From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Cooper Subject: Re: passing through SMBus Date: Fri, 22 Nov 2013 11:20:35 +0000 Message-ID: <528F3E03.9000205@citrix.com> References: <5284AC2D.6060304@citrix.com> <528547DE.3050704@citrix.com> <5289FB0D.4@citrix.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Shailesh Kumar Cc: "xen-devel@lists.xen.org" List-Id: xen-devel@lists.xenproject.org On 22/11/13 00:05, Shailesh Kumar wrote: > On Mon, Nov 18, 2013 at 3:33 AM, Andrew Cooper > wrote: >> On 18/11/13 06:40, Shailesh Kumar wrote: >> >> Andrew, >> >> Did I miss to share any details . >> Let me know if you have more questions for me . >> >> waiting for your suggestions. >> >> thanks >> TSK >> >> >> Once again, please do not top post. >> >> Qemu has its own chipset emulation. Attempting to shoehorn part of a host >> chipset into a virtualised guest chipset is unlikely to work. >> >> If the Win7 vm is on a headless server, why do you need the local mouse if >> you are connecting via VNC (I am assuming VNC as you only specified 'ip'). >> >> ~Andrew >> > Andrew > > I got the SMBus pass-through to my guest and I don't see any problems yet. > I am still curious to understand what kind of problem I expect to observe. > > AFAIK Qemu is designed to pass-through all the sub-functions of a > given device only if function '0' of that device is passed through. > Which is targeting to multi-function devices. > where as if one want to pass-through a specific sub-function on a > bridge it won't happen coz qemu or xl are not handling it. > > BTW: I cleaned up the top post mess and also did some reading on > etiquette and will be practicing that. > > -tsk They were concerns, rather than explicit points which wouldn't work. It might be worth using pciback.hide on dom0's boot command line to try and prevent any dom0 drivers from binding against the SMBus device. There are certainly issues passing through device functions. Enumeration of the PCI devices require function 0 to exist and advertise itself as multifunction before you can look for other functions. Therefore, to pass through a single function which is not function 0, there needs to be an additional function 0 faked up. I am not aware of any safe/sane way of doing this. ~Andrew