On Tue, 25 Apr 2000, Raghavendra R wrote:
Hi,
My BIOS recognises the 17 GB hard drive as one with 33416 cylinders, 16 heads and 63 sectors. So, this comes to approx. 512 MB for the 1024 cylinder limit. But, i was able to install linux 6.3 after the 4 GB partition (i,e from E:, C: and D: each taking 2 GB) and the linux kernel reported the hard drive configuration as one with 2096 cylinders, 255 heads and 63 sectors. So, according to the kernel the 1024 cylinder limit falls at 8 GB. Is it true?? I read in one of the HOWTO docs. (i think it is Large-Disk), that the specs. for large disk (i,e above 8 GB) will be reported as C/H/S = 16383/16/63. I think there is some kind of translation between the kernel and the BIOS.
It looks like your BIOS is recognising a "large" configuration, while Linux is recognising an "lba" configuration. LBA is much more friendly to old operating systems, because the combination of BIOS and IDE allows 1024/255/63 which is 8 meg per cylinder - over 8 gig within the 1024 cylinders. Software which thinks it is talking to old-style BIOS (e.g. the FAT16 file system in Win95, or lilo) can thus use the first 8 gig of the disk, rather than the first 512M. LBA must be supported by the drive *and* by whatever BIOS-level software is being used on the computer, in order to be used. This is because the drive actually does the translation, but only if the BIOS-level software gives it permission to do so. Most modern BIOSes support it, as does the BIOS-replacement code in the Linux kernel. I think the Windows NT BIOS-replacement code does too. (Any true, stable multitasking OS *must* replace the BIOS disk IO routines, because they aren't reliable in multitasking.) Lilo uses the BIOS, not replacement code. For situations where the BIOS does not recognise LBA but the Linux kernel does - and therefore the geometries do not agree - lilo internally will translate between either LBA or whatever the BIOS says, and a special internal mode called "linear" - one head, one cylinder, up to 16-million-plus sectors per track. Any given sector will always get the same address in this system, no matter how the geometries differ. You enable this by putting the word "linear" (without quotes) in the global section of lilo.conf. A "large" format is what the drive is actually doing, but is not compatible with old-style BIOSes and software which expects to talk to them. (What idiots picked the BIOS specs in the first place? They put a great deal of effort into making sure that an absolute sector address could fit in 3 bytes, on a 16-bit processor. They didn't bother to check the IDE specs. If they had simply been lazy, the BIOS would support 65,536 cylinders, 256 heads, and 256 sectors per track - 2 terabytes, with 512-byte sectors. Instead we get some restrictions that are entirely different from those of IDE, causing all sorts of complications: a celing at 512M, and another at 8-point-something gig, in addition to whatever the limit of IDE is and the limits of the actual drive in consideration at the moment.) (But, the history of standards can be interesting. The standard for the distance between railroad tracks, for example, is what it is because of a couple of horse's asses. Literally.) -- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/Support/Doku/FAQ/
Hi, Thanks for the extensive and very helpful information regarding hard drives and partition. I went ahead and created a primary partition of 990 Megs (C:) with FAT16 type and created an extended partition for the rest of the disk wherein i created another partition of 990 Megs (D: FAT16) and installed Win 95 on the primary partition (C:). I booted thro' Linux Yast and parititioned the extended drive into 3 logical partitions (after D: ) of 4 Gigs, 3 Gigs and 256 Megs for swap. I installed Linux 6.3 on the root paritition of 4 Gigs. I was able to boot from the CD later. I also installed LILO on the root partition of the Linux partition ( 4 gigs). Now, whenever i boot the system, it goes directly to Win 95. How can i make LILO as the boot manager?? What is the procedure for this?? Raghav Warrl wrote:
On Tue, 25 Apr 2000, Raghavendra R wrote:
Hi,
My BIOS recognises the 17 GB hard drive as one with 33416 cylinders, 16 heads and 63 sectors. So, this comes to approx. 512 MB for the 1024 cylinder limit. But, i was able to install linux 6.3 after the 4 GB partition (i,e from E:, C: and D: each taking 2 GB) and the linux kernel reported the hard drive configuration as one with 2096 cylinders, 255 heads and 63 sectors. So, according to the kernel the 1024 cylinder limit falls at 8 GB. Is it true?? I read in one of the HOWTO docs. (i think it is Large-Disk), that the specs. for large disk (i,e above 8 GB) will be reported as C/H/S = 16383/16/63. I think there is some kind of translation between the kernel and the BIOS.
