Hi Mike,
Thank you for your replies!
On Fri Oct 2, 2026 at 3:08 AM UTC, Mike Larkin wrote:
> 1. provide the entire vmd log not just a tiny cut-paste please.
That's about all there is to it. Here's the rest.
Oct 2 00:42:31 bragi vmd[45485]: startup
Oct 2 00:42:31 bragi vmd[45485]: started bit (vm 1) successfully, tty
/dev/ttyp0
Oct 2 00:42:31 bragi vmd[45485]: started potato (vm 2) successfully, tty
/dev/ttyp1
Oct 2 00:42:31 bragi vmd[83512]: decode_prefix: invalid index
Oct 2 00:42:31 bragi vmd[83512]: decode_opcode: invalid index
Oct 2 00:42:31 bragi vmd[83512]: decode_modrm: invalid index
Oct 2 00:42:31 bragi vmd[83512]: decode_modrm: invalid state or null insn
Oct 2 00:42:31 bragi vmd[83512]: insn_decode: error decoding modrm
Oct 2 00:42:31 bragi vmd[93773]: decode_prefix: invalid index
Oct 2 00:42:31 bragi vmd[93773]: decode_opcode: invalid index
Oct 2 00:42:31 bragi vmd[93773]: decode_modrm: invalid index
Oct 2 00:42:31 bragi vmd[93773]: decode_modrm: invalid state or null insn
Oct 2 00:42:31 bragi vmd[93773]: insn_decode: error decoding modrm
Oct 2 00:43:31 bragi vmd[45485]: started farm (vm 3) successfully, tty
/dev/ttyp0
Oct 2 00:43:31 bragi vmd[41046]: decode_prefix: invalid index
Oct 2 00:43:31 bragi vmd[41046]: decode_opcode: invalid index
Oct 2 00:43:31 bragi vmd[41046]: decode_modrm: invalid index
Oct 2 00:43:31 bragi vmd[41046]: decode_modrm: invalid state or null insn
Oct 2 00:43:31 bragi vmd[41046]: insn_decode: error decoding modrm
Oct 2 00:55:24 bragi vmd[45485]: started potato (vm 2) successfully, tty
/dev/ttyp1
Oct 2 00:55:24 bragi vmd[48168]: decode_prefix: invalid index
Oct 2 00:55:24 bragi vmd[48168]: decode_opcode: invalid index
Oct 2 00:55:24 bragi vmd[48168]: decode_modrm: invalid index
Oct 2 00:55:24 bragi vmd[48168]: decode_modrm: invalid state or null insn
Oct 2 00:55:24 bragi vmd[48168]: insn_decode: error decoding modrm
Oct 2 00:55:41 bragi vmd[45485]: started potato (vm 2) successfully, tty
/dev/ttyp1
Oct 2 00:55:41 bragi vmd[15401]: decode_prefix: invalid index
Oct 2 00:55:41 bragi vmd[15401]: decode_opcode: invalid index
Oct 2 00:55:41 bragi vmd[15401]: decode_modrm: invalid index
Oct 2 00:55:41 bragi vmd[15401]: decode_modrm: invalid state or null insn
Oct 2 00:55:41 bragi vmd[15401]: insn_decode: error decoding modrm
> 2. provide the exact guest configuration. what openbsd version, what
> archicture, what kernel version build number, how you started it,
> vm configuration, etc.
On bragi, I have:
2x 7.9
1x 8.0 snapshot
socket owner :_vmdusers
staggered start parallel 2 delay 60
vm "bit" {
memory 3G
disk "/var/vmm/bit.raw"
interface tap1 {
locked lladdr 00:55:05:00:00:01
}
owner parsnip
disable
}
vm "potato" {
memory 768M
disk "/var/vmm/potato.raw"
interface tap3 {
locked lladdr 00:55:05:00:00:03
}
owner roflcopter
disable
}
vm "farm" {
memory 768M
disk "/var/vmm/farm.raw"
interface tap4 {
locked lladdr 00:55:05:00:00:04
}
owner roflcopter
disable
}
All amd64.
