RE: Getting the nfs_fh of a specific file/dir from the kernel

RE: Getting the nfs_fh of a specific file/dir from the kernel

[Menny Hamburger]

We implement our own nfsd in user space - so the  kernel nfsd (as well
as the lockd) are disabled.
We need the handle in order to associate a kernel file handle with our
own file id.

M.

-----Original Message-----
From: linux-kernel-owner@vger.kernel.org
[mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of J. Bruce Fields
Sent: Thursday, February 08, 2007 10:12 PM
To: Menny Hamburger
Cc: linux-kernel@vger.kernel.org
Subject: Re: Getting the nfs_fh of a specific file/dir from the kernel

On Thu, Feb 08, 2007 at 06:38:48PM +0200, Menny Hamburger wrote:
> I have a piece of code in my 2.6 kernel that associates an ioctl 
> file_operation to nfs in file.c and dir.c.
> This ioctl sends the nfs_fh to a userland application.

Doesn't /proc/fs/nfsd/filehandle do what you want already?

See nfs-util/utils/mountd/cache.c:cache_get_filehandle(), or, for the
kernel side, linux/fs/nfsd/nfsctl.c:write_filehandle().

(Just out of curiosity--why are you doing this?)

--b.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel"
in the body of a message to majordomo@vger.kernel.org More majordomo
info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

RE: Getting the nfs_fh of a specific file/dir from the kernel

[Trond Myklebust]

On Sun, 2007-02-11 at 10:06 +0200, Menny Hamburger wrote:
> We implement our own nfsd in user space - so the  kernel nfsd (as well
> as the lockd) are disabled.
> We need the handle in order to associate a kernel file handle with our
> own file id.

Filehandles are not part of any API that is exported to userland. There
are many reasons why that would be a bad idea.

Cheers
  Trond

> M.
> 
> -----Original Message-----
> From: linux-kernel-owner@vger.kernel.org
> [mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of J. Bruce Fields
> Sent: Thursday, February 08, 2007 10:12 PM
> To: Menny Hamburger
> Cc: linux-kernel@vger.kernel.org
> Subject: Re: Getting the nfs_fh of a specific file/dir from the kernel
> 
> On Thu, Feb 08, 2007 at 06:38:48PM +0200, Menny Hamburger wrote:
> > I have a piece of code in my 2.6 kernel that associates an ioctl 
> > file_operation to nfs in file.c and dir.c.
> > This ioctl sends the nfs_fh to a userland application.
> 
> Doesn't /proc/fs/nfsd/filehandle do what you want already?
> 
> See nfs-util/utils/mountd/cache.c:cache_get_filehandle(), or, for the
> kernel side, linux/fs/nfsd/nfsctl.c:write_filehandle().
> 
> (Just out of curiosity--why are you doing this?)
> 
> --b.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel"
> in the body of a message to majordomo@vger.kernel.org More majordomo
> info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: Getting the nfs_fh of a specific file/dir from the kernel

[J. Bruce Fields]

On Tue, Feb 13, 2007 at 03:13:01PM -0800, Trond Myklebust wrote:
> On Sun, 2007-02-11 at 10:06 +0200, Menny Hamburger wrote:
> > We implement our own nfsd in user space - so the  kernel nfsd (as well
> > as the lockd) are disabled.
> > We need the handle in order to associate a kernel file handle with our
> > own file id.
> 
> Filehandles are not part of any API that is exported to userland.

/proc/fs/nfsd/filehandle ???

--b.

Re: Getting the nfs_fh of a specific file/dir from the kernel

[Trond Myklebust]

On Tue, 2007-02-13 at 18:16 -0500, J. Bruce Fields wrote:
> On Tue, Feb 13, 2007 at 03:13:01PM -0800, Trond Myklebust wrote:
> > On Sun, 2007-02-11 at 10:06 +0200, Menny Hamburger wrote:
> > > We implement our own nfsd in user space - so the  kernel nfsd (as well
> > > as the lockd) are disabled.
> > > We need the handle in order to associate a kernel file handle with our
> > > own file id.
> > 
> > Filehandles are not part of any API that is exported to userland.
> 
> /proc/fs/nfsd/filehandle ???

That is only the information needed by mountd. It should not be usable
for opening random files etc.

Cheers
  Trond

Re: Getting the nfs_fh of a specific file/dir from the kernel

[J. Bruce Fields]

On Tue, Feb 13, 2007 at 03:19:40PM -0800, Trond Myklebust wrote:
> On Tue, 2007-02-13 at 18:16 -0500, J. Bruce Fields wrote:
> > On Tue, Feb 13, 2007 at 03:13:01PM -0800, Trond Myklebust wrote:
> > > On Sun, 2007-02-11 at 10:06 +0200, Menny Hamburger wrote:
> > > > We implement our own nfsd in user space - so the  kernel nfsd (as well
> > > > as the lockd) are disabled.
> > > > We need the handle in order to associate a kernel file handle with our
> > > > own file id.
> > > 
> > > Filehandles are not part of any API that is exported to userland.
> > 
> > /proc/fs/nfsd/filehandle ???
> 
> That is only the information needed by mountd. It should not be usable
> for opening random files etc.

Oh, right--I guess it only gives filehandles for export roots--I'd
missed that.

--b.

Re: Getting the nfs_fh of a specific file/dir from the kernel

[J. Bruce Fields]

On Thu, Feb 08, 2007 at 06:38:48PM +0200, Menny Hamburger wrote:
> I have a piece of code in my 2.6 kernel that associates an ioctl
> file_operation to nfs in file.c and dir.c.
> This ioctl sends the nfs_fh to a userland application.

Doesn't /proc/fs/nfsd/filehandle do what you want already?

See nfs-util/utils/mountd/cache.c:cache_get_filehandle(), or, for the
kernel side, linux/fs/nfsd/nfsctl.c:write_filehandle().

(Just out of curiosity--why are you doing this?)

--b.

Getting the nfs_fh of a specific file/dir from the kernel

[Menny Hamburger]

Hi,

I have a piece of code in my 2.6 kernel that associates an ioctl
file_operation to nfs in file.c and dir.c.
This ioctl sends the nfs_fh to a userland application.
I have been trying to remove this ugly code by creating my own device
and implementing the ioctl but I keep getting junk instead.
Please tell me if I'm doing anything wrong here:

1) Define the I/O structure:
    struct nfs_getfh {
        int fd;
        struct nfs_fh *fh;
    };
2) Receive the  fd of the wanted file/dir.
3) inside the ioctl function:

    struct file *filp;
    struct dentry *dentp;
    struct inode *inop;
    struct nfs_getfh getfh;
    struct files_struct *files;
    struct nfs_getfh __user *user_getfh = (struct nfs_getfh __user *)
arg;

    files = current->files;
    if (files) {
        spin_lock(&files->file_lock);
        filp = fcheck(getfh.fd);
        if (filp) {
            dentp = dget(filp->f_dentry);
	if (dentp) {
	    inop = dentp->d_inode;
	    if (inop) {
	        getfh.fh = NFS_FH(inop);
	        if (getfh.fh) {
                        copy_to_user((void*) user_getfh->fh, getfh.fh,
getfh.fh->size + ((int) &((struct nfs_fh*)0)->data)))
                    }
                }
            }
        }
        spin_unlock(&files->file_lock);
     }

Best Regards,
Menny