mac80211: handle failed restart/resume better

When the driver fails during HW restart or resume, the whole
stack goes into a very confused state with interfaces being
up while the hardware is down etc.

Address this by shutting down everything; we'll run into a
lot of warnings in the process but that's better than having
the whole stack get messed up.
Reviewed-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

include/net/cfg80211.h

@@ -4771,6 +4771,20 @@ int cfg80211_iter_combinations(struct wiphy *wiphy,
 void cfg80211_stop_iface(struct wiphy *wiphy, struct wireless_dev *wdev,
 			 gfp_t gfp);
 
+/**
+ * cfg80211_shutdown_all_interfaces - shut down all interfaces for a wiphy
+ * @wiphy: the wiphy to shut down
+ *
+ * This function shuts down all interfaces belonging to this wiphy by
+ * calling dev_close() (and treating non-netdev interfaces as needed).
+ * It shouldn't really be used unless there are some fatal device errors
+ * that really can't be recovered in any other way.
+ *
+ * Callers must hold the RTNL and be able to deal with callbacks into
+ * the driver while the function is running.
+ */
+void cfg80211_shutdown_all_interfaces(struct wiphy *wiphy);
+
 /* Logging, debugging and troubleshooting/diagnostic helpers. */
 
 /* wiphy_printk helpers, similar to dev_printk */

net/mac80211/ieee80211_i.h

@@ -1459,6 +1459,7 @@ __ieee80211_request_sched_scan_start(struct ieee80211_sub_if_data *sdata,
 int ieee80211_request_sched_scan_start(struct ieee80211_sub_if_data *sdata,
 				       struct cfg80211_sched_scan_request *req);
 int ieee80211_request_sched_scan_stop(struct ieee80211_sub_if_data *sdata);
+void ieee80211_sched_scan_end(struct ieee80211_local *local);
 void ieee80211_sched_scan_stopped_work(struct work_struct *work);
 
 /* off-channel helpers */

net/mac80211/scan.c

@@ -1076,12 +1076,8 @@ void ieee80211_sched_scan_results(struct ieee80211_hw *hw)
 }
 EXPORT_SYMBOL(ieee80211_sched_scan_results);
 
-void ieee80211_sched_scan_stopped_work(struct work_struct *work)
+void ieee80211_sched_scan_end(struct ieee80211_local *local)
 {
-	struct ieee80211_local *local =
-		container_of(work, struct ieee80211_local,
-			     sched_scan_stopped_work);
-
 	mutex_lock(&local->mtx);
 
 	if (!rcu_access_pointer(local->sched_scan_sdata)) {
@@ -1099,6 +1095,15 @@ void ieee80211_sched_scan_stopped_work(struct work_struct *work)
 	cfg80211_sched_scan_stopped(local->hw.wiphy);
 }
 
+void ieee80211_sched_scan_stopped_work(struct work_struct *work)
+{
+	struct ieee80211_local *local =
+		container_of(work, struct ieee80211_local,
+			     sched_scan_stopped_work);
+
+	ieee80211_sched_scan_end(local);
+}
+
 void ieee80211_sched_scan_stopped(struct ieee80211_hw *hw)
 {
 	struct ieee80211_local *local = hw_to_local(hw);

net/mac80211/util.c

@@ -1457,6 +1457,44 @@ void ieee80211_stop_device(struct ieee80211_local *local)
 	drv_stop(local);
 }
 
+static void ieee80211_handle_reconfig_failure(struct ieee80211_local *local)
+{
+	struct ieee80211_sub_if_data *sdata;
+	struct ieee80211_chanctx *ctx;
+
+	/*
+	 * We get here if during resume the device can't be restarted properly.
+	 * We might also get here if this happens during HW reset, which is a
+	 * slightly different situation and we need to drop all connections in
+	 * the latter case.
+	 *
+	 * Ask cfg80211 to turn off all interfaces, this will result in more
+	 * warnings but at least we'll then get into a clean stopped state.
+	 */
+
+	local->resuming = false;
+	local->suspended = false;
+	local->started = false;
+
+	/* scheduled scan clearly can't be running any more, but tell
+	 * cfg80211 and clear local state
+	 */
+	ieee80211_sched_scan_end(local);
+
+	list_for_each_entry(sdata, &local->interfaces, list)
+		sdata->flags &= ~IEEE80211_SDATA_IN_DRIVER;
+
+	/* Mark channel contexts as not being in the driver any more to avoid
+	 * removing them from the driver during the shutdown process...
+	 */
+	mutex_lock(&local->chanctx_mtx);
+	list_for_each_entry(ctx, &local->chanctx_list, list)
+		ctx->driver_present = false;
+	mutex_unlock(&local->chanctx_mtx);
+
+	cfg80211_shutdown_all_interfaces(local->hw.wiphy);
+}
+
 static void ieee80211_assign_chanctx(struct ieee80211_local *local,
 				     struct ieee80211_sub_if_data *sdata)
 {
@@ -1520,9 +1558,11 @@ int ieee80211_reconfig(struct ieee80211_local *local)
 	 */
 	res = drv_start(local);
 	if (res) {
-		WARN(local->suspended, "Hardware became unavailable "
-		     "upon resume. This could be a software issue "
-		     "prior to suspend or a hardware issue.\n");
+		if (local->suspended)
+			WARN(1, "Hardware became unavailable upon resume. This could be a software issue prior to suspend or a hardware issue.\n");
+		else
+			WARN(1, "Hardware became unavailable during restart.\n");
+		ieee80211_handle_reconfig_failure(local);
 		return res;
 	}
 

net/wireless/core.c

@@ -210,15 +210,12 @@ void cfg80211_stop_p2p_device(struct cfg80211_registered_device *rdev,
 	}
 }
 
-static int cfg80211_rfkill_set_block(void *data, bool blocked)
+void cfg80211_shutdown_all_interfaces(struct wiphy *wiphy)
 {
-	struct cfg80211_registered_device *rdev = data;
+	struct cfg80211_registered_device *rdev = wiphy_to_rdev(wiphy);
 	struct wireless_dev *wdev;
 
-	if (!blocked)
-		return 0;
-
-	rtnl_lock();
+	ASSERT_RTNL();
 
 	list_for_each_entry(wdev, &rdev->wdev_list, list) {
 		if (wdev->netdev) {
@@ -234,7 +231,18 @@ static int cfg80211_rfkill_set_block(void *data, bool blocked)
 			break;
 		}
 	}
+}
+EXPORT_SYMBOL_GPL(cfg80211_shutdown_all_interfaces);
 
+static int cfg80211_rfkill_set_block(void *data, bool blocked)
+{
+	struct cfg80211_registered_device *rdev = data;
+
+	if (!blocked)
+		return 0;
+
+	rtnl_lock();
+	cfg80211_shutdown_all_interfaces(&rdev->wiphy);
 	rtnl_unlock();
 
 	return 0;