When no partitions on a DOS-partitioned disk are mounted, the kernel
automatically rescans the partition table when the file handle to the
raw device is released. Currently the code opens and closes the raw
device when writing the primary partition table and when writing each
extended partition table segment. As the extended partition table
segments form a linked list, this allows the kernel to get in and
rescan the table when the list is not in a coherent state. This patch
changes the code to open the raw device before writing the primary
partition table and to close it only after writing the last extended
partition table segment.

The behaviour for other partition table types is unchanged.

v2 (tvignaud): simplify by moving copies of noop funcs into the base class

they're obsolete since switching from CVS/SVN to git...

This commit consitsts of the following subversion commits:

------------------------------------------------------------------------
r431 | dmorgan | 2011-02-06 01:58:02 +0000 (Sun, 06 Feb 2011) | 1 line

Add drakx structure on the SVN
------------------------------------------------------------------------
r432 | dmorgan | 2011-02-06 01:59:38 +0000 (Sun, 06 Feb 2011) | 1 line

Import cleaned advertising
------------------------------------------------------------------------
r446 | dmorgan | 2011-02-06 22:37:08 +0000 (Sun, 06 Feb 2011) | 1 line

Import Clean perl-install
------------------------------------------------------------------------
r448 | dmorgan | 2011-02-06 23:02:36 +0000 (Sun, 06 Feb 2011) | 1 line

Add cleaned files
------------------------------------------------------------------------
r449 | dmorgan | 2011-02-06 23:03:52 +0000 (Sun, 06 Feb 2011) | 1 line

Import cleaned kernel
------------------------------------------------------------------------
r450 | dmorgan | 2011-02-06 23:23:30 +0000 (Sun, 06 Feb 2011) | 1 line

Import cleaned tools
------------------------------------------------------------------------
r451 | dmorgan | 2011-02-07 00:01:56 +0000 (Mon, 07 Feb 2011) | 1 line

Import stage1
------------------------------------------------------------------------
r473 | dmorgan | 2011-02-07 11:20:50 +0000 (Mon, 07 Feb 2011) | 1 line

Import images
------------------------------------------------------------------------
r497 | ennael | 2011-02-09 21:02:06 +0000 (Wed, 09 Feb 2011) | 2 lines

readd missing file

------------------------------------------------------------------------
r502 | pterjan | 2011-02-09 23:30:32 +0000 (Wed, 09 Feb 2011) | 1 line

Import rescue
------------------------------------------------------------------------
r603 | ennael | 2011-02-24 13:32:54 +0000 (Thu, 24 Feb 2011) | 2 lines

import cleaned rescue for Mageia

------------------------------------------------------------------------
r2061 | tv | 2011-10-20 20:32:19 +0100 (Thu, 20 Oct 2011) | 1 line

import (cleaned) doc
------------------------------------------------------------------------

(useful to have a more standard partition_table::lvm)

  o do not show partition types which have no associated filesystem
    for LVM LV (#32326)

nb: we still would need to have pt_type per partition_table types

  {pt_type} is always a number, {fs_type} is always a string
- introduce set_isFormatted()
  (to ensure {notFormatted} but also {fs_type_from_magic} and {bad_fs_type_magic} are updated)
- don't use 0x483 for ext3 anymore (same for reiserfs...),
  the type_name gives both a pt_type and a fs_type

- many accessors from partition_table removed (type2fs(), fs2pt_type())
- remove isThisFs() (not useful anymore since we can use directly {fs_type})
- remove isFat() (inline the function)
- other isXXX() from partition_table are moved to fs::type
- part2name() is now fs::type::part2type_name
- name2pt_type() is now fs::type::type_name2subpart()
- partition_table::important_types() is now fs::type::type_names()
- fsedit::typeOfPart() is now fs::type::fs_type_from_magic()

- no need to truncate type_name since they are shorter

this will allow defining {fs_type} which will always be a string
whereas {pt_type} will always be a number