Re: [PATCH v5 10/13] xen/pvcalls: implement recvmsg

[Boris Ostrovsky]

On 10/19/2017 09:38 PM, Stefano Stabellini wrote:
> On Tue, 17 Oct 2017, Boris Ostrovsky wrote:
>>> +
>>> +int pvcalls_front_recvmsg(struct socket *sock, struct msghdr *msg, size_t len,
>>> +		     int flags)
>>> +{
>>> +	struct pvcalls_bedata *bedata;
>>> +	int ret;
>>> +	struct sock_mapping *map;
>>> +
>>> +	if (flags & (MSG_CMSG_CLOEXEC|MSG_ERRQUEUE|MSG_OOB|MSG_TRUNC))
>>> +		return -EOPNOTSUPP;
>>> +
>>> +	pvcalls_enter();
>>> +	if (!pvcalls_front_dev) {
>>> +		pvcalls_exit();
>>> +		return -ENOTCONN;
>>> +	}
>>> +	bedata = dev_get_drvdata(&pvcalls_front_dev->dev);
>>> +
>>> +	map = (struct sock_mapping *) sock->sk->sk_send_head;
>>> +	if (!map) {
>>> +		pvcalls_exit();
>>> +		return -ENOTSOCK;
>>> +	}
>>> +
>>> +	mutex_lock(&map->active.in_mutex);
>>> +	if (len > XEN_FLEX_RING_SIZE(PVCALLS_RING_ORDER))
>>> +		len = XEN_FLEX_RING_SIZE(PVCALLS_RING_ORDER);
>>> +
>>> +	while (!(flags & MSG_DONTWAIT) && !pvcalls_front_read_todo(map)) {
>>> +		wait_event_interruptible(map->active.inflight_conn_req,
>>> +					 pvcalls_front_read_todo(map));
>>> +	}
>>> +	ret = __read_ring(map->active.ring, &map->active.data,
>>> +			  &msg->msg_iter, len, flags);
>>> +
>>> +	if (ret > 0)
>>> +		notify_remote_via_irq(map->active.irq);
>>> +	if (ret == 0)
>>> +		ret = -EAGAIN;
>> Why not 0? The manpage says:
>>
>>        EAGAIN or EWOULDBLOCK
>>               The  socket  is  marked nonblocking and the receive
>> operation would block, or a receive timeout
>>               had been set and the timeout expired before data was
>> received.  POSIX.1 allows either error  to
>>               be  returned  for  this case, and does not require these
>> constants to have the same value, so a
>>               portable application should check for both possibilities.
>>
>>
>> I don't think either of these conditions is true here.
>>
>> (Again, should have noticed this earlier, sorry)
> In case the socket is MSG_DONTWAIT, then we should return -EAGAIN here.
> However, it is true that if the socket is not MSG_DONTWAIT, then
> returning 0 would make more sense.
>
> So I'll do:
>
> if (ret == 0)
>     ret = (flags & MSG_DONTWAIT) ? -EAGAIN : 0;

Sure. With that

Reviewed-by: Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx>