RE: [Xen-users] Sharing PCI devices

["Petersson, Mats" <Mats.Petersson@xxxxxxx>]

> This means that I could make my device working in a kind of 
> degraded mode, 
> where it would only be slave on the PCI bus (i.e. no DMA with 
> bus mastering, 
> in either direction, and no interrupts) within a HVM. Is this correct 
> (leaving aside the concurrent access problems) ?

Yes that would, technically, work [assuming the modifications are made
to the HVM portion of Xen and related tools to allow you to pass the
information of which hardware you want to pass to the HVM domain and
"unblock" the related IO ports for the device - currently there is no
support for this, although I think the patches for supporting PCI
passthrough from Intel and Neocleus may help on this - neither of these
patch-sets have been accepted yet]. 

However, you would probably struggle to find much hardware that works in
this way - at the very least you'll need to write your own drivers, as
almost all drivers at the very least use interrupts to know what's going
on.

--
Mats



_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users