Re: libparted ped_disk_get_max_partition_geometry() gives wrong size

[Jim Meyering]

ben wrote:

> With libparted, version 2.3, I have been having a problem with
> ped_disk_get_max_partition_geometry(). The function returns values that are
> correct for the beginning of a partition but too small for the end. This only
> seems to happen with MS-DOS disk labels. The returned value can be short by
> over 2 megabytes and can even be *smaller* than the partition that already
> exists there.
>
> I have seen the same behavior with disks of 40 GB too 1 TB and even flash
> drives. GParted reports the correct sizes and often shows substantial
> unallocated space after a partition that was supposed to be maximized.
>
> I am running Debian Linux.

Thank you for the report.

If you provide more detail, it will be easier to understand and fix
the problem.  The best would be to post code showing how to reproduce
the failure, assuming it is easy to demonstrate with most disks.
Can you do that?  Currently, parted has no unit test to exercise
ped_disk_get_max_partition_geometry, so it'd be great to add a test
with a simple main program, like the existing tests/print-align.c or
tests/print-max.c, and use that to demonstrate your problem.

Otherwise, please post at least the actual numbers, including existing
partition layout (sector numbers) and the actual inputs and outputs for
your use of ped_disk_get_max_partition_geometry.