[pnfs] filehandles and GETDEVICELIST/GETDEVICEINFO

[david m. richter]

On Thu, May 1, 2008 at 11:36 AM, Benny Halevy <bhalevy at panasas.com> wrote:
> On May. 01, 2008, 16:56 +0300, "david m. richter" <richterd at gmail.com> wrote:
>  > hello,
>  >
>  > i found in the server code that both GETDEVICELIST and GETDEVICEINFO
>  > get a flag set that permits the operation to proceed without a current
>  > filehandle.
>
>  Can you please be more specific.
>  I'm not sure what you're referring to...

sure, sure -- it's just in one place.  in nfsd/nfs4proc.c, struct
nfsd4_operation nfsd4_ops has ALLOWED_WITHOUT_FH set for
OP_GETDEVICELIST (and it's only checked once, in
nfsd4_proc_compound()).

>  Benny
>
>
>  >
>  > that sounds correct for GETDEVICEINFO, but it seems that GETDEVICELIST
>  > would need a filehandle so the server can id the filesystem in
>  > question (and the draft confirms that).  so i'm thinking that i should
>  > remove the flag for -LIST, yes?
>
>  GETDEVICEINFO doesn't use the current filehandle.
>  GETDEVICELIST must have a valid filehandle.

ok, that's what i wanted to confirm.

>  >
>  > relatedly, the draft says that a possible error value for
>  > GETDEVICEINFO is NFS4ERR_NOFILEHANDLE (but not NFS4ERR_FHEXPIRED, at
>  > least) -- that error should be removed from the draft, yes?
>
>  Hmm, that's wrong.  GETDEVICEINFO needs no file handle
>  [nfsv4 mailing list Cc'ed]
>  I think that NFS4ERR_FHEXPIRED should allowed for all ops that
>  are allowed to return NFS4ERR_NOFILEHANDLE.  This seems like
>  something that will be fixed when the editors make the final scrubbing
>  of ops errors codes.

good, just wanted to make sure.

thanks, benny,

  d
  .

>  Benny
>
>  >
>  >
>  > thanks kindly,
>  >
>  >   d
>  >   .
>  > _______________________________________________
>  > pNFS mailing list
>  > pNFS at linux-nfs.org
>  > http://linux-nfs.org/cgi-bin/mailman/listinfo/pnfs
>
>