It looks like your BIOS is recognising a "large" configuration, while Linux is recognising an "lba" configuration.
LBA is much more friendly to old operating systems, because the combination of BIOS and IDE allows 1024/255/63 which is 8 meg per cylinder - over 8 gig within the 1024 cylinders. Software which thinks it is talking to old-style BIOS (e.g. the FAT16 file system in Win95, or lilo) can thus use the first 8 gig of the disk, rather than the first 512M.
LBA must be supported by the drive *and* by whatever BIOS-level software is being used on the computer, in order to be used. This is because the drive actually does the translation, but only if the BIOS-level software gives it permission to do so. Most modern BIOSes support it, as does the BIOS-replacement code in the Linux kernel. I think the Windows NT BIOS-replacement code does too.
(Any true, stable multitasking OS *must* replace the BIOS disk IO routines, because they aren't reliable in multitasking.)
Lilo uses the BIOS, not replacement code. For situations where the BIOS does not recognise LBA but the Linux kernel does - and therefore the geometries do not agree - lilo internally will translate between either LBA or whatever the BIOS says, and a special internal mode called "linear" - one head, one cylinder, up to 16-million-plus sectors per track. Any given sector will always get the same address in this system, no matter how the geometries differ. You enable this by putting the word "linear" (without quotes) in the global section of lilo.conf.
A "large" format is what the drive is actually doing, but is not compatible with old-style BIOSes and software which expects to talk to them.
(What idiots picked the BIOS specs in the first place? They put a great deal of effort into making sure that an absolute sector address could fit in 3 bytes, on a 16-bit processor. They didn't bother to check the IDE specs. If they had simply been lazy, the BIOS would support 65,536 cylinders, 256 heads, and 256 sectors per track - 2 terabytes, with 512-byte sectors. Instead we get some restrictions that are entirely different from those of IDE, causing all sorts of complications: a celing at 512M, and another at 8-point-something gig, in addition to whatever the limit of IDE is and the limits of the actual drive in consideration at the moment.)
(But, the history of standards can be interesting. The standard for the distance between railroad tracks, for example, is what it is because of a couple of horse's asses. Literally.)
-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/Support/Doku/FAQ/
Raghavendra R wrote:
Hi,
Thanks for the extensive and very helpful information regarding hard drives and partition. I went ahead and created a primary partition of 990 Megs (C:) with FAT16 type and created an
/dev/hda1 (990 MB, FAT16)
extended partition for the rest of the disk wherein i created another partition of 990 Megs (D: FAT16) and installed Win 95 on the primary partition (C:). I booted thro' Linux Yast and
/dev/hda2 (extended)
parititioned the extended drive into 3 logical partitions (after D: ) of 4 Gigs, 3 Gigs and 256
/dev/hda5 (4 GB, ext2) linux boot with LILO /dev/hda6 (3 GB, <free>) /dev/hda7 (256 MB, swap)
Megs for swap. I installed Linux 6.3 on the root paritition of 4 Gigs. I was able to boot from the CD later. I also installed LILO on the root partition of the Linux partition ( 4 gigs).
Now, whenever i boot the system, it goes directly to Win 95. How can i make LILO as the boot manager?? What is the procedure for this??
