From 9799c50372b23ed774791bdb87d700f1286ee8a9 Mon Sep 17 00:00:00 2001 From: Linus Lüssing Date: Thu, 6 Oct 2016 01:43:08 +0200 Subject: batman-adv: fix splat on disabling an interface MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit As long as there is still a reference for a hard interface held, there might still be a forwarding packet relying on its attributes. Therefore avoid setting hard_iface->soft_iface to NULL when disabling a hard interface. This fixes the following, potential splat: batman_adv: bat0: Interface deactivated: eth1 batman_adv: bat0: Removing interface: eth1 cgroup: new mount options do not match the existing superblock, will be ignored batman_adv: bat0: Interface deactivated: eth3 batman_adv: bat0: Removing interface: eth3 ------------[ cut here ]------------ WARNING: CPU: 3 PID: 1986 at ./net/batman-adv/bat_iv_ogm.c:549 batadv_iv_send_outstanding_bat_ogm_packet+0x145/0x643 [batman_adv] Modules linked in: batman_adv(O-) <...> CPU: 3 PID: 1986 Comm: kworker/u8:2 Tainted: G W O 4.6.0-rc6+ #1 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.7.5-20140531_083030-gandalf 04/01/2014 Workqueue: bat_events batadv_iv_send_outstanding_bat_ogm_packet [batman_adv] 0000000000000000 ffff88001d93bca0 ffffffff8126c26b 0000000000000000 0000000000000000 ffff88001d93bcf0 ffffffff81051615 ffff88001f19f818 000002251d93bd68 0000000000000046 ffff88001dc04a00 ffff88001becbe48 Call Trace: [] dump_stack+0x67/0x90 [] __warn+0xc7/0xe5 [] warn_slowpath_null+0x18/0x1a [] batadv_iv_send_outstanding_bat_ogm_packet+0x145/0x643 [batman_adv] [] ? __lock_is_held+0x32/0x54 [] process_one_work+0x2a8/0x4f5 [] ? process_one_work+0x15c/0x4f5 [] worker_thread+0x1d5/0x2c0 [] ? process_scheduled_works+0x2e/0x2e [] ? process_scheduled_works+0x2e/0x2e [] kthread+0xc0/0xc8 [] ret_from_fork+0x22/0x40 [] ? __init_kthread_worker+0x55/0x55 ---[ end trace 647f9f325123dc05 ]--- What happened here is, that there was still a forw_packet (here: a BATMAN IV OGM) in the queue of eth3 with the forw_packet->if_incoming set to eth1 and the forw_packet->if_outgoing set to eth3. When eth3 is to be deactivated and removed, then this thread waits for the forw_packet queued on eth3 to finish. Because eth1 was deactivated and removed earlier and by that had forw_packet->if_incoming->soft_iface, set to NULL, the splat when trying to send/flush the OGM on eth3 occures. Fixes: c6c8fea29769 ("net: Add batman-adv meshing protocol") Signed-off-by: Linus Lüssing [sven@narfation.org: Reduced size of Oops message] Signed-off-by: Sven Eckelmann Signed-off-by: Simon Wunderlich --- net/batman-adv/hard-interface.c | 1 - 1 file changed, 1 deletion(-) (limited to 'net/batman-adv') diff --git a/net/batman-adv/hard-interface.c b/net/batman-adv/hard-interface.c index 08ce36147c4c..e034afbd1bb0 100644 --- a/net/batman-adv/hard-interface.c +++ b/net/batman-adv/hard-interface.c @@ -652,7 +652,6 @@ void batadv_hardif_disable_interface(struct batadv_hard_iface *hard_iface, batadv_softif_destroy_sysfs(hard_iface->soft_iface); } - hard_iface->soft_iface = NULL; batadv_hardif_put(hard_iface); out: -- cgit