From: Keir Fraser Subject: Re: [rfc 00/18] ioemu: use devfn instead of slots as the unit for passthrough Date: Thu, 05 Mar 2009 12:44:08 +0000 Message-ID: References: <20090305101258.GB18949@verge.net.au> Mime-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20090305101258.GB18949@verge.net.au> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Simon Horman , "Jiang, Yunhong" Cc: Yuji Shimada , "xen-devel@lists.xensource.com" , Ian Jackson List-Id: xen-devel@lists.xenproject.org On 05/03/2009 10:13, "Simon Horman" wrote: >> So it is entirely possible that MSI won't be used on an MSI-capable >> function? Is this true of SR-IOV (virtual) functions too? > > Thinking more. One of the arguments surounding why 32 GSI is enough, > is that systems with more functions than that can reasonably be > expected to be using MSI. And thus will not actually use more than 32 GSI. I think that is a reasonable assumption. But the fallback to GSI does need to work, and I wonder why sharing of GSIs among passed-thru devices needs to be absolutely disallowed? We'd have an easier time if it worked. ;-) -- Keir