virtio: rpmsg: make it clear that virtqueue_add_buf() no longer retur…

…ns > 0

We simplified virtqueue_add_buf(), make it clear in the callers.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

drivers/rpmsg/virtio_rpmsg_bus.c

@@ -764,7 +764,7 @@ int rpmsg_send_offchannel_raw(struct rpmsg_channel *rpdev, u32 src, u32 dst,
 
 	/* add message to the remote processor's virtqueue */
 	err = virtqueue_add_buf(vrp->svq, &sg, 1, 0, msg, GFP_KERNEL);
-	if (err < 0) {
+	if (err) {
 		/*
 		 * need to reclaim the buffer here, otherwise it's lost
 		 * (memory won't leak, but rpmsg won't use it again for TX).
@@ -776,8 +776,6 @@ int rpmsg_send_offchannel_raw(struct rpmsg_channel *rpdev, u32 src, u32 dst,
 
 	/* tell the remote processor it has a pending message to read */
 	virtqueue_kick(vrp->svq);
-
-	err = 0;
 out:
 	mutex_unlock(&vrp->tx_lock);
 	return err;
@@ -980,7 +978,7 @@ static int rpmsg_probe(struct virtio_device *vdev)
 
 		err = virtqueue_add_buf(vrp->rvq, &sg, 0, 1, cpu_addr,
 								GFP_KERNEL);
-		WARN_ON(err < 0); /* sanity check; this can't really happen */
+		WARN_ON(err); /* sanity check; this can't really happen */
 	}
 
 	/* suppress "tx-complete" interrupts */