Re: whazzup with fat16 ?

[Andrew Clausen]

On Mon, Aug 02, 2004 at 11:11:52PM +0200, B.Hakvoort wrote:
> After i did some playing with parted/libparted i have a few questions
> about the fat16 filesystem.
> 
> - I can only create fat16 up to 1023 MB. Shouldn't that be 2048 MB ?

Should it?  I remember DOS/Windows having problems with 2Gig Fat16
partitions, but I could be wrong.  (I really should have documented
my experiments, sorry!)

> - When i try to resize a fat16 filesystem the results are erratic,     
> shrinking goes fine (mostly), but growing results in a
>        "No Implementation: GNU Parted can not resize this partition   to this
> size.  We're working on it!"
> or the filesystem is changed from fat16 to fat32 without asking the
> user!
> OR.. it goes fine...
> 
> Hopefully someone can shed some light on this!

The problem is that FAT has strict requirements on the number of
clusters, the size of clusters and the size of file allocation tables.
Microsoft operating systems often ignore these values stored in the
superblock (boot sector), and compute them directly from the size of the
file system!!!

To maintain these strict relationships, Parted often needs to change
the cluster size.  Unfortunately, Parted can't increase the cluster
size - only reduce.  So, the Parted resizer is constrained by Microsoft
weirdness + it's own cluster resizing limitations.

Cheers,
Andrew