Re: [PATCH] [RESEND] VT-d: Fix iommu map page for mmio pages

[Muli Ben-Yehuda <muli@il.ibm.com>]

On Thu, Oct 02, 2008 at 02:56:55PM +0300, Avi Kivity wrote:
> Han, Weidong wrote:
>> From 61028d958dc7c57ee02de32ea89b025dccb9650d Mon Sep 17 00:00:00 2001
>> From: Weidong Han <weidong.han@intel.com>
>> Date: Thu, 25 Sep 2008 23:32:02 +0800
>> Subject: [PATCH] Map mmio pages into VT-d page table
>>
>> Assigned device could DMA to mmio pages, so also need to map mmio pages
>> into VT-d page table.
>>
>>   
>
> Well, Muli says at least on one machine this allows on guest to kill
> the host.  What are we doing with this?
>
> If it's a hardware bug which is planned to be fixed (or is already
> fixed), great, but I need to know.

Unfortunately I don't have access to the machine any more. We did
spend some time perusing the PCIe spec on this point, and although it
is pretty vague, the bottom line appears to be that peer-to-peer
traffic (device-to-device traffic) is allowed. I'm fine with applying
the patch.

Cheers,
Muli
-- 
The First Workshop on I/O Virtualization (WIOV '08)
Dec 2008, San Diego, CA, http://www.usenix.org/wiov08/
                      xxx
SYSTOR 2009---The Israeli Experimental Systems Conference
http://www.haifa.il.ibm.com/conferences/systor2009/