kern/55681: Kernel can panic on shutdown if accounting is enabled due to deadfs call

[thorpej%me.com@localhost]

>Number:         55681
>Category:       kern
>Synopsis:       Kernel can panic on shutdown if accounting is enabled due to deadfs call
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Sep 23 17:15:01 +0000 2020
>Originator:     Jason Thorpe
>Release:        NetBSD 9.99.72
>Organization:
RISCy Business
>Environment:
NetBSD 9.99.72 (multiprocessor Alpha, in this case, but this is a MI issue)
>Description:
The kernel can panic due to a deadfs call during halt or reboot if accounting was enabled and not disabled before the reboot() system call.

[ 68841.1748690] syncing disks... done
[ 68842.3274481] unmounting file systems...
[ 68858.3562613] Skipping crash dump on recursive panic
[ 68858.4129006] panic: dead fs operation used
[ 68858.4627047] cpu1: Begin traceback...
[ 68858.5076248] alpha trace requires known PC =eject=
[ 68858.5662176] cpu1: End traceback...
Stopped in pid 0.762 (system) at        netbsd:cpu_Debugger+0x4:        ret
zero,(ra)
db{1}> bt
cpu_Debugger() at netbsd:cpu_Debugger+0x4
db_panic() at netbsd:db_panic+0xc8
vpanic() at netbsd:vpanic+0x10c
panic() at netbsd:panic+0x58
dead_panic() at netbsd:dead_panic+0x20
VFS_STATVFS() at netbsd:VFS_STATVFS+0x34
acct_chkfree() at netbsd:acct_chkfree+0x4c
acctwatch() at netbsd:acctwatch+0xb0
--- kernel thread backstop ---
db{1}>

>How-To-Repeat:
Enable accounting, and use "halt" or "reboot" to shut down.  Issue is not 100% reproducible, but will happen eventually.
>Fix:
Not provided.