Re: Assertion (cyl_size <= 255 * 63) at disk_dos.c:546 in function probe_partition_for_geom() failed.

[Sven Luther]

On Sun, Dec 19, 2004 at 09:15:44AM +1100, Andrew Clausen wrote:
> On Sat, Dec 18, 2004 at 03:36:10PM +0100, Sven Luther wrote:
> > Maybe that code could be reused to generate an optimal geometry for
> 
> This code isn't relevant to "optimal geometry".  It is very simple:
> 
>       if (A > (C+1) * cyl_size)
>               return 0;
> 
> That's all!  A == end address, C == end cylinder.  These cryptic names
> are used to make the maths readable.

Ok.

> > partitioning scheme like the amiga partition table, which support partition
> > granularity only on cylinder boundaries ? For the creation of new partition
> > tables this is, since the current scheme of using the <whatever> proposed
> > geometry often means the waste of disk space at the end of the disk.
> 
> We're only talking about 8 megabytes, worst case, right?
> Does anyone really care?

He, i hadn't thought at it this way, it was obviously important back in the
years of <1GB disks, but useless here. We may try to obtimize for smallest
cylinder size for partition placement granularity, but even that may not be
worth it.

> Is the geometry on Amiga relevant for booting, etc.?

Well, the pegasos OF knows about the partition table format, and boots
directly from an elf file on one of the partition, so the partition table
needs to be readable, and since it contains a copy of the geometry locally, it
is the only relevant info. The other geometry is totally ignored except for
the creation of new partition tables.

Friendly,

Sven Luther