On Sun, Aug 28, 2011 at 10:50:49AM +0300, Avi Kivity wrote: > On 08/26/2011 12:43 PM, Michael S. Tsirkin wrote: > >On Thu, Aug 18, 2011 at 08:15:43AM -0700, Avi Kivity wrote: > >> It's correct but insufficient, the filtering code > >> (pci_bridge_filter) needs to be updated to use the memory API. > >> > >> Basically it gets simpler and correcter. > > > >I've been struggling with the following problem: bridges have two memory > >ranges: prefetcheable and non-prefetcheable. > > > >Memory in the device can be behind the prefetcheable and > >non-prefetcheable memory range, but things only work correctly if > >non-prefetcheable memory on the device is put behind a non-prefetcheable > >range. Prefetcheable memory can go anywhere I think. > > > >This didn't work correctly before the memory API change, > >but it was easy to fix ... Now I'm not sure how. > > If it really matters, you can add a prefetchability attribute to > MemoryRegions. Does it though?
Well, its another one of these things that guests *probably* won't in practice use. But I don't see a way to be sure. If the guest puts a prefetcheable memory BAR behind a non-prefetcheable range in the bridge, it won't be able to access that BAR, and it should. Prefetcheable BARs on devices are less common than non-prefetcheable, but they do exist: I have a system with 2 devices: a VGA controller from Matrox and an ethernet card from Mellanox have prefetcheable BARs. I'm not sure how prefetcheability attribute will help. Could you explain pls? > -- > I have a truly marvellous patch that fixes the bug which this > signature is too narrow to contain.
