BUG: sleeping function called from invalid context at kernel/sched/completion.c:90
in_atomic(): 1, irqs_disabled(): 0, pid: 17, name: ksoftirqd/1
1 lock held by ksoftirqd/1/17:
 #0:  (rcu_callback){......}, at: [<ffffffff812878ee>] __rcu_reclaim kernel/rcu/rcu.h:108 [inline]
 #0:  (rcu_callback){......}, at: [<ffffffff812878ee>] rcu_do_batch kernel/rcu/tree.c:2789 [inline]
 #0:  (rcu_callback){......}, at: [<ffffffff812878ee>] invoke_rcu_callbacks kernel/rcu/tree.c:3053 [inline]
 #0:  (rcu_callback){......}, at: [<ffffffff812878ee>] __rcu_process_callbacks kernel/rcu/tree.c:3020 [inline]
 #0:  (rcu_callback){......}, at: [<ffffffff812878ee>] rcu_process_callbacks+0x98e/0x12b0 kernel/rcu/tree.c:3037
Preemption disabled at:[   53.446511] [<ffffffff83a036cb>] __do_softirq+0xdb/0x937 kernel/softirq.c:261
CPU: 1 PID: 17 Comm: ksoftirqd/1 Not tainted 4.9.117-g8b21e85 #71
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
 ffff8801d9abf868 ffffffff81eb4be9 ffffffff83a036cb 0000000000000000
 0000000000000100 ffff8801d9aa3000 ffff8801d9aa3000 ffff8801d9abf8a0
 ffffffff81424620 ffff8801d9aa3000 ffffffff83c59f60 000000000000005a
Call Trace:
 [<ffffffff81eb4be9>] __dump_stack lib/dump_stack.c:15 [inline]
 [<ffffffff81eb4be9>] dump_stack+0xc1/0x128 lib/dump_stack.c:51
 [<ffffffff81424620>] ___might_sleep.cold.123+0x1bc/0x1f5 kernel/sched/core.c:7988
 [<ffffffff811bb6c5>] __might_sleep+0x95/0x1a0 kernel/sched/core.c:7945
 [<ffffffff839f0359>] __wait_for_common kernel/sched/completion.c:90 [inline]
 [<ffffffff839f0359>] wait_for_common kernel/sched/completion.c:101 [inline]
 [<ffffffff839f0359>] wait_for_completion+0x89/0x2e0 kernel/sched/completion.c:122
 [<ffffffff8127be77>] __wait_rcu_gp+0x137/0x1b0 kernel/rcu/update.c:369
 [<ffffffff8128410a>] synchronize_rcu.part.62+0xfa/0x110 kernel/rcu/tree_plugin.h:684
 [<ffffffff81284147>] synchronize_rcu+0x27/0x90 kernel/rcu/tree_plugin.h:685
 [<ffffffff836bfd8a>] __l2tp_session_unhash+0x38a/0x520 net/l2tp/l2tp_core.c:1796
 [<ffffffff836c0117>] l2tp_tunnel_closeall+0x1f7/0x350 net/l2tp/l2tp_core.c:1364
 [<ffffffff836c09d2>] l2tp_tunnel_destruct+0x2f2/0x590 net/l2tp/l2tp_core.c:1324
 [<ffffffff83023095>] __sk_destruct+0x55/0x590 net/core/sock.c:1428
 [<ffffffff8128780e>] __rcu_reclaim kernel/rcu/rcu.h:118 [inline]
 [<ffffffff8128780e>] rcu_do_batch kernel/rcu/tree.c:2789 [inline]
 [<ffffffff8128780e>] invoke_rcu_callbacks kernel/rcu/tree.c:3053 [inline]
 [<ffffffff8128780e>] __rcu_process_callbacks kernel/rcu/tree.c:3020 [inline]
 [<ffffffff8128780e>] rcu_process_callbacks+0x8ae/0x12b0 kernel/rcu/tree.c:3037
 [<ffffffff83a037fb>] __do_softirq+0x20b/0x937 kernel/softirq.c:284
 [<ffffffff8114b00e>] run_ksoftirqd+0x2e/0x60 kernel/softirq.c:676

