When the network interface goes down while a scan request is still
pending that can't be stopped due to firmware hickups, wcn->scan_req
remains set, even though the hardware is deinitialized. This results
in -EBUSY for all scan requests after the interface was brought up
again.

Fix this by explicitly completing pending scan requests in
wcn36xx_stop().

Signed-off-by: Daniel Mack <dan...@zonque.org>
---
 drivers/net/wireless/ath/wcn36xx/main.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/drivers/net/wireless/ath/wcn36xx/main.c 
b/drivers/net/wireless/ath/wcn36xx/main.c
index 1b17c35a7944..e5ef2cbb622f 100644
--- a/drivers/net/wireless/ath/wcn36xx/main.c
+++ b/drivers/net/wireless/ath/wcn36xx/main.c
@@ -353,6 +353,19 @@ static void wcn36xx_stop(struct ieee80211_hw *hw)
 
        wcn36xx_dbg(WCN36XX_DBG_MAC, "mac stop\n");
 
+       cancel_work_sync(&wcn->scan_work);
+
+       mutex_lock(&wcn->scan_lock);
+       if (wcn->scan_req) {
+               struct cfg80211_scan_info scan_info = {
+                       .aborted = true,
+               };
+
+               ieee80211_scan_completed(wcn->hw, &scan_info);
+       }
+       wcn->scan_req = NULL;
+       mutex_unlock(&wcn->scan_lock);
+
        wcn36xx_debugfs_exit(wcn);
        wcn36xx_smd_stop(wcn);
        wcn36xx_dxe_deinit(wcn);
-- 
2.14.3

Reply via email to