Re: kern/12404 (panic: ffs_alloccg: map corrupted)

[David Laight <david%l8s.co.uk@localhost>]

The following reply was made to PR kern/12404; it has been noted by GNATS.

From: David Laight <david%l8s.co.uk@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: kern/12404 (panic: ffs_alloccg: map corrupted)
Date: Thu, 17 Jul 2008 17:52:53 +0100

 On Thu, Jul 17, 2008 at 06:40:02AM +0000, Havard Eidnes wrote:
 > Subject: Re: kern/12404 (panic: ffs_alloccg: map corrupted)
 >  
 >  I had another occurrance, and per request, here's "show reg"
 >  output together with a backtrace.  This is with 4.99.70 on the
 >  new system (dual-Xeon 2.2GHz, 2GB memory, 205GB RAID on ciss(4)):
 >  
 ...
 >  Looking at the code, it appears that the backtrace is misleading,
 >  the panic() is in ffs_mapsearch(), not in ffs_alloccgblk().  I'll
 >  admit that I don't understand why ddb gets the backtrace wrong.
 
 Almost certainly due to leaf subroutines and/or static functions
 not being in the symbol table.
 
 Without having looked at how this code gets called, but which mutex
 controls access to the allocation maps in each 'cylinder group' ?
 
        David
 
 -- 
 David Laight: david%l8s.co.uk@localhost