Skip to content
Commits on Apr 26, 2011
  1. add udevadm settle call after triggering uevents

    This also adds a fun little timer to show how long udev processing took.
    catch it if you can!
Commits on Apr 25, 2011
  1. fixup waiting for root

    this was potentially waiting for 100s for root by default.
Commits on Apr 22, 2011
  1. refactor delete_contents to avoid heap allocation

    we don't need a whole separate function to do the path concatenation --
    just use a static buffer and append with snprintf.
  2. add support for non-udev environments

    $root can be specified as:
    * a block device => /dev/sda2
    * a major/minor pair => 8:2
    * a hex encoded major/minor => 802
  3. refactor and improve wait_for_root

    * don't wait for something that isn't a /dev path
    * change wait delta to .1s instead of 1s
    * remove bail to shell on failure. this isn't necessarily the end.
    * ensure that what shows up is actually a block device
    * return something meaningful
Commits on Apr 21, 2011
Commits on Apr 20, 2011
  1. refactor sanitize_var

    be more strict about what we accept
  2. introduce movemount for moving tmpfs's

    The target root may not have the mount point we're moving. if it doesn't
    exist, umount the tmpfs instead of moving it.
  3. use vfork over fork

    as per vfork(3), since we're only using fork as a prelude to exec, this
    is a viable, lighter, alternative.
  4. use _exit to leave a child process

    _exit(3) will explicitly avoid flushing stream buffers on exit, which is
    desirable for forked processes which may inherit unrelated data in the
    copied FDs. Flushing these buffers would result in erroneously
    duplicated output.
  5. rootdelay <=0 is invalid

Commits on Apr 19, 2011
  1. overhaul later logic to be more robust and simpler.

    * adds support for rootfstype=
  2. udev best be comin down, yo

  3. fix output in err()

Commits on Apr 18, 2011
  1. add missing udev msg

  2. wrap at 80 cols

  3. optimize last_char_is

Commits on Apr 17, 2011
  1. initial commit. works for me!

Something went wrong with that request. Please try again.