On Fri 03-10-14 19:36:23, Vladimir Davydov wrote: > On Thu, Oct 02, 2014 at 08:07:48AM -0400, Johannes Weiner wrote: [...] > > The barriers are implied in change-return atomics, which is why there > > is an xchg. But it's clear that this needs to be documented. This?: > > With the comments it looks correct to me, but I wonder if we can always > rely on implicit memory barriers issued by atomic ops. Are there any > archs where it doesn't hold? xchg is explcitly mentioned in Documentation/memory-barriers.txt so it is expected to be barrier on all archs. Besides that not all atomic ops imply memory barriers. Only those that "modifies some state in memory and returns information about the state" do. -- Michal Hocko SUSE Labs -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>