Re: [PATCH] Re: kernel BUG in eth_alloc_tx_desc_index at drivers/net/mv643xx_eth.c:1069!

["Dale Farnsworth" <dale@farnsworth.org>]

Jarek and Thibaut,

Thank you both very much for your work finding and fixing this bug.
Jarek, can you verify that the following patch fixes the problem you
were seeing?

-Dale

----- Patch follows -----

From: Dale Farnsworth <dale@farnsworth.org>

mv643xx_eth: Fix race condition in mv643xx_eth_free_tx_descs

The bug was found and isolated by Thibaut VARENE <T-Bone@parisc-linux.org>
and Jarek Poplawski <jarkao2@o2.pl>.  This patch is a modification of their
fixes.  We acquire and release the lock for each descriptor that is freed
to minimize the time the lock is held.

---

 drivers/net/mv643xx_eth.c |    7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/net/mv643xx_eth.c b/drivers/net/mv643xx_eth.c
index c41ae42..0d32381 100644
--- a/drivers/net/mv643xx_eth.c
+++ b/drivers/net/mv643xx_eth.c
@@ -314,6 +314,13 @@ int mv643xx_eth_free_tx_descs(struct net
 
 	while (mp->tx_desc_count > 0) {
 		spin_lock_irqsave(&mp->lock, flags);
+
+		/* maybe tx_desc_count changed before the lock was acquired */
+		if (mp->tx_desc_count <= 0) {
+			spin_unlock_irqrestore(&mp->lock, flags);
+			return released;
+		}
+
 		tx_index = mp->tx_used_desc_q;
 		desc = &mp->p_tx_desc_area[tx_index];
 		cmd_sts = desc->cmd_sts;