On Thu, 2013-04-04 at 15:31 +0200, Daniel Vetter wrote:
I'm a bit confused about the different classes you're talking about. Since the ticket queue is currently a global counter there's only one class of ww_mutexes.
Right, so that's not something that's going to fly.. we need to support multiple users, including nesting of those. Again, you're adding something to the generic kernel infrastructure, it had better be usable :-)
I guess we could change that once a second user shows up
No, we fix that before it all goes in. I would _so_ hate to find out it cannot be 'fixed' and be stuck with a half-arsed sync primitive in the core kernel that's only every usable by the one existent user.
So for now, forget all about TTM, DMA-BUF and other such coolness except to verify that whatever we end up with does indeed work for the case you need it for ;-)