Hi, when I installed lilo in Gentoo (I eventually changed it to grub) I kept gettiing an error message from lilo that said the following:

I really didn't make much of it at the time, as things seemed to work ok anyway. However after my gentoo system recently crashed and I suffered some file system corruption (another story) I opted to reinstall gentoo with an ext3 file system in preference to reiserfs file sytem - as my experiences of it in the past had been that it was fairly robust. However I have noticed after every 20 reboots or so that ext3 file systems are automatically checked for consistency. Which is where my problem comes in, because every time this file system checking takes place it gets passed the first part of checking the file system for consistancy, but then borks and says something like 'the file system could not be repaired...' Then it asks for the admin password and drops to a bash prompt. Now my guess is that this is because the / file sytem (on hdb3) is mounted - and fsck doesn't like file systems to be mounted before it can run checks. I can get very similar results when I check the drive in Knoppix and leave it mounted. It will complain and refuse to work in more or less exactly the same way. The question is, why is fsck attempting every 20 reboots, to run repairs on a mounted disk? I am certain I have seen this behaviour before (after a hard reboot in RedHat for example) and the system has always seemed very able to carry out any needed repairs to the file system without any of these problems. What I wonder is going on? Surely there must be a way to fix it so that the file sytem can be checked, repaired and then mounted without these strange complaints? That is of course assuming that this is all that is causing the problem. Also why is lilo reporting these weird problems? What does it mean? It seems my bios is reporting the wrong disk geometry to lilo (and potentially now grub and the system as a whole hence the apparent need for regular active file system repairs) where as my bios seems to be telling the system that my harddrive is 1024 cylanders in length and the system is seeing it as 59131 cylanders. Obviously it is the latter which is accurate. But this still doesn't answer why this would happen, or give me any clues at all about how to fix it.

So if anyone has any suggestions at all or can make sense of anything above, or have encountered similar problems in the past, please let me know your views. Your input would be extremely welcome.

Thanks in advance.

Q

Oh, BTW just to clarify, I have tried this on 4 seperate, very modern fairly large discs recently, so it isn't the disk that is faulty. Just to clarify this though, here is the output from my current disk fdrom the gentoo live CD.

Which looks fine to me. Anyway, I hope this helps.

Q

Try the jumper settings on your hard drive...some large drives have a jumper setting that limit the number of cylinders reported to the BIOS...this is a feature that is designed for older BIOS versions that cannt recognize the full capacity of a modern large HD. This may not be set, or is set and Linux is trying to tell you it's lying...this is just a guess, and don't forget that once you change those jumpers or any HD geometry settings, you WILL have to re-format and install...

Hope this helps...but it sounds like you already have the problem already almost solved.

I think I know what your getting at... There are some jumbers on some of my drives that can limit them to 32GB - presumably for fat32. But I have always taken care to remove any jumpers that limited the capacity of my disks in any way. (That is to say I always simply set my drives up as masters). In any case it has always worked in the 8 years I have used computers. It is only with this one specific rig that I have this problem. Gentoo and Grub are very happy on all the others. (Baring in mind that drives have been swapped and reformated/reused between these systems. So if it works in one, why not the other?)

I don't know about being closer to an answer, I don't even know if my interpritation is correct, since so far nothing has been resolved. I still have to mount the disk manually in Knoppix or Gentoo Live CD every 20 reboots to run fsck -f.

Normally you might think this is not such a big problem, but it is 'why' that peturbs me. I don't like not knowing and I smell something very odd going on. It is quite frustrating.

Anyway thanks for the input. If anyone has anything futher they feel may be of use, please feel free to post your comment.

Q

Linux can run for years w/o a reboot. Don't reboot so offten.

Lol yeah, I thought someone would say that. But when you're fixing problems, especially problems with a FS or a bios glitch just after an install, it can't really be avoided. In any case I'm afraid that's not a fix. Just not turning your computer off doesn't make the problem go away. It's still there and while its still there it's still in my mind, niggling at me.
Also I should make it clear, I do reboot my computer. Partly to save energy, partly because I can't stand the noise as it's in a room which is also my bedroom, partly because well what the hey, if it's not doing anything, what use is it being on in the first place?

I know lots of linux users are in love with the idea of uptime, but personally I prefer a somewhat more pragmatic approach. My 'uptime' as it were, is probably better spent elsewhere.

In any case if anybody can give a direct interpitation of what this means and it's likely causes, I would be very keen to hear it.

