Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly
Appearance settings

Comments

Close side panel

Workaround crash in track-extension#402

Merged
jgrosseo merged 2 commits intomasterAliceO2Group/O2Physics:masterfrom
ktf-patch-3AliceO2Group/O2Physics:ktf-patch-3Copy head branch name to clipboard
Dec 8, 2021
Merged

Workaround crash in track-extension#402
jgrosseo merged 2 commits intomasterAliceO2Group/O2Physics:masterfrom
ktf-patch-3AliceO2Group/O2Physics:ktf-patch-3Copy head branch name to clipboard

Conversation

@ktf
Copy link
Member

@ktf ktf commented Dec 8, 2021

Without this we get a segfault on exit (probably due to a double delete) on the grid. Moreover, when compiling with -fsanitize=address it complains about:

[47474:track-extension]: ==47474==ERROR: AddressSanitizer: SEGV on unknown address 0x7f893300006d (pc 0x00000041ef1c bp 0x60b000236000 sp 0x7fffaf0c9f48 T0)
[47474:track-extension]: ==47474==The signal is caused by a READ memory access.
[47474:track-extension]:     #0 0x41ef1c in std::_Sp_counted_ptr<TGeoManager*, (__gnu_cxx::_Lock_policy)2>::_M_dispose() /home/alitrain/eulisse/sw2/ubuntu1804_x86-64/GCC-Toolchain/v10.2.0-alice2-1/include/c++/10.2.0/bits/shared_ptr_base.h:380

With this change in, address sanitizer does not complain.

@ktf ktf requested review from a team, alibuild, iarsene and jgrosseo as code owners December 8, 2021 16:19
@ktf
Copy link
Member Author

ktf commented Dec 8, 2021

@shahor02 @sawenzel any idea what could be the issue?

@ktf
Copy link
Member Author

ktf commented Dec 8, 2021

cc @jgrosseo

@shahor02
Copy link
Contributor

shahor02 commented Dec 8, 2021

It does not complain because the geometry is not loaded. Will check what is the problem with loading, the TGeoManager is quite a beast...

@victor-gonzalez
Copy link
Collaborator

victor-gonzalez commented Dec 8, 2021

Could it be caused by the BasicCCDBManager being created within the task scope?
When the task gets out of scope the CCDB manager has to be destroyed and the double delete could happen
Perhaps having it at the global scope will fix it?

@shahor02
Copy link
Contributor

shahor02 commented Dec 8, 2021

Hi @victor-gonzalez
no, the problem is in the caching of gGeoManager by the BasicCCDBManager, apparently on exit the root garbage collector deletes the gGeoManager before the destructor of the BasicCCDBManager (and its cache) is called. Will see what can be done.

@jgrosseo jgrosseo enabled auto-merge (squash) December 8, 2021 18:19
@victor-gonzalez
Copy link
Collaborator

victor-gonzalez commented Dec 8, 2021

@shahor02 I see!
Thanks!
Will the propagation work fine without loading the geometry?

@shahor02
Copy link
Contributor

shahor02 commented Dec 8, 2021

Will the propagation work fine without loading the geometry?

Depends: with yet unmerged AliceO2Group/AliceO2#7810 one can instantiate the Propagator w/o creating geometry, but the propagation will crash if called with USEMatCorrTGeo

@shahor02
Copy link
Contributor

shahor02 commented Dec 8, 2021

In any case, I would not merge this PR, I am trying to find a solution. As a workaround, for the moment you can set ccdb.setCach(false) when setting up the manager.

@jgrosseo jgrosseo merged commit cfc2e3d into master Dec 8, 2021
@jgrosseo jgrosseo deleted the ktf-patch-3 branch December 8, 2021 19:44
@shahor02 shahor02 mentioned this pull request Dec 8, 2021
Merged
jgrosseo added a commit that referenced this pull request Dec 9, 2021
@jgrosseo
Revert "Workaround crash in track-extension (#402)"
d89cb0f 
This reverts commit cfc2e3d.
@jgrosseo jgrosseo mentioned this pull request Dec 9, 2021
Closed
jgrosseo added a commit to jgrosseo/O2Physics that referenced this pull request Dec 9, 2021
@jgrosseo
Revert "Workaround crash in track-extension (AliceO2Group#402)"
8db3b55 
This reverts commit cfc2e3d.
jgrosseo added a commit that referenced this pull request Dec 9, 2021
@jgrosseo
Revert "Workaround crash in track-extension (#402)" (#406)
8269ebf 
This reverts commit cfc2e3d.
jezwilkinson pushed a commit to jezwilkinson/O2Physics that referenced this pull request Dec 14, 2021
@ktf @jezwilkinson
Workaround crash in track-extension (AliceO2Group#402)
8e3d24b
jezwilkinson pushed a commit to jezwilkinson/O2Physics that referenced this pull request Dec 14, 2021
@jgrosseo @jezwilkinson
Revert "Workaround crash in track-extension (AliceO2Group#402)" (Alic…
099087b 
…eO2Group#406)

This reverts commit cfc2e3d.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@jgrosseo jgrosseo jgrosseo approved these changes

@alibuild alibuild Awaiting requested review from alibuild alibuild is a code owner

@iarsene iarsene Awaiting requested review from iarsene iarsene is a code owner

Assignees

No one assigned

Labels

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@ktf @shahor02 @victor-gonzalez @jgrosseo
Morty Proxy This is a proxified and sanitized view of the page, visit original site.