Re: [PATCH] fix NULL pointer + success return in route lookup path

[Jarek Poplawski <jarkao2@gmail.com>]

On Mon, Jun 22, 2009 at 12:59:50PM +0400, Alexey Kuznetsov wrote:
> On Mon, Jun 22, 2009 at 05:43:15AM +0000, Jarek Poplawski wrote:
> > Maybe it can work, but it needs a thorough checking now and adds a new
> > code path to track later while looking for bugs. So, I wonder if it's
> > not better to link such dsts in rt_intern_hash anyway, probably as a
> > separate list, scanned only for expired entries.
> 
> Such a list already exists, it is gc list in core/dst.c.
> 
> The fix to the problem could be replacing rt_drop() with rt_free()
> (adding rt_free() after the patch, which deleted rt_drop()), something like:
> 
> 	if (!rt_caching(dev_net(rt->u.dst.dev))) {
> 		/* ..... */
> +		rt_free(rt);
> 		goto report_and_exit;
> 	}
> 
> rt_free() will put the route on that gc list and it will be releases
> as soon as refcnt becomes 0.

One little doubt would be RCU: if it's currently used in rt_free, and
some code depends on it, there would be a change: rt could be freed
just after atomic dec, without waiting for RCU yet. The second one is
about timing: freeing this always from a workqueue could probably
make a problem if softirqs are often disabled.

Jarek P.