On Mon, Nov 1, 2021 at 3:46 PM Sven Eckelmann sven@narfation.org wrote:
On Monday, 1 November 2021 05:01:02 CET Dongliang Mu wrote:
Call Trace: __cancel_work_timer+0x1c9/0x280 kernel/workqueue.c:3170 batadv_v_ogm_free+0x1d/0x50 net/batman-adv/bat_v_ogm.c:1076 batadv_mesh_free+0x35/0xa0 net/batman-adv/main.c:244 batadv_mesh_init+0x22a/0x240 net/batman-adv/main.c:226 batadv_softif_init_late+0x1ad/0x240 net/batman-adv/soft-interface.c:804 register_netdevice+0x15d/0x810 net/core/dev.c:10229
This is definitely not a backtrace of the current code and its error handling. Please check the current code [1] and explain the situation against this version.
Yes, you're right. The error handling code in the upstream is not prone to this bug.
My local syzkaller instance is fuzzing on 5.14-rc5
Kind regards, Sven
[1] https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/tree/net...