From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751282AbaI3XVe (ORCPT ); Tue, 30 Sep 2014 19:21:34 -0400 Received: from www.linutronix.de ([62.245.132.108]:33975 "EHLO Galois.linutronix.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750850AbaI3XVd (ORCPT ); Tue, 30 Sep 2014 19:21:33 -0400 Date: Wed, 1 Oct 2014 01:21:23 +0200 (CEST) From: Thomas Gleixner To: Andy Lutomirski cc: Sebastian Lackner , X86 ML , Ingo Molnar , "H. Peter Anvin" , Anish Bhatt , "linux-kernel@vger.kernel.org" , Chuck Ebbert Subject: Re: [PATCH 1/2] x86_64,entry: Filter RFLAGS.NT on entry from userspace In-Reply-To: Message-ID: References: <542B2305.9060805@fds-team.de> User-Agent: Alpine 2.11 (DEB 23 2013-08-11) MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Linutronix-Spam-Score: -1.0 X-Linutronix-Spam-Level: - X-Linutronix-Spam-Status: No , -1.0 points, 5.0 required, ALL_TRUSTED=-1,SHORTCIRCUIT=-0.0001 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, 30 Sep 2014, Andy Lutomirski wrote: > On Tue, Sep 30, 2014 at 3:27 PM, Thomas Gleixner wrote: > > On Tue, 30 Sep 2014, Andy Lutomirski wrote: > >> It would certainly be possible to clear NT and retry IRET if IRET > >> fails with NT set. This would have no overhead for anything relevant. > >> That would be this alternative from my 0/2 email: > >> > >> - Don't filter NT on sysenter. Instead, filter it on EFI entry > >> and modify the IRET code to retry without NT set if NT was set. > >> > >> Thomas hpa, etc: any thoughts? > > > > Filter it right away. That's solid and obvious. Anything else is just > > complex and prone for future brown paperbag failures. > > Yeah, agreed. That's exactly what these patches do, although, if you > put them in -tip and want to keep the stable CC, it's probably worth > fixing the address (oops). Even more oops as you failed to update it in your reply again .... > > We get the context switch benefit from it, so there is some > > compensation for the extra cycles. > > If we ever want those cycles back, I bet that the compat sysenter path > could be trimmed down a lot. For example, I think that all of the > zero-extension stuff is unnecessary now that we have the magic syscall > wrappers for all (?) syscalls. Emphasis on "(?)". So yes, once we verified that .... Thanks, tglx