On Fri, Sep 5, 2014 at 11:53 AM, David Miller [off-list ref] wrote:

From: Sabrina Dubroca <sd@queasysnail.net>
Date: Tue, 2 Sep 2014 10:29:29 +0200

quoted

Calling setsockopt with IPV6_JOIN_ANYCAST or IPV6_LEAVE_ANYCAST
triggers the assertion in addrconf_join_solict()/addrconf_leave_solict()

ipv6_sock_ac_join(), ipv6_sock_ac_drop(), ipv6_sock_ac_close() need to
take RTNL before calling ipv6_dev_ac_inc/dec. Same thing with
ipv6_sock_mc_join(), ipv6_sock_mc_drop(), ipv6_sock_mc_close() before
calling ipv6_dev_mc_inc/dec.

This patch moves ASSERT_RTNL() up a level in the call stack.

Signed-off-by: Cong Wang <redacted>
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Reported-by: Tommi Rantala <redacted>

Applied and queued up for -stable, thanks.

I believe you applied a wrong version, at least the following
is not correct:

+       if (!dev)
+               return -ENODEV;

Sabrina took that from my draft patch, but they all don't
realize this is wrong.

(I did provide a correct version which is just ignored by you.)