[PATCH] b43: work around a locking issue in ->set_tim()

[PATCH] b43: work around a locking issue in ->set_tim()

[Felix Fietkau]

ops->set_tim() must be atomic, so b43 trying to acquire a mutex leads
to a kernel crash. This patch trades an easy to trigger crash in AP
mode for an unlikely race condition. According to Michael, the real
fix would be to allow set_tim() to sleep, since b43 is not the only
driver that needs to sleep in all callbacks.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
---

--- a/drivers/net/wireless/b43/main.c
+++ b/drivers/net/wireless/b43/main.c
@@ -4534,9 +4534,8 @@ static int b43_op_beacon_set_tim(struct
 {
 	struct b43_wl *wl = hw_to_b43_wl(hw);

-	mutex_lock(&wl->mutex);
+	/* FIXME: add locking */
 	b43_update_templates(wl);
-	mutex_unlock(&wl->mutex);

 	return 0;
 }

Re: [PATCH] b43: work around a locking issue in ->set_tim()

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 311 bytes --]

On Sat, 2009-11-07 at 18:37 +0100, Felix Fietkau wrote:
> According to Michael, the real
> fix would be to allow set_tim() to sleep, since b43 is not the only
> driver that needs to sleep in all callbacks.

According to me, that's pretty much impossible... But patches welcome,
of course. :)

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 801 bytes --]