[PATCH] gianfar: Fix TX ring processing on SMP machines

Esben Haabendal esbenhaabendal at gmail.com
Fri Jun 11 18:45:24 EST 2010


On Wed, Mar 3, 2010 at 8:18 PM, Anton Vorontsov
<avorontsov at ru.mvista.com> wrote:
> Starting with commit a3bc1f11e9b867a4f49505 ("gianfar: Revive SKB
> recycling") gianfar driver sooner or later stops transmitting any
> packets on SMP machines.
>
> start_xmit() prepares new skb for transmitting, generally it does
> three things:
>
> 1. sets up all BDs (marks them ready to send), except the first one.
> 2. stores skb into tx_queue->tx_skbuff so that clean_tx_ring()
>   would cleanup it later.
> 3. sets up the first BD, i.e. marks it ready.
>
> Here is what clean_tx_ring() does:
>
> 1. reads skbs from tx_queue->tx_skbuff
> 2. checks if the *last* BD is ready. If it's still ready [to send]
>   then it it isn't transmitted, so clean_tx_ring() returns.
>   Otherwise it actually cleanups BDs. All is OK.
>
> Now, if there is just one BD, code flow:
>
> - start_xmit(): stores skb into tx_skbuff. Note that the first BD
>  (which is also the last one) isn't marked as ready, yet.
> - clean_tx_ring(): sees that skb is not null, *and* its lstatus
>  says that it is NOT ready (like if BD was sent), so it cleans
>  it up (bad!)
> - start_xmit(): marks BD as ready [to send], but it's too late.
>
> We can fix this simply by reordering lstatus/tx_skbuff writes.
>
> Reported-by: Martyn Welch <martyn.welch at ge.com>
> Bisected-by: Paul Gortmaker <paul.gortmaker at windriver.com>
> Signed-off-by: Anton Vorontsov <avorontsov at ru.mvista.com>
> Tested-by: Paul Gortmaker <paul.gortmaker at windriver.com>
> Tested-by: Martyn Welch <martyn.welch at ge.com>
> Cc: Sandeep Gopalpet <Sandeep.Kumar at freescale.com>
> Cc: Stable <stable at vger.kernel.org> [2.6.33]
> ---
>  drivers/net/gianfar.c |    5 ++++-
>  1 files changed, 4 insertions(+), 1 deletions(-)
>
> diff --git a/drivers/net/gianfar.c b/drivers/net/gianfar.c
> index 8bd3c9f..cccb409 100644
> --- a/drivers/net/gianfar.c
> +++ b/drivers/net/gianfar.c
> @@ -2021,7 +2021,6 @@ static int gfar_start_xmit(struct sk_buff *skb, struct net_device *dev)
>        }
>
>        /* setup the TxBD length and buffer pointer for the first BD */
> -       tx_queue->tx_skbuff[tx_queue->skb_curtx] = skb;
>        txbdp_start->bufPtr = dma_map_single(&priv->ofdev->dev, skb->data,
>                        skb_headlen(skb), DMA_TO_DEVICE);
>
> @@ -2053,6 +2052,10 @@ static int gfar_start_xmit(struct sk_buff *skb, struct net_device *dev)
>
>        txbdp_start->lstatus = lstatus;
>
> +       eieio(); /* force lstatus write before tx_skbuff */
> +
> +       tx_queue->tx_skbuff[tx_queue->skb_curtx] = skb;
> +
>        /* Update the current skb pointer to the next entry we will use
>         * (wrapping if necessary) */
>        tx_queue->skb_curtx = (tx_queue->skb_curtx + 1) &

This patch also makes gianfar work stable on mpc8313 with 2.6.33/RT_PREEMPT.
WIthout it, I see exactly the same problems as reported by Anton on SMP.

/Esben
-- 
Esben Haabendal, Senior Software Consultant
DoréDevelopment ApS, Ved Stranden 1, 9560 Hadsund, DK-Denmark
Phone: +45 51 92 53 93, E-mail: eha at doredevelopment.dk
WWW: http://www.doredevelopment.dk


More information about the Linuxppc-dev mailing list