From mboxrd@z Thu Jan  1 00:00:00 1970
From: Radim =?utf-8?B?S3LEjW3DocWZ?= <rkrcmar@redhat.com>
Subject: Re: [PATCH 2/4] KVM: nSVM: do not forward NMI window singlestep VM
 exits to L1
Date: Mon, 19 Jun 2017 19:46:29 +0200
Message-ID: <20170619174629.GA10318@potion>
References: <20170615112032.15812-1-lprosek@redhat.com>
 <20170615112032.15812-3-lprosek@redhat.com>
 <20170616132648.GF2224@potion>
 <CABdb734PVjgvJr9qSLraF1W+3goJ09ku9tgC=BEnBckWJW11Dg@mail.gmail.com>
 <CABdb7355wnHaUaKe7kOXDFxjNXCJV5N+i7BBbnsHQe1AbR7vNg@mail.gmail.com>
 <20170619161740.GA13549@potion>
 <b1132747-75bd-6399-48fb-edf9277e2468@redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit
Cc: Ladi Prosek <lprosek@redhat.com>, KVM list <kvm@vger.kernel.org>
To: Paolo Bonzini <pbonzini@redhat.com>
Return-path: <kvm-owner@vger.kernel.org>
Received: from mx1.redhat.com ([209.132.183.28]:54990 "EHLO mx1.redhat.com"
        rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
        id S1751024AbdFSRqc (ORCPT <rfc822;kvm@vger.kernel.org>);
        Mon, 19 Jun 2017 13:46:32 -0400
Received: from smtp.corp.redhat.com (int-mx02.intmail.prod.int.phx2.redhat.com [10.5.11.12])
        (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits))
        (No client certificate requested)
        by mx1.redhat.com (Postfix) with ESMTPS id 44D7D85A07
        for <kvm@vger.kernel.org>; Mon, 19 Jun 2017 17:46:32 +0000 (UTC)
Content-Disposition: inline
In-Reply-To: <b1132747-75bd-6399-48fb-edf9277e2468@redhat.com>
Sender: kvm-owner@vger.kernel.org
List-ID: <kvm.vger.kernel.org>

2017-06-19 19:17+0200, Paolo Bonzini:
> On 19/06/2017 18:17, Radim Krčmář wrote:
> > Right, we only need the single step over IRET and interrupt shadow.
> > 
> > Btw. instead of single-stepping over IRET/interrupt shadow, could we set
> > INTERRUPT_SHADOW in VMCB, inject the NMI, and let it execute?
> > This mechanism would explain why AMD didn't provide a trap for IRET ...
> 
> You mean they didn't provide a trap-like VMEXIT for IRET, only fault-like?

Yes.  SVM has trap-like VM exit, so I didn't understand why it was not
used for IRET.  Forcing the hypervisor to have two VM exits and a clumsy
single-step felt out of place when the rest was designed nicely ...