RE: linux kernel 2.4.10 possibly breaks LILO

RE: linux kernel 2.4.10 possibly breaks LILO

[Rui Ribeiro]

I'm using LILO 21.6.1 w/ 2.4.10 and Reiserfs in my two notebooks: a Compaq Armada 1500c and a Compaq Armada 100S.

Rui

-----Original Message-----
From:	Petr Titera [SMTP:owl@volny.cz]
Sent:	Terça-feira, 2 de Outubro de 2001 18:55
To:	haiquy@yahoo.com
Cc:	linux-kernel@vger.kernel.org
Subject:	Re: linux kernel 2.4.10 possibly breaks LILO

Possitive here. Just try to change to new kernel image. I was forced to
install grub after upgrading to 2.4.10.

Petr

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

Re: linux kernel 2.4.10 possibly breaks LILO

[Petr Titera]

> I'm using LILO 21.6.1 w/ 2.4.10 and Reiserfs in my two notebooks: a Compaq
Armada 1500c and a Compaq Armada
> 100S.

May be, that it is filesystem dependant (using ext3), but I dont know...
Looks like effect of blkdev-in-pagecache. I saw something similiar in past
when I played with loop-like devices.

Petr
>
> Rui

Re: linux kernel 2.4.10 possibly breaks LILO

[Andreas Dilger]

On Oct 02, 2001  20:10 +0200, Petr Titera wrote:
> May be, that it is filesystem dependant (using ext3), but I dont know...
> Looks like effect of blkdev-in-pagecache. I saw something similiar in past
> when I played with loop-like devices.

Yes, it is blkdev-in-pagecache, and it _should_ be fixed in 2.4.11-pre2.
Please upgrade and give it a try.

Cheers, Andreas
--
Andreas Dilger  \ "If a man ate a pound of pasta and a pound of antipasto,
                 \  would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/               -- Dogbert

RE: linux kernel 2.4.10 possibly breaks LILO

[Rui Ribeiro]

I did have to upgrade LILO to a new version, when I *manually* upgrade from ext2 to Reiserfs. 

Rui

-----Original Message-----
From:	Petr Titera [SMTP:owl@volny.cz]
Sent:	Terça-feira, 2 de Outubro de 2001 19:11
To:	rui.ribeiro@case.pt
Cc:	linux-kernel@vger.kernel.org
Subject:	Re: linux kernel 2.4.10 possibly breaks LILO



> I'm using LILO 21.6.1 w/ 2.4.10 and Reiserfs in my two notebooks: a Compaq
Armada 1500c and a Compaq Armada
> 100S.

May be, that it is filesystem dependant (using ext3), but I dont know...
Looks like effect of blkdev-in-pagecache. I saw something similiar in past
when I played with loop-like devices.

Petr
>
> Rui

Re: linux kernel 2.4.10 possibly breaks LILO

[Petr Titera]

Possitive here. Just try to change to new kernel image. I was forced to
install grub after upgrading to 2.4.10.

Petr

Re: linux kernel 2.4.10 possibly breaks LILO

[David Ford]

Lilo is fine, back down to the kernel image prior to your current one, 
run lilo with your new image, then boot into the new image.

Lilo can now be run just fine.

Something in the pre10 series hosed something which caused the first 
instance of lilo to segfault then the next to deadlock waiting on a 
page.  2.4.10 proper fixed that.  In my case I reverted to 2.4.7-pre6, 
ran lilo, then booted into 2.4.10 and ran lilo.  Worked like a charm. 
 ('course I had a boot disk at the ready)

David

Petr Titera wrote:

>Possitive here. Just try to change to new kernel image. I was forced to
>install grub after upgrading to 2.4.10.
>

Re: linux kernel 2.4.10 possibly breaks LILO

[Jonathan Walther]

-----BEGIN PGP SIGNED MESSAGE-----

Although its not a bad idea to switch to GRUB in any event. I've
found it to be far superior in terms of userfriendliness and power.

Jonathan

On Tue, 2 Oct 2001, David Ford wrote:
> Lilo is fine, back down to the kernel image prior to your current one,
> run lilo with your new image, then boot into the new image.

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3ia
Charset: noconv

iQCVAwUBO7pRnMK9HT/YfGeBAQHY/wQAmLadHS8w3KdU5L5GMda2REchyWHYCMCa
Nqm2WJD9GC493VGzmIlnCK05//k1tDmIkACOFJ1Bnw8FO71dNWYyVQOwOO/Mg4pw
Ydd9xgMOvBlrRkeq0ghPG4X8ZfIf1Zw9IFk+q2wDbK8ZmfNq4dt5GPKIKN1E1jII
l6nA5209ry8=
=Zfex
-----END PGP SIGNATURE-----

Re: linux kernel 2.4.10 possibly breaks LILO

[Dave Cinege]

On Tuesday 02 October 2001 13:55, Petr Titera wrote:
> Possitive here. Just try to change to new kernel image. I was forced to
> install grub after upgrading to 2.4.10.

