Skip to contentSkip to conversation

patch b43-workaround-circular-locking-in-hw-tkip-key-update-callback.patch added to 2.6.33-stable tree

DISCUSSION Stefan Lippers-Hollmann 2 participants 2 messages Mar 19, 2010 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw

Hi

Quoted 5 lines · On Friday 19 March 2010, [email protected] wrote:

This is a note to let you know that we have just queued up the patch titled

Subject: b43: Workaround circular locking in hw-tkip key update callback

to the 2.6.33-stable tree. Its filename is

[...]

Quoted 11 lines

From: Larry Finger <[email protected]>
Date: Wed, 10 Mar 2010 22:10:32 -0600
Subject: b43: Workaround circular locking in hw-tkip key update callback
To: Greg Kroah-Hartman <[email protected]>
Cc: Michael Buesch <[email protected]>
Message-ID: <4b986d38.aOKVIPS3U9/aYsOP%[email protected]>

From: Michael Buesch <[email protected]>

commit 96869a39399269a776a94812e9fff3d38b47d838 upstream

Quoted 7 lines

The TKIP key update callback is called from the RX path, where the driver mutex is already locked. This results in a circular locking bug. Avoid this by removing the lock.

Johannes noted that there is a separate bug: The callback still breaks on SDIO hardware, because SDIO hardware access needs to sleep, but we are not allowed to sleep in the callback due to mac80211's RCU locking.

[...]

This patch breaks compiling 2.6.33.1 + the current stable queue (and likely queue-2.6.32 as well):

CC [M] drivers/net/wireless/b43/main.o
drivers/net/wireless/b43/main.c: In function 'b43_op_update_tkip_key':
drivers/net/wireless/b43/main.c:868: error: 'sta' undeclared (first use in this function) drivers/net/wireless/b43/main.c:868: error: (Each undeclared identifier is reported only once drivers/net/wireless/b43/main.c:868: error: for each function it appears in.)

as it is based on the mac80211 API change of

Gitweb:     http://git.kernel.org/linus/b3fbdcf49f940d0703c356441e0daf045e64e076
Commit:     b3fbdcf49f940d0703c356441e0daf045e64e076
Parent:     e4fca007b06165900d0e44e8d5e251376819bf5d
Author:     Johannes Berg <[email protected]>
AuthorDate: Thu Jan 21 11:40:47 2010 +0100
Committer:  John W. Linville <[email protected]>
CommitDate: Fri Jan 22 16:08:55 2010 -0500

mac80211: pass vif and station to update_tkip_key

When a TKIP key is updated, we should pass the station pointer instead of just the address, since drivers can use that to store their own data. We also need to pass the virtual interface pointer.

which also touches, besides the API mac80211 tkip handling, iwl-agn as well as b43.

Regards
Stefan Lippers-Hollmann

Quoted 29 lines
drivers/net/wireless/b43/main.c +7 -7
@@ -852,19 +852,19 @@ static void b43_op_update_tkip_key(struc
 	if (B43_WARN_ON(!modparam_hwtkip))
 		return;
 
-	mutex_lock(&wl->mutex);
-
+	/* This is only called from the RX path through mac80211, where
+	 * our mutex is already locked. */
+	B43_WARN_ON(!mutex_is_locked(&wl->mutex));
 	dev = wl->current_dev;
-	if (!dev || b43_status(dev) < B43_STAT_INITIALIZED)
-		goto out_unlock;
+	B43_WARN_ON(!dev || b43_status(dev) < B43_STAT_INITIALIZED);
 
 	keymac_write(dev, index, NULL);	/* First zero out mac to avoid race */
 
 	rx_tkip_phase1_write(dev, index, iv32, phase1key);
+	/* only pairwise TKIP keys are supported right now */
+	if (WARN_ON(!sta))
+		return;
 	keymac_write(dev, index, addr);
-
-out_unlock:
-	mutex_unlock(&wl->mutex);
 }
 
 static void do_key_write(struct b43_wldev *dev,
#2 lore raw
Quoted 50 lines · Stefan Lippers-Hollmann

Hi

Quoted 5 lines · On Friday 19 March 2010, [email protected] wrote:

This is a note to let you know that we have just queued up the patch titled

Subject: b43: Workaround circular locking in hw-tkip key update callback

to the 2.6.33-stable tree. Its filename is

[...]

Quoted 11 lines

From: Larry Finger <[email protected]>
Date: Wed, 10 Mar 2010 22:10:32 -0600
Subject: b43: Workaround circular locking in hw-tkip key update callback
To: Greg Kroah-Hartman <[email protected]>
Cc: Michael Buesch <[email protected]>
Message-ID: <4b986d38.aOKVIPS3U9/aYsOP%[email protected]>

From: Michael Buesch <[email protected]>

commit 96869a39399269a776a94812e9fff3d38b47d838 upstream

Quoted 7 lines

The TKIP key update callback is called from the RX path, where the driver mutex is already locked. This results in a circular locking bug. Avoid this by removing the lock.

Johannes noted that there is a separate bug: The callback still breaks on SDIO hardware, because SDIO hardware access needs to sleep, but we are not allowed to sleep in the callback due to mac80211's RCU locking.

[...]

This patch breaks compiling 2.6.33.1 + the current stable queue (and likely queue-2.6.32 as well):

CC [M] drivers/net/wireless/b43/main.o
drivers/net/wireless/b43/main.c: In function 'b43_op_update_tkip_key':
drivers/net/wireless/b43/main.c:868: error: 'sta' undeclared (first use in this function) drivers/net/wireless/b43/main.c:868: error: (Each undeclared identifier is reported only once drivers/net/wireless/b43/main.c:868: error: for each function it appears in.)

as it is based on the mac80211 API change of

Gitweb:     http://git.kernel.org/linus/b3fbdcf49f940d0703c356441e0daf045e64e076
Commit:     b3fbdcf49f940d0703c356441e0daf045e64e076
Parent:     e4fca007b06165900d0e44e8d5e251376819bf5d
Author:     Johannes Berg <[email protected]>
AuthorDate: Thu Jan 21 11:40:47 2010 +0100
Committer:  John W. Linville <[email protected]>
CommitDate: Fri Jan 22 16:08:55 2010 -0500

mac80211: pass vif and station to update_tkip_key

Quoted 7 lines

When a TKIP key is updated, we should pass the station pointer instead of just the address, since drivers can use that to store their own data. We also need to pass the virtual interface pointer.

which also touches, besides the API mac80211 tkip handling, iwl-agn as well as b43.

Yeah, I backed it out right away and told the authors about it, but didn't push the stable-queue changes out, sorry about that.

greg k-h