From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Kallol Biswas" Subject: RE: NIC: Queue assignment to guest OS Date: Tue, 19 Apr 2011 16:24:59 -0700 Message-ID: <005801cbfee9$00ef50f0$02cdf2d0$@com> References: <003801cbfb0a$de67a410$9b36ec30$@com> <20110419184750.GB7427@dumpdata.com> <8DE07D65-675C-4AF7-8B4F-AA643173250F@nucleodyne.com> <20110419204532.GA13003@dumpdata.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20110419204532.GA13003@dumpdata.com> Content-Language: en-us List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: 'Konrad Rzeszutek Wilk' Cc: xen-devel@lists.xensource.com List-Id: xen-devel@lists.xenproject.org If it is ok to modify host the network driver how should we do it in Xen? What modules/driver should be changed. The idea is to avoid backend/frontend implementation, and give direct HW access to guest for maximum performance. There are certain issues with guests accessing HW directly, one of them is guest to guest communication, and another is VM migration. Sorry again, legal agreement prevents us to disclose anything in previous implementation; we overcame these issues in a certain way. Xen NIC folks should be able to address the issues. > > > > How did you do this in other virtualization frameworks? I am still at loss here how you guys managed this without ripping the network driver of a NIC card and passed the MSI-X to a guest and the full PCI device to _all_ of them and managed to negotiate some form of locking. _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel