[GCC14]Fix array bound warnings #47634

smuzaffar · 2025-03-20T08:40:34Z

workaround to suppress the array-bound warning in LTO builds

This is an attempt to fix the array bound warning we get in GCC14 IBs [a].
Size of array I m_v[S]; at https://github.com/cms-sw/cmssw/blob/master/HeterogeneousCore/AlpakaInterface/interface/FlexiStorage.h#L21 is 257 via

totbins() at https://github.com/cms-sw/cmssw/blob/master/HeterogeneousCore/CUDAUtilities/interface/HistoContainer.h#L125 return 257 and then https://github.com/cms-sw/cmssw/blob/master/RecoVertex/PixelVertexFinding/plugins/alpaka/clusterTracksByDensity.h#L61-L63 tries to initialize/access m_v[257] which is out of array bound.

Either totbins() should return NHISTS * NBINS instead of NHISTS * NBINS +1 or we loop over 1 less element.

@fwyzard, if this is not the correct fix then feel free to propose a correct fix.

[a] https://cmssdt.cern.ch/SDT/cgi-bin/buildlogs/el8_amd64_gcc14/CMSSW_15_1_X_2025-03-18-2300/RecoVertex/PixelVertexFinding

In member function 'operator[]',
    inlined from 'clusterTracksByDensity' at src/RecoVertex/PixelVertexFinding/plugins/alpaka/clusterTracksByDensity.h:62:17,
    inlined from 'operator().isra' at src/RecoVertex/PixelVertexFinding/plugins/alpaka/clusterTracksByDensity.h:247:29:
  [src/HeterogeneousCore/AlpakaInterface/interface/FlexiStorage.h:14](https://github.com/cms-sw/cmssw/blob/CMSSW_15_1_X_2025-03-18-2300/HeterogeneousCore/AlpakaInterface/interface/FlexiStorage.h#L14):48: warning: array subscript 257 is above array bounds of 'unsigned int[257]' [-Warray-bounds=]
    14 |     constexpr I& operator[](int i) { return m_v[i]; }
      |                                                ^
src/HeterogeneousCore/AlpakaInterface/interface/FlexiStorage.h: In function 'operator().isra':
src/HeterogeneousCore/AlpakaInterface/interface/FlexiStorage.h:21:7: note: while referencing 'm_v'
   21 |     I m_v[S];
      |       ^
In member function 'operator[]',
    inlined from 'clusterTracksByDensity' at src/RecoVertex/PixelVertexFinding/plugins/alpaka/clusterTracksByDensity.h:62:17,
    inlined from 'operator().isra' at src/RecoVertex/PixelVertexFinding/plugins/alpaka/clusterTracksByDensity.h:247:29:
  [src/HeterogeneousCore/AlpakaInterface/interface/FlexiStorage.h:14](https://github.com/cms-sw/cmssw/blob/CMSSW_15_1_X_2025-03-18-2300/HeterogeneousCore/AlpakaInterface/interface/FlexiStorage.h#L14):48: warning: array subscript 257 is above array bounds of 'unsigned int[257]' [-Warray-bounds=]
    14 |     constexpr I& operator[](int i) { return m_v[i]; }

cmsbuild · 2025-03-20T08:41:02Z

cms-bot internal usage

cmsbuild · 2025-03-20T08:43:45Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47634/44166

There are other open Pull requests which might conflict with changes you have proposed:
- File RecoVertex/PixelVertexFinding/plugins/alpaka/clusterTracksByDensity.h modified in PR(s): Level-1 Scouting DQM #47627

cmsbuild · 2025-03-20T08:44:06Z

A new Pull Request was created by @smuzaffar for master.

It involves the following packages:

RecoVertex/PixelVertexFinding (reconstruction)

@cmsbuild, @jfernan2, @mandrenguyen can you please review it and eventually sign? Thanks.
@GiacomoSguazzoni, @VinInn, @VourMa, @dgulhan, @fabiocos, @martinamalberti, @missirol, @mmusich, @mtosi, @rovere this is something you requested to watch as well.
@antoniovilela, @mandrenguyen, @rappoccio, @sextonkennedy you are the release manager for this.

cms-bot commands are listed here

smuzaffar · 2025-03-20T08:48:18Z

please test for el8_amd64_gcc14

fwyzard · 2025-03-20T09:07:44Z

@fwyzard, if this is not the correct fix then feel free to propose a correct fix.

I'll have a look.

cmsbuild · 2025-03-20T09:48:19Z

-1

Failed Tests: UnitTests RelVals AddOn
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-f42cc9/45098/summary.html
COMMIT: a27faa3
CMSSW: CMSSW_15_1_X_2025-03-19-2300/el8_amd64_gcc14
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week1/cms-sw/cmssw/47634/45098/install.sh to create a dev area with all the needed externals and cmssw changes.

Unit Tests

I found 2 errors in the following unit tests:

---> test deviceVertexFinderByDensity_tSerialSync had ERRORS
---> test deviceVertexFinderOneKernel_tSerialSync had ERRORS

RelVals

145.408A fatal system signal has occurred: abort signal
145.713A fatal system signal has occurred: abort signal
145.301A fatal system signal has occurred: abort signal

Expand to see more relval errors ...

AddOn Tests

hlt_mc_GRun

A fatal system signal has occurred: abort signal

hlt_mc_GRun

A fatal system signal has occurred: abort signal

hlt_mc_HIon

A fatal system signal has occurred: abort signal

Expand to see more addon errors ...

fwyzard · 2025-03-21T13:57:42Z

I'm still going through the details of the indices and ranges, but to first order I think GCC is wrong.

I've run the deviceVertexFinderByDensity_t tests for different back-ends after instrumenting the code of the clusterTracksByDensity function with

    for (auto j : cms::alpakatools::uniform_elements(acc, Hist::totbins())) {
      printf("set hist.off[%d] to 0\n", j);
      hist.off[j] = 0;
    }

On the CPU back-end:

./deviceVertexFinderByDensity_tSerialSync | grep 'set hist.off' | sort -k2.11,2.13g -u
set hist.off[0] to 0
set hist.off[1] to 0
set hist.off[2] to 0
set hist.off[3] to 0
set hist.off[4] to 0
...
set hist.off[253] to 0
set hist.off[254] to 0
set hist.off[255] to 0
set hist.off[256] to 0

On the CUDA back-end:

set hist.off[0] to 0
set hist.off[1] to 0
set hist.off[2] to 0
set hist.off[3] to 0
set hist.off[4] to 0
...
set hist.off[253] to 0
set hist.off[254] to 0
set hist.off[255] to 0
set hist.off[256] to 0

We never try to write to index 257.

fwyzard · 2025-03-21T14:05:54Z

In fact, with the proposed changes, the tests fail :-/

smuzaffar · 2025-03-24T09:13:17Z

@fwyzard , yes I also have added some debug print outs and index 257 was never accessed. Yes this seems like compiler bug

Previously we have suppressed this for RecoTracker/PixelSeeding via removing the array-bounds warnings from the compilation flags ( using pragma to suppress this warnings did not work for LTO). So I can update this PR to add <use name="no-array-bounds-flag" for="alpaka/serial"/>

cmsbuild · 2025-03-24T09:19:00Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47634/44209

There are other open Pull requests which might conflict with changes you have proposed:
- File RecoVertex/PixelVertexFinding/plugins/BuildFile.xml modified in PR(s): Remove legacy CUDA modules for pixel track and vertex reconstruction #45853
- File RecoVertex/PixelVertexFinding/test/BuildFile.xml modified in PR(s): Remove legacy CUDA modules for pixel track and vertex reconstruction #45853

cmsbuild · 2025-03-24T09:19:23Z

Pull request #47634 was updated. @cmsbuild, @jfernan2, @mandrenguyen can you please check and sign again.

smuzaffar · 2025-03-24T09:23:56Z

please test for el8_amd64_gcc14

fwyzard · 2025-03-24T10:37:56Z

+heterogeneous

smuzaffar · 2025-03-24T13:20:55Z

please test

lets run the tests for production arch too

cmsbuild · 2025-03-24T15:35:45Z

+1

Size: This PR adds an extra 20KB to repository
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-f42cc9/45175/summary.html
COMMIT: 8643ca7
CMSSW: CMSSW_15_1_X_2025-03-24-1100/el8_amd64_gcc12
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/47634/45175/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

You potentially added 1 lines to the logs
Reco comparison results: 579 differences found in the comparisons
DQMHistoTests: Total files compared: 52
DQMHistoTests: Total histograms compared: 4137425
DQMHistoTests: Total failures: 2744
DQMHistoTests: Total nulls: 0
DQMHistoTests: Total successes: 4134661
DQMHistoTests: Total skipped: 20
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: 0.0 KiB( 51 files compared)
Checked 223 log files, 194 edm output root files, 52 DQM output files
TriggerResults: no differences found

jfernan2 · 2025-03-24T17:38:59Z

+1

cmsbuild · 2025-03-24T17:39:19Z

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

cmsbuild · 2025-03-24T18:17:16Z

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-f42cc9/45166/summary.html
COMMIT: 8643ca7
CMSSW: CMSSW_15_1_X_2025-03-21-1100/el8_amd64_gcc14
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week1/cms-sw/cmssw/47634/45166/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

