linux-kernel - RE: [PATCH] staging: comedi: cast to (unsigned int *)

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <abc2c1d0a23847c9aa581eb9bf9c5161@AcuMS.aculab.com>
Date:   Fri, 19 Feb 2021 09:36:48 +0000
From:   David Laight <David.Laight@...LAB.COM>
To:     'Dan Carpenter' <dan.carpenter@...cle.com>
CC:     'Ian Abbott' <abbotti@....co.uk>,
        Greg KH <gregkh@...uxfoundation.org>,
        Atul Gopinathan <atulgopinathan@...il.com>,
        "devel@...verdev.osuosl.org" <devel@...verdev.osuosl.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH] staging: comedi: cast to (unsigned int *)

From: Dan Carpenter
> Sent: 19 February 2021 09:26
> 
> On Fri, Feb 19, 2021 at 09:03:59AM +0000, David Laight wrote:
> > > It's kind of moot anyway because the patch is outdated.  But the reason
> > > for the ___force is that the same `struct comedi_cmd` is used in both
> > > user and kernel contexts.  In user contexts, the `chanlist` member
> > > points to user memory and in kernel contexts it points to kernel memory
> > > (copied from userspace).
> >
> > Can't you use a union of the user and kernel pointers?
> > (Possibly even anonymous?)
> > Although, ideally, keeping them in separate fields is better.
> > 8 bytes for a pointer isn't going make a fat lot of difference.
> >
> 
> Creating a union is worse than adding casts.  With the casts, at least
> you know that you're doing something dangerous.  It's good that it looks
> scary because it is scary.
> 
> Keeping them in separate fields is a good idea, but this is part of the
> user space API so it's not possible.
> 
> The best we can do is adding some more comments so people know why we
> are doing the scary casts.

Another option is to use a longer structure in the kernel with the kernel
pointer in the 'extension'.
So you could have:
struct kernel_foo {
	struct foo;
	void *kernel_pointer;
};

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)