Re: amiga partitioning support for parted

[Sven Luther]

On Wed, Jul 30, 2003 at 05:41:35PM +1000, Andrew Clausen wrote:
> On Wed, Jul 23, 2003 at 08:22:22AM +0200, Sven Luther wrote:
> > Well, there is the partition table reading, which is in the file i
> > wrote, and the normal partition accessing, that is the reading of the
> > file system. But i think i now understand what you meant, the values i
> > read from the partition list will be multiplied by two before being put
> > into the partition data (and constraints and so on) and divided by two
> > before being accessed to write to the partition list.
> 
> Right.

:)

> > I still think having a way to specify sector size, and not fix it to 512
> > would be nice.
> 
> I think it's too complicated.  I think it would be nicer to remove
> the whole concept of sector all together.  Flat linear address spaces
> are good!

Well, i find the way of reading the sectors quite easy, and i would hate
to loose that. That said, the amiga partition scheme enables you to be
able to read the block/sector size in the partition table, which can be
on any of the first 16 sectors. Old 256 byte sectors are obsoleted, 512
are currently in use, but i guess that some people use bigger sector
sizes, not sure though. I would hate to destroy their current existing
partitions though because i don't handle this.

> > > > Should a read wrapper not be provided in the partition specific
> > > > layer, or the partition size not be a variable one can set ?
> > > 
> > > Do you understand what PedConstraint does?
> > 
> > Yep, i have battled with it enough to know perfectly what it means.
> 
> Was the battle unpleasant?

Well, i faced two problems. Somehow, my partition table used a geometry
different that the one reported by the disk, a geometry which could
cover a bigger part of the harddisk, it seems, and this on all 4 amiga
harddisk i have tried it one, each being partitioned by a different tool
(the linux amiga-fdisk, the original amiga partitioner and the MorphOS
scsi-config partitioner). I suppose they do that because the amiga
partitioning scheme use cylinder values as partition boundaries (which
you can express nicely as constraint) and thus having lower sector/head
values gives you higher granularity. Also, on the disk partitioned by
the old amiga OS partitioner 6 years ago, the blocks/cylinder value was
equal to head*sectors-1, which was quite strange, and made me go crazy
before i noticed it. It would really be nice to have some easy debuging
output about geometries and constraints builtin if you set DEBUG to
true, instead of having to code this by hand.

Friendly,

Sven Luther