On 10/17/2018 04:14 PM, Paolo Bonzini wrote:
On 16/10/2018 18:40, Emilio G. Cota wrote:
+#define SMP_CACHE_BYTES 64
+#define ____cacheline_aligned_in_smp \
+ __attribute__((__aligned__(SMP_CACHE_BYTES)))
You could use QEMU_ALIGNED() here.
Yes, you are right.
+
+#define WRITE_ONCE(ptr, val) \
+ (*((volatile typeof(ptr) *)(&(ptr))) = (val))
+#define READ_ONCE(ptr) (*((volatile typeof(ptr) *)(&(ptr))))
Why not atomic_read/set, like in the rest of the QEMU code base?
Or even atomic_rcu_read/atomic_rcu_set, which includes the necessary
barriers.
Okay, will fix it, thank you and Emilio for pointing the
issue out.
Also, please do not use __ identifiers in QEMU code.
____cacheline_aligned_in_smp can become just QEMU_ALIGNED(SMP_CACHE_BYTES).
Sure, will keep that in my mind. :)
