At Sat, 20 Dec 2025 12:24:03 +0000 (UTC), RVP <rvp%SDF.ORG@localhost> wrote:
Subject: Re: NetBSD still fails to boot on a MacBook7,1
>
> Looking at that screenshot again however, it still looks very much like what
> would happen if there was some mis-agreement between the bootloader and the
> kernel.
The bootloader seems to be running with the default mini vga emulation
done by the EFI firmware.
When the kernel starts it takes over the whole display and I would
expect that's where it changes something in the hardware configuration.
I see something similar when macos boots with full boot message
verbosity enabled ("nvram boot-args="-v"). First you get the
small-box/window VGA emulated EFI boot messages, then when the kernel
starts it switches to full-screen and the kernel and startup messages
appear at a higher resolution and with many more lines and columns.
Something changed between NetBSD-7.0 and NetBSD-8.0 in the way the
kernel resets the display hardware and it hasn't worked since 7.0. I
still don't think anyone has pointed directly at the code which does
this, and I haven't easily been able to find it, and so I still don't
know where to bisect/blame it to find the bug.
> Is the bootloader the one built along with the kernel, or a very old one?
> (the kernel version it was built against should be printed in the banner.)
It's all from the same "build.sh" invocation. I make an install image,
copy it to the USB disk fob, and try to boot from it (using rEFInd).
In this latest test case the boot "version" command reports:
>> NetBSD/x86 EFI Boot (x64), Revision 1.2 (Sat Dec 6 08:56:07 UTC 2025) (from NetBSD 11.99.4)
>> Memory: 572/2882560 K
(BTW, the boot loader scrolls half the "help" response off the top of
the screen, and before I drop to the boot prompt there appears to be
some text displayed and quickly replaced with the boot menu (which has
the options of (1) install, (2) drop to boot prompt). I truly hate
screen-based firmware and bootloaders -- give me a serial port and plain
line-by-line input and output!!! Maybe the recent ucom console will
eventually be usable for better debugging of laptops, but perhaps not,
depending on the built-in firmware -- in this case I would at least be
able to read what the kernel prints before it hangs!)
--
Greg A. Woods <gwoods%acm.org@localhost>
Kelowna, BC +1 250 762-7675 RoboHack <woods%robohack.ca@localhost>
Planix, Inc. <woods%planix.com@localhost> Avoncote Farms <woods%avoncote.ca@localhost>
Attachment:
pgpbAaecmET7B.pgp
Description: OpenPGP Digital Signature