Hi Claudio, others,
First of, let me apologize for the severe lack of details in my
previous post. I thought to quickly check if anyone else had seen
what I'm seeing.
I've gone to my archive of snapshots and extracted all the pxeboot's
from every snapshot I have, a total 1972 snapshot over the last 4+
years. I found 91 different pxe bootloaders, so at most 7 attempts to
find the breaking one :) As I only have an archive of amd64
installers, I used the amd64 pxeboot (even though ALIX is an i386
platform, the bootloader from amd64 has worked fine - and I did verify
the i386 pxeboot from 6.5 and the latest snapshot have the same
behaviour).
My method was using the pxeboot loader to boot bsd.rd from the local
storage. The only change I made between reboots was installing a
different version of pxeboot on my tftp server.
With a reasonable starting guess, I brought it down to 5 attempts.
First to fail is the pxeboot from the snapshot dated 2019-04-10 at
18:10:42, kernel build number 817. This shows some extra information
during boot though:
>> OpenBSD/amd64 PXEBOOT 3.43
boot> boot hd0a:/bsd.rd
booting hd0a:/bsd.rd: 3107327+1352704+3362824+0+458752
[363419+98+289008+28303]=0x8cc8a0
64 bit entry point at 0x2000d4
entry = 0x2000d4
kern_pml4 = 0
kern_pml3 = 0
kern_pml2 = 1
kern_pml1 = 0
end of bootstrap page tables = 0xa0000
The pxeboot from the snapshot before that (kernel build time
2019-04-10 at 11:52:59, with kernel build number 816) shows:
>> OpenBSD/amd64 PXEBOOT 3.42
boot> boot hd0a:/bsd.rd
booting hd0a:/bsd.rd: 3107327+1352704+3362824+0+458752
[363419+98+289008+28303]=0x8cc8a0
entry point at 0x2000d4
Copyright (c) 1982, 1986, 1989, 1991, 1993
The Regents of the University of California. All rights reserved.
Copyright (c) 1995-2018 OpenBSD. All rights reserved. https://www.OpenBSD.org
The next (different) pxeboot from the snapshot with the kernel built
on 2019-04-12 at 20:40:53 (kernel build number 0) gets stuck at the
`entry point at 0x2000d4` again:
>> OpenBSD/amd64 PXEBOOT 3.43
boot> boot hd0a:/bsd.rd
booting hd0a:/bsd.rd: 3107327+1352704+3362824+0+458752
[363419+98+289008+28303]=0x8cc8a0
entry point at 0x2000d4
So, I was looking at commits to the boot code at or shortly after
april 10. The only one I see is this one, where Florian brings
sys/stand/boot/boot.c to version 1.48:
----------------------------------------------------------------------
Modified files:
sys/stand/boot : boot.c
Log message:
Unbreak "boot bsd.up" line in /etc/boot.conf
Found the hard way by Raf Czlonka (rczlonka AT gmail), thanks!
OK deraadt
----------------------------------------------------------------------
But I don't see how Florian's change could break things this way.
I'll try a revert, but it's going to take a bit of time to configure
my build environment and to figure out how to make the pxe bootloader.
If anyone has any suggestions in the mean time, I'm eager to hear
them.
Thanks,
Paul
On Wed, Jun 19, 2019 at 08:55:06AM +0200, Claudio Jeker wrote:
| On Wed, Jun 19, 2019 at 08:37:28AM +0200, Paul de Weerd wrote:
| > Morning folks,
| >
| > I ran into a problem after upgrading my ALIX to a more recent snapshot
| > in that it won't boot anymore. It gets to "entry point 0x20000d0" and
| > then stops. I tried using the PXE bootloader to load the local kernel
| > from disk (both bsd and bsd.rd) and to load kernels from tftp, but all
| > fails in similar ways with the entry point being the last output.
| >
| > I grabbed another ALIX to test, but I'm afraid I screwed that one up
| > and now that one doesn't boot either anymore. This is probably user
| > error, but now I'd like to confirm: has anyone successfully upgraded
| > their ALIX to a recent snapshot?
| >
| > It could be that my hardware is dying on me (I should find my piggy
| > bank for some nickels), so confirmation that this still works for
| > others is appreciated.
| >
|
| There were some boot(8) changes so try some older pxeboot from 6.4, 6.5 or
| the snapshot archive to see when the breakage was introduced.
--
>++++++++[<++++++++++>-]<+++++++.>+++[<------>-]<.>+++[<+
+++++++++++>-]<.>++[<------------>-]<+.--------------.[-]
http://www.weirdnet.nl/
