Re: xfs mknod regression

[David Chinner <dgc@sgi.com>]

On Tue, Dec 18, 2007 at 05:36:42PM +0000, Christoph Hellwig wrote:
> 
> This was broken by my '[XFS] simplify xfs_create/mknod/symlink prototype',
> which assigned the re-shuffled ondisk dev_t back to the rdev variable in
> xfs_vn_mknod.  Because of that i_rdev is set to the ondisk dev_t instead
> of the linux dev_t later down the function.
> 
> Fortunately the fix for it is trivial:  we can just remove the
> assignment because xfs_revalidate_inode has done the proper job before
> unlocking the inode.
> 
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> 
> Index: linux-2.6-xfs/fs/xfs/linux-2.6/xfs_iops.c
> ===================================================================
> --- linux-2.6-xfs.orig/fs/xfs/linux-2.6/xfs_iops.c	2007-12-18 18:23:32.000000000 +0100
> +++ linux-2.6-xfs/fs/xfs/linux-2.6/xfs_iops.c	2007-12-18 18:23:43.000000000 +0100
> @@ -345,9 +345,7 @@ xfs_vn_mknod(
>  		ASSERT(vp);
>  		ip = vn_to_inode(vp);
>  
> -		if (S_ISCHR(mode) || S_ISBLK(mode))
> -			ip->i_rdev = rdev;
> -		else if (S_ISDIR(mode))
> +		if (S_ISDIR(mode))
>  			xfs_validate_fields(ip);
>  		d_instantiate(dentry, ip);
>  		xfs_validate_fields(dir);

Thanks for this, Christoph - I'll run some tests on it and check it in.

Rafael - this is a regression introduced in 2.6.24-rc1 if you want to
track it.

Cheers,

Dave.
-- 
Dave Chinner
Principal Engineer
SGI Australian Software Group