Bug 249226 - Freeze during early boot
Summary: Freeze during early boot
Status: Closed FIXED
Alias: None
Product: Base System
Classification: Unclassified
Component: kern (show other bugs)
Version: 12.1-RELEASE
Hardware: amd64 Any
: --- Affects Some People
Assignee: freebsd-bugs (Nobody)
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2020-09-10 00:54 UTC by Mason Loring Bliss
Modified: 2021-10-09 19:31 UTC (History)
4 users (show)

See Also:


Attachments
Photo of freeze. (963.37 KB, image/jpeg)
2020-09-10 00:54 UTC, Mason Loring Bliss
no flags Details
dmesg from linux (55.17 KB, text/plain)
2020-09-10 00:55 UTC, Mason Loring Bliss
no flags Details
dmesg from netbsd (15.24 KB, text/plain)
2020-09-10 00:56 UTC, Mason Loring Bliss
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Mason Loring Bliss freebsd_triage 2020-09-10 00:54:54 UTC
Created attachment 217861 [details]
Photo of freeze.

Hi, all. I'd like to see FreeBSD running on a new class of box I've got
here. Not new hardware. These are Atom chips on Micro-ITX motherboards, and
are interesting in that they are low-power and have dual gigabit NICs.
They're UEFI-only.

These boxes seem to not like the FreeBSD 12.1 .iso files as written to USB
sticks, but I could boot the installer with an .img.

That said, the resulting system as installed seems to freeze in precisely 
the same place as the .iso-files-written-to-USB froze. I took a photo of 
the freeze, and then realized that it was the same as when I was trying to 
boot from the USB stick the first time.

Photo attached, along with dmesg output on this hardware from NetBSD and 
Linux (two different systems, identical but for MAC addresses) in case that 
helps.

What's different between the .iso and the .img files, and how might that 
translate to the installed system, if that's not a red herring? And how 
might I get these boxes to boot FreeBSD? 

The boxes don't have build-in storage so I'm installing and booting from 
USB drives, so making modifications from another system to test things 
ought to be fairly straightforward.

Addendum: To try -current in case it was a known issue, I downloaded the 
mini-memstick.img, but it freezes in the same place.
Comment 1 Mason Loring Bliss freebsd_triage 2020-09-10 00:55:48 UTC
Created attachment 217862 [details]
dmesg from linux
Comment 2 Mason Loring Bliss freebsd_triage 2020-09-10 00:56:14 UTC
Created attachment 217863 [details]
dmesg from netbsd
Comment 3 Mikael Urankar freebsd_committer freebsd_triage 2020-09-10 07:44:57 UTC
It looks similar to bug #209821
Comment 4 Ed Maste freebsd_committer freebsd_triage 2021-10-05 13:57:48 UTC
Could you try again with a snapshot from 20210930 or later?
Comment 5 Mason Loring Bliss freebsd_triage 2021-10-05 17:36:42 UTC
I'm exceedingly glad you're asking, as I've got a box here and I was 
already going to do exactly that, only I was going to use 13 again and
explore. Knowing that there's reason to use something newer to test changes
is exciting. I'll have feedback by this evening (EST).
Comment 6 Mason Loring Bliss freebsd_triage 2021-10-05 18:05:29 UTC
For reference:

The board: https://www.aaeon.com/en/p/mini-itx-emb-cv1

Will try stable/13, caaf4ae21e0600, October 1 or later
Comment 7 Mason Loring Bliss freebsd_triage 2021-10-05 23:24:23 UTC
I tried -current as the fix didn't make its way into 12-stable in time for 
the most recent snapshot, but I'll try FreeBSD-13.0-STABLE-amd64-
20210930-94ad8d7c7a3-247474 tonight and set 'copy_staging disable' at the
loader menu.

With -current it booted reliably. I'll report on how it goes with -stable later.
Comment 8 Mason Loring Bliss freebsd_triage 2021-10-05 23:24:45 UTC
And by 12-stable I mean 13-stable, sorry.
Comment 9 Mason Loring Bliss freebsd_triage 2021-10-09 19:31:41 UTC
This has been corrected in 14-current and MFC'd to 13-stable.

14-CURRENT commit: 6032b6ba9596927aba15a8004ade521a593a7d58
13-STABLE MFC: caaf4ae21e0600844aa723f87c57dcff37c27a39

I've tested both on my hardware and both are fine out of the box.

Thank you. Closing this now.