From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [PATCH v2] core, x86: Add user return notifiers Date: Tue, 22 Sep 2009 20:08:59 +0300 Message-ID: <4AB904AB.3020307@redhat.com> References: <4AB3CDEC.5050902@zytor.com> <1253342422-13811-1-git-send-email-avi@redhat.com> <4AB8980D.5070803@redhat.com> <20090922143223.GA31702@elte.hu> <4AB8E2F1.6030401@redhat.com> <4AB8F256.2080205@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Peter Zijlstra , "H. Peter Anvin" , linux-kernel@vger.kernel.org, kvm@vger.kernel.org, Marcelo Tosatti To: Ingo Molnar Return-path: Received: from mx1.redhat.com ([209.132.183.28]:34941 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751232AbZIVRJL (ORCPT ); Tue, 22 Sep 2009 13:09:11 -0400 In-Reply-To: <4AB8F256.2080205@redhat.com> Sender: kvm-owner@vger.kernel.org List-ID: On 09/22/2009 06:50 PM, Avi Kivity wrote: > On 09/22/2009 05:45 PM, Avi Kivity wrote: >>> Would be nice to convert some existing open-coded return-to-user-space >>> logic to this facility. One such candidate would be lockdep_sys_exit? >> >> I only implemented this for x86, while lockdep is arch independent. >> If arch support is added, it should be trivial. >> > > The lockdep_sys_exit bit is actually x86/s390 only, and can easily be > adapted to use the new functionality on x86 only. I'll try it out. Unfortunately it doesn't work out well. The notifier is called until explicitly unregistered (since it relies on a bit in TIF_NOTIFY_MASK), so we have to disarm it on the first return to usersspace or it spins forever. We could re-arm it on the next kernel entry, but we don't have a kernel entry notifier so we'll just be moving hooks from one point to another. -- error compiling committee.c: too many arguments to function