4.4-stable review patch.  If anyone has any objections, please let me know.


From: Michael Chan <michael.c...@broadcom.com>

[ Upstream commit f5992b72ebe0dde488fa8f706b887194020c66fc ]

The driver's ndo_get_stats64() method is not always called under RTNL.
So it can race with driver close or ethtool reconfigurations.  Fix the
race condition by taking tp->lock spinlock in tg3_free_consistent()
when freeing the tp->hw_stats memory block.  tg3_get_stats64() is
already taking tp->lock.

Reported-by: Wang Yufen <wangyu...@huawei.com>
Signed-off-by: Michael Chan <michael.c...@broadcom.com>
Signed-off-by: David S. Miller <da...@davemloft.net>
Signed-off-by: Sasha Levin <alexander.le...@verizon.com>
Signed-off-by: Greg Kroah-Hartman <gre...@linuxfoundation.org>
 drivers/net/ethernet/broadcom/tg3.c |    3 +++
 1 file changed, 3 insertions(+)

--- a/drivers/net/ethernet/broadcom/tg3.c
+++ b/drivers/net/ethernet/broadcom/tg3.c
@@ -8722,11 +8722,14 @@ static void tg3_free_consistent(struct t
+       /* Protect tg3_get_stats64() from reading freed tp->hw_stats. */
+       tg3_full_lock(tp, 0);
        if (tp->hw_stats) {
                dma_free_coherent(&tp->pdev->dev, sizeof(struct tg3_hw_stats),
                                  tp->hw_stats, tp->stats_mapping);
                tp->hw_stats = NULL;
+       tg3_full_unlock(tp);

Reply via email to