[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <202406201158.346A2BE@keescook>
Date: Thu, 20 Jun 2024 11:59:12 -0700
From: Kees Cook <kees@...nel.org>
To: Anjali K <anjalik@...ux.ibm.com>
Cc: mpe@...erman.id.au, linuxppc-dev@...ts.ozlabs.org, npiggin@...il.com,
naveen@...nel.org, christophe.leroy@...roup.eu,
gustavoars@...nel.org, linux-hardening@...r.kernel.org,
vishalc@...ux.ibm.com
Subject: Re: [PATCH] powerpc/pseries: Whitelist dtl slub object for copying
to userspace
On Thu, Jun 20, 2024 at 10:58:49PM +0530, Anjali K wrote:
> However given that:
> (i) The dtl buffer is read-only. The dtl trace is a set of metrics which
> are collected to be read by privileged users.
> (ii) Users usually reads all the dtl entries, not a subset.
> (iii) Read overflows are unlikely to expose anything useful to attackers
> since we are whitelisting the complete slub object and there are no
> contiguous memory locations which need to be hidden.
> Can we go ahead with the whitelisting using kmem_cache_create_usercopy()
> approach?
> Or are there other reasons to prefer the bounce buffer approach?
Yeah, based on this and what mpe said, I have no objection to just
allowing it in kmem_cache_create_usercopy(). I was mainly just curious
what the threat model was. :)
Reviewed-by: Kees Cook <kees@...nel.org>
--
Kees Cook
Powered by blists - more mailing lists