On 01/11/2016 15:33, Pranith Kumar wrote: > We can use __atomic_load_n() saving a store and load from the _val. > > Signed-off-by: Pranith Kumar <[email protected]> > --- > include/qemu/atomic.h | 4 +--- > 1 file changed, 1 insertion(+), 3 deletions(-) > > diff --git a/include/qemu/atomic.h b/include/qemu/atomic.h > index 878fa07..6775603 100644 > --- a/include/qemu/atomic.h > +++ b/include/qemu/atomic.h > @@ -144,9 +144,7 @@ > #define atomic_load_acquire(ptr) \ > ({ \ > QEMU_BUILD_BUG_ON(sizeof(*ptr) > sizeof(void *)); \ > - typeof_strip_qual(*ptr) _val; \ > - __atomic_load(ptr, &_val, __ATOMIC_ACQUIRE); \ > - _val; \ > + __atomic_load_n(ptr, __ATOMIC_ACQUIRE); \ > }) > > #define atomic_store_release(ptr, i) do { \ >
Can you do the same for atomic_rcu_read__nocheck (both implementations)? Thanks, Paolo
