Skip to content
Commits on Feb 5, 2013
  1. @rwmjones

    THIS REPOSITORY HAS MOVED!

    rwmjones committed
    The tool formerly called 'febootstrap' is now called 'supermin'
    and the new repository is here:
    
    https://github.com/libguestfs/supermin
    
    The new home page is here:
    
    http://people.redhat.com/~rjones/supermin/
    
    - RWMJ 2013-02-05
Commits on Jan 22, 2013
  1. @rwmjones

    helper: Drop supplemental groups when using -g command line option.

    rwmjones committed
    If we don't call setgroups then the febootstrap-supermin-helper
    process will inherit a random set of supplemental groups from the
    parent process.
    
    (Thanks Steve Grubb)
Commits on Dec 22, 2012
  1. @rwmjones

    febootstrap 3.21.

    rwmjones committed
  2. @rwmjones
  3. @rwmjones

    Copy editing of --help output.

    rwmjones committed
  4. @rwmjones

    Use --packager-config option to replace --yum-config and --pacman-con…

    rwmjones committed
    …fig.
    
    This makes it easier to use this option from libguestfs.
Commits on Dec 21, 2012
  1. @rwmjones

    archlinux: Add pacman configuration option.

    Evaggelos Balaskas committed with rwmjones
Commits on Dec 14, 2012
  1. @rwmjones

    arch: Change location of packages.

    Evaggelos Balaskas committed with rwmjones
    Arch User Repository - https://aur.archlinux.org/
    has changed the package location
    
    eg.
    old location:
    http://aur.archlinux.org/packages/hivex/hivex.tar.gz
    
    new location:
    https://aur.archlinux.org/packages/hi/hivex/hivex.tar.gz
Commits on Aug 31, 2012
  1. @rwmjones

    Version 3.20.

    rwmjones committed
  2. @rwmjones

    Add febootstrap-supermin-helper --copy-kernel option.

    rwmjones committed
    Because libvirt wants to set an SELinux label on the kernel, we can't
    just symlink to it (because we don't want to change the SELinux label
    on the kernel in /boot, and in any case it's owned by root so we
    cannot).
    
    Add the --copy-kernel option so that libguestfs can indicate that it
    wants to copy the kernel instead of symlinking it.
Commits on Aug 22, 2012
  1. @rwmjones
Commits on Jul 31, 2012
  1. @rwmjones

    Version 3.19.

    rwmjones committed
Commits on Jul 30, 2012
  1. @rwmjones

    Update gnulib to latest.

    rwmjones committed
  2. @rwmjones

    yum-rpm: Don't automatically exclude the kernel.

    rwmjones committed
    People might want an appliance that includes the kernel.  Also Fedora
    now has a package "kernel-debug" that didn't match this rule so it was
    being pulled into the libguestfs appliance by accident.
    
    Better to exclude all unwanted kernel packages in libguestfs instead.
Commits on Jun 29, 2012
  1. @rwmjones
Commits on Jun 26, 2012
  1. @rwmjones

    helper: Don't print usage after error messages.

    rwmjones committed
    It obscures the error.
  2. @rwmjones

    Fix dependencies for OCaml 4.00.0.

    rwmjones committed
    ocamldep seems to put a space before the ':' character.
  3. @rwmjones

    helper: Fix -u and -g options when host uses LDAP or NIS (RHBZ#835130).

    rwmjones committed
    The getpwnam/getgrnam interfaces are not very well specified in the
    case where we are looking up a name that does not exist.
    
    Previously the code assumed that if errno was set, that would
    distinguish an error from the not found case.  However this is
    certainly not true for LDAP, where errno == ENOENT might indicate not
    found.  Other errno values could be returned too.
    
    The specific problem is that if the user specifies a numeric -u or -g
    option, then the first lookup (eg) getpwnam ("42") could return NULL
    with a non-zero errno, and that would not indicate an error.
    
    Change the parsing of both -u and -g options as follows:
    
    (1) If getpwnam/getgrnam returns an error, record it in saved_errno,
    but do NOT fail at this point (as we did before).
    
    (2) If the name is numeric, return the numeric value regardless of
    whether getpwnam/getgrnam returned any error.
    
    (3) Otherwise print an error message and fail.  Ensure that
    saved_errno is printed for debugging purposes.
Commits on Jun 18, 2012
  1. @rwmjones

    Version 3.18.

    rwmjones committed
  2. @rwmjones

    RHEL 5: Use mke2fs -T ext2 or -t ext2 as appropriate.

    rwmjones committed
    On RHEL 5, mke2fs used the -T option to specify the filesystem type.
    
    Change the configure script so it can detect either method, and then
    use the correct method in febootstrap-supermin-helper.
Commits on Jun 17, 2012
  1. @rwmjones

    RHEL 5: Don't fail if YumBase.setCacheDir method doesn't exist.

    rwmjones committed
    This method didn't exist in RHEL 5-era yum.
Commits on Jun 15, 2012
  1. @rwmjones

    helper: Add IBM Virtual SCSI (ibmvscsic.ko) driver to the initramfs.

    rwmjones committed
    This allows libguestfs to use virtio-scsi over the spapr-vscsi
    transport on ppc64.
Commits on Jun 14, 2012
  1. @rwmjones

    Version 3.17.

    rwmjones committed
  2. @rwmjones

    Increase scsi wait time again to ~ 10 minutes.

    rwmjones committed
    Enumerating 255 SCSI disks in Koji really does take a long time.
Commits on Jun 13, 2012
  1. @rwmjones

    Version 3.16.

    rwmjones committed
  2. @rwmjones

    Make SCSI wait longer.

    rwmjones committed
    On slow machines (ie. Koji) a 2 second wait isn't nearly enough for
    the SCSI disks to become available.
  3. @rwmjones

    Don't access /proc/uptime after switching root.

    rwmjones committed
    /proc isn't mounted at this point.
Commits on Jun 12, 2012
  1. @rwmjones

    Version 3.15.

    rwmjones committed
  2. @rwmjones

    Update gnulib to latest.

    rwmjones committed
  3. @rwmjones

    init: Add 'root=...' parameter to specify the appliance.

    rwmjones committed
    This is required when using virtio-scsi with libguestfs.
Commits on Apr 27, 2012
  1. @rwmjones
Commits on Mar 30, 2012
  1. @rwmjones
  2. @rwmjones
  3. @rwmjones
  4. @rwmjones

    febootstrap 3.14.

    rwmjones committed
Something went wrong with that request. Please try again.