[plug] NFS problem

[Greg Mildenhall]

On Tue, 1 Dec 1998, John Summerfield wrote:
> On Tue, 1 Dec 1998, Greg Mildenhall wrote:
> > So it's entirely userspace.
> > But, imagining it wasn't:
> > If there was an incompatibility, the kernel writers would already know
> > about it. The NFS maintainers would have alerted them.
> Only if it was already found.
A bug preventing completely standard NFS mounts on the most popular
distro?
They'd know about it.

> If there's a choice between reporting to the Kernel folk and not reporting
> it...
Then you should not report it. One should always report a bug to the
furthest-downstream maintainer that is relevant. I say this not for your
benefit, John, but for the benefit of those unfamiliar with the
process of reporting bugs. It is something that new users will do very
well to remember.

In this case, the maintainer of the RedHat package, or, if you can't do
that, the maintainer of the NFS mountd. Bug reports usually only flow
upstream. 
You see, the NFS maintainer will not care about a bug in the redhat
package, but the redhat packager will indeed care about a bug in the NFS
code, and will make sure it gets sorted out. Likewise, the kernel
developers won't care if the NFS mountd is broken, but if the kernel
suddenly breaks, the NFS maintainer will be directly affected, and will
make sure things are sorted out.

> I'm now inclined to think the two versions of NFS are mutually incompatible:
> perhaps because of some security change, misalignment of fields on workd
> boundaries, string overrun....
I'd say the probability is incredibly slim. The NFS protocol is not
something that changes. On the other hand, the likelihood of their config
files being incompatible is regrettably high.

-Greg Mildenhall