On 11/13/2010 02:48 AM, Eric Dumazet wrote:
> Le samedi 13 novembre 2010 Ã 18:17 +0800, AmÃrico Wang a Ãcrit :
>> On Wed, Nov 03, 2010 at 10:59:47AM -0400, Jeremy Fitzhardinge wrote:
>>
>>> + union {
>>> + struct __raw_tickets tickets;
>>> + __ticketpair_t slock;
>>> + } tmp, new;
>>> + int ret;
>>> +
>>> + tmp.tickets = ACCESS_ONCE(lock->tickets);
>>> + if (tmp.tickets.head != tmp.tickets.tail)
>>> + return 0;
>>> +
>>> + new.slock = tmp.slock + (1 << TICKET_SHIFT);
>>> +
>>> + ret = cmpxchg(&lock->ticketpair, tmp.slock, new.slock) == tmp.slock;
>>> + barrier(); /* just make nothing creeps before lock is claimed */
>> This one should be smp_wmb(), right? No CONFIG_X86_OOSTORE protected.
> cmpxchg() is a full memory barrier, no need for smp_wmb() or barrier()
Agreed.
J
_______________________________________________
Virtualization mailing list
Virtualization@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linux-foundation.org/mailman/listinfo/virtualization
