[readonly/geninitrd] Updated tag 10000.32

[draenog]

The tag '10000.32' was replaced with a new tag. It previously
pointed to:

 954f1ca... Release 10000.32 (experimental).

NOTE: People pulling from the repository will not get the new tag.

New tag information:

Tagger: Arkadiusz Miśkiewicz <arekm at maven.pl>
Date: 1298104102 +0000

    Release 10000.32 (experimental)
    
    git-svn-id: http://svn.pld-linux.org/svn/geninitrd/tags/10000.32@12165 11da6c6d-24c7-0310-97e0-b6d17e038e62

Changes since the last tag '10000.30':

Arkadiusz Miśkiewicz (18):
      Allow tuxonice snapshot device be device mapper device(LUKS, LVM, dmraid). Patch by Krzysztof Krakowiak.
      Mention xz and lzma for initrd compresion.
      Add COMPRESS=none/no support (uncompressed initrd).
      Use fastest compressor first.
      Handle /dev/mdXpY as root device.
      Try to find main device in case when rootfs in on a md partition (instead of whole md device).
      By default rootdev_new is the same as rootdev.
      We have /dev/urandom, so also store /dev/random (luks can use it for example).
      Mount rootfs from initramfs with rootfsflags options.
      ccis needs sd_mod (areq).
      Revert, that was wrong.
      Try to activate luks only if not yet active (because cryptsetup is run twice).
      debugshell after failing to chroot.
      root dev finding: fallback to calculating based on initrd /dev file
      Report RCSID for loaded modules, too.
      Revert, this wasn't meant to be in previous commit.
      Nicer way to display rcsid.
      Release 10000.32 (experimental).

Artur Frysiak (8):
      Allow rootfs on crypted device
      Simplify dm_longname()
      Use DM subsystem name not UUID for LVM/crypt detection
      Simplify dm_name()
      Check source device before cryptsetup run
      Don't drop to shell if LVM is used but root device isn't LV
      Try to setup encryption after soft-RAID and LVM setup
      Make addons executable

Elan Ruusamäe (26):
      - more generic and more exact finding code
      - pcimap not used when lspci invoked with -n
      - as uknown what to do for multiple module match at least load them for now
      import extract-ikconfig from kernel sources
      recognize xz compression (patch by arekm)
      use copy of extract-ikconfig from kernel sources what can process any file     containing it (vmlinux or kernel mod with or without compressions)
      include extract-ikconfig
      - use mktemp
      don't leak parentDir variable
      - 10000.31 second experimental (can extract all compressed kernels for CONFIG)
      - refentry2man dep
      Consistent timestamp in release tarball files. in fact, package all files to tarball
      - multipath install lib64 safe (why it was just lib anyway?)
      rename dm_* helpers to avoid confision what they do and make them more generic
      Use double quotes (more readable in this context)
      Simplify main code by moving state checks to modules itself
      partial testcase
      hack up test using actual functions
      add setup so detection results could be later tweaked for testing
      switch to parsing in shell instead of awk
      - compare device bitwise. LP#715930
      update test
      there is no bitwise matching only DEVICE_ANY
      sort modules that ones who depend on libata are first
      - update test
      setup TZ in export

sparky (2):
      Renamed rootdev_alternate and rootdev_alternative to rootdev_alias to prevent     further typos.
      Fixed one more typo.