2008-03-13 07:14:01

by Divy Le ray

Subject: [PATCH 2.6.25] - Fix transmit queue stop mechanism

From: Divy Le Ray <[email protected]>

The last change in the Tx queue stop mechanism opens a window
where the Tx queue might be stopped after pending credits
returned.

Tx credits are returned via a control message generated by the HW.
It returns tx credits on demand, triggered by a completion bit
set in selective transmit packet headers.

The current code can lead to the Tx queue stopped
with all pending credits returned, and the current frame
not triggering a credit return. The Tx queue will then never be
awaken.

The driver could alternatively request a completion for packets
that stop the queue. It's however safer at this point to go back
to the pre-existing behaviour.

Signed-off-by: Divy Le Ray <[email protected]>
---

drivers/net/cxgb3/sge.c | 12 +++++++++---
1 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/drivers/net/cxgb3/sge.c b/drivers/net/cxgb3/sge.c
index 979f3fc..db58687 100644
--- a/drivers/net/cxgb3/sge.c
+++ b/drivers/net/cxgb3/sge.c
@@ -1107,9 +1107,15 @@ int t3_eth_xmit(struct sk_buff *skb, struct net_device *dev)
}

q->in_use += ndesc;
- if (unlikely(credits - ndesc < q->stop_thres))
- if (USE_GTS || !should_restart_tx(q))
- t3_stop_queue(dev, qs, q);
+ if (unlikely(credits - ndesc < q->stop_thres)) {
+ t3_stop_queue(dev, qs, q);
+
+ if (should_restart_tx(q) &&
+ test_and_clear_bit(TXQ_ETH, &qs->txq_stopped)) {
+ q->restarts++;
+ netif_wake_queue(dev);
+ }
+ }

gen = q->gen;
q->unacked += ndesc;

2008-03-17 12:09:37

by Jeff Garzik

Subject: Re: [PATCH 2.6.25] - Fix transmit queue stop mechanism

Divy Le Ray wrote:
> From: Divy Le Ray <[email protected]>
>
> The last change in the Tx queue stop mechanism opens a window
> where the Tx queue might be stopped after pending credits
> returned.
>
> Tx credits are returned via a control message generated by the HW.
> It returns tx credits on demand, triggered by a completion bit
> set in selective transmit packet headers.
>
> The current code can lead to the Tx queue stopped
> with all pending credits returned, and the current frame
> not triggering a credit return. The Tx queue will then never be
> awaken.
>
> The driver could alternatively request a completion for packets
> that stop the queue. It's however safer at this point to go back
> to the pre-existing behaviour.
>
> Signed-off-by: Divy Le Ray <[email protected]>
> ---
>
> drivers/net/cxgb3/sge.c | 12 +++++++++---
> 1 files changed, 9 insertions(+), 3 deletions(-)

applied -- please include "cxgb3: " prefix in your subject line next time.

See http://linux.yyz.us/patch-format.html for more info.

Jeff