Thanks again.

Q

"It is only with this one specific rig that I have this problem."

Gotta be in the jumpersettings or BIOS.

Just 4fun, trying setting the HD geometry to the settings that fn 48 gives you (fn 48: 59131 cylinders, 16 heads, 255 sectors) and see if the problem persists or disappears. I went back and looked at your original post...it still reminds of some issues I had a while ago with translation BIOS settings, and I had to set the geometry manually to get past the issue....

I don't know if I can, it's a Gigabyte KT400-A board, and the bios options on Gigabyte boards suck.

Anyway, I'm caught in a multi gigabyte download atm and it can't be resumed, so as soon as it finishes I will take a look and report back on what I find. That particular drive is no longer in my machine and I will have to see if I can interprit your instructions for my larger WD 120Gig drive.

Thanks very much for the input.

Q

Ok, scratch that...read this, taken from the Western Digital support site:

"Why all EIDE drive specifications for drives 8.4GB and larger list the same cylinder, head, and sector information:
Drives 8.4 GB and larger all use the same parameters: 16383 cylinders, 16 heads, and 63 sectors. This is due to a limitation designed into the original IDE interface. Drives identify themselves to the system BIOS using the "Identify Drive Data" bit. The original design of the IDE interface allowed for values up to 16383 x 16 x 63 (around 8.4 GB) to be used in this bit. In order for drives to identify themselves as larger than 8.4 GB, a new method was needed. This new method was added on to the Identify Drive Data bit, and is known as the "Extended Functions."

The extended functions communicate drive size in a straightforward manner, simply telling the BIOS how many sectors are available on the drive (the capacity can then be calculated by multiplying the number of sectors times 512 bytes). Once the extended functions are used, the cylinders, heads, and sectors on the drive are no longer important. All drive manufacturers agreed that drives larger than 8.4 GB would report the maximum number of cylinders, heads and sectors, which is 16383 x 16 x 63. The actual size of the drive must be determined by the extended functions. Some manufacturers may list larger parameters with their drives, but these drives will still only be able to report 16383 x 16 x 63 in the Identify Drive Data.

In order to make use of drives larger than 8.4 GB, both the system BIOS and the operating system must be able to interpret extended functions."

More stuff from the WD site:

"When a drive is not detected by the system BIOS, there are several possible causes – a physical configuration problem, BIOS limitation, or defective hardware.

Configuration Problems

The BIOS will not detect your drive if the cable connection is not correct. If you are using an Ultra ATA cable (80-conductor) make sure that the Master or Single drive is on the black end connector. Connect any Slave device present to the gray middle connector. Finally, the blue end connector should be connected to the motherboard or EIDE controller card. If you are using a standard 40-conductor data cable, connect the Master device to one end, the Slave to the middle and the other end to the motherboard. In all cases, make sure that pin 1 on the data cable is aligned with pin 1 on the motherboard and pin 1 on the drive. For Western Digital drives, pin one is located closest to the power cable.


Incorrect Jumper Settings will also prevent the BIOS from detecting your drive. The most common incorrect jumper setting used is for a drive that is alone on the data cable. Keep in mind that the concept of Master/Slave applies to a given EIDE channel. If there are two drives on the data cable, one must be Master and one designated as Slave. If the drive is the only device on the cable, it should be jumpered as a Single drive, NOT a Master. To do this, place a jumper shunt on pins 4 & 6, or remove the jumper shunt completely to set the drive as a Single drive."

And more good stuff:

"Drives larger than 8.4GB
Drives larger than 8.4GB use the same logical cylinder, head, and sector values (16383, 16, and 63 respectively). Setting values greater than these in the BIOS WILL NOT allow a larger drive to work in a system that does not have extended interrupt 13 support. For more information, see Answer ID 240."

I know...some of this stuff you are reading is like, "duh!" I know this already, but the settings and the extended int 13 info might be of service to you....OK, now i will just shut up...:~)

No, that is al very.... informational... thanks.

I'm not sure it explains what's happening here, since I have two other IBM Deskstar 120GB drives (well one's Hittachi who I think bought out IBM's harddrive business) and they all experience the same problem in this machine. But it will certainly help me undestand what I'm doing a little better when I'm looking in the bios. Now as soon as this damn downloand finishes, I'll take a look and let you know.As I said, thanks a lot for the input, it is very much appreciated.

Q