[SSSD] [PATCH 6/6] Fix segfault in GSSAPI reconnect code

Stephen Gallagher sgallagh at redhat.com
Fri May 7 13:03:29 UTC 2010


On 05/07/2010 02:33 AM, Eugene Indenbom wrote:
> Dear Stephen and Dmitri,
>
> We are going in rounds. I have already addressed all this copy-paste
> hell in my patches concerning GSSAPI reconnect.
>
> I would appreciate very much if you look into my patches now and give me
> feedback. The patches can be found at:
> https://fedorahosted.org/pipermail/sssd-devel/2010-April/003113.html
>
> 0003-Fixed-recursive-sdap_handle-disconnect-sequence-from.patch is
> already superseded by Simo's clever cancel-destruction solution and can
> be dropped, but other patches are still valid.
>
> 0004-The-LDAP-ID-backend-connection-logic-has-been-refact.patch actually
> addresses multitude of problems in LDAP request processing including
> copy-paste hell you are trying to fix now.
>
> All patches apply on top off SSSD-1.1.1. If you find them useful as a
> base for further discussion, I will rebase them to sssd-1.2 or master on
> your choice.
>
> I have also posted tickets for all issues resolved in my patch as
> promised earlier.
>
> Eugene Indenbom
>

Thank you for posting those tickets. I discussed them with the rest of 
the team today, and a decision was made to make these the core of the 
SSSD 1.3.0 release, now tentatively scheduled for the end of June. (One 
exception: the expiring LDAP context will be included in 1.2.1, at the 
end of May).

We decided that the best place for these fixes would be 1.3.0 so that we 
would not need to write them twice. (The interface to the sysdb changed 
significantly in the master branch, and it is very difficult to port 
between them right now).

I will work with you personally to get your changes into 1.3.0. I see 
plenty of value in them; all we need to do is get them in line with our 
style guidelines (which I will be updating as we go, to be more clear 
for future contributors).

As for this particular set of changes, I'm still going to resolve this 
particular issue in the way that Dmitri suggests for 1.2.0. That's 
because this is a segfault and needs to be eliminated before we can ship 
1.2.0 (the preview tarball of which is targeted to be released today). 
So please understand that this is not me ignoring your patches, just me 
opting to make a less-disruptive change at the end of a dev cycle. (And 
also taking your advice about reducing code duplication)


-- 
Stephen Gallagher
RHCE 804006346421761

Delivering value year after year.
Red Hat ranks #1 in value among software vendors.
http://www.redhat.com/promo/vendor/



More information about the sssd-devel mailing list