Re: [PATCH] Fix root hole in raw device

[Al Viro <viro@parcelfarce.linux.theplanet.co.uk>]

On Tue, May 17, 2005 at 09:03:05AM +0200, Willy Tarreau wrote:
> On Tue, May 17, 2005 at 05:57:48AM +0100, Al Viro wrote:
> > On Mon, May 16, 2005 at 09:37:48PM -0700, Greg KH wrote:
> > > @@ -122,7 +122,7 @@
> > >  {
> > >  	struct block_device *bdev = filp->private_data;
> > >  
> > > -	return ioctl_by_bdev(bdev, command, arg);
> > > +	return blkdev_ioctl(bdev->bd_inode, filp, command, arg);
> > >  }
> > 
> > That is not quite correct.  You are passing very odd filp to ->ioctl()...
> > Old variant gave NULL, which is also not too nice, though.
> 
> 2.4 already does it in a cleaner manner :
> 
>         err = -EINVAL;
>         if (b && b->bd_inode && b->bd_op && b->bd_op->ioctl) {
>                 err = b->bd_op->ioctl(b->bd_inode, NULL, command, arg);
>         }
>         return err;
> 
> So may be something like this would be better (hand-written) :

> +	if (bdev && bdev->bd_inode)
> +		err = blkdev_ioctl(bdev->bd_inode, filp, command, arg);
> +	return err;

Huh?  These checks are useless and you are still passing a struct file *
that has nothing to do with what ->ioctl() expects.