net: ipmr, ip6mr: fix vif/tunnel failure race condition
authorNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Tue, 24 Nov 2015 16:09:30 +0000 (17:09 +0100)
committerDavid S. Miller <davem@davemloft.net>
Tue, 24 Nov 2015 22:15:56 +0000 (17:15 -0500)
commitfbdd29bfd2da979b7ac6a0084af56624156c1069
treec019c641c0c6130a769a596561529253e49bec13
parent33c40e242ce681092ab778c238f3fff5a345ee0e
net: ipmr, ip6mr: fix vif/tunnel failure race condition

Since (at least) commit b17a7c179dd3 ("[NET]: Do sysfs registration as
part of register_netdevice."), netdev_run_todo() deals only with
unregistration, so we don't need to do the rtnl_unlock/lock cycle to
finish registration when failing pimreg or dvmrp device creation. In
fact that opens a race condition where someone can delete the device
while rtnl is unlocked because it's fully registered. The problem gets
worse when netlink support is introduced as there are more points of entry
that can cause it and it also makes reusing that code correctly impossible.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Reviewed-by: Cong Wang <cwang@twopensource.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/ipmr.c
net/ipv6/ip6mr.c