nnpfs usage

christopher@0x90.org christopher at 0x90.org
Fri Sep 10 07:52:25 CEST 2004


On Fri, Sep 10, 2004 at 12:03:42AM +0200, Tomas Olsson wrote:
> Just some quick comments for now:
> I don't think it matters, but the 1.1.1.1 /afs/fakenfile/.. looks odd.

That was a typo when I was copying it down, its actually 0.1.1.1 in the code,
the rationale trying to be:
    0.x.x.x -> 'root' node, same as /afs
    1.x.x.x -> cell nodes beneath

I am still not sure if it matters that second level directories are a different
cell than the root, I was trying to emulate the regular afs behavior of
/afs/cell.hostname.tld/

> Maybe you want to compare to the sequence of messages you get when mounting
> a similar volume as AFS root. For details of what happens inside nnpfs, the
> 'fs nnpfsdebug' command helps a lot. s/nnpfs/xfs/ where applicable.

After comparing them yet again, ended up getting a whole new set of errors,
turns out it was a thinko after all; missed a memset in the getnode handler.

Can anyone explain why the root node needs an inode number >0 in the dirent
directory files, yet passing anything to nnpfs OTHER THAN ZERO causes problems?

Thanks,
Christopher

--
christopher at 0x90.org






More information about the Arla-drinkers mailing list