Messages in this thread | | | Date | Wed, 22 Sep 2010 20:42:49 -0700 (PDT) | From | Sage Weil <> | Subject | Re: [PATCH] fs/locks.c: prepare for BKL removal |
| |
On Tue, 21 Sep 2010, Arnd Bergmann wrote:
> On Tuesday 21 September 2010 18:12:07 Sage Weil wrote: > > On Mon, 20 Sep 2010, Stephen Rothwell wrote: > > > > > On Sun, 19 Sep 2010 20:59:01 -0700 (PDT) Sage Weil <sage@newdream.net> wrote: > > > > > > > > I suspect the easiest thing is to leave Ceph out of this stage of your > > > > series, I'll switch lock_kernel() to lock_flocks() once that exists > > > > upstream. Unless there is a better way? > > > > > > Maybe someone could write a trivial implementation of lock_flocks() (i.e. > > > one that does not make any changes to behaviour) and ask Linus to take it > > > now in preparation for the next merge window (he has done that before). > > > That way, more of this could be put into individual other trees and avoid > > > more conflicts ... > > > > This sounds like the easiest solution to me. Something as simple as > > > > #define lock_flocks lock_kernel > > #define unlock_flocks unlock_kernel > > > > in fs.h? > > Sounds fine to me. I don't think it's necessary but if you prefer to do > it, you can have my Ack.
Okay, the lock/unlock_flocks() stubs on in Linus' tree now, and the Ceph for-next branch is rebased and updated to fix the memory allocations and switch to the new interface.
Unfortunately you still need to #include smp_lock.h for now since the stubs are just #defines, so we'll need to remember to clean that up later.
sage
| |