Re: bug report

[Andrew Clausen]

Hi Joern,

Joern Tappe wrote:
> Thank you kindly for your very fast and most informative reply!

:-)

> Btw, the command
> 
> (parted) mkpartfs logical FAT 8032.500 19077.187      ;FAT32
>                                    ^^^
> resulted in Parted's placing the partition at '8032.562 thru 19077.187',
>                                                     ^^^
> thus creating an additional offset of 1 track (63 sectors, 0.031 MB),
> making the partition start at head #2 of the respective cylinder.
> There was another FAT32 partition of the same size on my disk, and that
> partition started at head #1 of its first cylinder. On the other hand,
> I spotted a partition offset of 0.063 MB in example 1 of Parted's
> documentation, so I guess that once again this is not a bug :-)

Yeah, for logical partitions (and the first primary partition),
you need to start on the second head (in some cases the third
head...where the extended partition starts at the start of the
disk), because you need to keep room for the partition table(s).

> If so: What might be the reason for the offset of 2 tracks?

No idea!  Could you print out the partition table from
Parted?

> > Does this make sense?  Feel like trying this out?  *grin*
> 
> It certainly does make sense, and although your code leaves no doubt
> that it will do the trick, I'll be sure to try it out ;-)

BTW, I would be very happy if you went through the partition
alignment code, and checked it... and comment on how readable
it is, etc.  Look at everything (recursively) that
msdos_partition_align() calls.  You can look in the docs for
a bit of info about how the constraint solver works, though
I hope that's straightforward also ;-)

Thanks,
Andrew Clausen