On Thu, Sep 10, 2009 at 01:04:32PM -0500, Scott Wood wrote:
> Anton Vorontsov wrote:
> >MPC8360 QE UCC ethernet controllers hang when changing link duplex
> >under a load (a bit of NFS activity is enough).
> >
> > PHY: m...@e0102120:00 - Link is Up - 1000/Full
> > sh-3.00# ethtool -s eth0 speed 100 duplex half autoneg off
> > PHY: m...@e0102120:00 - Link is Down
> > PHY: m...@e0102120:00 - Link is Up - 100/Half
> > NETDEV WATCHDOG: eth0 (ucc_geth): transmit queue 0 timed out
> > ------------[ cut here ]------------
> > Badness at c01fcbd0 [verbose debug info unavailable]
> > NIP: c01fcbd0 LR: c01fcbd0 CTR: c0194e44
> > ...
> >
> >The cure is to disable the controller before changing speed/duplex
> >and enable it afterwards.
> >
> >Since ugeth_graceful_stop_{tx,rx} now may be called from an atomic
> >context, switch the two functions from msleep() to mdelay().
>
> Ouch.
Yeah, right... delaying for 10ms with irqs off isn't good.
> Can we put this in a workqueue or something?
adjust_link() itself isn't called from an atomic context.
It's we are grabbing ugeth->lock, i.e. a spinlock. I don't see
why the lock is needed in adjust_link() in its current form,
but if we're going to disable the controller for some time,
we'll have to make sure that no start_xmit() or NAPI is running,
scheduled or will be scheduled until we say so.
I think that lock-less, and thus completely sleep-able variant
of adjust_link is doable.
Thanks,
--
Anton Vorontsov
email: cbouatmai...@gmail.com
irc://irc.freenode.net/bd2
_______________________________________________
Linuxppc-dev mailing list
Linuxppc-dev@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/linuxppc-dev
