PF_NETLINK, SOCK_DGRAM, 2 => EPROTONOSUPPORT

PF_NETLINK, SOCK_DGRAM, 2 => EPROTONOSUPPORT

[Thomas Voegtle]

Hello,


This code snippet:
#include <sys/socket.h>
#include <linux/netlink.h>
int main() {
         socket(PF_NETLINK, SOCK_DGRAM, NETLINK_USERSOCK);
         return 0;
}

ends up with:

socket(PF_NETLINK, SOCK_DGRAM, 2) = -1 EPROTONOSUPPORT (Protocol not 
supported)

since 2.6.33.

I bisected it down to:
974c37e9d88c3e5a3e56eb98cb9c84232eb2bdcb is the first bad commit
commit 974c37e9d88c3e5a3e56eb98cb9c84232eb2bdcb
Author: Alexey Dobriyan <adobriyan@gmail.com>
Date:   Sat Jan 30 10:05:05 2010 +0000

     netlink: fix for too early rmmod


This is strange, because the code snippet even fails on a monolithic 
kernel.
When I revert this commit on top of 2.6.36-rc3, it works.

Was this intended, that this code snippet doesn't work this way anymore or
is this just a bug?


       Thomas

Re: PF_NETLINK, SOCK_DGRAM, 2 => EPROTONOSUPPORT

[David Miller]

From: Thomas Voegtle <tv@lio96.de>
Date: Mon, 30 Aug 2010 12:02:51 +0200 (CEST)

> Was this intended, that this code snippet doesn't work this way
> anymore or is this just a bug?

It's an unintended regression.

But nobody noticed it, because almost no-one uses NETLINK_USER.

This might fix it, give it a test:

netlink: Make NETLINK_USERSOCK work again.

Once we started enforcing the a nl_table[] entry exist for
a protocol, NETLINK_USERSOCK stopped working.  Add a dummy
table entry so that it works again.

Reported-by: Thomas Voegtle <tv@lio96.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index 980fe4a..cd96ed3 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c
@@ -2102,6 +2102,26 @@ static void __net_exit netlink_net_exit(struct net *net)
 #endif
 }
 
+static void __init netlink_add_usersock_entry(void)
+{
+	unsigned long *listeners;
+	int groups = 32;
+
+	listeners = kzalloc(NLGRPSZ(groups) + sizeof(struct listeners_rcu_head),
+			    GFP_KERNEL);
+	if (!listeners)
+		panic("netlink_add_usersock_entry: Cannot allocate listneres\n");
+
+	netlink_table_grab();
+
+	nl_table[NETLINK_USERSOCK].groups = groups;
+	nl_table[NETLINK_USERSOCK].listeners = listeners;
+	nl_table[NETLINK_USERSOCK].module = THIS_MODULE;
+	nl_table[NETLINK_USERSOCK].registered = 1;
+
+	netlink_table_ungrab();
+}
+
 static struct pernet_operations __net_initdata netlink_net_ops = {
 	.init = netlink_net_init,
 	.exit = netlink_net_exit,
@@ -2150,6 +2170,8 @@ static int __init netlink_proto_init(void)
 		hash->rehash_time = jiffies;
 	}
 
+	netlink_add_usersock_entry();
+
 	sock_register(&netlink_family_ops);
 	register_pernet_subsys(&netlink_net_ops);
 	/* The netlink device handler may be needed early. */

Re: PF_NETLINK, SOCK_DGRAM, 2 => EPROTONOSUPPORT

[Thomas Voegtle]

On Mon, 30 Aug 2010, David Miller wrote:

> From: Thomas Voegtle <tv@lio96.de>
> Date: Mon, 30 Aug 2010 12:02:51 +0200 (CEST)
>
>> Was this intended, that this code snippet doesn't work this way
>> anymore or is this just a bug?
>
> It's an unintended regression.
>
> But nobody noticed it, because almost no-one uses NETLINK_USER.
>
> This might fix it, give it a test:


Yes, that fixes it. Tested with 2.6.35.4 and 2.6.36-rc3.

Thank you,

  Thomas

Re: PF_NETLINK, SOCK_DGRAM, 2 => EPROTONOSUPPORT

[David Miller]

From: Thomas Voegtle <tv@lio96.de>
Date: Tue, 31 Aug 2010 12:15:26 +0200 (CEST)

> On Mon, 30 Aug 2010, David Miller wrote:
> 
>> From: Thomas Voegtle <tv@lio96.de>
>> Date: Mon, 30 Aug 2010 12:02:51 +0200 (CEST)
>>
>>> Was this intended, that this code snippet doesn't work this way
>>> anymore or is this just a bug?
>>
>> It's an unintended regression.
>>
>> But nobody noticed it, because almost no-one uses NETLINK_USER.
>>
>> This might fix it, give it a test:
> 
> 
> Yes, that fixes it. Tested with 2.6.35.4 and 2.6.36-rc3.

Thanks for testing.