kernel oops: unable to handle kernel NULL pointer dereference; EIP is at rpc_shutdown_client+0x20/0xd0 [sunrpc]

[Jon Skanes]

Binary package hint: linux-image-2.6.24-18-server

I've been getting oopses after upgrading my kernel on a Hardy server to 2.6.24-18 which seem to be related to NFS v4. I have included log entries in the dmesg.txt file. This seems to be a regression from 2.6.24-17 and earlier.

Thanks,
Jon

[44519.624597] BUG: unable to handle kernel NULL pointer dereference at virtual address 0000000c
[44519.624662] printing eip: f0b5e220 *pdpt = 000000002ed76001 *pde = 0000000000000000
[44519.624723] Oops: 0000 [#1] SMP
...
[44519.625344] EIP: 0060:[<f0b5e220>] EFLAGS: 00010246 CPU: 0
[44519.625495] EIP is at rpc_shutdown_client+0x20/0xd0 [sunrpc]

[Sense Egbert Hofstede (sense)]

Thank you for your report. However, we still need some more information before it can be passed on to the developers.
When did the oops happen and can it be reproduced(how)?
Please attach the output of the following commands:
uname -a
cat /proc/version_signature
sudo lspci -vvnn

Thanks in advance.

[Sense Egbert Hofstede (sense)]

And one thing I forgot: do you get any problems from this bug? Does something goes wrong or doesn't work anymore after the oops?

[Jon Skanes (jon-skanes)]

No, this bug doesn't seem to be critical. I only noticed the log output on the console a couple days after it happened. There are a few things I can try to reproduce it. I'll have a look through the logs on the NFS client machines to see if there was anything interesting happening on their end to cause it.

[Jon Skanes (jon-skanes)]

The only client activity seems to be a network adapter change. The wired ethernet was unplugged and the client switched to wireless. There were no suspicious log entries on the client.

[Sense Egbert Hofstede (sense)]

OK, thank you for your information. Can you please try to reproduce this? Try to recreate the situation at the moment as exact as possible and eliminate as much things as possible to isolate the cause.

[Jon Skanes (jon-skanes)]

I'm invalidating this one. It turns out that out of the 4G of ram in the system there was one bad bit. Since replacing the ram, there hasn't been an issue.

[Jon Skanes (jon-skanes)]

Hardware issue.