If this is the case, I'd call that a feature.  ; >

Dave
<Well known absolute block offset bootloader hater>

-- 
The time is now 22:19 (Totalitarian)  -  http://www.ccops.org/clock.html

Re: linux kernel 2.4.10 possibly breaks LILO

[Petr Titera]

>
> If this is the case, I'd call that a feature.  ; >
>
> Dave
> <Well known absolute block offset bootloader hater>

I hate it too now. I even booted from image I deleted. God knows what would
happen if I did not noticed it early :))

>
> --

Petr

linux kernel 2.4.10 possibly breaks LILO

[Ken Zalewski]

Summary:

Modifying /etc/lilo.conf and running "lilo" when using kernel version
2.4.10 does not appear to modify the boot sector accordingly, even
though I receive no errors or warnings.  On next reboot, LILO is
configured as it was previous to my changes.  I am therefore unable to
modify LILO configuration in any way while using 2.4.10.


Machine summary:

AMD Athlon 1 GHz
VIA 686b southbridge
512 MB PC133 SDRAM
/dev/hda = 30 GB Western Digital ATA-100 drive
/dev/hda1 is root partition and active partition
Linux 2.4.10
LILO 21.7-5


Details:

I recently built the latest 2.4.10 kernel from source and installed it
(upgraded from 2.4.7).  Ran "lilo", and the new image booted and I was
happily running 2.4.10.  I then made some changes to /etc/lilo.conf,
re-ran "lilo", and rebooted.  On reboot, the my changes were not
apparent, and the previous configuration was used.

I began experimenting more, making drastic changes to /etc/lilo.conf. 
No matter what I did to lilo.conf (even changing from "linear" to
"lba32" and back), running "lilo" produced no change to the startup.

I used a floppy to boot into 2.4.7, ran "lilo", rebooted, and noticed
that my changes had registered.

I then built kernel version 2.4.9, and did the same experiment.  My
change to /etc/lilo.conf registered fine on the next reboot.

As a result, I have concluded that something has changed between 2.4.9
and 2.4.10 that causes LILO to be unable to overwrite the boot sector of
the active partition.  I also tried using LILO 22.0-beta, and the
problem persisted.

My kernel configurations for both 2.4.9 and 2.4.10 are almost identical.

I do not know if this is something specific to my own machine, but it is
consistently reproducible.  I have used versions 2.4.5, 2.4.7, 2.4.9,
and 2.4.10, and LILO fails to work ONLY on 2.4.10.  No errors are
printed by LILO, and doing "lilo -q -v" before rebooting seems to
indicate that everything happened properly (though I think that only
reads the map file, not the actual boot sector).

I would appreciate a response regarding this issue.  I have not seen
anything posted about this yet, so it could be something specific to my
machine - it's just strange that it just started happening on 2.4.10.

Thanks for any assistance you can provide.

-- 
Ken Zalewski -- Director of Information Technology
Document Development Corporation
1223 Peoples Avenue, Troy, NY 12180
kennyz@nycap.rr.com   http://www.cs.rpi.edu/~kennyz/

Re: linux kernel 2.4.10 possibly breaks LILO

[Andreas Dilger]

On Sep 26, 2001  23:48 -0400, Ken Zalewski wrote:
> Modifying /etc/lilo.conf and running "lilo" when using kernel version
> 2.4.10 does not appear to modify the boot sector accordingly, even
> though I receive no errors or warnings.  On next reboot, LILO is
> configured as it was previous to my changes.  I am therefore unable to
> modify LILO configuration in any way while using 2.4.10.

It is likely the blkdev-in-pagecache change which caused this problem.
If you have a mounted filesystem on the same device as the bootsector
you are writing (e.g. /dev/hda1) and the bootsector shares a block with
the filesystem (e.g. 2kB or 4kB blocks and ext2/ext3) then any changes
made to the bootsector are overwritten as soon as you make any change
to the filesystem.

To test this theory (if possible) you could remount the filesystem
read-only before running LILO.  This would only work if you had a
setup like /dev/hda1 is / and /dev/hdX is /boot, because you still
need to be able to write into /boot.  If /boot is in the root fs, it
is not possible to do this.  If you CAN do this, then you need to
reboot after running LILO.

Cheers, Andreas
--
Andreas Dilger  \ "If a man ate a pound of pasta and a pound of antipasto,
                 \  would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/               -- Dogbert

Re: linux kernel 2.4.10 possibly breaks LILO

[Steve Kieu]

 --- Ken Zalewski <kennyz@nycap.rr.com> wrote: >
Summary:
> 
> Modifying /etc/lilo.conf and running "lilo" when
> using kernel version
> 2.4.10 does not appear to modify the boot sector
.tux.org/lkml/ 

Negative here, I just modify lilo.conf, add the option
prompt; run lilo and reboot. It works fine.

I use 2.4.10 with preemt patch...

I believe the others options will work too.



=====
S.KIEU

http://travel.yahoo.com.au - Yahoo! Travel
- Got Itchy feet? Get inspired!