Baslinux mailing list

[baslinux AT lists.ibiblio.org]

>
>>> I zeroed out the entire drive with a Seagate/Maxtor hard disk utility on
>>> the Ultimate Boot CD, and repartitioned with BL3 fdisk and mke2fs and
DOS
>>> format.  sys'ed it.  It still acts the same as when I used PQMAGIC.
>>> Won't boot (I can boot to floppy disk and cd to it) on two computers.
>>> I think yesterday it was booting on the third computer (but D: and E:
were
>>> full of gibberish).
>>
>> These are sympthoms of bad zero tracks!

>Bad physical partition table area?

Yes! The boot sector is at track #0

>After my third repartition and reformat, PQMagic 8 is complaining again
>about drive geometry so I gave up on repartitioning it again.  It works as
>a non-booting master, but if I make it a slave, linux cannot find any of
>the linux partitions. (DOS works).  So I will boot from floppy disk (or it
>could be secondary master - on this computer I cannot boot from anything
>except primary master so I can't boot from primary slave and use this as
>non-booting master).

>>> Linux won't mount the two logical DOS partitions that fdisk made.  If I
>>> copy files to those partitions via DOS laplink pro, they turn into
>>> gibberish.  The programs in C: work so I will enlarge C: and put all my
>>> DOS programs in it.
>>
>>> Linux works fine once I boot to it from DOS.  LILO stops at LI
>>> when booting from lilo on floppy disk.  I doubt it would work from MBR.
>>
My 486 IBM does is not booting from floppy. It is giving me a 620 error.
I don't know exactly what it means and I tried another floppy drive with the
same results.
But it boots normally from HD. I have a BL3 DOS-version there, running in a
DOS loop.

>>> The solution is to boot from DOS floppy disk to DOS, and then with
loadlin
>>> to linux.  I can't use this drive in the ECS computer where loadlin
stops
>>> the linux keyboard from working.  Is there some other way to boot from
DOS
>>> to linux?

I was also thinking on loadlin. or using the BL3 DOS-version, which runs in
a loop.
But it's not the correct option for a powerful PC like yours.

>>
>> ECS manufactures  the same Pc Chips motherboards, with other names.
>> Many Pc Chips MBs are known for conflicting with Linux somehow.

>SiS motherboard.  An identical ECS board works with the keyboard but not
>with a scanner (it used to work) and the USB went bad.  So I think
>something broke on this board and it is not a built-in conflict.

Sis is the chipset brand, not the motherboard itself.

>>
>>> Any idea why linux behaves so much better than DOS in logical
partitions?

No, I just was told it.
BL3 uses EXT2 file system, actually considered "less efficient" than EXT3 ou
ReiserFS, but We'd better have in mind that EXT2 was used successfully for
many years on every GNU/Linux distros.

>>> Sindi
>>
>> At this point I can affirm that your HDs surely have bad "zero tracks",
>> which are not fully damaged but won't boot.
>> However they can be useful for a second drive (for data storage).

>zero tracks are where it boots from and keeps partition tables?

Zero track is the outer track on the first cylinder.
Every time the head lands on it, so the possibility of damage there is much
higher than on other tracks.
MBR and the partition tables (or FAT, in DOS) are located there.

>I might try it as secondary master with a primary master.  Experiments
>show that mixing two speeds of drive do not slow down the faster one, but
>sharing a cable with CD-ROM drive might.  I will test that.

>> DOS is extremely zero-track dependable because it keeps the system files
>> strictly there, while Linux doesn't.

>dependent or dependable?  Do you think lilo would boot from this hard
>disk?  From floppy disk it stops as LI.

Hmmm, I've been trapped by a word!
Sorry, I'm not an English native writer.
I mean DOS depends on zero-track, since one cannot place that system
elsewhere.
I'm not very sure of this, but I think since fdisk can start a Linux
partition from any given location,  so it would be possible to boot from
there, wherever it could be.
It is possible to put hda1 on the last sectors of the HD and hda5 in the
beginning (just for example).
What I don't know is if I still depend on the track #0, since I can place
the boot sector in a partition instead of the MBR.

>>>>> Fdisk. cfdisk, qparted, etc. are unable to handle this kind of
viruses.
>> ... ... ...
>>>>> (edited for short)
>>>>>
>>
>>> I used the Maxtor utility.  I think the partition table area is worn
out.
>>
>> That's why I say your HD have bad zero tracks.

>We agree then.  Better than a mysterious virus.  People give us lots of
>worn out hard disks.

>>>>
>>>> To erase (overwrite with zeros) the MBR I can also
>>>> dd if=/dev/zero  of=/dev/hda  bs=512  count=1
>>>> Would this be just as good?  Can I then repartition and reformat?
>>>> Should I overwrite more than this in case the virus is hiding some
place
>>>> else?
>>>>
>>
>> I do believe this will work fine and makes the same debug does (and its
more
>> concise).

>Easier to remember.

>> The focus is the master boot record.
>> Sparse bad sectors on other tracks don't affect the booting.

>I think the Maxtor utility that wrote zeros also stopped using any bad
>sectors.  It may have been the same as low-level format.

Low-level formatting IS really dangerous to the HD, but it differs from
logical formatting (that's what we do).

>>> I found a site warning against using FDISK/MBR, which can cause more
>>> problems than it cures.
>>>
>>
>> FDISK/MBR is a poor solution, and has to be done from a floppy known as
>> "free of viruses".
>> If you aren't very sure of this, really don't use it.
>> This option just replaces the HD's MBR with  the floppy disk's one.

>Zeroing destroys the MBR and you replace it by partitioning?

Yes, that's it!
A virus partition can  manipulate the MBR information and trick the
partitioner, making us to think the disk is defective.
Since MBR is very small, a partition virus has to be even smaller.

>>
>>>
>>> Would spinrite be adequate to determine if the drive is simply worn out?
>>>
>>
>> I'm not sure. I've never used it.

>Maxtor diagnostic test found no problems.

>>>
>> Since zero-track is much more sensitive to failures, I think HD
>> manufacturers should provide some kind of adjustment so that track zero
>> could be "pulled" to another location, but unhappily this doesn't happen.
>> Naturally this would change the geography of the HD.

>We were talking about that.  Maybe you could flash the BIOS to change the
>location of MBR?  This way they sell more drives.

I believe it won't work so, because the Hd has its own firmware.
And yes, they are not interested on selling less units.

>Should I expect anything else on this drive to fail?  In the past few
>months I lost several drives which I why I am using this one.  One of them
>left scorch marks that we had to clean off the controller pins.

Normally when a HD starts failing it doesn't live any longer.
The magnetic painting on the discs (cylinders) is too weak.
There are some paths for doing some fixing, but there are no warranties of a
perfect working from there on.
When a HD becomes noisy it's announcing its death to come.

>Sindi
>>>>
 Teixeira
>>
>>