[Parted-maintainers] A possible parted bug

[Tapio Kelloniemi]

Hi

I would have filled a bug report on your bug tracking system, but it 
uses JavaScript, so I e-mail.
I'm not sure if this bug is in parted, since parted and BIOS seem
to disagree about the disk geometry.

System configuration:
parted 1.6.24
linux 2.6.11.12
disk Samsung Spinpoint SerialAta 80GB
Disk geometry for /dev/sda: 0,0,0 - 9733,116,62
BIOS cylinder,head,sector geometry: 9733,255,63.  Each cylinder is 8225kB.

The following steps result in an assertion failure:
> mklabel msdos
> mkpart extended 0 -1
> mkpart logical ext2 0 5gb
> mkpart logical ext2 5gb 5.5gb
Assertion (start <= end) at parted.c:251 in function 
snap_to_boundaries() failed.

After Ignoring or Cancelling:
Error: Can't have the end before the start!

But in any case:
> print
Disk geometry for /dev/sda: 0kB - 80GB
Disk label type: msdos
Number  Start   End     Size    Type      File system  Flags
1       1kB     80GB    80GB    extended
5       32kB    5001MB  5001MB  logical
6       5001MB  5009MB  8193kB  logical

Reversing the order of the two mkpart logicals fixes the problem, but 
the resulting partition table looks identical (in compact unit mode).

Another assertion failure is triggered by:
> mklabel msdos
> mkpart extended 0 -1
> mkpart logical ext2 0 3072mb
> mkpart logical ext2 3072mb 5120mb
Assertion (metadata_length > 0) at disk_dos.c:1962 in function
add_logical_part_metadata() failed.

If printing is attempted, the same assertion fails again.

Hope this was useful. I'm not a member of list, so please CC to me.

-- 
Tapio