You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@kudu.apache.org by "Todd Lipcon (JIRA)" <ji...@apache.org> on 2018/05/08 15:32:00 UTC

[jira] [Commented] (KUDU-2433) Possible stack-smashing bug

    [ https://issues.apache.org/jira/browse/KUDU-2433?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16467562#comment-16467562 ] 

Todd Lipcon commented on KUDU-2433:
-----------------------------------

I commented on the review, but also noting here: I dont think this is necessarily indicative of a corrupt stack. More likely it was just an unwinding failure due to something in the stack from a library without frame pointers. The unwinder can give a NULL frame in that case. So, the UBSAN error is just a false positive

> Possible stack-smashing bug
> ---------------------------
>
>                 Key: KUDU-2433
>                 URL: https://issues.apache.org/jira/browse/KUDU-2433
>             Project: Kudu
>          Issue Type: Bug
>            Reporter: Mike Percy
>            Priority: Major
>         Attachments: client-negotiation-failover-itest.txt
>
>
> Edit: ClientFailoverOnNegotiationTimeoutITest.Kudu1580ConnectToTServer failed in ASAN mode with what appeared to be a flake, but after more investigation (see attached log file) there appears to be some case where the PC was set to NULL when we were trying to symbolize the stack. Unfortunately a UBSAN error, triggered when we decremented the NULL pointer, masked the original problem in this case.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)