Jakub Kicinski <kuba@xxxxxxxxxx> wrote: > netfilter... rtnl.. workqueue... let's CC Florian.. Thanks. > > HEAD commit: a4e6f95a891a Merge tag 'pinctrl-v5.15-2' of git://git.kern.. This HEAD doesn't include 7970a19b71044bf4dc2c1becc200275bdf1884d4 netfilter: nf_nat_masquerade: defer conntrack walk to work queue so, with a bit of luck this is already resolved. > > context_switch kernel/sched/core.c:4940 [inline] > > __schedule+0x940/0x26f0 kernel/sched/core.c:6287 > > schedule+0xd3/0x270 kernel/sched/core.c:6366 > > schedule_preempt_disabled+0xf/0x20 kernel/sched/core.c:6425 > > __mutex_lock_common kernel/locking/mutex.c:669 [inline] > > __mutex_lock+0xa34/0x12f0 kernel/locking/mutex.c:729 > > reg_check_chans_work+0x83/0xe10 net/wireless/reg.c:2423 > > process_one_work+0x9bf/0x16b0 kernel/workqueue.c:2297 workqueue tries to rtnl_lock()... > > task:syz-executor.0 state:D stack:27152 pid:17047 ppid: 13518 flags:0x00004004 > > Call Trace: > > context_switch kernel/sched/core.c:4940 [inline] > > __schedule+0x940/0x26f0 kernel/sched/core.c:6287 > > schedule+0xd3/0x270 kernel/sched/core.c:6366 > > schedule_preempt_disabled+0xf/0x20 kernel/sched/core.c:6425 > > __mutex_lock_common kernel/locking/mutex.c:669 [inline] > > __mutex_lock+0xa34/0x12f0 kernel/locking/mutex.c:729 > > rtnl_lock net/core/rtnetlink.c:72 [inline] > > rtnetlink_rcv_msg+0x3be/0xb80 net/core/rtnetlink.c:5569 ... syz-executor as well ... > > task:syz-executor.4 state:D stack:27216 pid:17052 ppid: 6665 flags:0x00004004 > > Call Trace: > > context_switch kernel/sched/core.c:4940 [inline] > > __schedule+0x940/0x26f0 kernel/sched/core.c:6287 > > schedule+0xd3/0x270 kernel/sched/core.c:6366 > > schedule_preempt_disabled+0xf/0x20 kernel/sched/core.c:6425 > > __mutex_lock_common kernel/locking/mutex.c:669 [inline] > > __mutex_lock+0xa34/0x12f0 kernel/locking/mutex.c:729 > > rtnl_lock net/core/rtnetlink.c:72 [inline] > > rtnetlink_rcv_msg+0x3be/0xb80 net/core/rtnetlink.c:5569 > > netlink_rcv_skb+0x153/0x420 net/netlink/af_netlink.c:2504 ... and another syz-executor instance ... > > __local_bh_enable_ip+0xcd/0x120 kernel/softirq.c:371 > > local_bh_enable include/linux/bottom_half.h:32 [inline] > > get_next_corpse net/netfilter/nf_conntrack_core.c:2252 [inline] > > nf_ct_iterate_cleanup+0x15a/0x450 net/netfilter/nf_conntrack_core.c:2275 > > nf_ct_iterate_cleanup_net net/netfilter/nf_conntrack_core.c:2363 [inline] > > nf_ct_iterate_cleanup_net+0x236/0x400 net/netfilter/nf_conntrack_core.c:2347 > > masq_device_event+0xae/0xe0 net/netfilter/nf_nat_masquerade.c:88 > > notifier_call_chain+0xb5/0x200 kernel/notifier.c:83 ... and rtnl is held by notifier call chain. This is no longer the case in current net head, nf_ct_iterate_cleanup() runs from workqueue without rtnl locked.
