Re: Partitions not detected

[Ignaz Forster]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

(See inline text below)

Ignaz Forster schrieb am 10.01.2011 21:25 Uhr:
> Jim Meyering schrieb am 10.01.2011 10:05 Uhr:
>> Ignaz Forster wrote:
>>> I fear I need some help with Parted.
>>> The two partitions of a disk are not detected, instead a message about
>>> a third partition is printed:
>>>
>>>
>>>> # parted /dev/sdc unit s print
>>>> No Implementation: Partition 3 isn't aligned to cylinder boundaries.
>>>> This is still unsupported.
>>>
>>>
>>> If I remember correctly the disk was partitioned using cfdisk, using
>>> byte values (instead of sectors) to specify the partition sizes. fdisk
>>> reports the following layout:
>>>
>>>
>>>> # fdisk -l -u /dev/sdc
>>>>
>>>> Disk /dev/sdc: 1000.2 GB, 1000204886016 bytes
>>>> 255 heads, 63 sectors/track, 121601 cylinders, total 1953525168 sectors
>>>> Units = sectors of 1 * 512 = 512 bytes
>>>> Sector size (logical/physical): 512 bytes / 512 bytes
>>>> I/O size (minimum/optimal): 512 bytes / 512 bytes
>>>> Disk identifier: 0x14762e21
>>>>
>>>>    Device Boot      Start         End      Blocks   Id  System
>>>> /dev/sdc1              63   976559219   488279578+  83  Linux
>>>> /dev/sdc2       976559220  1953520064   488480422+  83  Linux
>>>
>>>
>>> The first partition contains a XFS, the second one a JFS filesystem.
>>>
>>> If my calculation is right at least the first partition should be
>>> aligned to cylinder boundaries, but temporarily deleting the second
>>> partition did not help either.
>>> Any hint why Parted doesn't see those partitions?
> [...]
>> If you're using a relatively recent version of parted,
>> then the fact that you're seeing that diagnostic means
>> that device has a PC98 partition table (or at least that
>> parted detects it as such -- which may be a bug in parted).
>> [...]
>>
>> If you are using one of those, please make a copy of
>> the first few sectors of your disk and send them to this list.
>> I.e., you could run this and then attach the file F:
>>
>>     dd if=/dev/sdc count=34 > F
>>
> 
> This is interesting. The dump contains the string "Apple_partition_map" - see 
> the attached file.
> The disk was advertised with it's Mac compatibility and originally contained 
> a HFS file system which was deleted (as far as I remember also using cfdisk) 
> to create the new partitions / filesystems.

I've played around a bit this weekend and it seems I have found the error:

Parted seems to be confused if both a MBR partition table in sector 0 and an 
Apple Partition Map starting at sector 1 exist. To test this I filled the first 
cylinder (except for the first sector with the MBR) with zeros to delete the 
Apple Partition Map; after that parted could see the partitions, so my problem 
is solved.
I'm still wondering however: Shouldn't Parted use the MBR partition table if 
both an MBR and APM table are present? If I understand the specs of the APM 
correctly the first sector should contain a map of driver information, but not 
a MBR - which would indicate that the APM is not valid any more.

Is Parted able to handle other partition tables than MBR (like GUID Partition 
Table (GPT) or Apple Partitions Maps (APM)) anyway?

Greetings, and thanks for the hint in the right direction,
- -- 
Ignaz Forster <address@hidden>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk0zfyQACgkQ/2+SjQTlQJkh2ACZAfjq4p6iYJ6BvyqzFJF0ZcrB
rGcAoJlMHtvInRYcV+biwYzCktMMbhJS
=pftf
-----END PGP SIGNATURE-----