On Wed, Jan 23, 2008 at 01:51:23PM +0100, Gerd Hoffmann wrote: > Jumping in here, looks like this could develop into a direction useful > for Xen. > > Background: Xen has a mechanism called "grant tables" for page sharing. > Guest #1 can issue a "grant" for another guest #2, which in turn then > can use that grant to map the page owned by guest #1 into its address > space. This is used by the virtual network/disk drivers, i.e. typically > Domain-0 (which has access to the real hardware) maps pages of other > guests to fill in disk/network data.
This is extremely similar to what XPMEM is providing. > That would render the notifies useless for Xen too. Xen needs to > intercept the actual pte clear and instead of just zapping it use the > hypercall to do the unmap and release the grant. We are tackling that by having our own page table hanging off the structure representing our seg (thing created when we do the equiv of your grant call). > Current implementation uses a new vm_ops operation which is called if > present instead of doing a ptep_get_and_clear_full(). It is in the > XenSource tree only, mainline hasn't this yet due to implementing only > the DomU bits so far. When adding Dom0 support to mainline we'll need > some way to handle it, and I'd like to see the notifies be designed in a > way that Xen can simply use them. Would the callouts Christoph proposed work for you if you maintained your own page table and moved them after the callouts the mmu_notifiers are using. Thanks, Robin ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ kvm-devel mailing list kvm-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/kvm-devel