Raghav
Hi Ragav, what happened to NT?? Usual boot procedure is to boot from the active partition. Win98 is on the active partition (hda1 / c:\). The master boot sector contains a list for 4 partitions. (As Warrl explained) This list contains first/last cylinder, part type and, for *ONE* partition the *active* flag. On bootup standard procedure is to load the MBR (with a short 440-somthing bytes) machine code program. This program looks for the active partition, loads the first sector there and executes the code from there. This usually starts further programs (ibmdos.sys or whatever they are called) With LILO, you can jump to selected MBR's of a partition. I am not so overexperianced in boot configurations, but I guess activating the activity flag for the linux partition would fire up linux (rather lilo), you could start DOS / Win98 probably from there. I use this sort of config with Compaqs: Some meanwhile elder deskpros have a service partition, accessed by pressing F10 on bootup. If you place LILO into MBR it's as reacheable as your linux now. ;-( At this time I'd experiment. Place LILO to the MBR and create boot entries for Win98 (=boot DOS) and DOS (=boot DOS) from yast. Things get interesting if you still need to put NT (?? to the 3 GB hda6 partition??) on top of that. I'd try the NT bootloader then: "dd if=/dev/hda5 of=/tmp/linux.bot bs=512 count=1" to write the linux bootsector to a file. Copy that file to c:\ and create the entry in boot.ini (SDB!!) I do not realy feel happy advising on that, its different if I do it to my own machine. I can correct most of my errors. BTW. its always a good idea (particulary with WinXX on the machine as we learned over the weekend) to keep an emergency copy of the MBR and partition MBR's somewhere: dd if=/dev/hda[X] of=somefile count=1 bs=512 just to have them handy to put them back in emergency: dd if=somefile of=/dev/hda[X] count=1 bs=512 I forgot to look into my boot.ini at work today. The example in my book could be a SCSII system; but I think the boot descriptions for NT look all like beeing SCSI. Juergen -- =========================================== __ _ Juergen Braukmann juergen.braukmann@gmx.de| -o)/ / (_)__ __ ____ __ Tel: 0201-743648 dk4jb@db0qs.#nrw.deu.eu | /\\ /__/ / _ \/ // /\ \/ / ===========================================_\_v __/_/_//_/\_,_/ /_/\_\ -- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/Support/Doku/FAQ/
Raghavendra R wrote:
Thanks for the extensive and very helpful information regarding hard drives and partition. I went ahead and created a primary partition of 990 Megs (C:) with FAT16 type and created an extended partition for the rest of the disk wherein i created another partition of 990 Megs (D: FAT16) and installed Win 95 on the primary partition (C:). I booted thro' Linux Yast and parititioned the extended drive into 3 logical partitions (after D: ) of 4 Gigs, 3 Gigs and 256 Megs for swap. I installed Linux 6.3 on the root paritition of 4 Gigs. I was able to boot from the CD later. I also installed LILO on the root partition of the Linux partition ( 4 gigs).
Now, whenever i boot the system, it goes directly to Win 95. How can i make LILO as the boot manager?? What is the procedure for this??
All you have to do is fire up YaST and go to the 'LILO configuration' section (or whatever its name is). Set up all the partitions you want to boot (including Win9x (boot DOS/Win), NT (boot DOS/Win) and, of course, Linux (boot Linux)) and set LILO to install into the MBR. Next time the computer boots, LILO should run and you can use it to select which partition to boot.
Warrl wrote:
On Tue, 25 Apr 2000, Raghavendra R wrote:
Hi,
My BIOS recognises the 17 GB hard drive as one with 33416 cylinders, 16 heads and 63 sectors. So, this comes to approx. 512 MB for the 1024 cylinder limit. But, i was able to install linux 6.3 after the 4 GB partition (i,e from E:, C: and D: each taking 2 GB) and the linux kernel reported the hard drive configuration as one with 2096 cylinders, 255 heads and 63 sectors. So, according to the kernel the 1024 cylinder limit falls at 8 GB. Is it true?? I read in one of the HOWTO docs. (i think it is Large-Disk), that the specs. for large disk (i,e above 8 GB) will be reported as C/H/S = 16383/16/63. I think there is some kind of translation between the kernel and the BIOS.