=================================
[ INFO: inconsistent lock state ]
4.9.117-g8b21e85 #71 Not tainted
---------------------------------
inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage.
ksoftirqd/0/3 [HC0[0]:SC1[1]:HE1:SE0] takes:
 (&(&rnp->exp_lock)->rlock){+.?...}, at: [<ffffffff81282a4b>] spin_lock include/linux/spinlock.h:302 [inline]
 (&(&rnp->exp_lock)->rlock){+.?...}, at: [<ffffffff81282a4b>] exp_funnel_lock kernel/rcu/tree_exp.h:271 [inline]
 (&(&rnp->exp_lock)->rlock){+.?...}, at: [<ffffffff81282a4b>] _synchronize_rcu_expedited+0x1fb/0x840 kernel/rcu/tree_exp.h:569
  mark_irqflags kernel/locking/lockdep.c:2941 [inline]
  __lock_acquire+0xdd0/0x4070 kernel/locking/lockdep.c:3302
  lock_acquire+0x130/0x3e0 kernel/locking/lockdep.c:3756
  __raw_spin_lock include/linux/spinlock_api_smp.h:144 [inline]
  _raw_spin_lock+0x36/0x50 kernel/locking/spinlock.c:151
  spin_lock include/linux/spinlock.h:302 [inline]
  rcu_exp_wait_wake+0x2d2/0x5e0 kernel/rcu/tree_exp.h:504
  rcu_exp_sel_wait_wake kernel/rcu/tree_exp.h:535 [inline]
  _synchronize_rcu_expedited+0x73c/0x840 kernel/rcu/tree_exp.h:575
  synchronize_rcu_expedited kernel/rcu/tree_exp.h:687 [inline]
  synchronize_rcu.part.62+0xd3/0x110 kernel/rcu/tree_plugin.h:682
  synchronize_rcu+0x27/0x90 kernel/rcu/tree_plugin.h:685
  rcu_test_sync_prims+0x9/0x30 kernel/rcu/update.c:829
  rcu_scheduler_starting+0x51/0x80 kernel/rcu/tree.c:4016
  rest_init+0xf/0x189 init/main.c:388
  start_kernel+0x67e/0x6b2 init/main.c:664
  x86_64_start_reservations+0x29/0x2b arch/x86/kernel/head64.c:196
  x86_64_start_kernel+0x13f/0x162 arch/x86/kernel/head64.c:177
irq event stamp: 757388
hardirqs last  enabled at (757388): [<ffffffff811498da>] __local_bh_enable_ip+0x6a/0xd0 kernel/softirq.c:186
hardirqs last disabled at (757387): [<ffffffff811498a2>] __local_bh_enable_ip+0x32/0xd0 kernel/softirq.c:163
softirqs last  enabled at (757076): [<ffffffff83a03a5c>] __do_softirq+0x46c/0x937 kernel/softirq.c:310
softirqs last disabled at (757081): [<ffffffff8114b00e>] run_ksoftirqd+0x2e/0x60 kernel/softirq.c:676

other info that might help us debug this:
 Possible unsafe locking scenario:

       CPU0
       ----
  lock(&(&rnp->exp_lock)->rlock);
  <Interrupt>
    lock(&(&rnp->exp_lock)->rlock);

 *** DEADLOCK ***

1 lock held by ksoftirqd/0/3:
 #0:  (rcu_callback){......}, at: [<ffffffff812878ee>] __rcu_reclaim kernel/rcu/rcu.h:108 [inline]
 #0:  (rcu_callback){......}, at: [<ffffffff812878ee>] rcu_do_batch kernel/rcu/tree.c:2789 [inline]
 #0:  (rcu_callback){......}, at: [<ffffffff812878ee>] invoke_rcu_callbacks kernel/rcu/tree.c:3053 [inline]
 #0:  (rcu_callback){......}, at: [<ffffffff812878ee>] __rcu_process_callbacks kernel/rcu/tree.c:3020 [inline]
 #0:  (rcu_callback){......}, at: [<ffffffff812878ee>] rcu_process_callbacks+0x98e/0x12b0 kernel/rcu/tree.c:3037

stack backtrace:
 [<ffffffff811a82c9>] smpboot_thread_fn+0x5c9/0x900 kernel/smpboot.c:163
 [<ffffffff8119d0ed>] kthread+0x26d/0x300 kernel/kthread.c:211
 [<ffffffff839fcddc>] ret_from_fork+0x5c/0x70 arch/x86/entry/entry_64.S:373
CPU: 0 PID: 3 Comm: ksoftirqd/0 Not tainted 4.9.117-g8b21e85 #71
BUG: scheduling while atomic: ksoftirqd/1/17/0x00000101
INFO: lockdep is turned off.
Modules linked in:
Preemption disabled at:
[<ffffffff83a036cb>] __do_softirq+0xdb/0x937 kernel/softirq.c:261