From: Haynes, Tom (thomas@netapp.com)
Date: 11/19/02-11:31:38 AM Z
From: "Haynes, Tom" <thomas@netapp.com> Message-Id: <200211191731.gAJHVcq01528@verona.eng.netapp.com> Subject: Re: Should PUTPUBFH and PUTROOTFH check that fh is a dir Date: Tue, 19 Nov 2002 09:31:38 -0800 (PST) > Carl Burnett wrote: > > It seems that it would be a good idea for PUTPUBFH and PUTROOTFH to check > > that the provided filehandle represents a directory . If not, then these > > ops would return NFS4ERR_NOTDIR. The spec currently does not show these as > > error returns for these operations. > > > > What do people think? > > In many common NFS server solutions it is possible to export just a > single file, not a directory. The old SunOS 4.x diskless swap > files used this feature. Would this interpretation force a > server that wanted to export just a single file to also export its > parent directory as well? > > -David > In the case of PUTROOTFH, you cannot export a single file - you need to provide a root of a pseudo-fs from the spec. This root need not be the parent of the single file, indeed it would be more flexible to build a pseudo-directory chain from the top of the fs. As for the PUTPUBFH, yes it can be a FH to a file and not a directory, at least as the spec states: Unlike the ROOT filehandle, the PUBLIC filehandle may be bound or represent an arbitrary filesystem object at the server. Most implementations may chose to map the PUTPUBFH to the PUTROOTFH, but assuming that it is the PUTROOTFH violates the spec. Also, assuming it is a directory also violates the spec as written. -- Tom Haynes, cfb thomas@netapp.com
This archive was generated by hypermail 2.1.2 : 03/04/05-01:50:28 AM Z CST