Re: Page fault early in boot

[Taylor R Campbell <riastradh%NetBSD.org@localhost>]

> Date: Mon, 14 Apr 2025 07:03:34 +0000
> From: Emmanuel Dreyfus <manu%netbsd.org@localhost>
> 
> Your patch avoids the panic. now it hangs later during the boot
> 
> Last kernel message is swwdog0: software watchdog initilized
> I can break into ddb. It shows init waiting on lbolt. Almost all
> CPU are waiting on idle, one of softclk/0. Process 0.0 (swapper)
> is on: cv_wait/biowait/validate_label/readdisklabel/dk_getdisklabel/
> dk_open/bdev_open/spec_open/VOP_OPEN/dk_open_parent/dkopen/bdev_open/
> spec_open/VOP_OPEN/opendisk/match_bootwedge/findroot/cpu_bootconf/
> rf_autoconf/config_finalize/main
> 
> [...]
> 
> The kernel detected a mfii controller with no disk attached. The
> only dk devices are the GPT partitions from the USB key I used
> to boot. I will try a kernel built without raidframe.

Can you `boot -c' to enter userconf and disable mfii?

> boot -c
userconf> disable mfii
userconf> quit

If that works, can you file a PR about mfii(4) hanging with no disks?
I wouldn't be surprised if that's an edge case that nobody developing
the driver had tested.  It's probably just a matter of a missing
xs->error = XS_WHATEVER; scsipi_done(xs) somewhere, maybe conditional
on `if (sc->sc_doesn't_actually_have_disks_plugged_in)'.