Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Subject: Re: x86/retpoline: Fill RSB on context switch for affected CPUs
To:     Andi Kleen <ak@linux.intel.com>
Cc:     "Woodhouse, David" <dwmw@amazon.co.uk>,
        Paul Turner <pjt@google.com>,
        LKML <linux-kernel@vger.kernel.org>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Greg Kroah-Hartman <gregkh@linux-foundation.org>,
        Tim Chen <tim.c.chen@linux.intel.com>,
        Dave Hansen <dave.hansen@intel.com>, tglx@linutronix.de,
        Kees Cook <keescook@google.com>,
        Rik van Riel <riel@redhat.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Andy Lutomirski <luto@amacapital.net>,
        Jiri Kosina <jikos@kernel.org>, gnomes@lxorguk.ukuu.org.uk,
        x86@kernel.org, thomas.lendacky@amd.com,
        Josh Poimboeuf <jpoimboe@redhat.com>
References: <1515779365-9032-1-git-send-email-dwmw@amazon.co.uk>
 <f2faf9ff-8955-d8b5-c521-91b9c273128b@maciej.szmigiero.name>
 <20180309151428.GE22087@tassilo.jf.intel.com>
From:   "Maciej S. Szmigiero" <mail@maciej.szmigiero.name>
Message-ID: <f3834e5b-ae79-79b0-f280-fbb3871e0030@maciej.szmigiero.name>
Date:   Fri, 9 Mar 2018 16:33:51 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
 Thunderbird/52.6.0
MIME-Version: 1.0
In-Reply-To: <20180309151428.GE22087@tassilo.jf.intel.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 09.03.2018 16:14, Andi Kleen wrote:
>> Shouldn't the RSB filling on context switch also be done on non-IBPB
>> CPUs to protect (retpolined) user space tasks from other user space
>> tasks?
> 
> The comment is actually incorrect. There's no risk to hit user space
> addresses if we have KPTI and NX (which is fairly universal).
> 
> It's mainly needed on Skylake era CPUs.
> 
> Should fix the comment. I'll send a patch.

But what about userspace-to-userspace attacks? - the ones that IBPB on 
context switches currently protects against (at least for high-value, or
as implemented currently, non-dumpable, processes)?

If understand the issue correctly, high-value user space processes can
be protected from other user space processes even on CPUs that lack
IBPB as long as they are recompiled with retpolines and there is no
danger of RSB entries from one process being used by another one after
a context switch.
For Skyklake this would not be enough, but there we'll (hopefully) have
the IBPB instead.

> -Andi
> 

Maciej