On Mon, Aug 19, 2013 at 07:04:19PM +0800, Ming Lei wrote: > This patch kills atomic_inc/atomic_dec operations on > urb->use_count in URB submit/complete path. Any reason only this patch was "RFC"? And you didn't kill them all, please look closer, this should have no affect on "speed", did you measure it? > The urb->use_count is only used for unlinking URB, and it isn't > necessary defined as atomic counter, so the variable is renamed > as urb->use_flag for this purpose, then reading/writing the > flag is still kept as atomic but ARCH's atomic operations(atomic_inc/ > atomic_dec) are saved. That sentence made no sense to me at all :( And how is atomic_set() any faster/slower than atomic_inc/dec? The same memory barriers kick in, right? confused, greg k-h -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html