On Fri, Oct 28, 2005 at 09:58:20AM -0400, Greg Troxel wrote:
>   "No buffer space available" problem still exist.
> 
> Note that this doesn't necessarily indicate a lack of memory.  It's
> very likely that the output queue is at the # of packets limit.  If
> I'm reading the code correctly, that's 256.  In other words, probably
> adding a new packet fails because the queue is at the defined limit,
> not because memory for the packet can't be obtained.

Yes, that's it. And this is because the chip can't send packets (or fails
to acknowledge the packets sent).

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--