It looks like your BIOS is recognising a "large" configuration, while Linux is recognising an "lba" configuration.
LBA is much more friendly to old operating systems, because the combination of BIOS and IDE allows 1024/255/63 which is 8 meg per cylinder - over 8 gig within the 1024 cylinders. Software which thinks it is talking to old-style BIOS (e.g. the FAT16 file system in Win95, or lilo) can thus use the first 8 gig of the disk, rather than the first 512M.
LBA must be supported by the drive *and* by whatever BIOS-level software is being used on the computer, in order to be used. This is because the drive actually does the translation, but only if the BIOS-level software gives it permission to do so. Most modern BIOSes support it, as does the BIOS-replacement code in the Linux kernel. I think the Windows NT BIOS-replacement code does too.
(Any true, stable multitasking OS *must* replace the BIOS disk IO routines, because they aren't reliable in multitasking.)
Lilo uses the BIOS, not replacement code. For situations where the BIOS does not recognise LBA but the Linux kernel does - and therefore the geometries do not agree - lilo internally will translate between either LBA or whatever the BIOS says, and a special internal mode called "linear" - one head, one cylinder, up to 16-million-plus sectors per track. Any given sector will always get the same address in this system, no matter how the geometries differ. You enable this by putting the word "linear" (without quotes) in the global section of lilo.conf.
A "large" format is what the drive is actually doing, but is not compatible with old-style BIOSes and software which expects to talk to them.
(What idiots picked the BIOS specs in the first place? They put a great deal of effort into making sure that an absolute sector address could fit in 3 bytes, on a 16-bit processor. They didn't bother to check the IDE specs. If they had simply been lazy, the BIOS would support 65,536 cylinders, 256 heads, and 256 sectors per track - 2 terabytes, with 512-byte sectors. Instead we get some restrictions that are entirely different from those of IDE, causing all sorts of complications: a celing at 512M, and another at 8-point-something gig, in addition to whatever the limit of IDE is and the limits of the actual drive in consideration at the moment.)
(But, the history of standards can be interesting. The standard for the distance between railroad tracks, for example, is what it is because of a couple of horse's asses. Literally.)
Hope that helps some more, Chris -- __ _ -o)/ / (_)__ __ ____ __ Chris Reeves /\\ /__/ / _ \/ // /\ \/ / ICQ# 22219005 _\_v __/_/_//_/\_,_/ /_/\_\ -- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/Support/Doku/FAQ/
Hello folks, maybe you can help a friend of mine (Fabio) who was running NT and Linux. I say "was" as (I don't know why) NT bootloader started fooling and didn't let him boot his 4gig NT partition. He said to me that after activating the partion, NT shows a black screen at start up. So, to be able to work today on damned office applications, he deleted linux and installed win98 (sob sob). BTW: But this didn't help him booting NT. Is there any way to use linux to recover the MBR in order to enable NT bootloader to work again? (a software called BOOTPART is not working) I know that this is a NT question and wouldn't like to fire a flame war, only if there is anybody who can help (maybe on private mail) in order to recovery 4gb working data without reinstalling NT (i.e.: formatting), well thank you very much. Please save bandwidth rather than discuss over the NTity of the question. At the end I really hope me and my friend will get free of these OS. best regards, Stefano Papini Chris Reeves wrote:
Raghavendra R wrote:
Thanks for the extensive and very helpful information regarding hard drives and partition. I went ahead and created a primary partition of 990 Megs (C:) with FAT16 type and created an extended partition for the rest of the disk wherein i created another partition of 990 Megs (D: FAT16) and installed Win 95 on the primary partition (C:). I booted thro' Linux Yast and parititioned the extended drive into 3 logical partitions (after D: ) of 4 Gigs, 3 Gigs and 256 Megs for swap. I installed Linux 6.3 on the root paritition of 4 Gigs. I was able to boot from the CD later. I also installed LILO on the root partition of the Linux partition ( 4 gigs).
