From mboxrd@z Thu Jan 1 00:00:00 1970 From: Keir Fraser Subject: Re: granting access to MSI-X table and pending bit array Date: Fri, 9 Jul 2010 10:59:20 +0100 Message-ID: References: <4C370DDF020000780000A718@vpn.id2.novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4C370DDF020000780000A718@vpn.id2.novell.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Jan Beulich , Stefano Stabellini Cc: "xen-devel@lists.xensource.com" , Rzeszutek Wilk , Konrad List-Id: xen-devel@lists.xenproject.org On 09/07/2010 10:54, "Jan Beulich" wrote: > With the revert mentioned above being about PV guests, I wonder > what problems PV guests had with not being able to access the > table and PBA ranges. Linux doesn't (so far) use the PBA at all, > and according to my analysis of drivers/pci/msi-xen.c also doesn't > access the table in any way even when being run privileged. Keir, > with the description of c/s 20171 being rather brief - what was > the problem PV guests were having? Or was that more specifically > with stubdom wanting to do above mentioned writes on behalf of > the serviced guest? I was trying to pass through an Intel ixgbe VF to a PV guest. The guest driver attempts to map some resource range (I think indicated via a BAR) and that fails because one page in the range is disallowed, corresponding to an MSI-X iomem page. Reverting this patch made things work. :-) Perhaps a dummy page should get silently mapped in or something, but I didn't investigate further myself. -- Keir