linux-kernel - Re: [PATCH] x86/vsyscall: Remove CONFIG_LEGACY_VSYSCALL

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] [day] [month] [year] [list]

Message-ID: <202205111424.DEB0E3418@keescook>
Date:   Wed, 11 May 2022 14:25:26 -0700
From:   Kees Cook <keescook@...omium.org>
To:     Florian Weimer <fweimer@...hat.com>
Cc:     Andy Lutomirski <luto@...nel.org>, x86@...nel.org,
        LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] x86/vsyscall: Remove CONFIG_LEGACY_VSYSCALL_EMULATE

On Wed, May 11, 2022 at 08:36:48PM +0200, Florian Weimer wrote:
> * Kees Cook:
> 
> > On Wed, May 11, 2022 at 10:38:53AM -0700, Andy Lutomirski wrote:
> >> CONFIG_LEGACY_VSYSCALL_EMULATE is, as far as I know, only needed for the
> >> combined use of exotic and outdated debugging mechanisms with outdated
> >> binaries.  At this point, no one should be using it.  We would like to
> >> implement dynamic switching of vsyscalls, but this is much more
> >> complicated to support in EMULATE mode than XONLY mode.
> >> 
> >> So let's force all the distros off of EMULATE mode.  If anyone actually
> >> needs it, they can set vsyscall=emulate, and we can then get away with
> >> refusing to support newer security models if that option is set.
> >> 
> >> Cc: x86@...nel.org
> >> Cc: Kees Cook <keescook@...omium.org>
> >> Cc: Florian Weimer <fweimer@...hat.com>
> >> Signed-off-by: Andy Lutomirski <luto@...nel.org>
> >
> > Sounds legit! Can we switch the default to "none" while we're at it?
> 
> I think that's a userspace ABI break?
> 
> I do not have a *very* strong opinion about this.  For one thing, it
> would be handy example to have that such deliberate  do actually
> happen. 8-)

Right, I'm just suggesting switch the Kconfig default, so systems that
don't care will get that setting, and systems that do care can still
pick what they need.

(But yes, a separate patch, etc)

-- 
Kees Cook