From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: Oops for sale Date: Wed, 11 Jun 2008 06:02:36 +0300 Message-ID: <484F404C.2040708@qumranet.com> References: <48494341.1060007@redhat.com> <4849A2B7.1030305@qumranet.com> <484D3310.6070908@redhat.com> <484D5B4F.3080300@qumranet.com> <484D6029.1020209@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Cc: kvm@vger.kernel.org To: Gerd Hoffmann Return-path: Received: from il.qumranet.com ([212.179.150.194]:27192 "EHLO il.qumranet.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753802AbYFKDEn (ORCPT ); Tue, 10 Jun 2008 23:04:43 -0400 In-Reply-To: <484D6029.1020209@redhat.com> Sender: kvm-owner@vger.kernel.org List-ID: Gerd Hoffmann wrote: > Avi Kivity wrote: > >> The unshadow-on-userspace-access heuristic is broken. Since it is also >> bad for Xenner, try removing the following lines: >> >> spte |= PT_WRITABLE_MASK; >> > > Well, that line should better not be removed. I get a hang otherwise, > guess due to running into a page-fault loop ... > > Sorry, that was unintended. >> if (user_fault) { >> mmu_unshadow(vcpu->kvm, gfn); >> goto unshadowed; >> } >> > > Deactivating those four lines fixed it. > > Yeah, I'll drop them for kvm.git. This never triggered since OSes are conscientious about scrubbing page tables, and when they do that the flood detector unshadows the pages. -- I have a truly marvellous patch that fixes the bug which this signature is too narrow to contain.