From: Ahmad Masri amasri@codeaurora.org
[ Upstream commit 1a276003111c0404f6bfeffe924c5a21f482428b ]
This change fixes a rare race condition of handling WMI events after wmi_call expires.
wmi_recv_cmd immediately handles an event when reply_buf is defined and a wmi_call is waiting for the event. However, in case the wmi_call has already timed-out, there will be no waiting/running wmi_call and the event will be queued in WMI queue and will be handled later in wmi_event_handle. Meanwhile, a new similar wmi_call for the same command and event may be issued. In this case, when handling the queued event we got WARN_ON printed.
Fixing this case as a valid timeout and drop the unexpected event.
Signed-off-by: Ahmad Masri amasri@codeaurora.org Signed-off-by: Maya Erez merez@codeaurora.org Signed-off-by: Kalle Valo kvalo@codeaurora.org Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/net/wireless/ath/wil6210/wmi.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-)
diff --git a/drivers/net/wireless/ath/wil6210/wmi.c b/drivers/net/wireless/ath/wil6210/wmi.c index 63116f4b62c7..de52e532c105 100644 --- a/drivers/net/wireless/ath/wil6210/wmi.c +++ b/drivers/net/wireless/ath/wil6210/wmi.c @@ -3211,7 +3211,18 @@ static void wmi_event_handle(struct wil6210_priv *wil, /* check if someone waits for this event */ if (wil->reply_id && wil->reply_id == id && wil->reply_mid == mid) { - WARN_ON(wil->reply_buf); + if (wil->reply_buf) { + /* event received while wmi_call is waiting + * with a buffer. Such event should be handled + * in wmi_recv_cmd function. Handling the event + * here means a previous wmi_call was timeout. + * Drop the event and do not handle it. + */ + wil_err(wil, + "Old event (%d, %s) while wmi_call is waiting. Drop it and Continue waiting\n", + id, eventid2name(id)); + return; + }
wmi_evt_call_handler(vif, id, evt_data, len - sizeof(*wmi));