On Tue, 2012-12-11 at 14:34 +0100, Jakub Hrozek wrote:
> On Mon, Dec 10, 2012 at 12:19:06PM -0500, Simo Sorce wrote:
> > On Mon, 2012-12-10 at 18:10 +0100, Jakub Hrozek wrote:
> > > On Mon, Dec 10, 2012 at 09:06:16AM -0500, Simo Sorce wrote:
> > > > On Mon, 2012-12-10 at 08:39 +0100, Jakub Hrozek wrote:
> > > > > Coverity bug again.
> > > > >
> > > > >
https://fedorahosted.org/sssd/ticket/1704
> > > >
> > > > Any reason why you do not close the file anymore in a lock error ?
> > > >
> > > > I know the caller will also cleanup but I would expect that if the
> > > > create function fails no file is open and no file is on disk ...
> > > >
> > >
> > > Hmm, did I attach the wrong patch? In the version I have in tree, there
> > > is close(mc_ctx->fd); called in case sss_br_lock_file() fails. It's
not
> > > very readable from the diff itself, but in the source, the close would be
> > > on line 721 in src/responder/nss/nsssrv_mmap_cache.c
> > >
> > > Anyway, I'm attaching the patch again.
> >
> > Ah sorry, it is there, but I was also wondering if you shouldn't call
> > unlink ?
> > After all if you fail here you just have created an empty file, and I am
> > not sure we want to leave it around.
> > I know the code did not unlink but given you are fixing this already ...
> >
> > Simo.
>
> Yes, that makes sense. A new patch is attached.
ACK
Simo.