firewire: net: fix unicast reception RCODE in failure paths
authorStefan Richter <stefanr@s5r6.in-berlin.de>
Mon, 16 Aug 2010 21:45:54 +0000 (23:45 +0200)
committerStefan Richter <stefanr@s5r6.in-berlin.de>
Thu, 19 Aug 2010 18:28:25 +0000 (20:28 +0200)
The incoming request hander fwnet_receive_packet() expects subsequent
datagram handling code to return non-zero on errors.  However, almost
none of the failure paths did so.  Fix them all.

(This error reporting is used to send and RCODE_CONFLICT_ERROR to the
sender node in such failure cases.  Two modes of failure exist:  Out of
memory, or firewire-net is unaware of any peer node to which a fragment
or an ARP packet belongs.  However, it is unclear whether a sender can
actually make use of such information.  A Linux peer apparently can't.
Maybe it should all be simplified to void functions.)

Reported-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
drivers/firewire/net.c

index da17d40..33f8421 100644 (file)
@@ -579,7 +579,7 @@ static int fwnet_finish_incoming_packet(struct net_device *net,
                if (!peer) {
                        fw_notify("No peer for ARP packet from %016llx\n",
                                  (unsigned long long)peer_guid);
                if (!peer) {
                        fw_notify("No peer for ARP packet from %016llx\n",
                                  (unsigned long long)peer_guid);
-                       goto failed_proto;
+                       goto no_peer;
                }
 
                /*
                }
 
                /*
@@ -656,7 +656,7 @@ static int fwnet_finish_incoming_packet(struct net_device *net,
 
        return 0;
 
 
        return 0;
 
failed_proto:
no_peer:
        net->stats.rx_errors++;
        net->stats.rx_dropped++;
 
        net->stats.rx_errors++;
        net->stats.rx_dropped++;
 
@@ -664,7 +664,7 @@ static int fwnet_finish_incoming_packet(struct net_device *net,
        if (netif_queue_stopped(net))
                netif_wake_queue(net);
 
        if (netif_queue_stopped(net))
                netif_wake_queue(net);
 
-       return 0;
+       return -ENOENT;
 }
 
 static int fwnet_incoming_packet(struct fwnet_device *dev, __be32 *buf, int len,
 }
 
 static int fwnet_incoming_packet(struct fwnet_device *dev, __be32 *buf, int len,
@@ -701,7 +701,7 @@ static int fwnet_incoming_packet(struct fwnet_device *dev, __be32 *buf, int len,
                        fw_error("out of memory\n");
                        net->stats.rx_dropped++;
 
                        fw_error("out of memory\n");
                        net->stats.rx_dropped++;
 
-                       return -1;
+                       return -ENOMEM;
                }
                skb_reserve(skb, (net->hard_header_len + 15) & ~15);
                memcpy(skb_put(skb, len), buf, len);
                }
                skb_reserve(skb, (net->hard_header_len + 15) & ~15);
                memcpy(skb_put(skb, len), buf, len);
@@ -726,8 +726,10 @@ static int fwnet_incoming_packet(struct fwnet_device *dev, __be32 *buf, int len,
        spin_lock_irqsave(&dev->lock, flags);
 
        peer = fwnet_peer_find_by_node_id(dev, source_node_id, generation);
        spin_lock_irqsave(&dev->lock, flags);
 
        peer = fwnet_peer_find_by_node_id(dev, source_node_id, generation);
-       if (!peer)
-               goto bad_proto;
+       if (!peer) {
+               retval = -ENOENT;
+               goto fail;
+       }
 
        pd = fwnet_pd_find(peer, datagram_label);
        if (pd == NULL) {
 
        pd = fwnet_pd_find(peer, datagram_label);
        if (pd == NULL) {
@@ -741,7 +743,7 @@ static int fwnet_incoming_packet(struct fwnet_device *dev, __be32 *buf, int len,
                                  dg_size, buf, fg_off, len);
                if (pd == NULL) {
                        retval = -ENOMEM;
                                  dg_size, buf, fg_off, len);
                if (pd == NULL) {
                        retval = -ENOMEM;
-                       goto bad_proto;
+                       goto fail;
                }
                peer->pdg_size++;
        } else {
                }
                peer->pdg_size++;
        } else {
@@ -755,9 +757,9 @@ static int fwnet_incoming_packet(struct fwnet_device *dev, __be32 *buf, int len,
                        pd = fwnet_pd_new(net, peer, datagram_label,
                                          dg_size, buf, fg_off, len);
                        if (pd == NULL) {
                        pd = fwnet_pd_new(net, peer, datagram_label,
                                          dg_size, buf, fg_off, len);
                        if (pd == NULL) {
-                               retval = -ENOMEM;
                                peer->pdg_size--;
                                peer->pdg_size--;
-                               goto bad_proto;
+                               retval = -ENOMEM;
+                               goto fail;
                        }
                } else {
                        if (!fwnet_pd_update(peer, pd, buf, fg_off, len)) {
                        }
                } else {
                        if (!fwnet_pd_update(peer, pd, buf, fg_off, len)) {
@@ -768,7 +770,8 @@ static int fwnet_incoming_packet(struct fwnet_device *dev, __be32 *buf, int len,
                                 */
                                fwnet_pd_delete(pd);
                                peer->pdg_size--;
                                 */
                                fwnet_pd_delete(pd);
                                peer->pdg_size--;
-                               goto bad_proto;
+                               retval = -ENOMEM;
+                               goto fail;
                        }
                }
        } /* new datagram or add to existing one */
                        }
                }
        } /* new datagram or add to existing one */
@@ -794,14 +797,13 @@ static int fwnet_incoming_packet(struct fwnet_device *dev, __be32 *buf, int len,
        spin_unlock_irqrestore(&dev->lock, flags);
 
        return 0;
        spin_unlock_irqrestore(&dev->lock, flags);
 
        return 0;
-
- bad_proto:
+ fail:
        spin_unlock_irqrestore(&dev->lock, flags);
 
        if (netif_queue_stopped(net))
                netif_wake_queue(net);
 
        spin_unlock_irqrestore(&dev->lock, flags);
 
        if (netif_queue_stopped(net))
                netif_wake_queue(net);
 
-       return 0;
+       return retval;
 }
 
 static void fwnet_receive_packet(struct fw_card *card, struct fw_request *r,
 }
 
 static void fwnet_receive_packet(struct fw_card *card, struct fw_request *r,