Re: KGDB on virt68k?

On Apr 27, 2026, at 3:56=E2=80=AFPM, Jason Thorpe <[email protected]> =
wrote:
> So maybe VirtIO serial might be a better approach for KGDB. (I assume =
if I create an appropriate node in /dev and run getty on it that I'd be =
able to connect to that port and get a login...)
>=20
> Yah, should:
>=20
> device-major viocon    char 364            viocon

So, I tried this, but it appears that VirtIO serial/console on NetBSD =
isn=E2=80=99t yet ready for prime time. Looks like this is noted in BUGS =
in the viocon(4) man page:

     Use as a kernel console for NetBSD is not yet supported.

     The multiport feature is not yet supported.

So I built a QEMU with a tweaked hw/m68k/virt.c that provides four =
Goldfish TTYs 0x200 bytes apart, a tweaked NetBSD that expects the same =
(there doesn=E2=80=99t appear to be a way to pass the count or stride of =
the Goldfish TTY devices, only one of which is passed in the Linux =
bootinfo), and then configured the appropriate devices after booting a =
newly-generated live image and that Just Worked.

That makes me wonder whether it=E2=80=99d be a better investment of my =
time to get Goldfish TTY usable by KGDB or to get viocon(4) working and =
usable by KGDB. It doesn=E2=80=99t seem like switching QEMU virt68k to =
support VirtIO console would be verry difficult, and it already allows =
attaching any number of them (all of which NetBSD will at least already =
detect, even if they don=E2=80=99t currently work).

  -- Chris