From mboxrd@z Thu Jan 1 00:00:00 1970 From: Neo Jia Subject: Re: [PATCH 0/2] extend page_track for external usage Date: Wed, 26 Oct 2016 01:03:36 -0700 Message-ID: <20161026080336.GA13559@nvidia.com> References: <1477381843-18375-1-git-send-email-jike.song@intel.com> <20161025164224.GA27343@nvidia.com> <58101290.7070802@intel.com> <20161026073939.GA12839@nvidia.com> <45e4f895-5195-3048-6ace-4d9d4e699692@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Cc: Jike Song , , , , , To: Paolo Bonzini Return-path: Received: from hqemgate16.nvidia.com ([216.228.121.65]:14413 "EHLO hqemgate16.nvidia.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932267AbcJZIDl (ORCPT ); Wed, 26 Oct 2016 04:03:41 -0400 Content-Disposition: inline In-Reply-To: <45e4f895-5195-3048-6ace-4d9d4e699692@redhat.com> Sender: kvm-owner@vger.kernel.org List-ID: On Wed, Oct 26, 2016 at 09:44:05AM +0200, Paolo Bonzini wrote: > > > On 26/10/2016 09:39, Neo Jia wrote: > > > The page_track implementation is almost ready for external usage, my > > > changes here are quite trivial. I guess the external user like your > > > vendor driver can call them directly. > > > > Unfortunately, I don't think the current implementation is accessible to us, > > could you please change? > > How so? > > > > BTW, do you also have to shadow the device page-tables? Per my previous > > > understanding you don't have to. > > > > It is not for the page table, we have other features will require such wp > > capabilities to trap guest memory access. > > Can you please expand? Sure, this wp capabilities will allow us to check GPU command stream for VMs. Thanks, Neo > > Paolo