um: Error handling fixes in vector drivers

With the addition of bess support which uses connection
oriented SEQPACKET sockets the vector routines can now
encounter a "remote end closed the connection" scenario.

This adds handling code to detect it in the TX path and
the legacy RX path. There is no way to detect it in the
vector RX path because that can legitimately return 0
even if the remote end has not closed the connection. As
a result the detection is delayed until the first TX
event after the close.

Signed-off-by: Anton Ivanov <anton.ivanov@cambridgegreys.com>
Signed-off-by: Richard Weinberger <richard@nod.at>

3 files changed, 35 insertions, 7 deletions

diff --git a/arch/um/drivers/vector_kern.c b/arch/um/drivers/vector_kern.c
index 45ec9bc2430f..af27d5c41776 100644
--- a/arch/um/drivers/vector_kern.c
+++ b/arch/um/drivers/vector_kern.c
@@ -76,6 +76,7 @@ static void vector_eth_configure(int n, struct arglist *def);
 #define DEFAULT_VECTOR_SIZE 64
 #define TX_SMALL_PACKET 128
 #define MAX_IOV_SIZE (MAX_SKB_FRAGS + 1)
+#define MAX_ITERATIONS 64
 
 static const struct {
 	const char string[ETH_GSTRING_LEN];
@@ -418,6 +419,7 @@ static int vector_send(struct vector_queue *qi)
 					if (net_ratelimit())
 						netdev_err(vp->dev, "sendmmsg err=%i\n",
 							result);
+					vp->in_error = true;
 					result = send_len;
 				}
 				if (result > 0) {
@@ -845,6 +847,10 @@ static int vector_legacy_rx(struct vector_private *vp)
 	}
 
 	pkt_len = uml_vector_recvmsg(vp->fds->rx_fd, &hdr, 0);
+	if (pkt_len < 0) {
+		vp->in_error = true;
+		return pkt_len;
+	}
 
 	if (skb != NULL) {
 		if (pkt_len > vp->header_size) {
@@ -891,12 +897,16 @@ static int writev_tx(struct vector_private *vp, struct sk_buff *skb)
 
 	if (iov_count < 1)
 		goto drop;
+
 	pkt_len = uml_vector_writev(
 		vp->fds->tx_fd,
 		(struct iovec *) &iov,
 		iov_count
 	);
 
+	if (pkt_len < 0)
+		goto drop;
+
 	netif_trans_update(vp->dev);
 	netif_wake_queue(vp->dev);
 
@@ -911,6 +921,8 @@ static int writev_tx(struct vector_private *vp, struct sk_buff *skb)
 drop:
 	vp->dev->stats.tx_dropped++;
 	consume_skb(skb);
+	if (pkt_len < 0)
+		vp->in_error = true;
 	return pkt_len;
 }
 
@@ -939,6 +951,9 @@ static int vector_mmsg_rx(struct vector_private *vp)
 	packet_count = uml_vector_recvmmsg(
 		vp->fds->rx_fd, qi->mmsg_vector, qi->max_depth, 0);
 
+	if (packet_count < 0)
+		vp->in_error = true;
+
 	if (packet_count <= 0)
 		return packet_count;
 
@@ -1008,15 +1023,18 @@ static int vector_mmsg_rx(struct vector_private *vp)
 static void vector_rx(struct vector_private *vp)
 {
 	int err;
+	int iter = 0;
 
 	if ((vp->options & VECTOR_RX) > 0)
-		while ((err = vector_mmsg_rx(vp)) > 0)
-			;
+		while (((err = vector_mmsg_rx(vp)) > 0) && (iter < MAX_ITERATIONS))
+			iter++;
 	else
-		while ((err = vector_legacy_rx(vp)) > 0)
-			;
+		while (((err = vector_legacy_rx(vp)) > 0) && (iter < MAX_ITERATIONS))
+			iter++;
 	if ((err != 0) && net_ratelimit())
 		netdev_err(vp->dev, "vector_rx: error(%d)\n", err);
+	if (iter == MAX_ITERATIONS)
+		netdev_err(vp->dev, "vector_rx: device stuck, remote end may have closed the connection\n");
 }
 
 static int vector_net_start_xmit(struct sk_buff *skb, struct net_device *dev)
@@ -1024,6 +1042,13 @@ static int vector_net_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	struct vector_private *vp = netdev_priv(dev);
 	int queue_depth = 0;
 
+	if (vp->in_error) {
+		deactivate_fd(vp->fds->rx_fd, vp->rx_irq);
+		if ((vp->fds->rx_fd != vp->fds->tx_fd) && (vp->tx_irq != 0))
+			deactivate_fd(vp->fds->tx_fd, vp->tx_irq);
+		return NETDEV_TX_BUSY;
+	}
+
 	if ((vp->options & VECTOR_TX) == 0) {
 		writev_tx(vp, skb);
 		return NETDEV_TX_OK;
@@ -1134,6 +1159,7 @@ static int vector_net_close(struct net_device *dev)
 	vp->fds = NULL;
 	spin_lock_irqsave(&vp->lock, flags);
 	vp->opened = false;
+	vp->in_error = false;
 	spin_unlock_irqrestore(&vp->lock, flags);
 	return 0;
 }
@@ -1501,7 +1527,8 @@ static void vector_eth_configure(
 		.transport_data		= NULL,
 		.in_write_poll		= false,
 		.coalesce		= 2,
-		.req_size		= get_req_size(def)
+		.req_size		= get_req_size(def),
+		.in_error		= false
 		});
 
 	dev->features = dev->hw_features = (NETIF_F_SG | NETIF_F_FRAGLIST);
diff --git a/arch/um/drivers/vector_kern.h b/arch/um/drivers/vector_kern.h
index 0b0a767b9076..684d2e60ad86 100644
--- a/arch/um/drivers/vector_kern.h
+++ b/arch/um/drivers/vector_kern.h
@@ -116,6 +116,7 @@ struct vector_private {
 	bool rexmit_scheduled;
 	bool opened;
 	bool in_write_poll;
+	bool in_error;
 
 	/* ethtool stats */
 
diff --git a/arch/um/drivers/vector_user.c b/arch/um/drivers/vector_user.c
index 4e068beb9f66..f92e05ad145f 100644
--- a/arch/um/drivers/vector_user.c
+++ b/arch/um/drivers/vector_user.c
@@ -618,7 +618,7 @@ int uml_vector_writev(int fd, void *hdr, int iovcount)
 	int n;
 
 	CATCH_EINTR(n = writev(fd, (struct iovec *) hdr,  iovcount));
-	if ((n < 0) && (errno == EAGAIN))
+	if ((n < 0) && ((errno == EAGAIN) || (errno == ENOBUFS)))
 		return 0;
 	if (n >= 0)
 		return n;
@@ -635,7 +635,7 @@ int uml_vector_sendmmsg(
 	int n;
 
 	CATCH_EINTR(n = sendmmsg(fd, (struct mmsghdr *) msgvec, vlen, flags));
-	if ((n < 0) && (errno == EAGAIN))
+	if ((n < 0) && ((errno == EAGAIN) || (errno == ENOBUFS)))
 		return 0;
 	if (n >= 0)
 		return n;