Now, whenever i boot the system, it goes directly to Win 95. How can i make LILO as the boot manager?? What is the procedure for this??
All you have to do is fire up YaST and go to the 'LILO configuration' section (or whatever its name is). Set up all the partitions you want to boot (including Win9x (boot DOS/Win), NT (boot DOS/Win) and, of course, Linux (boot Linux)) and set LILO to install into the MBR. Next time the computer boots, LILO should run and you can use it to select which partition to boot.
Warrl wrote:
On Tue, 25 Apr 2000, Raghavendra R wrote:
Hi,
My BIOS recognises the 17 GB hard drive as one with 33416 cylinders, 16 heads and 63 sectors. So, this comes to approx. 512 MB for the 1024 cylinder limit. But, i was able to install linux 6.3 after the 4 GB partition (i,e from E:, C: and D: each taking 2 GB) and the linux kernel reported the hard drive configuration as one with 2096 cylinders, 255 heads and 63 sectors. So, according to the kernel the 1024 cylinder limit falls at 8 GB. Is it true?? I read in one of the HOWTO docs. (i think it is Large-Disk), that the specs. for large disk (i,e above 8 GB) will be reported as C/H/S = 16383/16/63. I think there is some kind of translation between the kernel and the BIOS.
It looks like your BIOS is recognising a "large" configuration, while Linux is recognising an "lba" configuration.
LBA is much more friendly to old operating systems, because the combination of BIOS and IDE allows 1024/255/63 which is 8 meg per cylinder - over 8 gig within the 1024 cylinders. Software which thinks it is talking to old-style BIOS (e.g. the FAT16 file system in Win95, or lilo) can thus use the first 8 gig of the disk, rather than the first 512M.
LBA must be supported by the drive *and* by whatever BIOS-level software is being used on the computer, in order to be used. This is because the drive actually does the translation, but only if the BIOS-level software gives it permission to do so. Most modern BIOSes support it, as does the BIOS-replacement code in the Linux kernel. I think the Windows NT BIOS-replacement code does too.
(Any true, stable multitasking OS *must* replace the BIOS disk IO routines, because they aren't reliable in multitasking.)
Lilo uses the BIOS, not replacement code. For situations where the BIOS does not recognise LBA but the Linux kernel does - and therefore the geometries do not agree - lilo internally will translate between either LBA or whatever the BIOS says, and a special internal mode called "linear" - one head, one cylinder, up to 16-million-plus sectors per track. Any given sector will always get the same address in this system, no matter how the geometries differ. You enable this by putting the word "linear" (without quotes) in the global section of lilo.conf.
A "large" format is what the drive is actually doing, but is not compatible with old-style BIOSes and software which expects to talk to them.
(What idiots picked the BIOS specs in the first place? They put a great deal of effort into making sure that an absolute sector address could fit in 3 bytes, on a 16-bit processor. They didn't bother to check the IDE specs. If they had simply been lazy, the BIOS would support 65,536 cylinders, 256 heads, and 256 sectors per track - 2 terabytes, with 512-byte sectors. Instead we get some restrictions that are entirely different from those of IDE, causing all sorts of complications: a celing at 512M, and another at 8-point-something gig, in addition to whatever the limit of IDE is and the limits of the actual drive in consideration at the moment.)
(But, the history of standards can be interesting. The standard for the distance between railroad tracks, for example, is what it is because of a couple of horse's asses. Literally.)
Hope that helps some more, Chris -- __ _ -o)/ / (_)__ __ ____ __ Chris Reeves /\\ /__/ / _ \/ // /\ \/ / ICQ# 22219005 _\_v __/_/_//_/\_,_/ /_/\_\
-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/Support/Doku/FAQ/
-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/Support/Doku/FAQ/
On Wed, 26 Apr 2000, Raghavendra R wrote:
Thanks for the extensive and very helpful information regarding hard drives and partition. I went ahead and created a primary partition of 990 Megs (C:) with FAT16 type and created an extended partition for the rest of the disk wherein i created another partition of 990 Megs (D: FAT16) and installed Win 95 on the primary partition (C:). I booted thro' Linux Yast and parititioned the extended drive into 3 logical partitions (after D: ) of 4 Gigs, 3 Gigs and 256 Megs for swap. I installed Linux 6.3 on the root paritition of 4 Gigs. I was able to boot from the CD later. I also installed LILO on the root partition of the Linux partition ( 4 gigs).
