On 02/01/2016 04:02 PM, Wei Xu wrote:
[...] >> >>> + return NULL; >>> + } >>> + >>> + chain->proto = proto; >>> + chain->do_receive = virtio_net_rsc_receive4; >>> + >>> + QTAILQ_INIT(&chain->buffers); >>> + QTAILQ_INSERT_TAIL(&n->rsc_chains, chain, next); >>> + return chain; >>> +} >> Better to split the chain initialization from lookup. And we can >> initialize ipv4 chain during initialization. > > Since the allocation happens really seldom, is it ok to keep the > mechanism to make the logic clean? Ok for now.