>>> On 03.08.16 at 15:11, <andrew.coop...@citrix.com> wrote: > On 03/08/16 14:04, Jan Beulich wrote: >> On x86 there's no need for full barriers in loops waiting for some >> memory location to change. Nor do we need full barriers between two >> reads and two writes - SMP ones fully suffice (and I actually think >> they could in fact be dropped, since atomic_*() operations should >> already provide enough ordering). > > Missing a SoB,
Oops. > Which "ones" are you referring to? atomic_*() is only ordered with > respect to the atomic_t used. Oh, right - the one followed by atomic_inc() could be converted to plain barrier() (but not dropped entirely), while the other one follows an atomic_read() only, and hence can't be dropped. I'll remove that part of the description. > Overall, I think the change is correct, so Reviewed-by: Andrew Cooper > <andrew.coop...@citrix.com> Thanks. Jan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org https://lists.xen.org/xen-devel
