[PATCH] b43: Fix possible a data race in b43_op_tx()
From: Jia-Ju Bai
Date: Sat Jan 11 2020 - 11:17:40 EST
The functions b43_op_tx() and b43_tx_work() may be concurrently executed.
In b43_tx_work(), the variable wl->tx_queue_stopped[queue_num] is
accessed with holding a mutex lock wl->mutex. But in b43_op_tx(), the
identical variable wl->tx_queue_stopped[skb->queue_mapping] is accessed
without holding this mutex lock. Thus, a possible data race may occur.
To fix this data race, in b43_op_tx(), the variable
wl->tx_queue_stopped[skb->queue_mapping] is accessed with holding the
mutex lock wl->mutex.
This data race is found by the runtime testing of our tool DILP-2.
Signed-off-by: Jia-Ju Bai <baijiaju1990@xxxxxxxxx>
---
drivers/net/wireless/broadcom/b43/main.c | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/drivers/net/wireless/broadcom/b43/main.c b/drivers/net/wireless/broadcom/b43/main.c
index 39da1a4c30ac..adedb38f50f2 100644
--- a/drivers/net/wireless/broadcom/b43/main.c
+++ b/drivers/net/wireless/broadcom/b43/main.c
@@ -3625,6 +3625,11 @@ static void b43_op_tx(struct ieee80211_hw *hw,
struct sk_buff *skb)
{
struct b43_wl *wl = hw_to_b43_wl(hw);
+ bool stopped;
+
+ mutex_lock(&wl->mutex);
+ stopped = wl->tx_queue_stopped[skb->queue_mapping];
+ mutex_unlock(&wl->mutex);
if (unlikely(skb->len < 2 + 2 + 6)) {
/* Too short, this can't be a valid frame. */
@@ -3634,7 +3639,7 @@ static void b43_op_tx(struct ieee80211_hw *hw,
B43_WARN_ON(skb_shinfo(skb)->nr_frags);
skb_queue_tail(&wl->tx_queue[skb->queue_mapping], skb);
- if (!wl->tx_queue_stopped[skb->queue_mapping]) {
+ if (!stopped) {
ieee80211_queue_work(wl->hw, &wl->tx_work);
} else {
ieee80211_stop_queue(wl->hw, skb->queue_mapping);
--
2.17.1