[BL] LILO challenge
baslinux at lists.ibiblio.org
baslinux at lists.ibiblio.org
Fri May 2 23:04:24 EDT 2008
>>> The problem was fixed by starting PQMAGIC 8.0 (I had used 5.0 the first
>>> time and then BL3 fdisk) which said the extended partition crossed the
>>> 1024 cylinder boundary, which would cause problems, and changed it to
>>> ExtendedX partition, and now all the DOS partitions work -
>The problem recurred (fourth time) and is probably physical.
Yes, I think so.
>> As far as I know, BL3 uses only primary partitions.
>/dev/hda1 through hda9 but I think it needs to boot from a primary
>partition because it was set up that way.
That's what I understand, too.
>>> The extended partition contains all the logical partitions and crosses
>>> the 1024 cylinder boundary (the same one that BL3 lilo cannot handle
>>> unless you copy boot.b and the kernel to the first 1024 cylinders).
>> In DOS it is costumary to make a lot of logical partitions to handle a HD
>> that cannot be fully used by the Pc's BIOS.
>> This means if one has a 20gb HD but the system "sees" only 4gb for
>> someone needs to make a 4gb primary partition and 4 (or more) extended or
>> logical partitions up-to-4gb each.
>I don't want 25GB partitions even in linux. They take longer to check.
>I put several operating systems on one drive - DOS, BL3, BL2.
>> PC BIOSes provide booting directly from IDE0, the first floppy driver,
>> CD/DVD rom, or a master HD on IDE1, and even a pendrive on USB, but not
>> from a SLAVE HD, whatever the IDE can be.
>Some newer computers allow booting from hdd-0 hdd-1 hdd-2 hdd-3. I can
>switch between booting XP and DOS from slave (hdd-1) and master (hdd-0).
>LS120 ZIP100 boot too.
Very good! (I am not up to date).
>> Linux allows redirecting the boot after booting from a "legal" device.
>With loadlin? Lilo?
loadlin, I suppose.
>>> I can try it as a slave, with DOS on a 2GB master, but that is said to
>>> slow down the larger drive.
>> It really slows down if the two HDs are of different architectures and
>> installed on the same IDE.
>2GB UDMA-2 and 100GB UDMA-6 on one cable both ran at normal speeds (the
>board can do UDMA-5 in linux with correct driver, and UDMA-4 in DOS).
>But the 100GB as slave would not boot linux. Linux stopped and complained
>it could not find the partition it was on. As master, it worked fine (I
>edited fstab between a and b).
Why don't you try formatting the first partition after 1024 or 2048 bytes
from the beginning?
In theory it is possible to do so. It's just an experiment, but maybe it
>> Normally they are. The system does not assume a medium, but the lower
>On this newer (2003) computer the faster drive did not slow down. I
>tested with hdparm and spinrite. File transfer between them would of
>course be at the slower speed.
What really slows down is if one of the HDs is capable of LBA and the other
doesn't (very old HDs).
Or in case of other differences like this.
>> It really doesn't concern to the size itself, but to the architecture.
>> But you can try it as follows:
>> Master on IDE0
>> CD drive (master) on IDE1
>> HD (slave) on IDE1
>My 100GB won't work as slave. At least not on IDE0.
>The drive was given to me as secondary slave (with XP booting from primary
>slave - the jumpers were all mixed up, one as CS, one on some jumper that
>was not even listed.)
A 100GB HD is not set to "slave" without a good reason.
Maybe it was defective a long time before the donation, and that was an
effort to avoid discarding it.
>> This way it's supposed not to slow down the CD neither the HDs.
>I will live with a floppy disk boot for a while and then maybe try this as
>hdc with a different hda that boots. I don't know if the CD would slow
>down the hard drive.
CDs have a hibrid structure (IDE / SCSI) and this way they can slow down a
high performance HD installed on the same IDE cable.
More information about the BasLinux