Re: [Bug #14925] sky2 panic under load
From: Jarek Poplawski <hidden>
Date: 2010-01-11 22:08:00
On Mon, Jan 11, 2010 at 08:45:04AM -0800, Stephen Hemminger wrote:
On Mon, 11 Jan 2010 23:03:41 +0900 Mike McCormack [off-list ref] wrote:quoted
Perhaps only sky2_tx_done should wake the queue? Does the patch below fix the problem too? thanks, MikeThe idea is good,
I don't agree: you both try to make this check more specific. Actually, since this problem is quite tricky, I wondered about making it more genaral and visible for other maintainers by adding something like netif_wake_queue_present() with the check and some comment. Jarek P.
quoted hunk ↗ jump to hunk
but what if transmit queue was full (stopped), and TX FIFO gets stuck. Then Transmit timeout happens and the reset logic clears the queue. What will start the queue? Something like this: -------------------------------------------------------------- a/drivers/net/sky2.c 2010-01-11 08:36:42.617426300 -0800 +++ b/drivers/net/sky2.c 2010-01-11 08:42:51.295237661 -0800@@ -1843,9 +1843,6 @@ static void sky2_tx_complete(struct sky2 sky2->tx_cons = idx; smp_mb(); - - if (tx_avail(sky2) > MAX_SKB_TX_LE + 4) - netif_wake_queue(dev); } static void sky2_tx_reset(struct sky2_hw *hw, unsigned port)@@ -2416,8 +2413,12 @@ static inline void sky2_tx_done(struct n { struct sky2_port *sky2 = netdev_priv(dev); - if (netif_running(dev)) + if (netif_running(dev) && netif_device_present(dev)) { sky2_tx_complete(sky2, last); + + if (tx_avail(sky2) > MAX_SKB_TX_LE + 4) + netif_wake_queue(dev); + } } static inline void sky2_skb_rx(const struct sky2_port *sky2,@@ -3197,6 +3198,7 @@ static int sky2_reattach(struct net_devi } else { netif_device_attach(dev); sky2_set_multicast(dev); + netif_start_queue(dev); } }--