You potentially added 239 lines to the logs
ROOTFileChecks: Some differences in event products or their sizes found
Reco comparison results: 107708 differences found in the comparisons
DQMHistoTests: Total files compared: 52
DQMHistoTests: Total histograms compared: 4137425
DQMHistoTests: Total failures: 503001
DQMHistoTests: Total nulls: 418
DQMHistoTests: Total successes: 3633986
DQMHistoTests: Total skipped: 20
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: 8.831000000000001 KiB( 51 files compared)
DQMHistoSizes: changed ( 10224.0 ): 0.415 KiB SiStrip/MechanicalView
DQMHistoSizes: changed ( 13034.0 ): 0.938 KiB SiStrip/MechanicalView
DQMHistoSizes: changed ( 140.045,... ): -0.004 KiB JetMET/SUSYDQM
DQMHistoSizes: changed ( 141.042 ): 0.043 KiB JetMET/SUSYDQM
DQMHistoSizes: changed ( 145.014,... ): 0.008 KiB JetMET/SUSYDQM
DQMHistoSizes: changed ( 145.202 ): 0.023 KiB JetMET/SUSYDQM
DQMHistoSizes: changed ( 145.408 ): -0.016 KiB JetMET/SUSYDQM
DQMHistoSizes: changed ( 145.604 ): 0.090 KiB JetMET/SUSYDQM
DQMHistoSizes: changed ( 145.713 ): -0.008 KiB JetMET/SUSYDQM
DQMHistoSizes: changed ( 17034.0 ): 3.490 KiB SiStrip/MechanicalView
DQMHistoSizes: changed ( 250202.181 ): ...
Checked 223 log files, 194 edm output root files, 52 DQM output files
TriggerResults: found differences in 23 / 50 workflows

mandrenguyen · 2025-03-25T19:21:15Z

+1

cmsbuild added this to the CMSSW_15_1_X milestone Mar 20, 2025

cmsbuild added reconstruction-pending pending-signatures tests-pending orp-pending code-checks-pending tracking labels Mar 20, 2025

cmsbuild added code-checks-approved and removed code-checks-pending labels Mar 20, 2025

cmsbuild added tests-started and removed tests-pending labels Mar 20, 2025

cmsbuild added tests-rejected and removed tests-started labels Mar 20, 2025

PixelVertexFinding: workaround for -Warray-bounds/LTO builds

8643ca7

smuzaffar force-pushed the gcc14-array-bound-clusterTracksByDensity branch from a27faa3 to 8643ca7 Compare March 24, 2025 09:16

cmsbuild added tests-pending code-checks-pending and removed tests-rejected code-checks-approved labels Mar 24, 2025

cmsbuild removed the code-checks-pending label Mar 24, 2025

cmsbuild added the code-checks-approved label Mar 24, 2025

cmsbuild added tests-started and removed tests-pending labels Mar 24, 2025

cmsbuild added tests-approved and removed tests-started labels Mar 24, 2025

cmsbuild added reconstruction-approved fully-signed and removed reconstruction-pending pending-signatures labels Mar 24, 2025

cmsbuild added orp-approved and removed orp-pending labels Mar 25, 2025

cmsbuild merged commit 538ac87 into cms-sw:master Mar 25, 2025
17 checks passed

[GCC14]Fix array bound warnings #47634

[GCC14]Fix array bound warnings #47634

Uh oh!

Conversation

smuzaffar commented Mar 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cmsbuild commented Mar 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cmsbuild commented Mar 20, 2025

Uh oh!

cmsbuild commented Mar 20, 2025

Uh oh!

smuzaffar commented Mar 20, 2025

Uh oh!

fwyzard commented Mar 20, 2025

Uh oh!

cmsbuild commented Mar 20, 2025

Unit Tests

RelVals

AddOn Tests

Uh oh!

fwyzard commented Mar 21, 2025

Uh oh!

fwyzard commented Mar 21, 2025

Uh oh!

smuzaffar commented Mar 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cmsbuild commented Mar 24, 2025

Uh oh!

cmsbuild commented Mar 24, 2025

Uh oh!

smuzaffar commented Mar 24, 2025

Uh oh!

fwyzard commented Mar 24, 2025

Uh oh!

smuzaffar commented Mar 24, 2025

Uh oh!

cmsbuild commented Mar 24, 2025

Comparison Summary

Uh oh!

jfernan2 commented Mar 24, 2025

Uh oh!

cmsbuild commented Mar 24, 2025

Uh oh!

cmsbuild commented Mar 24, 2025

Comparison Summary

Uh oh!

mandrenguyen commented Mar 25, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

smuzaffar commented Mar 20, 2025 •

edited

Loading

cmsbuild commented Mar 20, 2025 •

edited

Loading

smuzaffar commented Mar 24, 2025 •

edited

Loading