From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754457AbcGZAVg (ORCPT ); Mon, 25 Jul 2016 20:21:36 -0400 Received: from mail-yw0-f174.google.com ([209.85.161.174]:36562 "EHLO mail-yw0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752108AbcGZAVf (ORCPT ); Mon, 25 Jul 2016 20:21:35 -0400 MIME-Version: 1.0 In-Reply-To: References: <8bacd629dcdbb97ee0626912c27a4a09f991b5dd.1466464928.git.luto@kernel.org> <20160725063825.GB12474@gmail.com> From: Andy Lutomirski Date: Mon, 25 Jul 2016 17:21:15 -0700 Message-ID: Subject: Re: [PATCH v3 1/3] x86/ptrace: Stop setting TS_COMPAT in ptrace code To: Ingo Molnar Cc: Pedro Alves , Oleg Nesterov , Kees Cook , Borislav Petkov , "linux-kernel@vger.kernel.org" , X86 ML Content-Type: text/plain; charset=UTF-8 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Jul 25, 2016 at 9:57 AM, Andy Lutomirski wrote: > On Jul 24, 2016 11:38 PM, "Ingo Molnar" wrote: >> >> >> * Andy Lutomirski wrote: >> >> > On Mon, Jun 20, 2016 at 4:39 PM, Andy Lutomirski wrote: >> > > Setting TS_COMPAT in ptrace is wrong: if we happen to do it during >> > > syscall entry, then we'll confuse seccomp and audit. (The former >> > > isn't a security problem: seccomp is currently entirely insecure if a >> > > malicious ptracer is attached.) As a minimal fix, this patch adds a >> > > new flag TS_I386_REGS_POKED that handles the ptrace special case. >> > >> > Hi Ingo- >> > >> > Could you apply this one patch for 4.8? While I don't think it's a >> > significant security issue in 4.7 or earlier, leaving it unfixed in >> > 4.8 will introduce a potentially unpleasant interaction with some >> > seccomp changes that are queued up in the >> > security tree for 4.8. >> > >> > It will have a trivially-resolvable conflict with -mm. >> > >> > The rest of the series this is in can wait. >> >> I don't mind the rest of the series either - could you please repost it (with the >> review feedback addressed)? > > I'm nervous about it for a couple reasons involving the fact that it's > user visible. > > 1. It doesn't make gdb work right in all the cases that gdb currently > gets wrong. I haven't had time to think about whether there's a > minimal tweak that would fix this. After re-reading the whole thread, I think that the rest of the series needs a good self-test to make sure that we're providing whatever behavior we think we're providing. So I really don't what to apply it yet. --Andy