Re: [PATCH -V11 1/9] exportfs: Return the minimum required handle size

[Neil Brown <neilb@suse.de>]

On Sat, 22 May 2010 20:57:50 +0530
"Aneesh Kumar K. V" <aneesh.kumar@linux.vnet.ibm.com> wrote:

> On Fri, 21 May 2010 18:15:16 -0400, "J. Bruce Fields" <bfields@fieldses.org> wrote:
> > On Thu, May 20, 2010 at 01:05:30PM +0530, Aneesh Kumar K.V wrote:
> > > The exportfs encode handle function should return the minimum required
> > > handle size. This helps user to find out the handle size by passing 0
> > > handle size in the first step and then redoing to the call again with
> > > the returned handle size value.
> > 
> > The encode_fh() interface is a little confusing.  (Not your fault,
> > really, mainly it's the return value (and the special use of 255) that I
> > always find odd.)
> > 
> > But maybe it would help to have a little more documention in the
> > export_encode_fh() kerneldoc comment and/or in
> > Documentation/filesystems/nfs/Exporting?
> > 
> 
> Kernel documentation says 
> 
>  * encode_fh:
>  *    @encode_fh should store in the file handle fragment @fh (using at most
>  *    @max_len bytes) information that can be used by @decode_fh to recover the
>  *    file refered to by the &struct dentry @de.  If the @connectable flag is
>  *    set, the encode_fh() should store sufficient information so that a good
>  *    attempt can be made to find not only the file but also it's place in the
>  *    filesystem.   This typically means storing a reference to de->d_parent in
>  *    the filehandle fragment.  encode_fh() should return the number of bytes
>  *    stored or a negative error code such as %-ENOSPC
>  *
> 
> Clearly the file system encode_fh is not returning the correct return
> values. Should i fix the kernel to follow the documentation or should
> the kernel documentation should be fixed. I would prefer code, because
> the documentation look more easy/clear to follow that returning value 255.
>

The documentation is wrong in that it never returns the number of bytes.
The number of bytes is stored back in the 'max_len' by-reference argument.
The return value is a 'type' which is stored in the 4th byte of the
filehandle.

Error return is by a magic type number (255) simply because it is easier if
this is stored temporarily in fb_fileid_type which is __u8.  However it
doesn't need to be stored there.
code like
		_fh_update(fhp, fhp->fh_export, dentry);
		if (fhp->fh_handle.fh_fileid_type == 255)
			return nfserr_opnotsupp;

could be changed to
		err = _fh_update(fhp, fhp->fh_export, dentry);
		if (err < 0)
			return nfserr_opnotsupp;


and _fh_update could be changed from
		fhp->fh_handle.fh_fileid_type =
			exportfs_encode_fh(dentry, fid, &maxsize, subtreecheck);
to
		type = exportfs_encode_fh(dentry, fid, &maxsize, subtreecheck);
		if (type == 255) type = -ENOSPC; /* temp until filesystems changed*/
		if (type > 0)
			fhp-.fh_filehandle.fh_fileid_type = type;
		...
		return type;


And the documentation should be changed to report how the size is returned
and that the return value is a type, or an error.

NeilBrown