busy inodes and segfault?

[Gregory Davis]

Hi, I've been having a consistent problem with both arla and OpenAFS that I 
think is a behavior problem rather than a bug.  I just can't figure out what 
causes my problem.  I'm using linux kernel 2.4.19 btw.  Whenever I fill up my 
memory cache, and I presume the kernel starts recycling memory pages, the AFS 
program goes awry.  In the case of arla, I can still read files in the afs 
cache, but when I try to write to an afs volume, the program that made the 
write call hangs.  The xfs device then becomes busy and I cannot umount afs.  
The same happens with OpenAFS, though their setup seems to cause a kernel 
panic in addition to the hang, and I get an error message in my logs (see 
below) about busy inodes.  I'm guessing virtual address ffffffff line means a 
segfault occured.  Does anyone know if this is a bug in the kernel?  Is this 
a known problem?

Thanks,
Greg

Dec 13 22:46:12 uranus kernel: Can't open inode 5037
Dec 13 22:46:12 uranus kernel: Unable to handle kernel paging request at 
virtual address ffffffff

... dumped contents of cpu registers ...

Dec 13 22:46:12 uranus kernel: Call Trace:    [<e0b3f65e>] [<e0b2c3de>] 
[<e0b3f65e>] [<e0aeb956>] [filldir64+206/272]
Dec 13 22:46:12 uranus kernel:   [<e0af61c0>] [<e0b0e7f7>] [<e0b2e3c9>] 
[<e0b2e2f1>] [cp_new_stat64+243/288] [vfs_readdir+157/160]
Dec 13 22:46:12 uranus kernel:   [filldir64+0/272] [sys_getdents64+91/192] 
[filldir64+0/272] [sys_fcntl64+93/192] [system_call+51/56]
Dec 13 22:46:12 uranus kernel:
Dec 13 22:46:12 uranus kernel: Code: c6 05 ff ff ff ff 2a 83 c4 1c c3 90 b8 41 
f3 b3 e0 eb d8 90