I also have another host with an even simpler setup. Very vanilla 7.9
install that was working before I upgraded to 8.0-snapshot.
I installed it from the 7.9 ISO. I think it doesn't even have syspatches.
It behaves identical to bragi.
slowservers0# grep vmd /var/log/daemon
Sep 16 23:57:03 slowservers0 vmd[14635]: startup
Sep 16 23:57:50 slowservers0 vmd[14635]: started sevennine (vm 1) successfully,
tty /dev/ttyp1
Sep 16 23:59:05 slowservers0 vmd[14635]: started sevennine (vm 1) successfully,
tty /dev/ttyp1
Sep 17 01:43:13 slowservers0 vmd[14635]: started sevennine (vm 1) successfully,
tty /dev/ttyp1
Sep 17 01:45:37 slowservers0 vmd[30061]: control exiting, pid 30061
Sep 17 01:45:37 slowservers0 vmd[4740]: vmm exiting, pid 4740
Sep 17 01:45:37 slowservers0 vmd[89179]: priv exiting, pid 89179
Sep 17 01:45:37 slowservers0 vmd[97233]: agentx exiting, pid 97233
Sep 17 01:45:37 slowservers0 vmd[14635]: terminating
Oct 2 02:24:23 slowservers0 vmd[52398]: startup
Oct 2 02:24:23 slowservers0 vmd[52398]: started sevennine (vm 1) successfully,
tty /dev/ttyp0
Oct 2 02:28:49 slowservers0 vmd[79082]: vmm exiting, pid 79082
Oct 2 02:28:49 slowservers0 vmd[95295]: control exiting, pid 95295
Oct 2 02:28:49 slowservers0 vmd[71428]: agentx exiting, pid 71428
Oct 2 02:28:49 slowservers0 vmd[14146]: priv exiting, pid 14146
Oct 2 02:28:49 slowservers0 vmd[52398]: terminating
Oct 2 02:29:59 slowservers0 vmd[47771]: startup
Oct 2 02:29:59 slowservers0 vmd[47771]: started sevennine (vm 1) successfully,
tty /dev/ttyp0
Oct 2 02:44:10 slowservers0 vmd[29064]: vmm exiting, pid 29064
Oct 2 02:44:10 slowservers0 vmd[96504]: priv exiting, pid 96504
Oct 2 02:44:10 slowservers0 vmd[36315]: control exiting, pid 36315
Oct 2 02:44:10 slowservers0 vmd[81355]: agentx exiting, pid 81355
Oct 2 02:44:10 slowservers0 vmd[47771]: terminating
Oct 2 02:48:05 slowservers0 vmd[84069]: startup
Oct 2 02:48:05 slowservers0 vmd[84069]: started sevennine (vm 1) successfully,
tty /dev/ttyp0
Oct 2 02:48:05 slowservers0 vmd[42720]: decode_prefix: invalid index
Oct 2 02:48:05 slowservers0 vmd[42720]: decode_opcode: invalid index
Oct 2 02:48:05 slowservers0 vmd[42720]: decode_modrm: invalid index
Oct 2 02:48:05 slowservers0 vmd[42720]: decode_modrm: invalid state or null
insn
Oct 2 02:48:05 slowservers0 vmd[42720]: insn_decode: error decoding modrm
Oct 2 03:21:25 slowservers0 vmd[59816]: vmm exiting, pid 59816
Oct 2 03:21:25 slowservers0 vmd[98407]: control exiting, pid 98407
Oct 2 03:21:25 slowservers0 vmd[39576]: agentx exiting, pid 39576
Oct 2 03:21:25 slowservers0 vmd[69767]: priv exiting, pid 69767
Oct 2 03:21:25 slowservers0 vmd[84069]: terminating
Oct 2 18:16:49 slowservers0 vmd[2271]: startup
Oct 2 18:16:49 slowservers0 vmd[2271]: started sevennine (vm 1) successfully,
tty /dev/ttyp0
Oct 2 18:16:49 slowservers0 vmd[8089]: decode_prefix: invalid index
Oct 2 18:16:49 slowservers0 vmd[8089]: decode_opcode: invalid index
Oct 2 18:16:49 slowservers0 vmd[8089]: decode_modrm: invalid index
Oct 2 18:16:49 slowservers0 vmd[8089]: decode_modrm: invalid state or null insn
Oct 2 18:16:49 slowservers0 vmd[8089]: insn_decode: error decoding modrm
slowservers0# cat /etc/vm.conf
vm "sevennine" {
memory 768M
disk "/var/vmm/sevennine.raw"
cdrom "/var/vmm-installers/install79.iso"
}
> 3. when does it fail like this? immediately at boot or sometime later?
