Re: [PATCH] Revert: audit: Fix check of return value of strnlen_user()

[Paul Moore <pmoore@redhat.com>]

On Wednesday, July 08, 2015 12:29:43 PM Steven Rostedt wrote:
> On Wed, 8 Jul 2015 12:02:49 -0400
> 
> Steven Rostedt <rostedt@goodmis.org> wrote:
> > Well, any testing will have to wait. The reason I found this is because
> > it caused my own tests to fail for a bug fix I'm testing (unrelated to
> > this) that I'm getting ready to send to you. My box to run this on is
> > back to running those tests, which can take several hours.
> 
> Oh well, my tests just stumbled over another unrelated 4.2-rc1 bug (I
> need to dig into this one now :-( ). But that freed up my machine to
> test this.
> 
> I tested the following patch. Feel free to author it yourself and just
> add my reported/tested-by tags, or give it to me. Either way, I don't
> care. I just want it fixed so that it doesn't make my own tests fail.
> 
> Thanks!

Acked-by: Paul Moore <pmoore@redhat.com>

Sorry to be late in replying, and I see that this is already in Linus' tree so 
the ack above it probably a bit pointless, but thanks for reporting this and 
providing a fix.

> diff --git a/kernel/auditsc.c b/kernel/auditsc.c
> index 09c65640cad6..e85bdfd15fed 100644
> --- a/kernel/auditsc.c
> +++ b/kernel/auditsc.c
> @@ -1021,8 +1021,7 @@ static int audit_log_single_execve_arg(struct
> audit_context *context, * for strings that are too long, we should not have
> created
>  	 * any.
>  	 */
> -	if (unlikely((len == 0) || len > MAX_ARG_STRLEN - 1)) {
> -		WARN_ON(1);
> +	if (WARN_ON_ONCE(len < 0 || len > MAX_ARG_STRLEN - 1)) {
>  		send_sig(SIGKILL, current, 0);
>  		return -1;
>  	}

-- 
paul moore
security @ redhat