Bug in parted 1.4.22 (I think)

[Frazer Williams]

Hi,

I seem to have found a bug in parted.  Whether the bug is in parted or
in my head, I'd appreciate any advice or help.  Here's my side of the story.

I have a 6GB IDE drive that has a 2GB Windows partion in partition 1.
Partition 2 is extended, and contatins 4 Linux logical partions, 5, 6,
7, and 8.  Partition 8 is swap, the rest are ext2.  I'm running RedHat
7.1 with no known problems from the disk.  fdisk, cfdisk, and sfdisk all
find no problem with the disk, but parted first complains about a
partition alignment problem, and then proceeds to get the minor numbers
reversed for the logical partions 5, 6, 7, and 8.

Here's the output of /sbin/sfdisk -uS -l /dev/hda.

=====sfdisk -uS -l /dev/hda=====

Disk /dev/hda: 784 cylinders, 255 heads, 63 sectors/track
Units = sectors of 512 bytes, counting from 0

   Device Boot    Start       End  #sectors  Id  System
/dev/hda1   *        63   4192964   4192902   6  FAT16
/dev/hda2       4192965  12594959   8401995   5  Extended
/dev/hda3             0         -         0   0  Empty
/dev/hda4             0         -         0   0  Empty
/dev/hda5       4193028   4514264    321237  83  Linux
/dev/hda6       4514328   8498384   3984057  83  Linux
/dev/hda7       8498448  12482504   3984057  83  Linux
/dev/hda8      12482568  12594959    112392  82  Linux swap

================================

And, here's the output of parted /dev/hda.  (This is version 1.4.16, but
I get the same result with 1.4.22.)

========parted /dev/hda=========

GNU Parted 1.4.16
[copyright stuff deleted]

Using /dev/hda
Warning: The operating system thinks the geometry on /dev/hda is 784/255/63.
You should check that this matches the BIOS geometry before using this program.
(parted) print
Warning: Unable to align partition properly.  This probably means that another
partitioning tool generated an incorrect partition table, because it didn't have
the correct BIOS geometry.  It is safe to ignore,but ignoring may cause
(fixable) problems with some boot loaders.
Ignore Cancel ? Ignore
Disk geometry for /dev/hda: 0.000-6149.882 megabytes
Disk label type: msdos
Minor    Start       End     Type      Filesystem  Flags
1          0.031   2047.346  primary   FAT         boot
2       2047.346   6149.882  extended              
8       2047.377   2204.230  logical   ext2        
7       2204.262   4149.602  logical   ext2        
6       4149.633   6094.973  logical   ext2        
5       6095.004   6149.882  logical   linux-swap  
(parted) quit

=====================================

Note the minor numbers are backwards for 8, 7, 6, and 5.

I ran into this problem trying to upgrade RedHat Linux from 7.1 to 7.2.
The incorrect minor numbers cause the installer to give up with an error
message that it can't mount /dev/hda8.

I considered the possibility that parted could be correct and fdisk et
al. wrong.  To check that, from the installer shell I first created
/dev/hda[5-8] with commands like

mknod /dev/hda5 b 3 5

I was then able to mount /dev/hda5, and it contained the expected files.
I was not unable to mount /dev/hda8.  Thus the first partion in hda2
must indeed be the partition accessed with minor number 5 as sfdisk
reports, not 8 as parted claims.

Some further, probably useless, information:

1.  From the installer shell (so /dev/hda[5-8] were not mounted) I ran
    e2fsck on /dev/hda[5-8].  No problems were found.

2.  I dumped some of the partition tables, and to my novice eye they
    look fine.  For example, here is what was in the partition table for
        the first logical partition in /dev/hda2 (i.e. hda5), as seen
        through od:

========partition table dump for hda5============

    0001be 0000 1941 fe05 10bf e714 0004 caf8 003c
    0001ce 0100 0541 fe83 187f 003f 0000 e6d5 0004
    0001de 0000 0000 0000 0000 0000 0000 0000 0000
    *
    0001fe aa55
    000200

=================================================

I would appreciate any help or advice you could give me.  If it would be
helpful, I could probably track down and dump the other partition tables
on the disk.

Thanks.

Frazer