On Mon, Jul 1, 2013 at 9:20 PM, Anthony Liguori <anth...@codemonkey.ws> wrote: > Paolo Bonzini <pbonz...@redhat.com> writes: > >> Il 01/07/2013 15:00, Anthony Liguori ha scritto: >>>> I >>>> > cannot find the commit exactly, but I think mst added specific code for >>>> > that. >>> Right, I'm not questioning whether these functions have strong enough >>> semantics in their implementation, but asking what their contract should >>> be. >>> >>> Either we should document that these functions have atomic semantics or >>> we should introduce another variant that guarantee atomic access. >>> >>> I think the later makes more sense since the majority of users probably >>> don't need atomic semantics. >> >> I think many of these loads and stores do, actually; perhaps most. It >> also matches what hardware does. > > Hrm, I'm not sure if that's true. PCI has an explicit LOCK# bit to > enable exclusive access so my assumption would be that it doesn't by > default. > Is it a huge but another topic -- implement atomic on device register other than using? For virtio, using fence around but outside the st[u][wlb]_phys when needed?
Regards, Pingfan > But either way, we should document the semantics. > > Regards, > > Anthony Liguori > >> >> Paolo >
