linux-kernel - Kernel hang right before init on some i586 (bug report)

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-Id: <20070607155557.58D96CA0A4@ws5-11.us4.outblaze.com>
Date:	Thu, 7 Jun 2007 17:55:57 +0200
From:	"Lauri Kasanen" <curaga@...ramail.com>
To:	linux-kernel@...r.kernel.org
Subject: Kernel hang right before init on some i586 (bug report)

Hi.

Yesterday I encountered this bug, and after trying everything, thought
I'd let you know. I also googled a lot, and there is no known solution.
So now I tell my findings to the kernel developers.

On some i586 comps linux kernel hangs right after displaying "freeing unused 
kernel memory:". Here it's supposed to start init, but init never gets started.
On one site someone added debug statements to kernel's init.c and made /sbin/init a shell 
script echoing text. Guess what? kernel debug lines printed, that echo did not.

I've made a two-floppy linux system, it boots on every other comp but my old laptop.
After that kernel hang, I recompiled a few times, tried 2.6.16.52, 2.6.21.1 and 
2.6.16.21-ck2 with dynticks. Each had the bug. I also booted with every bootcode 
listed in kernel-parameters.txt as "BUGS=x86-32", and it still hung at that point.
This makes me believe there is no known solution.

Common points:

- always happens at some i586 systems. Mostly laptops, some desktops
- always hangs at the same point
- always prints no errors, and is not a total hang; boot messages can be scrolled
and ctrl-alt-del works
- has happened with both 2.4 and 2.6 series. No reports on 2.2 found on google..
- may be a bug in gcc, kernel or both
- they were mostly built on i686's, but not always
- the kernels were always configured for pentium-classic or lower. So they Should run..

One guy who also had this issue pressed sysrq-t at this bug and got 8 screenfuls of debug output.
Here: http://knierim.org/~justin/debug/

>From many hundred hits on google I found it hangs at execve() call for init.
I'm no programmer, but I think it may be gcc wrongly compiling that to need i686 code,
or a bug in kernel itself. But this doesn't affect all i586s. So go figure..

Here's also a long discussion about this bug:
http://wiki.linuxfromscratch.org/livecd/ticket/1638

My laptop is Toshiba Satellite 200cds. Pentium-classic with F00F bug and 40mb ram.

I hope I could help. Since I have the hardware where this bug occurs, I can test any patch you send me.
The list is very busy, so I haven't subscribed, so please CC me.

Lauri

PS: Linux rules!

-- 
_______________________________________________
Surf the Web in a faster, safer and easier way:
Download Opera 9 at http://www.opera.com

Powered by Outblaze
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/