Re: Kernel / atari-fdisk problem with IDE-HD > 8,4 GB
> (I'm author of atari-fdisk.)
>
> > On large (== every new Disk, yet) IDE-Harddisks the real disk size
> > is greater then the reported size by the HD itself in C/H/S
> > notation. The reported LBA- size is correct, but not used by the
> > Kernel so the atari-fdisk programm reports errors.
>
> Hmm... wait a moment. atari-fdisk uses the standard BLKGETSIZE ioctl
> to get the disk size from the kernel. You mean that the kernel's IDE
> driver reports that size wrong? So that would be a sole kernel
> problem, not atari-fdisk's one.
It would help tremendously to have a copy of the kernel log messages from
the IDE driver. The IDE driver probes for the disk size and geometry, and
these messages should help us sort out the problem.
> Ok, back to the kernel: Most parts of the IDE driver are nowadays
> shared between all archs, so I can't really believe that IDE on Atari
> should make things different than elsewhere. Or, wait a moment: It
> could be that using the LBA size is usually enabled by a corresponding
> setting in the PC BIOS, and that the default is using the CHS size. So
> the bug would be that some flag or the like isn't set on Atari...
> Would have to look into this, but I'm no real expert for IDE :-(
Given the small size of the Atari specific portion of the IDE driver, I
doubt it. I'll look into that though (won't be able to test it due to lack
of such a large disk, and my Falcon generally dying faster than it boots
these days).
> I think it's best if you write to the Linux/m68k kernel list for this
> issue.
CC: set; is there anyone from the Amiga crowd with similar problems?
> For the Debian part: If a patch is available, we should integrate it
> as fast as possible into our current kernel-patch package and
> recompile the Atari kernel. This could be considered release
> critical...
If someone sends a patch, we should give it a high priority. But as we can
just post the updated kernel at a later time, I'd not consider it a show
stopper.
Michael
Reply to: