On Tue, Apr 18, 2000 at 02:01:09PM +0200, Andre Oppermann wrote:
> Peter van Dijk wrote:
> >
> > On Mon, Apr 17, 2000 at 10:31:51PM +0300, Gabriel Ambuehl wrote:
> > > Hello Peter,
> > > >> code and not part of the stable branch, I'd avoid using it for any production
> > > >> servers till it's migrated to the stable tree.
> > > > 4.0 went STABLE a couple of weeks ago.
> > >
> > > So should I go for 4.0 stable directly instead of 4.0 Release?
> >
> > You can't. There is no 4.0-STABLE release. That's wat 4.0-RELEASE is called
> > RELEASE for.
> >
> > Get 4.0-RELEASE mfsroot and kern disks, FTP-install 4.0-RELEASE, and then
> > cvsup to 4.0-STABLE and make world and install a new kernel.
>
> There is a partial fix in FreeBSD-4.0-RELEASE.
>
> For more information see PR kern/17311:
>
> http://www.freebsd.org/cgi/query-pr.cgi?pr=17311
>
> CVS Log for src/sys/net/if.c:
> 1.85 Mon Feb 28 19:30:25 2000 UTC by guido
> CVS Tags: RELENG_4_0_0_RELEASE, RELENG_4_BP, HEAD, RELENG_4
> Diffs to 1.84
>
> This fixes a problem where the SIOCGIFCONF ioctl goes wrong. This
> is triggered when qmail is used with INET6 enabled. The bug
This has also happened to me before INET6 was committed to
-current IIRC. Thus this commit message is at least bogus..
> manifests itself in that the space variable can become negative
> and that in the comparison in the guards of the 2 loops, this was
> not noticed because sizeof() returns an unsigned and thus the signed
> variable gets promoted to unsigned. I decided not to make space
> unsigned because I think we should guard against this from happening.
> Thus panic() in case space becomes negative.
>
> --
> Andre
- Sascha
