You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@trafficserver.apache.org by "B Wyatt (JIRA)" <ji...@apache.org> on 2012/08/29 21:55:07 UTC

[jira] [Resolved] (TS-1425) Crash (rare) when an origin server disconnects mid POST request upload (client->tunnel->OS)

     [ https://issues.apache.org/jira/browse/TS-1425?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

B Wyatt resolved TS-1425.
-------------------------

    Resolution: Fixed

Git commit: 7c2a60873aab331ccd6bc02b5c93105aeb0c4ca3
                
> Crash (rare) when an origin server disconnects mid POST request upload (client->tunnel->OS) 
> --------------------------------------------------------------------------------------------
>
>                 Key: TS-1425
>                 URL: https://issues.apache.org/jira/browse/TS-1425
>             Project: Traffic Server
>          Issue Type: Bug
>          Components: HTTP, Network
>    Affects Versions: 3.2.0
>         Environment: ATS 3.2 proxying a POST request to an unstable server
>            Reporter: B Wyatt
>            Assignee: B Wyatt
>
> When an Origin server terminates the connection during a post upload, the cache must reset and re-create the tunnel in order to deliver the 50X error message. 
> In doing so we deallocate the buffers either being used to read the POST data from the client or support a lingering read for client side aborts without cancelling either of these reads.
> Best case scenario, the buffer is still deallocated the next time read_from_net is called for this vc and the fact that it has been zero'd out puts disables the read (because it thinks its out of buffers space).
> bad case scenario, the buffer has been re-alloc'd when read_from_net is called and we potentially corrupt another buffer with data from the wire
> worst case scenario, the buffer has been re-alloc'd when we enter read_from_net and modified/freed before the call to fill the buffer resulting in a segfault and trashed data.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira