[Pkg-openldap-devel] back..

[Steve Langasek]

Hi guys,

On Fri, Mar 02, 2007 at 05:29:21PM -0800, Russ Allbery wrote:
> Quanah Gibson-Mount <quanah at stanford.edu> writes:

> > I had a discussion with Russ, and I'm willing to give this a try again.

Welcome back. :)

> > I would really like to see the connection code issue in 2.3.30 be
> > patched. A diff is available here:

> > <http://www.openldap.org/devel/cvsweb.cgi/servers/slapd/connection.c.diff?hideattic=1&r1=text&tr1=1.296.2.21&r2=text&tr2=1.296.2.23&f=h>

> > for the colored diff or

> > <http://www.openldap.org/devel/cvsweb.cgi/servers/slapd/connection.c.diff?hideattic=1&r1=text&tr1=1.296.2.21&r2=text&tr2=1.296.2.23&f=c>

> > for the context diff

... and
<http://www.openldap.org/devel/cvsweb.cgi/servers/slapd/connection.c.diff?hideattic=1&r1=text&tr1=1.296.2.21&r2=text&tr2=1.296.2.23&f=u>
for a unified diff. :)

> > This particular issue plagued my servers for several months.  Without
> > this fix, slapd will periodically segfault due to a race condition.

> This patch looks fairly small and safe to me, but I'm not the one who
> would have to bless it to get into the release.  Steve, opinions with your
> release manager hat on?  Basically, it moves more of a checking operation
> inside a mutex lock.

The only thing I'm not clear on is why this particular check:

+			if( connections[i].c_struct_state == SLAP_C_USED ) {

is moved inside the mutex, given that there's another check against
c_struct_state immediately preceding the mutex.

It's pretty clear what this code change is meant to do in general, though,
so yeah, I wouldn't have a problem with letting it in.

Cheers,
-- 
Steve Langasek                   Give me a lever long enough and a Free OS
Debian Developer                   to set it on, and I can move the world.
vorlon at debian.org                                   http://www.debian.org/