[RFC/RFT] b43: A fix for DMA transmission sequence errors

[Chris Vine <chris@cvine.freeserve.co.uk>]

On Wed, 20 Mar 2013 12:26:23 +0100
ISE Development <isedev@gmail.com> wrote:

> On Friday 15 Mar 2013 11:48:42 Larry Finger wrote:
> > On 03/15/2013 01:37 AM, Rafa? Mi?ecki wrote:
> > > OK, so after early tests, this card doesn't support
> > > full-addressing at all.
> > > 
> > > I'll give this patch a try during the weekend.
> > 
> > Please follow Michael's suggestion, and change
> > 
> > 		if (slot == firstused + 2) {
> > 
> > into
> > 
> > 		if (slot == next_slot(ring, next_slot(ring,
> > firstused))) {
> > 
> > This change is needed just in case "firstused + 2" wraps around.
> > 
> > Thanks,
> > 
> > Larry
> 
> Did you get a chance to validate this patch?
> 
> So far on my system, it has been running for over week without
> problems (over 15GB of traffic).

If it helps as a data point, I have run the patch with the wrap-over
change suggested by Michael Busch and with the unlikely->likely
branching correction and it has been working fine for the last 5 days
on the 3.8.2 kernel, with approx 5GB of traffic. Before that, I put
about 10GB through the previous version of the patch as originally sent
by isedev.

Chris