> from the timestamps it looks like immediately.
Yes, immediately. It fails at boot. If i try to start it, it fails then.
slowservers0# vmctl show
ID PID VCPUS MAXMEM CURMEM TTY OWNER STATE NAME
1 - 1 768M - - root stopped sevennine
slowservers0# vmctl start -c sevennine
Connected to /dev/ttyp1 (speed 115200)
[EOT]
slowservers0# vmctl show
ID PID VCPUS MAXMEM CURMEM TTY OWNER STATE NAME
1 - 1 768M - - root stopped sevennine
slowservers0# tail -n 5 /var/log/daemon
Oct 2 18:19:45 slowservers0 vmd[1302]: decode_prefix: invalid index
Oct 2 18:19:45 slowservers0 vmd[1302]: decode_opcode: invalid index
Oct 2 18:19:45 slowservers0 vmd[1302]: decode_modrm: invalid index
Oct 2 18:19:45 slowservers0 vmd[1302]: decode_modrm: invalid state or null insn
Oct 2 18:19:45 slowservers0 vmd[1302]: insn_decode: error decoding modrm
> 4. did you make sure you did an fw update to get the latest vmm-firmware
> after/during sysupgrade?
Yes. sysupgrade -s upgrades firmware as part of the progress.
In fact, on "slowservers0" I upgraded the firmware before
upgrading to snapshot. VM started fine, so it wasn't that
the firmware broke it. But on the snapshot it has the newest
firmware.
> I just installed this exact build on an AMD machine here and don't see any
> problems.
>
> The vmd log snippet above looks like you somehow have a mismatched set of
> vmd and kernel.
I upgraded from 7.9 by using sysupgrade -s. Is it possible to
have vmd and kernel mismatched?
odin# sha256 /usr/sbin/vmd # This is a 7.9 host.
SHA256 (/usr/sbin/vmd) =
8b3760ad2974e19b5156b870735635560226d9d405575cab0042b3f3c1076c39
bragi# sha256 /usr/sbin/vmd # 8.0-snapshot
SHA256 (/usr/sbin/vmd) =
a65f825d22cc31db592f43398261432707aeb6423de0ee38da565d9c3e11ce51
slowservers0# sha256 /usr/sbin/vmd # 8.0-snapshot
SHA256 (/usr/sbin/vmd) =
a65f825d22cc31db592f43398261432707aeb6423de0ee38da565d9c3e11ce51
> the kernel build string can be obtained via 'strings' or 'what' on the
> guest kernel. since it's a raw image you can use vnconfig to loopback
> mount it on the host to get at that kernel.
I just booted the 7.9 amd64 ISO and it does he exact same thing,
without reading a kernel from disk.
> that narrows it down basically nothing as that window lands exactly
> when dv and I made massive changes. since you can bisect (it appears),
> please narrow the window further if you can.
Could you advise me on if it's possible to downgrade snapshots
and where I can find them versioned? I don't see that on ftp.openbsd.org.
Thank you!
