[Bug 273715] dumpon: Kernel panic on boot when enabling dumpon over IP

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=273715

--- Comment #11 from Zhenlei Huang <zlei@FreeBSD.org> ---
A minimal snip to repeat this bug:

```
# dumpon -i 0 -c 192.168.255.3 -g 192.168.255.1 -s 192.168.255.1 nonexist
```

> I see various options how to resolve this:

> 1. Always allow using the physical device name (so it would somehow poke around and find it)
> 2. Add a flag to dumpon (e.g., `-f` for force) to skip the device exists/device supports debugnet check
> 3. Allow specifying multiple devices netdump_client to try when dumping
> 4. Determine based on the device name if it can exist early (feels wonky)

The `dumpon` start quite early, even before `FILESYSTEMS`. At that time `netif`
(interface renaming) has no chances to run.

I guess it is intended to let dumpon run as early as possible so that almost
all kernel panics can be caught rightly.

Well for netdump(4), if a pubkey is requested, then dumpon requires
`FILESYSTEMS`.

I guess we want to defer the setup of netdump(4), until `FILESYSTEMS` (probably
also `netif`) is/are ready.

Or we can split the rc.d/dumpon into `rc.d/dumpon` and `rc.d/netdump`.

-- 
You are receiving this mail because:
You are the assignee for the bug.