[PATCH 1/2] appletalk: Fix atalk_proc_init return path

[PATCH 1/2] appletalk: Fix atalk_proc_init return path

[Vincent Duvert]

Add a missing return statement to atalk_proc_init so it doesn't return
-ENOMEM when successful. This allows the appletalk module to load
properly.

Signed-off-by: Vincent Duvert <vincent.ldev@duvert.net>
---
 net/appletalk/atalk_proc.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/net/appletalk/atalk_proc.c b/net/appletalk/atalk_proc.c
index 550c6ca007cc..9c1241292d1d 100644
--- a/net/appletalk/atalk_proc.c
+++ b/net/appletalk/atalk_proc.c
@@ -229,6 +229,8 @@ int __init atalk_proc_init(void)
 				     sizeof(struct aarp_iter_state), NULL))
 		goto out;
 
+	return 0;
+
 out:
 	remove_proc_subtree("atalk", init_net.proc_net);
 	return -ENOMEM;
-- 
2.20.1

[PATCH 2/2] appletalk: Improve handling of broadcast packets

[Vincent Duvert]

When a broadcast AppleTalk packet is received, prefer queuing it on the
socket whose address matches the address of the interface that received
the packet (and is listening on the correct port). Userspace
applications that handle such packets will usually send a response on
the same socket that received the packet; this fix allows the response
to be sent on the correct interface.

If a socket matching the interface's address is not found, an arbitrary
socket listening on the correct port will be used, if any. This matches
the implementation's previous behavior.

Fixes atalkd's responses to network information requests when multiple
network interfaces are configured to use AppleTalk.

Signed-off-by: Vincent Duvert <vincent.ldev@duvert.net>
---
 net/appletalk/ddp.c | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)

diff --git a/net/appletalk/ddp.c b/net/appletalk/ddp.c
index 15787e8c0629..9d19cd03076f 100644
--- a/net/appletalk/ddp.c
+++ b/net/appletalk/ddp.c
@@ -89,6 +89,7 @@ static struct sock *atalk_search_socket(struct sockaddr_at *to,
 					struct atalk_iface *atif)
 {
 	struct sock *s;
+	struct sock *def_socket = NULL;
 
 	read_lock_bh(&atalk_sockets_lock);
 	sk_for_each(s, &atalk_sockets) {
@@ -98,8 +99,20 @@ static struct sock *atalk_search_socket(struct sockaddr_at *to,
 			continue;
 
 		if (to->sat_addr.s_net == ATADDR_ANYNET &&
-		    to->sat_addr.s_node == ATADDR_BCAST)
-			goto found;
+		    to->sat_addr.s_node == ATADDR_BCAST) {
+			if (atif->address.s_node == at->src_node &&
+			    atif->address.s_net == at->src_net) {
+				/* This socket's address matches the address of the interface
+				 * that received the packet -- use it
+				 */
+				goto found;
+			}
+
+			/* Continue searching for a socket matching the interface address,
+			 * but use this socket by default if no other one is found
+			 */
+			def_socket = s;
+		}
 
 		if (to->sat_addr.s_net == at->src_net &&
 		    (to->sat_addr.s_node == at->src_node ||
@@ -116,7 +129,7 @@ static struct sock *atalk_search_socket(struct sockaddr_at *to,
 			goto found;
 		}
 	}
-	s = NULL;
+	s = def_socket;
 found:
 	read_unlock_bh(&atalk_sockets_lock);
 	return s;
-- 
2.20.1

Re: [PATCH 1/2] appletalk: Fix atalk_proc_init return path

[David Miller]

From: Vincent Duvert <vincent.ldev@duvert.net>
Date: Wed, 22 Jul 2020 13:37:51 +0200

> Add a missing return statement to atalk_proc_init so it doesn't return
> -ENOMEM when successful. This allows the appletalk module to load
> properly.
> 
> Signed-off-by: Vincent Duvert <vincent.ldev@duvert.net>

Need an appropriate "Fixes: " tag here, if the module hasn't loaded
it means nobody has been using it at all.

Re: [PATCH 2/2] appletalk: Improve handling of broadcast packets

[David Miller]

From: Vincent Duvert <vincent.ldev@duvert.net>
Date: Wed, 22 Jul 2020 13:37:52 +0200

> @@ -89,6 +89,7 @@ static struct sock *atalk_search_socket(struct sockaddr_at *to,
>  					struct atalk_iface *atif)
>  {
>  	struct sock *s;
> +	struct sock *def_socket = NULL;
>  
>  	read_lock_bh(&atalk_sockets_lock);
>  	sk_for_each(s, &atalk_sockets) {

Please use reverse christmas tree ordering for local variables.

Also, please post the next revision of this patch series with
a proper "[PATCH net 0/N]" header posting, explaining what the
patch series is doing, how it is doing it, and why it is
doing it this way.

Your Subject lines should all also indicate the proper target GIT tree
your changes are for.  This is indicated in the "[]" bracket area,
as either 'net' or 'net-next', f.e. "[PATCH net 1/2] ..."

Thank you.