Identifying cause of crash, problem with crash(8)?

[Timo Buhrmester <fstd.lkml%gmail.com@localhost>]

I went from 7-stable as of Mar 6 2016 to 7-stable as of Jul 24 2016
on an amd64 server and this morning it crashed without an obvious
reason.  It dropped (as configured) into ddb and executed sync,
savecore did the rest.

Now I'm trying to figure out what happened using crash(8), but:
1. The (gunzip'ed) kernel image in /var/crash seems awfully small:
| -rw-------  1 root  wheel  554839 Jul 26 09:48 /var/crash/netbsd.3
compared with the kernel that the machine booted
| -rwxr-xr-x  1 root  wheel  5498863 Jul 24 22:03 /DEDIKERN
Is this right?

2. I get bogus register information and a diagnostic from crash(8):
| crash> show registers
| ds          0
| es          0
| fs          0
| gs          0
| rdi         0
| rsi         0
| rbp         fffffe80402bf860
| rbx         0
| rdx         0
| rcx         0
| rax         0
| r8          0
| r9          0
| r10         0
| r11         0
| r12         0
| r13         0
| r14         0
| r15         0
| rip         0
| cs          0
| rflags      0
| rsp         fffffe80402bf820
| ss          0
| crash: kvm_read(0x0, 1): invalid translation (invalid PTE)
The last line comes on crash's stderr, I don't understand what
it's telling me; but nearly all the registers esp. rip being 0
can't be right, can it?