[linux-yocto] [PATCH 12/21] rionet: Fix the corruption of tx_cnt during Transmit.

[Cristian Bercaru]

From: Palani <palaniappan.ramanathan at intel.com>

The non-atomic variable "tx_cnt" can be updated simultaneously
from different threads using "rionet_start_xmit" (Tx) and the
"rionet_outb_msg_event" (callback) functions. These functions
use different locks: "tx_lock" for the Tx and "lock" for the
callback function. This corruption of tx_cnt results in a kernel
crash, particularly visible during high throughput data trasfers.

Fixed to use the same lock "tx_lock" so as to avoid corruption
of tx_cnt.

Signed-off-by: Palani <palaniappan.ramanathan at lsi.com>
---
 drivers/net/rionet.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/rionet.c b/drivers/net/rionet.c
index f433b59..00a8128 100644
--- a/drivers/net/rionet.c
+++ b/drivers/net/rionet.c
@@ -269,7 +269,7 @@ static void rionet_outb_msg_event(struct rio_mport *mport, void *dev_id, int mbo
 	struct net_device *ndev = dev_id;
 	struct rionet_private *rnet = netdev_priv(ndev);
 
-	spin_lock(&rnet->lock);
+	spin_lock(&rnet->tx_lock);
 
 	if (netif_msg_intr(rnet))
 		printk(KERN_INFO
@@ -288,7 +288,7 @@ static void rionet_outb_msg_event(struct rio_mport *mport, void *dev_id, int mbo
 	if (rnet->tx_cnt < RIONET_TX_RING_SIZE)
 		netif_wake_queue(ndev);
 
-	spin_unlock(&rnet->lock);
+	spin_unlock(&rnet->tx_lock);
 }
 
 static int rionet_open(struct net_device *ndev)
-- 
1.7.9.5