On 05/08/2012 07:48 PM, Christoph Hellwig wrote:
> On Tue, May 08, 2012 at 05:34:25PM +0800, Liu Yuan wrote:
>> + if (!uatomic_add_return(&zk_notify_blocked, 0)
>> + && uatomic_add_return(&nr_zk_levents, 0)) {
>
> Shouldn't these use uatomic_read?
>
>> + nr_levents = uatomic_sub_return(&nr_zk_levents, 1) + 1;
>
> The combination of a read and an atomic_sub later looks racy to me, what
> prevents nr_zk_levents from changing between the two calls?
>
>> + if (uatomic_add_return(&nr_zk_levents, 0) || rc == ZOK) {
>
> uatomic_read again?
>
>> + if (uatomic_add_return(&zk_notify_blocked, 0))
>
> here as well.
>
Ah, yea, thanks for reminding, I wasn't aware of this handful helper.
Thanks,
Yuan
--
sheepdog mailing list
[email protected]
http://lists.wpkg.org/mailman/listinfo/sheepdog
