Hi,

I've managed to install LFS on an external usbdrive.
Everything works ok, as long as the PC used, has only linux
partitions installed. If the partition is of another type,
booting will fail and comes with:

.
.
Either the superblock or the partition table is likely to be corrupt!

The only option the system gives to me is to shutdown it, no
repair shell available.

I've used a test PC with an empty harddisk (hda) and created
an empty linux partition , with fdisk. > booting went fine.
Creating a partition of another type causes the failure (tried
with 0x0c,0x00,0x07).

I also tried it with my laptop, having linux ext3,swap,fat32 and NTFS.
but then the system hangs at checking my internal drive.

hda1,hda2,hda3 <hang>

Does anyone have a suggestion?
Maybe I did something wrong while compiling the kernel?

TIA,

Gearloose

You might see if there is a boot time parameter for the kernel that relaxes these tests. If there is and it allows you to get past this, it might give you a clue what is happening. I can never remember where the complete list of boot time kernel parameters is, so you may have to google for it.

Thanks for your reply.

I've searched through the kernel documentation about kernel parameters but
couldn't find anything usefull
The only timing parameter I noticed was the one I allready use: rootdelay
without rootdelay, I'm not able to boot from usb.

I also digged through the kernel config , but I couldn't find anything there.
Couldn't it be something in the init?

I was focusing on your laptop results since that is where you have the most information..

As I understand things, the kernel itself does a partition check. I just checked a system of mine with DSL-N running a 2.6.12 kernel, and the output from dmesg was:

(The above was hand entered as it would have taken more work than normal to copy & paste; don't ask about the drive letter -- I didn't build the system. ) My interpretation of your post was that this was what you were seeing. So I don't think your problem could be init. (Feel free to find a flaw in my logic.)

I'm sorry. I don't think I have enough knowledge to suggest anything else. With my limited knowledge, I wouldn't rule out a kernel config issue, but I don't know what it would be. TMK, you don't have to have the drivers for all existing partitions to boot -- particularly since these are frequently handled (if at all) by modules loaded later in the process.

The reason why I was thinking it is the init, was that I thought that the kernel was loaded at
that stage, but I can also be wrong .
It's OK to focus on the laptop, however I think it is the same problem with different result.
The only partition where the system has to worry about is the ext3 partition with / on it.
The other partitions may be ignored, but that doesn't seem to happen.

If you click on the "Report" button, you can request that the moderator move your thread to the LFS subforum where it might get more germane responses.

Ah, I didn't know that..
However I don't know if this really related to LFS, since it can occur with every other distribution, but I will follow your advise.

Silly me, I just looked at fstab and it seemed to be corrupt.
I've created a new and correct fstab and tested it with the test PC.
Booting went fine and I could login!

Next , I tried the disk with my laptop, but with no luck
The error with "hda1 hda2 hda3 <" still appears.
Tomorrow I wil test the disk at work, if it will function there
then I suggest it would be a failure in the BIOS of my laptop.
It was one of the first systems with USB functionality.

I have moved this to the LFS forum - if it turns out to be more general, feel free to report the thread again and it can be moved elsewhere.

If tested it and also with no luck

The last piece of the :

hdc: samsung cd-rom .......
hdc: ATAPI 48X CD-ROM drive, 128kB Cache
Uniform CD-ROM driver Revision: 3.20
ide-floppy drive 0.99.newide
hda: 8032MB, CHS=1024/255/63
hdb: 10MB, CHS=306/4/17
hda: hda1 hda2
hdb:_ <HANG>

There is only one harddisk (hda?) about 28 GB in size, with 8 GB used. The filesystem is NTFS. So the message about hda=8 GB is
a little bit strange, maybe that's why hda2 and hdb appears.

I used to have a Fedora Core 6 installation on an external USB drive
which worked alright with this system, so the system would be able
to run linux at an external usb drive

A few posts back, I was discussing with blackhole54 if it is the kernel or init.
After fixing fstab, the test pc was able to boot, and it seemed
that problem was before init, so the problem was at kernel stage.

Since the lost Fedora installation had worked with my PC at work,
I was thinking about what could be the difference.

The Fedora installation was kernel 2.6.18-1.2798,
the LFS installation has kernel 2.6.16.27.
I also have the LFS-livecd which is able to boot with my laptop
without any problems, it has kernel 2.6.16.28.
The number of releases between 2.6.16.28 and 2.6.16.27 are
less than between 2.6.16.27 and 2.6.18-1.2798 so I studied
at the changelog of 2.6.16.28.

The only interesting part was this:

commit 3cfd644882784276280c07ad4b9eb086fa44222f
tree 5977cf96450e94c70e6d607096b92e8e427bafc7
parent d2b61ec602e7d7be485725bcda8f3b847cfc380e
author Al Boldi <a1426z@gawab.com> Fri, 18 Aug 2006 21:30:53 +0200
committer Adrian Bunk <bunk@stusta.de> Fri, 18 Aug 2006 21:30:53 +0200

ide-io: increase timeout value to allow for slave wakeup

During an STR resume cycle, the ide master disk times-out when there is
also a slave present (especially CD). Increasing the timeout in ide-io
from 10,000 to 100,000 fixes this problem.

Acked-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Adrian Bunk <bunk@stusta.de>


I don't know if this could be related to my problem.

Maybe I am misssing something here, but I concluded just the opposite. TMK the kernel must have booted and turned control over to init before fstab is even consulted. The only part of this where my knowledge is shaky (TMK! ) is how udev interacts with all of this. But I still think it has to be after init has taken over.

I believe STR = Suspend to Ram. So at first blush this wouldn't apply to you. But I suppose it could fix something that Al Boldi was even was aware was a problem. Have you checked out what the symptom would be for a timeout in ide-io? Specifically, would it print an error message?

=========

I noticed two posts ago that when you had the mysterious 8GB figure reported for your drive that the reported figure did in fact match the cylinders/heads/sectors data reported. In case you hadn't noticed that, I'm throwing it out in case that gives you some kind of logic thread you can follow. I know 8GB was once some kind of limit, but I couldn't make that logic go any place. I wondered if there was a kernel config option concerning LBA or something that wasn't set right. But I didn't think that kicked in until 32GB. This rambling is just thinking out loud in case anything here helps you think of something.

Some older drive have a jumper setting to limit the drive to 8Gb. Perhaps he should check his drive jumper settings.

It was that time-out that had my attention. I can imagine that if the disk has a timeout , the process would freeze, however I forgot that it was about str.
I can take a look at the kernelconfig once again, but actually I have all options enabled, the most as modules.
Only USB,SCSI and ext3 is build in, since I need that to boot from USB, it saves me an initrd.
I will search about the LBA limit, you mentioned. It's a little bit strange, I've tested the live cd at work
and that give any problems, so what could be different except for the kernel version and the including options.

Thanks for thinking with me

Unfortunately, I cannot change the jumpers
I have no fysical access to the DELL system, the other system is my laptop.