On Wed, Aug 17, 2022 at 02:48:22PM +0100, Will Deacon wrote: > On Mon, Aug 08, 2022 at 08:45:48AM -0400, Michael S. Tsirkin wrote: > > > > Also yes, I think it's a good idea to change crosvm anyway. While the > > > > work around I describe might make sense upstream I don't think it's a > > > > reasonable thing to do in stable kernels. > > > > I think I'll prepare a patch documenting the legal vhost features > > > > as a 1st step even though crosvm is rust so it's not importing > > > > the header directly, right? > > > > > > Documentation is a good idea regardless, so thanks for that. Even though > > > crosvm has its own bindings for the vhost ioctl()s, the documentation > > > can be reference or duplicated once it's available in the kernel headers. > > > > > So for crosvm change, I will post the documentation change and > > you guys can discuss? > > FYI, the crosvm patch is merged here: > > https://github.com/google/crosvm/commit/4e7d00be2e135b0a2d964320ea4276e5d896f426 > > Will Great thanks a lot! I'm on vacation next week but will work on it afterwards. -- MST
