Messages in this thread |  | | | Subject | Re: Q: smp.c && barriers (Was: [PATCH 1/4] generic-smp: remove single ipi fallback for smp_call_function_many()) | | From | Peter Zijlstra <> | | Date | Tue, 17 Feb 2009 13:40:55 +0100 |
| |
On Tue, 2009-02-17 at 10:29 +0100, Peter Zijlstra wrote: > > It would be nice to have a comment which explains how can we miss the > > first addition without read_barrier_depends(). And why only on alpha. > > Paul, care to once again enlighten us? The best I can remember is that > alpha has split caches, and the rmb is needed for them to become > coherent -- no other arch is crazy in exactly that way. > > But note that read_barrier_depends() is not quite a NOP for !alpha, it > does that ACCESS_ONCE() thing, which very much makes a difference, even > on x86.
I've been saying crazy stuff, read_barrier_depends() is a NOP, I got confused with rcu_dereference(). My bad.
|  |