Now, whenever i boot the system, it goes directly to Win 95. How can i make LILO as the boot manager?? What is the procedure for this??
You were talking about a triple boot with Windows NT added to the above mix. Right now you have a dual-OS system with Windows 95 and Linux. That system works best with lilo bootloader written to the MBR of the disk. However, from what I hear, lilo can't boot Windows NT. Instead you have to have Windows NT's bootloader in the MBR, with an option to start lilo from the boot sector of some partition (usually the linux root partition). If you can use the rescue CD to boot your existing Linux install, you can then change /etc/lilo.conf and rerun lilo to have its boot loader written to another location. I STRONGLY recommend that you have it on a floppy; this will allow you to get into Linux even if the hard-drive bootloading stuff is seriously hosed. You also need it in the root partition boot sector, for when you get the triple boot working (there's nothing wrong with having it in multiple places), and for now you may want it in the MBR as well. Installing or re-installing Win9x WILL clobber lilo in the MBR, so anyone trying to dual-boot with Windows definitely needs lilo on a floppy. I think that installing Windows NT will do the same thing. -- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/Support/Doku/FAQ/
I multi-boot using Lilo with NT on a laptop and have done so on desktops. You just have to install linux after you install windows. You may need to install and re-install a couple of times before it works correctly. They often do not want to play nicely (particulary Win2000) Peter Don Edwards wrote:
On Wed, 26 Apr 2000, Raghavendra R wrote:
Thanks for the extensive and very helpful information regarding hard drives and partition. I went ahead and created a primary partition of 990 Megs (C:) with FAT16 type and created an extended partition for the rest of the disk wherein i created another partition of 990 Megs (D: FAT16) and installed Win 95 on the primary partition (C:). I booted thro' Linux Yast and parititioned the extended drive into 3 logical partitions (after D: ) of 4 Gigs, 3 Gigs and 256 Megs for swap. I installed Linux 6.3 on the root paritition of 4 Gigs. I was able to boot from the CD later. I also installed LILO on the root partition of the Linux partition ( 4 gigs).
Now, whenever i boot the system, it goes directly to Win 95. How can i make LILO as the boot manager?? What is the procedure for this??
You were talking about a triple boot with Windows NT added to the above mix.
Right now you have a dual-OS system with Windows 95 and Linux. That system works best with lilo bootloader written to the MBR of the disk.
However, from what I hear, lilo can't boot Windows NT. Instead you have to have Windows NT's bootloader in the MBR, with an option to start lilo from the boot sector of some partition (usually the linux root partition).
If you can use the rescue CD to boot your existing Linux install, you can then change /etc/lilo.conf and rerun lilo to have its boot loader written to another location. I STRONGLY recommend that you have it on a floppy; this will allow you to get into Linux even if the hard-drive bootloading stuff is seriously hosed. You also need it in the root partition boot sector, for when you get the triple boot working (there's nothing wrong with having it in multiple places), and for now you may want it in the MBR as well.
Installing or re-installing Win9x WILL clobber lilo in the MBR, so anyone trying to dual-boot with Windows definitely needs lilo on a floppy.
I think that installing Windows NT will do the same thing.
-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/Support/Doku/FAQ/
-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/Support/Doku/FAQ/
participants (6)
-
chris.reeves@iname.com
-
juergen.braukmann@ruhr-west.de
-
philtz@mediaone.net
-
raghav@easi.soft.net
-
stefano.papini@intercai.etnoteam.it
-
warrl@blarg.net