> On Wednesday 17 April 2002 18:04, Narfi wrote: > > Right now, I get the following reports about the (c,h,s) triple: > > DiskDrake: (4865, 255, 63) > > fdisk: (4865, 255, 63) > > bios: (19158, 16, 255) > > ... > > Not only do these not agree, but none of them are the same as fdisk > > reported on the old motherboard: > > fisk on old mobo: (77545, 16, 63) > > > > Motherboard: MSI K7T266 Pro2. Bios: AMI version 3.4 > > Hard drive: WD400-BB (Yes, I know it's trash, I bought it before I > > knew) ... > > # cat /etc/lilo.conf > > default=linux > > boot=/dev/hde > > ... > > disk=/dev/hda bios=0x81 > > disk=/dev/hde bios=0x80 > > We share the same Mobo, and I found something similar resulted. > In my case I had a 40gig Maxtor hard drive, where all the different > means of defining hard drive size, came up with different answers, > no matter which partition tools were employed. I asked the hard drive > manufacturers, about this problem, and to cut a long story short, it > turns out there are more than one way of actually measuring hard drive > sizes, which I did not know. It all depends upon which formular is > employed. This can mean that bioses can vary the result In the end I > elected to use one partition tool, PQ Partition Magic, to do all the > partition work,on all my hard drives, this way at least , the size came > out equal no matter which means of listing the partition table you chose > to work with, that is the old dos fdisk, or linux fdisk , and there were > no missing bits between the partitions, that somehow get created , and > cyliners head and sectors get rounded off. Yes, but one would like consistency in these numbers and that the partition boundaries match with the (c,h,s) triple that is used. I have read that the only place where linux actually uses the (c,h,s) addressing instead of lba32 addressing are in lilo and fdisk (diskdrake as well?), so I'm not overly concerned since I'm not sharing the hard drive with any other OS. If I were sharing the hard drive with windows, which only uses (c,h,s), I would be concerned since windows might think it was writing inside a fat32 partition boundary and write in an area that linux thought was ext2 area.
I think I'll end up with adding lines to lilo.conf stating that the (c,h,s) count should be what fisk on the old mobo thought it was, i.e. (77545, 16, 63). Unless somebody on this list tells me that it wouldn't be wise to do so, of course. > I also notice your lilo.conf has additional stanzas:- > ignore-table > disk=/dev/hda bios=0x81 > disk=/dev/hde bios=0x80 > I would ask what these entries are there for ? Sure, I added the bios-lines this morning and I was finally able to boot from hard drive as opposed to only from floppy :-) The ignore-table line came with the installation, I have to look closer at that line and why the installer put it there. If the bios-lines are not present, LILO tries to guess the code that the bios uses for the hard drives. In my case, lilo failed in the first stage of booting and it was due to incorrect guesses. In short, the whole bios<->lilo communication/guessing was a little bit messed up and I had to correct the situation by hand. I'm not the only one, I had already guessed this to be the case when I found an message on the MSI message board where somebody had to do the same thing. I don't have raid controllers on my mobo, I have an extra IDE controller card in an PCI slot. I've heard so many silly things about these RAID controllers that I didn't want them. However, I would have liked the USB 2.0 controller but I couldn't find the k7t266 pro2-U version for sale any where, only the plain pro2 or the pro2-RU. At some point, I'll perhaps add linux software raid to my system, just for the fun of it but I don't have the time for it right now. Narfi.
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
