From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paolo Bonzini Subject: Re: [PATCH] kvm: x86: vmx: add checks on guest RIP Date: Thu, 04 Dec 2014 08:41:24 +0100 Message-ID: <54801024.2050504@redhat.com> References: <20141129152726.GA25370@gnote> <547C9700.6020407@redhat.com> <547F9513.3090600@amacapital.net> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit To: Andy Lutomirski , Eugene Korenevsky , kvm@vger.kernel.org Return-path: Received: from mx1.redhat.com ([209.132.183.28]:59345 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751110AbaLDHlc (ORCPT ); Thu, 4 Dec 2014 02:41:32 -0500 In-Reply-To: <547F9513.3090600@amacapital.net> Sender: kvm-owner@vger.kernel.org List-ID: On 03/12/2014 23:56, Andy Lutomirski wrote: > > This check is off by one. It is checking bits 63:47 instead of bits > > 63:48 (this quirk is intentionally part of the specification, so that > > you can reenter a guest at 0x800000000000 after e.g. a VMCALL vmexit and > > cause a general protection fault). > > Seriously? Intel did that for vmcall but not sysret? Yes, it is even tested by kvm-unit-tests. :) Paolo