MFP4(128855, 129015):

- Trying to eliminate another racing by replacing the timeout(9) with
  callout APIs.  In addition to that, the callout_drain() in an_detach()
  help us to avoid a possible panic-on-free due to the callout API tries
  to lock a destroyed mutex.
- In an_stats_update(), check the return value of an_read_record(). This
  should reduce the chance of device removal(PCCARD) panic [2].
- Adding a comment to state the fact that an_stats_update() is now called
  via callout(9) with a lock held [2].

Submitted by:	jhb [1], ambrisko [2]
Reviewed by:	jhb, ambrisko
Reported by:	dhw
Tested by:	dhw
MFC after:	3 days

1 files changed, 1 insertions, 1 deletions

diff --git a/sys/dev/an/if_anreg.h b/sys/dev/an/if_anreg.h
index 44a702a..0a96ee3 100644
--- a/sys/dev/an/if_anreg.h
+++ b/sys/dev/an/if_anreg.h
@@ -485,7 +485,7 @@ struct an_softc	{
 	int			an_have_rssimap;
 	struct an_ltv_rssi_map	an_rssimap;
 #endif
-	struct callout_handle	an_stat_ch;
+	struct callout		an_stat_ch;
 	struct mtx		an_mtx;
 	device_t		an_dev;
 	struct ifmedia		an_ifmedia;