Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 23 Jun 2003 19:16:05 -0700 (PDT)
From:      Nate Lawson <nate@root.org>
To:        Sam Leffler <sam@errno.com>
Cc:        cvs-all@FreeBSD.org
Subject:   Re: cvs commit: src/sys/dev/fxp if_fxp.c
Message-ID:  <20030623191423.X68807@root.org>
In-Reply-To: <8618913.1056386895@melange.errno.com>
References:  <200306232323.h5NNNnJZ092748@repoman.freebsd.org> <8618913.1056386895@melange.errno.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 23 Jun 2003, Sam Leffler wrote:
> I believe this also makes things work for bridging (and other similar
> "middle ware" which may cause start to be entered directly from a recv.
> However my approach to dealing with this problem in other drivers was to
> use two different locks, one for the xmit side and one for the recv side.
> I'm not certain what I've done handles everything but it certainly
> eliminates the up-down locking per-packet that's required if you stick to
> one lock.

I'd be interested in hearing how you address these along with the softc
lock case.  For example, both tx and rcv paths need to serialize with the
detach method if someone unloads a driver.  I think right now I will have
a race in that case since I drop the lock.

-Nate



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030623191423.X68807>