On Mon, Sep 30, 2013 at 09:26:21AM -0700, Linus Torvalds wrote: > On Mon, Sep 30, 2013 at 1:52 AM, Andrea Arcangeli <aarca...@redhat.com> wrote: > > > > Sorry having to break the party but the sleepable locks for anon_vma > > and i_mmap_mutex are now requirement for the "pageable RDMA" effort > > recently achieved upstream by mellanox with the MMU notifier. > > I'll happily break that.
Unless a solution is found that could allow to invalidate secondary MMUs with a spinlock/rwlock for anon_vma->lock/i_mmap_lock, would it be acceptable to switch between spinlock/rwlock mutex/rwsem through a config option? option CONFIG_SLEEPABLE_RMAP, implicitly selected by CONFIG_SLEEPABLE_MMU_NOTIFIER, in turn selected by the RDMA and nvidia drivers if they're built (ideally nvidia drivers will figure out how to avoid scheduling). I mean it only requires a wrapping header file, aside from the header file it wouldn't be a much bigger patch than the one posted already. It would be much easier to switch between spinning and sleeping locks to keep benchmarking different scenarios too. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
