Date: Tue, 04 Dec 2001 13:56:06 -0800 From: Andy Sparrow <spadger@best.com> To: Guido van Rooij <guido@gvr.org> Cc: wpaul@freebsd.org, freebsd-mobile@FreeBSD.ORG Subject: Re: pccard kernel config for OmniBook 500 Message-ID: <20011204215606.2C8DD3E25@CRWdog.demon.co.uk> In-Reply-To: Message from Guido van Rooij <guido@gvr.org> of "Wed, 28 Nov 2001 09:47:33 %2B0100." <20011128094733.B23373@gvr.gvr.org>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --]
Guido,
Whilst suspending & resuming repeatedly to test the Maestro3 suspend/resume
patch, I noted that I didn't observe a single hesitation or 'xl0 watchdog
reset' message with your patch applied (which I had applied at the same time).
It now seems to work perfectly from a suspend/resume, I'm very happy.
It almost seems churlish to mention that there's a declaration missing in
xl_suspend() in the patch which prevents it compiling :-)
Can we get a PR for this and get it commited?
Thanks again.
Cheers,
AS
> On Tue, Nov 27, 2001 at 12:06:13PM -0800, Andy Sparrow wrote:
> > Hmmm. On my 6000, sometimes the xl0 doesn't come back, and I need to suspend &
> > resume again to get a link light.
> >
> > Quite often, it won't respond post-resume until I get an 'xl0: watchdog
> > timeout' message, and then the driver re-sets and it's fine. The link light
> > seems to be the clue here, although I'm sure I've seen it recover from "no
> > link light" after a driver reset too... :)
> >
> > Infrequently, I need to reboot to get the NIC into a sane state (although
> > sound causes me to do this more often than the NIC).
> >
> > Fairly often under heavy (network) load, the xl0 will get watchdog timeouts.
> > Other than that, it seems to work OK. The 'fxp0' in the 6100 works like a
> > champ, however...
> >
>
> W.r.t suspend/resume probs: try the attached patch which is currently
> under review.
>
> -Guido
>
> Index: if_xl.c
> ===================================================================
> RCS file: /scratch/cvsup/freebsd/CVS/src/sys/pci/if_xl.c,v
> retrieving revision 1.72.2.8
> diff -u -r1.72.2.8 if_xl.c
> --- if_xl.c 2001/10/27 03:31:34 1.72.2.8
> +++ if_xl.c 2001/11/27 15:42:46
> @@ -220,6 +220,8 @@
> static void xl_stop __P((struct xl_softc *));
> static void xl_watchdog __P((struct ifnet *));
> static void xl_shutdown __P((device_t));
> +static int xl_suspend __P((device_t));
> +static int xl_resume __P((device_t));
> static int xl_ifmedia_upd __P((struct ifnet *));
> static void xl_ifmedia_sts __P((struct ifnet *, struct ifmediareq *));
>
> @@ -266,6 +268,8 @@
> DEVMETHOD(device_attach, xl_attach),
> DEVMETHOD(device_detach, xl_detach),
> DEVMETHOD(device_shutdown, xl_shutdown),
> + DEVMETHOD(device_suspend, xl_suspend),
> + DEVMETHOD(device_resume, xl_resume),
>
> /* bus interface */
> DEVMETHOD(bus_print_child, bus_generic_print_child),
> @@ -2516,6 +2520,7 @@
> printf("xl%d: initialization failed: no "
> "memory for rx buffers\n", sc->xl_unit);
> xl_stop(sc);
> + splx(s);
> return;
> }
>
> @@ -2961,4 +2966,37 @@
> xl_stop(sc);
>
> return;
> +}
> +
> +static int xl_suspend(dev)
> + device_t dev;
> +{
> + struct xl_softc *sc;
> +
> + sc = device_get_softc(dev);
> +
> + s = splimp();
> + xl_stop(sc);
> + splx(s);
> +
> + return(0);
> +}
> +
> +static int xl_resume(dev)
> + device_t dev;
> +{
> + struct xl_softc *sc;
> + struct ifnet *ifp;
> + int s;
> +
> + s = splimp();
> + sc = device_get_softc(dev);
> + ifp = &sc->arpcom.ac_if;
> +
> + xl_reset(sc);
> + if (ifp->if_flags & IFF_UP)
> + xl_init(sc);
> +
> + splx(s);
> + return(0);
> }
>
[-- Attachment #2 --]
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (FreeBSD)
Comment: Exmh version 2.5 07/13/2001
iD8DBQE8DUZ2PHh895bDXeQRAqVbAJ9N5noZ3JmbGuw7yaJ6e8B0lx5C9QCfQ1oE
zUhmgiDySdQBZTY/R15386I=
=uPln
-----END PGP SIGNATURE-----
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20011204215606.2C8DD3E25>