From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [PATCH 0/3] Remove interrupt stack table usage from x86_64 kernel Date: Sun, 28 Dec 2008 22:28:42 +0200 Message-ID: <4957E17A.9080203@redhat.com> References: <87vdt5vfxc.fsf@basil.nowhere.org> <4956A0B1.1060908@redhat.com> <20081227224029.GB496@one.firstfloor.org> <49573FE7.9090802@redhat.com> <20081228131605.GC496@one.firstfloor.org> <49578896.8030703@redhat.com> <20081228190840.GD496@one.firstfloor.org> <4957CE78.6030706@redhat.com> <4957D14B.60805@redhat.com> <4957DCC3.3060505@redhat.com> <20081228203425.GF496@one.firstfloor.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Ingo Molnar , x86@kernel.org, linux-kernel , KVM list To: Andi Kleen Return-path: Received: from mx2.redhat.com ([66.187.237.31]:39390 "EHLO mx2.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750748AbYL1U2o (ORCPT ); Sun, 28 Dec 2008 15:28:44 -0500 In-Reply-To: <20081228203425.GF496@one.firstfloor.org> Sender: kvm-owner@vger.kernel.org List-ID: Andi Kleen wrote: >> This makes the whole thing unworthwhile. The vmload/vmsave pair costs >> only 200 cycles (I should have started with this), and 120 cycles on the >> heavyweight path plus complexity are not worth 200 cycles on the >> lightweight path. >> > > Actually to switch ISTs you need to change the TSS, not the IDT. > But I suppose that won't be any faster. > I can't touch the TSS (that's the starting point of the exercise). The plan was to have a copy of the IDT with all IST pointers zeroed out (or to have per-cpu IDT and zero out the IST pointers when entering guest mode, restoring them on context switch). It's not worth it though. -- I have a truly marvellous patch that fixes the bug which this signature is too narrow to contain.