Date: Tue, 17 May 2005 08:35:27 -0700
From: Greg KH <greg@kroah.com>
To: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Cc: linux-kernel@vger.kernel.org, petero2@telia.com
Subject: Re: [PATCH] Fix root hole in pktcdvd
Message-ID: <20050517153527.GA23281@kroah.com>
References: <11163046681444@kroah.com> <11163046692974@kroah.com> <20050517050025.GP1150@parcelfarce.linux.theplanet.co.uk>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20050517050025.GP1150@parcelfarce.linux.theplanet.co.uk>
User-Agent: Mutt/1.5.8i
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1247
Lines: 26

On Tue, May 17, 2005 at 06:00:25AM +0100, Al Viro wrote:
> On Mon, May 16, 2005 at 09:37:49PM -0700, Greg KH wrote:
> > ioctl_by_bdev may only be used INSIDE the kernel.  If the "arg" argument
> > refers to memory that is accessed by put_user/get_user in the ioctl
> > function, the memory needs to be in the kernel address space (that's the
> > set_fs(KERNEL_DS) doing in the ioctl_by_bdev).  This works on i386 because
> > even with set_fs(KERNEL_DS) the user space memory is still accessible with
> > put_user/get_user.  That is not true for s390.  In short the ioctl
> > implementation of the pktcdvd device driver is horribly broken.
> 
> Same comment as for previous patch.  I'll take a look at that sucker,
> it might happen to be OK, seeing that most of the bdev ->ioctl() instances
> ignore file argument and we might get away with passing odd stuff to
> anything that could occur here.

Ok, so, do you suggest we just pass NULL in there, or do you have a
better suggestion?

thanks,

greg k-h
-
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/