port-evbarm/54224: panic: lock error: Mutex: mutex_vector_enter,552: locking against myself

[perlhaq%free.fr@localhost]

>Number:         54224
>Category:       port-evbarm
>Synopsis:       panic: lock error: Mutex: mutex_vector_enter,552: locking against myself
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    port-evbarm-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed May 22 21:10:00 +0000 2019
>Originator:     Stephen Takacs
>Release:        NetBSD 8.0
>Organization:
>Environment:
NetBSD armv7 8.0 NetBSD 8.0 (CUBIETRUCK) #0: Fri Jan 18 12:20:54 CET 2019 root@armv7:/usr/src/sys/arch/evbarm/compile/CUBIETRUCK evbarm
>Description:
This is a Cubietruck that's been running NetBSD for about 6 months, and is
seldom powered off.  The problem happened twice so far.  The first time was Feb
17, a bit before 03:26, according to my notes.  The system panic'd and this is
what was visible on the serial console:

Mutex error: mutex_vector_enter,552: locking against myself

lock address : 0x000000009ce091ec
current cpu  :                  0
current lwp  : 0x00000000a60e6980
owner field  : 0x00000000a60e6980 wait/spin:                0/0

panic: lock error: Mutex: mutex_vector_enter,552: locking against myself: lock 0x9ce091ec cpu 0 lwp 0xa60e6980
cpu0: Begin traceback...
0x9673dbf4: netbsd:db_panic+0xc
0x9673dc0c: netbsd:vpanic+0x1a4
0x9673dc24: netbsd:snprintf
0x9673dc5c: netbsd:lockdebug_abort+0x70
0x9673dc74: netbsd:mutex_abort+0x30
0x9673dcd4: netbsd:mutex_enter+0x388
0x9673dd14: netbsd:pmap_fault_fixup+0x40
0x9673dd9c: netbsd:data_abort_handler+0x164
0x9673de14: netbsd:address_exception_entry+0x5c
0x9673df44: netbsd:uvm_fault_internal+0x32c
0x9673dfac:

And that's where the output stopped.  The second event was last night, and this
time I was able to pinpoint it between 03:12 and 03:22, so the trigger appears
to be the 03:15 daily cron job (all of which is unchanged from defaults).
Here's what was on the serial console:

Mutex error: mutex_vector_enter,552: locking against myself

lock address : 0x00000000be82e93c
current cpu  :                  1
current lwp  : 0x00000000a798edc0
owner field  : 0x00000000a798edc0 wait/spin:                0/0

panic: lock error: Mutex: mutex_vector_enter,552: locking against myself: lock 0xbe82e93c cpu 1 lwp 0xa798edc0
cpu1: Begin traceback...
0xb1e83bf4: netbsd:db_panic+0xc
0xb1e83c0c: netbsd:vpanic+0x1a4
0xb1e83c24: netbsd:snprintf
0xb1e83c5c: netbsd:lockdebug_abort+0x70
0xb1e83c74: netbsd:mutex_abort+0x30
0xb1e83cd4: netbsd:mutex_enter+0x388
0xb1e83d14: netbsd:pmap_fault_fixup+0x40
0xb1e83d9c: netbsd:data_abort_handler+0x164
0xb1e83e14: netbsd:address_exception_entry+0x5c
0xb1e83f44: netbsd:uvm_fault_internal+0x32c
0xb1e83fac: netbsd:prefetch_abort_handler+0x218
cpu1: End traceback...

dumping to dev 16,1 offset 991
dump 0

And that's where the output stopped.  The dump didn't work, and this
error came up when I reset the system:

savecore: kvm_read: L1 entry is invalid: Undefined error: 0

I should also mention that the CUBIETRUCK kernel config is the default one with
this additional line:

pseudo-device    cgd                     # cryptographic disk devices

Perhaps the dump didn't work because device 16,1 (rwd0b) was assigned in
/etc/cgd/cgd.conf like so:

cgd1 /dev/wd0b /etc/cgd/swap
>How-To-Repeat:

>Fix: