Ubuntu Hardy, 2.6.24 kernel, "Oops" at rpcb_clnt.c:322. Is there a patch?

[Chuck Lever]

On Jul 30, 2008, at 8:20 PM, Nathan Patwardhan wrote:
> Hello.
>
> We're running NFSv4 (sec=krb5) with the following configuration:
>  - OS: Ubuntu 8.04.1
>  - kernel: 2.6.24-19-generic
>  - idmapd: 0.20-0
>  - nfs: 1:1.1.2-2ubuntu2.1
>
> and we've been able to [re]produce the following error message when we
> hit our NFSv4 mount with lots of I/O (dd, bonnie++, tar, MySQL
> queries, etc):
>
>  ------------[ cut here ]------------
>  Jul 30 19:05:50 prod-unix-shell04.kendall.corp.akamai.com kernel: [
> 3060.487743]
>   kernel BUG at /build/buildd/linux-2.6.24/net/sunrpc/rpcb_clnt.c:322!
>  Jul 30 19:05:50 prod-unix-shell04.kendall.corp.akamai.com kernel: [
> 3060.576132]
>   invalid opcode: 0000 [1] SMP
>  Jul 30 19:05:50 prod-unix-shell04.kendall.corp.akamai.com kernel: [
> 3060.623977]
>   CPU 3
>  Jul 30 19:05:50 prod-unix-shell04.kendall.corp.akamai.com kernel: [
> 3060.647914]
>   Modules linked in: ppdev container battery sbs sbshc ac dock video
> output des_g
>  ... etc ...
>
> This is the same message that is discussed here:
>
>  https://bugs.launchpad.net/ubuntu/+source/linux/+bug/212485
>
> *NOTE NOTE NOTE*: This is not the same bug that is discussed in bug-id
> 224750.  We are already patched as described in bug-id 224750.
>
> Per bug-id 212485 we will test a 2.6.26 kernel on our system like will
> be shipped with Intrepid (8.10), but we were wondering if a patch
> exists that addresses the bug in a *2.6.24* kernel as (a) we'd like to
> run the vendor-supplied 2.6.24 kernel in production AND (b) we're
> trying to qualify the existing system as being production-ready and
> the existing kernel issue won't allow us to do this.  :-)
>
> Please let me know if a patch exists for this issue for 2.6.24 or if I
> just missed it!

There may be, but...

We're mostly upstream developers here, so we don't follow the  
distributor's kernels that closely.

Can you tell what changed between the working 2.6.24-12-generic and  
the not working 2.6.24-14-generic Ubuntu kernels?  Doing a git bisect  
between those two would identify the patch that breaks rpcb_clnt.   
Even a simple diff between 12-generic, 13-generic, and 14-generic  
might be informative.

--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com