On Tue, Jan 17, 2006 at 10:17:12AM -0500, Amy Griffis wrote:
Hi Steve,
On Mon, Jan 16, 2006 at 12:16:46PM -0500, Steve Grubb wrote:
> On Friday 13 January 2006 20:07, David Woodhouse wrote:
> > This is kernel-2.6.15-1.1826.2.10.2.2_FC5.lspp.6
> >
> > It contains the contents of the audit git tree, Janak's unshare patch
> > sequence, some patches from Steve and Amy's "string fields
interface"
> > patches.
>
> This kernel Oopses immediately on boot. I think its the string1 &
> string2 patches. Here's what I get:
Thanks for the bug report.
I didn't have any issues on boot in my testing, but this kernel
includes some patches I wasn't testing. I'll look into it and get
back to you.
<snip>
> Then with gdb, I find this line:
>
> (gdb) list *0xffffffff801614c4
> 0xffffffff801614c4 is in audit_receive_filter (kernel/auditfilter.c:388).
> 383 audit_free_watch(krule->watch);
> 384 krule->watch = w;
> 385 list_add(&krule->rlist, &w->rules);
> 386 return;
> 387 }
> 388 INIT_LIST_HEAD(&krule->watch->rules);
> 389 list_add(&krule->rlist, &krule->watch->rules);
> 390 list_add(&krule->watch->mlist, &master_watchlist);
> 391 }
>
> So, seems that krule is NULL?
Actually it's krule->watch that is NULL. The code block that adds a
watch needs to be called conditionally based on the presence of
krule->watch.
Looks like I forgot to try adding rules *without* watches after making
some final changes. :-/
Amy