On (27/06/13 10:20), Pavel Březina wrote:
On 06/26/2013 05:32 PM, Lukas Slebodnik wrote:
>On (26/06/13 10:49), Pavel Březina wrote:
>>https://fedorahosted.org/sssd/ticket/1693
>>
>>The reason why sudo show different messages is that SSSD crashes
>>(SIGABRT from talloc) when out of band refresh is finished. For some
>>reason, Nikolai's test is much more likely to reveal this on rhel
>>systems than on fedora, however the SSSD crashes on both systems.
>>
>>How to test it:
>>1. run Nikolai's test attached to bugzilla
>>2. attach gdb to sssd_sudo and hit continue
>>3. wait for it
>>
>>Without the patch, you will receive SIGABRT from talloc. With the
>>patch, you'll get SIGTERM from the test.
>>
>>I'd like to get this to both 1.10 and 1.9.
>
>Patch fixes the crash. There is no other place, where
>tevent_req_set_callback is called with callback
>sudosrv_dp_req_done and last argument NULL.
>
>I have just a question.
>There are two callbacks:
> sudosrv_dp_send_acct_req_done
> sudosrv_dp_req_done
>with pattern "sudosrv_dp.*_done" and new callback has name
>sudosrv_oob_dp_req_done.
>
>Is there any particular reason for this name?
>
>LS
Thanks, I changed it. I also removed the dead code from
sudosrv_dp_req_done() - if cb_ctx == NULL then tevent obviously calls
abort.
sssd_sudo did not crash even with removed dead code.
ACK
LS