From: Johannes Berg <johannes@xxxxxxxxxxxxxxxx>
This patch clarifies the comment about locking in wiphy_unregister.
Signed-off-by: Johannes Berg <johannes@xxxxxxxxxxxxxxxx>
Signed-off-by: John W. Linville <linville@xxxxxxxxxxxxx>
---
I applied this after the "fix locking in wiphy_new" patch, but I
don't think the order really matters.
net/wireless/core.c | 19 +++++++++++++++----
1 files changed, 15 insertions(+), 4 deletions(-)
diff --git a/net/wireless/core.c b/net/wireless/core.c
index 24a21ad..7eabd55 100644
--- a/net/wireless/core.c
+++ b/net/wireless/core.c
@@ -91,7 +91,6 @@ int wiphy_register(struct wiphy *wiphy)
mutex_lock(&cfg80211_drv_mutex);
-
res = device_add(&drv->wiphy.dev);
if (res)
goto out_unlock;
@@ -114,14 +113,26 @@ void wiphy_unregister(struct wiphy *wiphy)
{
struct cfg80211_registered_device *drv = wiphy_to_dev(wiphy);
+ /* protect the device list */
mutex_lock(&cfg80211_drv_mutex);
- /* hold registered driver mutex during list removal as well
- * to make sure no commands are in progress at the moment */
+ BUG_ON(!list_empty(&drv->netdev_list));
+
+ /*
+ * Try to grab drv->mtx. If a command is still in progress,
+ * hopefully the driver will refuse it since it's tearing
+ * down the device already. We wait for this command to complete
+ * before unlinking the item from the list.
+ * Note: as codified by the BUG_ON above we cannot get here if
+ * a virtual interface is still associated. Hence, we can only
+ * get to lock contention here if userspace issues a command
+ * that identified the hardware by wiphy index.
+ */
mutex_lock(&drv->mtx);
- list_del(&drv->list);
+ /* unlock again before freeing */
mutex_unlock(&drv->mtx);
+ list_del(&drv->list);
device_del(&drv->wiphy.dev);
debugfs_remove(drv->wiphy.debugfsdir);
--
1.5.0.6
--
John W. Linville
linville@xxxxxxxxxxxxx
-
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
