Subject: Re: [PATCH] macb: RLE and BNA handling
From: Erik Waling <erik.waling@konftel.se>
To: Haavard Skinnemoen <haavard.skinnemoen@atmel.com>
CC: Erik Waling <Erik.Waling@konftel.com>,
       "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
In-Reply-To: <20090114111505.48d34949@hskinnemoen-d830>
References: <1231863713.10152.11.camel@konftel>
	 <20090114111505.48d34949@hskinnemoen-d830>
Content-Type: text/plain
Date: Wed, 14 Jan 2009 11:27:47 +0100
Message-ID: <1231928868.10152.31.camel@konftel>
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2715
Lines: 69

See comments below...

On Wed, 2009-01-14 at 11:15 +0100, Haavard Skinnemoen wrote:
> Erik Waling wrote:
> > This patch (against 2.6.28) solves two issues we have been experiencing
> > when connecting the EMAC module on our AT91SAM9260 board to an ethernet
> > repeater hub. When transfering large amounts of data we sometimes
> > experienced that the Retry Limit Exceeded (RLE) bit got set in TSR
> > during transmission attempts. When this happened the driver would stall
> > in a state that prevented any more data from being sent.
> 
> Right. This part of your patch looks good...
> 
> > The other issue experienced was in the RX part of the driver. Sometimes
> > the driver runs out of unused slots in the RX ring buffer. All slots
> > will be filled with data but the driver is not able to extract a
> > complete frame from the fragments in the buffer. When this happens the
> > Buffer Not Available (BNA) bit is set in RSR.
> 
> This part also looks like a real problem, however...
> 
> > @@ -527,18 +529,31 @@ static int macb_poll(struct napi_struct
> >         dev_dbg(&bp->pdev->dev, "poll: status = %08lx, budget = %d\n",
> >                 (unsigned long)status, budget);
> >
> > -       if (!(status & MACB_BIT(REC))) {
> > +       if (status & MACB_BIT(REC)) {
> > +               work_done = macb_rx(bp, budget);
> > +               if (work_done < budget)
> > +                       netif_rx_complete(dev, napi);
> > +       } else if (status & MACB_BIT(BNA)) {
> > +               /* No slots available in RX ring. Mark all slots
> > +                * as unused.
> > +                */
> > +               int i;
> > +
> > +               dev_warn(&bp->pdev->dev,
> > +                               "No free RX buffers. Marking all as unused.\n");
> > +
> > +               for (i = 0; i < RX_RING_SIZE; i++)
> > +                       bp->rx_ring[i].addr &= ~MACB_BIT(RX_USED);
> > +
> > +               wmb();
> > +               netif_rx_complete(dev, napi);
> 
> I'm not sure if nuking all the buffers that were received successfully
> is the right thing to do here...?
> 

This was the only solution I could think of since all slots are marked
as used and we are not able to assemble a complete frame. Do you think
there is a better solution?

> > +       } else if (!(status & MACB_BIT(REC))) {
> 
> This should probably be just "else", since you checked the REC bit
> above.

You are right. That is probably more correct.

> Haavard

Erik

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/