kern/163318: [ath] ath(4) stops working

[Adrian Chadd]

The following reply was made to PR kern/163318; it has been noted by GNATS.

From: Adrian Chadd <adrian.chadd at gmail.com>
To: Joel Dahl <joel at freebsd.org>
Cc: bug-followup at freebsd.org
Subject: Re: kern/163318: [ath] ath(4) stops working
Date: Sun, 11 Mar 2012 13:45:02 -0700

 On 11 March 2012 13:13, Joel Dahl <joel at freebsd.org> wrote:
 
 >> Nope. That just started shifting around locks. Can you please try
 >> 231852 with the debugging patch I threw you (that patched
 >> ieee80211_scan.c) and see if it complains?
 >
 > Sure, I'll try 231852 + your patch.
 >
 >> If that patch fixed anything, it just delayed things enough to hide
 >> what's going on...
 >
 > I see. Hiding a problem is not equal to fixing it. :-)
 
 Yup. The main thing that changed is the order of the TX/RX/interrupt
 taskqueue stop versus grabbing the PCU lock.
 
 Since the PCU lock is also grabbed when doing some other things
 (notably RX and interrupts, briefly in each situation), it's possible
 that we're just now not hitting whatever the race is, versus having
 fixed it.
 
 I'd really like to know where the race _is_...
 
 
 Adrian