[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: [Xen-devel] [PATCH v12] introduce XENMEM_reserved_device_memory_map
>>> On 22.07.15 at 14:17, <julien.grall@xxxxxxxxxx> wrote: > On 22/07/15 11:42, Jan Beulich wrote: > >> +struct xen_reserved_device_memory_map { >> +#define XENMEM_RDM_ALL 1 /* Request all regions (ignore dev union). */ >> + /* IN */ >> + uint32_t flags; >> + union { >> + struct physdev_pci_device pci; >> + } dev; > > The size of physdev_pci_device is 4 bytes. So the maximum size of the > union would be 4 bytes as we can't change the layout of the hypercall > (IIRC, memory hypercall should be stable). > > Let's say we want to add support for device describe in the device tree. > The only way to find a device is a path, so it means that we at least > need a size field and a XEN_GUEST_HANDLE for the path. > > That won't fit in the union and we won't even be able to specify a > XEN_GUEST_HANDLE to point to another structure. > > Would it be possible to either make sure that the union is at least the > size of a XEN_GUEST_HANDLE and/or move the union at the end of the > structure? Very good point! Yes, I'll move it to the end. Whether to also add a dummy handle right away I'm not sure. While not doing so now will require a bit more care when extending the structure, it keeps the interface uncluttered (and we don't really know at this point whether we'd want a handle or a bigger union member or both or yet something else). Jan _______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxx http://lists.xen.org/xen-devel
|
Lists.xenproject.org is hosted with RackSpace, monitoring our |