Re: [Qemu-devel] [PATCH 02/10] qemu-thread: introduce QemuLockCnt

[Stefan Hajnoczi <stefanha@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 1078 bytes --]

On Wed, Jan 04, 2017 at 02:26:17PM +0100, Paolo Bonzini wrote:
> +/* Decrement a counter, and return locked if it is decremented to zero.
> + * It is impossible for the counter to become nonzero while the mutex
> + * is taken.
> + */
> +bool qemu_lockcnt_dec_and_lock(QemuLockCnt *lockcnt)
> +{
> +    int val = atomic_read(&lockcnt->count);

I think I've figured out the answer to my question why this isn't an
atomic_mb_read():

The following accesses use atomic_cmpxchg() or atomic_fetch_dec().
There is no adverse effect if we read an outdated value here since the
cmpxchg/fetch_dec are sequentially consistent and explicitly require us
to handle the old value.

> +    while (val > 1) {
> +        int old = atomic_cmpxchg(&lockcnt->count, val, val - 1);
> +        if (old != val) {
> +            val = old;
> +            continue;
> +        }
> +
> +        return false;
> +    }
> +
> +    qemu_lockcnt_lock(lockcnt);
> +    if (atomic_fetch_dec(&lockcnt->count) == 1) {
> +        return true;
> +    }
> +
> +    qemu_lockcnt_unlock(lockcnt);
> +    return false;

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 455 bytes --]