dirty cluster shutdown oopses on failover mount

[Jer Jackson <jerj@xxxxxxxxxxxx>]

While testing EVMS clustering/failover on a shared scsi system, I
encountered the following BUG() with kernel 2.4.20 and XFS 1.2.0.

Please let me know how to produce a more useful dump and also if this is
a known bug.

How I got here: I did a cp -a /usr /cv1 where /cv1 is an XFS filesystem
on the shared disks (private container - only 1 node can access at a
time).  after about 3 seconds I hit reset.  The second node takes over
the disks and private container, but when it tries to mount the fs, the
oops below is generated.  

I have never sent a kernel bug report so please advise on what else I
may need.

Regards,

Jeremy Jackson

XFS mounting filesystem
device-mapper(254,16)                                                        
Starting XFS recovery on filesystem: device-mapper(254,16) (dev: 254/16)        
                                   Filesystem "device-mapper(254,16)": Detected 
a bogus zero next_unlinked field in incore inode buffer 0xc3bb9820.  .XFS 
assertion failed: !INT_ISZERO(dip->di_next_unlinked, ARCH_CONVERT), file: 
xfs_inode.c, line: 102               kernel BUG at debug.c:52!                  
                                                                        invalid 
operand: 0000                                                                   
                           CPU:    0                                            
                                                              EIP:    
0010:[<c021c439>]    Not tainted                                                
                           EFLAGS: 00010282                                     
                                                              eax: 00000065   
ebx: c20fa000   ecx: c2f1c000   edx: 00000001                                   
                   esi: 00000010   edi: c3bb9820   ebp: 00000020   esp: 
c2483b04                                                      ds: 0018   es: 
0018   ss: 0018                                                                 
                    Process mount (pid: 15514, stackpage=c2483000)              
                                                       Stack: c0311760 c0308580 
c03084e4 00000066 c01e5ce7 c0308580 c03084e4 00000066                           
                 c3bb9820 c25b55e0 00000089 00000000 c01f5068 c0ee3800 c3bb9820 
00000000                                            c25b55e0 c3978ea0 00000002 
c3bb9e20 00000000 c01f4f2d c3bb9e20 c0ee3800                                    
 Call Trace:    [<c01e5ce7>] [<c01f5068>] [<c01f4f2d>] [<c01f5cae>] 
[<c01f5df7>]                                      [<c01f5fbd>] [<c01f6d99>] 
[<c01f74c6>] [<c01f754a>] [<c01f775c>] [<c01eee60>]                             
         [<c01fbdc4>] [<c01faf87>] [<c01faf99>] [<c02059ab>] [<c0205ab5>] 
[<c021b5a9>]                                      [<c0138a7b>] [<c013924e>] 
[<c01482a6>] [<c0139404>] [<c0149129>] [<c01493c2>]                             
         [<c0149244>] [<c0149744>] [<c0108843>]                                 
                                                                                
                                                                             
Code: 0f 0b 34 00 8e 17 31 c0 83 c4 10 c3 8d 76 00 53 8b 1d 60 d9