Re: kern/46822: NetBSD 5.1.2 seems to deadlock in the filesystem (or other kernel parts) "sometimes"

[David Holland <dholland-bugs%netbsd.org@localhost>]

The following reply was made to PR kern/46822; it has been noted by GNATS.

From: David Holland <dholland-bugs%netbsd.org@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: kern/46822: NetBSD 5.1.2 seems to deadlock in the filesystem (or
 other kernel parts) "sometimes"
Date: Wed, 22 Aug 2012 05:55:46 +0000

 On Wed, Aug 22, 2012 at 05:40:04AM +0000, David Holland wrote:
  >  In all probability it's deadlocking in ufs_rename.
 
 Basically "tstile" means "I'm waiting on some mutex, but I'm not going
 to tell you which one".
 
 The procedure for diagnosing a deadlock like this is to pick a process
 in "tstile" state, find the mutex it's waiting for, find which process
 (or more accurately, in-kernel lwp) is holding it, and repeat until
 you find the cycle. While all these things can be done using ddb, none
 of them are particularly obvious or easy; without ddb or a crash dump,
 basically the only information you can get is "something deadlocked".
 
 Rename is by far the most likely culprit, though, inasmuch as it's
 known to be badly broken.
 
 (It is actually broken worse in -4 than in -5, but the symptoms are
 different.)
 
 -- 
 David A. Holland
 dholland%netbsd.org@localhost