Need to force IDE geometry

Need to force IDE geometry

[John Huttley]

Hello,

I have a problem with the way the kernel handles geometry.

AFAIK, if a disk has partitions which were created in LBA mode,
the kernel automagically puts the disk into LBA mode. (good)

If the disk has no partitions.. then
	If its the first disk it will be in LBA mode (good)
Else it will be in CHS mode. (ungood)

There seems to be no way to tell the kernel to put the second disk into
LBA mode. Sure, you can force the geometry as a boot option, but that is
not the same as putting it in LBA mode.

Any bios settings to put the disk into LBA are ignored by the kernel.


This causes problems when installing a RAID enabled linux onto a
dual IDE system. RedHat 7X suffers this and so does the Mondo Rescue
restore system. It makes it very difficult to setup 2 disks so they are
of identical geometry. Trust me, its a PINA.

Are there any ioctl's controlling this?
Can we create a boot option? Or simply read the bios setting? Or just
default to LBA rather than CHS?

Any help gratefully received.
As I'm not on this list, would you please CC me.

Regards

John

Re: Need to force IDE geometry

[Andries.Brouwer]

> I have a problem with the way the kernel handles geometry.

Your question is based on your assumptions about geometry
and LBA. But your assumptions are incorrect, and therefore
your questions do not make sense. Please tell what you do
and what error messages you get.

Andries


> Else it will be in CHS mode. (ungood)

False.

Re: Need to force IDE geometry

[Andre Hedrick]

On Fri, 15 Feb 2002 Andries.Brouwer@cwi.nl wrote:

> > I have a problem with the way the kernel handles geometry.
> 
> Your question is based on your assumptions about geometry
> and LBA. But your assumptions are incorrect, and therefore
> your questions do not make sense. Please tell what you do
> and what error messages you get.
> 
> Andries
> 
> 
> > Else it will be in CHS mode. (ungood)
> 
> False.

It is simple, the raid cards force LBA in the BIOS but Linux choose to
NUKE it and thus the offsets for finding the reserved sectors is all
wrong.  Again we (you) have forced drives to run in CHS by default and it
is wrong.  We can debate but if the drive supports LBA we must force LBA
in the kernel.

Now I am out of this argument.

Cheers,


Andre Hedrick
Linux Disk Certification Project                Linux ATA Development