link
Contributor

hold
looks like this did not fix the ASAN runtime errors

The failures are different from #42059 [...] that would be better to be followed up in a separate issue.

Agreed, the new failures are unrelated to the regex issue, this PR should not be kept on hold because of those new failures. The new ones come from

for (unsigned int i0 = 0; i0 < iParts.size(); i0++) {
if (i0 > fNParticles_)
break;
fPt_.get()[i0] = pt_t(iParts[i0].pt());
fEta_.get()[i0] = etaphi_t(iSeed.eta() - iParts[i0].eta());
etaphi_t lDPhi = etaphi_t(iSeed.phi()) - etaphi_t(iParts[i0].phi());
etaphi_t lMPI = 3.1415;
if (lDPhi > lMPI)
lDPhi = lDPhi - lMPI;
if (lDPhi < -lMPI)
lDPhi = lDPhi + lMPI;
fPhi_.get()[i0] = lDPhi;
fId_.get()[i0] = id_t(iParts[i0].id());
}

where the test at line 92 should '>=', not '>'. I don't know why we have not encountered this before, but I've confirmed that's the problem and will make a new PR.

@dan131riley
Copy link
Contributor

Fix is in #42233

@perrotta
Copy link
Contributor

please test for CMSSW_13_2_ASAN_X with #42233

@perrotta
Copy link
Contributor

test parameters:

@perrotta
Copy link
Contributor

please test for CMSSW_13_2_ASAN_X

@cmsbuild
Copy link
Contributor

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-3a1d9e/33647/summary.html
COMMIT: 07f077b
CMSSW: CMSSW_13_2_ASAN_X_2023-07-10-2300/el8_amd64_gcc11
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week1/cms-sw/cmssw/42077/33647/install.sh to create a dev area with all the needed externals and cmssw changes.

The following merge commits were also included on top of IB + this PR after doing git cms-merge-topic:

You can see more details here:
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-3a1d9e/33647/git-recent-commits.json
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-3a1d9e/33647/git-merge-result

@perrotta
Copy link
Contributor

@smuzaffar together with #42233 the runtime errors in ASAN are gone, as supposed by @dan131riley : I think this can be unheld and merged, then

@smuzaffar
Copy link
Contributor Author

unhold

@cmsbuild
Copy link
Contributor

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, @rappoccio (and backports should be raised in the release meeting by the corresponding L2)

@perrotta
Copy link
Contributor

+1

@cmsbuild cmsbuild merged commit 5aae029 into cms-sw:master Jul 12, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[ASAN] global-buffer-overflow in L1HPSPFTauBuilder constructor

7 participants