linux-kernel - Re: 2.6.22-git17 boot failure

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

Date:	Mon, 23 Jul 2007 12:25:01 +0200
From:	Tilman Schmidt <tilman@...p.cc>
To:	Jeremy Fitzhardinge <jeremy@...p.org>
CC:	LKML <linux-kernel@...r.kernel.org>, andi@...stfloor.org,
	nix.or.die@...glemail.com
Subject: Re: 2.6.22-git17 boot failure

Jeremy Fitzhardinge schrieb:
> Tilman Schmidt wrote:
>>   
>>> On my test machine with a Pentium D 940 processor, Intel DQ965GF
>>> mainboard and SATA disks, kernel 2.6.22-git17 (32 bit build) fails
>>> to come up because it cannot find the disk drives. Same issue
>>> with a clone of the 'linus' git tree. [...]
>> Turns out it wasn't a good idea to answer 'Y' to
>> the new XEN config option during "make oldconfig". After changing
>> it to CONFIG_XEN=n, all's well again. Sorry for the noise.
> 
> That seems very unlikely.  If you're booting a Xen-enabled kernel on
> bare hardware, the Xen code should have no effect at all.  It certainly
> shouldn't prevent you from probing devices.
> 
> Could you send me your .config and dmesg output?

Attached are .config and dmesg (as captured in SuSE's /var/log/boot.msg)
of a CONFIG_XEN=n git-current build. The .config diff of the previous
kernel (which didn't detect the SATA disks) and this one is:

ts@...on:~/kernel/linux-2.6> diff .config.old .config
4c4
< # Mon Jul 23 01:51:16 2007
---
> # Mon Jul 23 02:06:38 2007
129c129
< CONFIG_XEN=y
---
> # CONFIG_XEN is not set
762d761
< CONFIG_XEN_BLKDEV_FRONTEND=m
1105d1103
< CONFIG_XEN_NETDEV_FRONTEND=m
1377d1374
< CONFIG_HVC_XEN=y
ts@...on:~/kernel/linux-2.6>

The .config I used for the -git17 kernel differs from the attached one
in two respects:
- It has CONFIG_MPENTIUM4=y instead of CONFIG_MCORE2=y following
  the answers to my recent question about Pentium D.
- It has CONFIG_IDE=m and most of the actual IDE drivers compiled
  as modules even though they are not used on that system. I
  disabled these for the git-current test because I suspected they
  might be the cause of the problem, but that didn't change anything.
The working/nonworking diff is the same as above.

>  Are you saying that
> leaving PARAVIRT enabled but XEN disabled will allow your system to
> boot?

Exactly.

> What happens if you enable VMI and/or LGUEST?

As you can see in the attached .config, both are enabled already,
without doing any harm.

Thanks,
Tilman

PS: I see that 2.6.23-rc1 is out. I'll try to reproduce the behaviour
with that this evening.

-- 
Tilman Schmidt                    E-Mail: tilman@...p.cc
Bonn, Germany
Diese Nachricht besteht zu 100% aus wiederverwerteten Bits.
Ungeöffnet mindestens haltbar bis: (siehe Rückseite)

View attachment "working.config" of type "text/plain" (55355 bytes)

View attachment "dmesg.out" of type "text/plain" (41696 bytes)

Download attachment "signature.asc" of type "application/pgp-signature" (251 bytes)