[
https://issues.apache.org/jira/browse/YARN-7857?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16347175#comment-16347175
]
Jim Brennan commented on YARN-7857:
-----------------------------------
Thanks [[email protected]] for the detailed analysis!
I am not suggesting that we revert the fix from YARN-7796 - it clearly resolves
the failure we were seeing and works in all cases.
The proposal in this Jira is to remove the {{-fstack-check}} flag because it
has been shown to cause binary incompatibility issues, depending on the version
of gcc a binary is compiled with and the OS it is run on.
{quote}As a conclusion, the stack check code seems to be legitimate. However,
the code might address the same memory later ending up with the same crash
without stack checking.
{quote}
I'm not sure I follow this? It sounds like you've shown that the size of the
buffer matters - 110KB buffer compiled on RHEL 6 with {{-fstack-check}} works
on RHEL 7 while it fails for a 128 KB buffer. But the 128 KB buffer works when
compiled on RHEL 7 (with or without {{-fstack-check}}). I agree that it may
be that the RHEL 6 version of the stack checking code is tripping some kernel
protection when the buffer is big enough. The RHEL 7 version of the stack
checking code does not. That seems like an incompatibility.
My concern is that if we leave the {{-fstack-check}} flag there, some future
change may cause a similar problem to the one we fixed in YARN-7796.
> -fstack-check compilation flag causes binary incompatibility for
> container-executor between RHEL 6 and RHEL 7
> -------------------------------------------------------------------------------------------------------------
>
> Key: YARN-7857
> URL: https://issues.apache.org/jira/browse/YARN-7857
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
> Affects Versions: 3.0.0
> Reporter: Jim Brennan
> Assignee: Jim Brennan
> Priority: Major
>
> The segmentation fault in container-executor reported in [YARN-7796] appears
> to be due to a binary compatibility issue with the {{-fstack-check}} flag
> that was added in [YARN-6721]
> Based on my testing, a container-executor (without the patch from
> [YARN-7796]) compiled on RHEL 6 with the -fstack-check flag always hits this
> segmentation fault when run on RHEL 7. But if you compile without this flag,
> the container-executor runs on RHEL 7 with no problems. I also verified this
> with a simple program that just does the copy_file.
> I think we need to either remove this flag, or find a suitable alternative.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
