Re: Big Problem with Winodows Me

[Andrew Clausen]

Sven Backhausen wrote:
> 
> Andrew Clausen schrieb:
> > Could you try creating partitions of different sizes?  Perhaps 0xfa
> > should have the same meaning as 1023 (too big for CHS).
> >
> Hi!
> 
> I did some testing with ME's fdisk and here are the results:
> 
> End-Cyl (real)  End-Cyl in partition table
> 1071            46
> 1275            250
> 2550            501
> 3825            752
> 4982            885
> 
> all values are decimal, I used Linux fdisk to get the real end cyl.
> Looks like ME's fdisk writes the real cylinder number modulo 1024 minus
> 1 into the partition table.

Thanks for that!
 
> Hope this helps getting parted work on these ME machines, we need it.

Yep.  BTW, it's easy to hack Parted to get it to work, by commenting
out all but the last line of fat_probe().  However, this breaks all
other disk labels.  (Just commenting out the sanity check breaks PC98
only)

The problem with PC98 & MSDOS is, it's possible to have a complete
PC98 table and a complete MSDOS table, at the same time.  In fact,
any PC98 table will also have signature for MSDOS tables.  So, Parted
does some sanity checks on the MSDOS table, to try to rule out the
possibility that it's really a PC98 table.

So, we've got to come up with a sanity check, that handles this
ME case also.  Urghhh

> All these tests were done with primary partitions, do you think we need
> to test logical volumes too?

I suspect they do the same thing, but with this stuff, you never know?
If you've got time, I'd be interested to know!

Thanks!
Andrew Clausen