Received-SPF: softfail (google.com: domain of transitioning linux-kernel-owner@vger.kernel.org does not designate 23.128.96.19 as permitted sender) client-ip=23.128.96.19;
Date:   Wed, 20 Apr 2022 09:07:30 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     Josh Poimboeuf <jpoimboe@redhat.com>
Cc:     x86@kernel.org, brgerst@gmail.com, jiangshanlai@gmail.com,
        Andrew.Cooper3@citrix.com, linux-kernel@vger.kernel.org,
        Andy Lutomirski <luto@kernel.org>
Subject: Re: [PATCH 2/2] x86,entry: Use PUSH_AND_CLEAR_REGS for compat
Message-ID: <20220420070730.GB2731@worktop.programming.kicks-ass.net>
References: <20220419204109.520779286@infradead.org>
 <20220419205241.339242797@infradead.org>
 <20220420032123.6c344rjr4poockjr@treble>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220420032123.6c344rjr4poockjr@treble>
Precedence: bulk

On Tue, Apr 19, 2022 at 08:21:23PM -0700, Josh Poimboeuf wrote:
> On Tue, Apr 19, 2022 at 10:41:11PM +0200, Peter Zijlstra wrote:
> > Since the upper regs don't exist for ia32 code, preserving them
> > doesn't hurt and it simplifies the code.
> 
> But an attacker can still control those registers, so clearing them on
> the stack is better, as it reduces user control over the kernel stack.
> 
> 64-bit syscalls *do* have to save those registers to the stack, so
> whether it truly matters if compat mode is made equally insecure, I
> can't say.  But without evidence to the contrary, my feeling is that we
> should err on the side of caution.

Right, so earlier Brian said simpler might be better, and I figured I'd
try to see if I could make that stick, because I too like simpler ;-)

Also, since int80 already has to do this, attackers already have their
attack surface.