similar to: supermin and dpkg-divert

--- configure.ac | 1 + src/config.ml.in | 1 + src/dpkg.ml | 1 + 3 files changed, 3 insertions(+) diff --git a/configure.ac b/configure.ac index 2141540..99ea913 100644 --- a/configure.ac +++ b/configure.ac @@ -92,6 +92,7 @@ AC_PATH_PROG(APT_GET,[apt-get],[no]) AC_PATH_PROG(DPKG,[dpkg],[no]) AC_PATH_PROG(DPKG_DEB,[dpkg-deb],[no]) AC_PATH_PROG(DPKG_QUERY,[dpkg-query],[no])

--- src/build.ml | 43 ++++++++++++++++++++++++++----------------- src/chroot.ml | 12 +++++++----- src/dpkg.ml | 17 +++++++++++++++-- src/ext2.ml | 8 +++++++- 4 files changed, 55 insertions(+), 25 deletions(-) diff --git a/src/build.ml b/src/build.ml index 9225184..205701b 100644 --- a/src/build.ml +++ b/src/build.ml @@ -106,11 +106,7 @@ let rec build debug *) let files =

* Richard W.M. Jones: > This seems to change the result of this (very important and complex) > function. Since you didn't answer my followup message, I take it that I failed to convince you. I noticed that I can get away with leaving the structure of the function as-is, just changing the type... What do you think about this diff? Cheers, -Hilko diff --git a/src/build.ml

--- src/dpkg.ml | 3 ++- src/package_handler.ml | 1 + src/package_handler.mli | 4 ++++ src/pacman.ml | 2 +- src/rpm.ml | 2 +- 5 files changed, 9 insertions(+), 3 deletions(-) diff --git a/src/dpkg.ml b/src/dpkg.ml index c28354a..5a650b8 100644 --- a/src/dpkg.ml +++ b/src/dpkg.ml @@ -166,7 +166,8 @@ let dpkg_get_all_files pkgs = let config = try

I have done some printf profiling and found that supermin's calls to dpkg for individual packages are quite expensive. Here are some patches that gather all information on demand where possible. On my Debian/unstable-based workstation at home, preparing a minimal appliance using $ ./supermin --use-installed --prepare bash -o supermin.d now takes ~3.5s (previously ~15s). Turning that

Hi, let's make supermin use /etc/os-release as primary source instead of the various release files in /etc; apparently distros (e.g. openSUSE) are starting removing them. Thanks, Pino Toscano (2): Add simple handling of /etc/os-release Use os-release to detect the distro src/Makefile.am | 3 +++ src/dpkg.ml | 3 ++- src/os_release.ml | 78

--- src/chroot.ml | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/src/chroot.ml b/src/chroot.ml index b5c1e53..9e522d9 100644 --- a/src/chroot.ml +++ b/src/chroot.ml @@ -26,7 +26,9 @@ let build_chroot debug files outputdir = List.iter ( fun file -> try - let path = file.ft_source_path in + let path = if file_exists file.ft_source_path +

Allow an alternate libc, such as dietlibc, to be used for the init binary in the supermin appliance. Rich.

Hi, this work makes supermin use the rpm library instead of invoking rpm directly. This, together with a needed refactoring of the dependency resolution, should help in make supermin faster on rpm-based systems. Surely the patches will still need polishing, especially for behaviours of newly added stuff, but at least it's a good starting point. Noting that you need rpm-devel on most of rpm

See patch #2 for more explanation. Pino Toscano (2): prepare: keep config_files available for longer prepare: create a really empty base.tar.gz with no config files src/mode_prepare.ml | 87 +++++++++++++++++++++++++++------------------ 1 file changed, 52 insertions(+), 35 deletions(-) -- 2.17.2

v1 -> v2: - If we split out the init program into a separate init/ directory, that makes it much easier to build against an alternate libc. I tried to build against uClibc, but uClibc requires an entire build chain, which looked like it was going to be a massive ballache. Rich.

Move the removal of the files which don't exist or cannot be read, doing it after the filtering with excludefiles and hostfiles files. This avoid stat'ing files which will be excluded later anyway, hence reducing the I/O during a supermin build phase. --- src/build.ml | 32 ++++++++++++++++---------------- 1 file changed, 16 insertions(+), 16 deletions(-) diff --git a/src/build.ml

This is a "merge" of two previous series: https://www.redhat.com/archives/libguestfs/2018-December/msg00015.html https://www.redhat.com/archives/libguestfs/2018-December/msg00020.html The goal is to handle better situations like: - empty file in the appliance directory - no config files available in the packages to include in an appliance Compared to the two series, the changes are: -

dpkg used to ship with a symlink /usr/sbin/dpkg-divert -> ../bin/dpkg-divert that recently got removed nd caused breakage (libguestfs build failures) because configure had chosen the wrong binary. --- configure.ac | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/configure.ac b/configure.ac index 3a0d25d..72ffb92 100644 --- a/configure.ac +++ b/configure.ac @@ -110,10

Hi, to ease debugging issues with appliances (e.g. when used in libguestfs), using --include-packagelist will add a file containing the list of all the packages used. Thanks, Pino Toscano (3): ext2: add ext2fs_chmod and ext2fs_chown chroot: factor out file copy code Add --include-packagelist src/build.ml | 42 +++++++++++++++++++++++++------ src/chroot.ml | 29

When building the mini initrd, previously we copied the modules into the initrd as-is, so for example if the module was xz-compressed, we copied the foo.ko.xz file to the initrd. This requires that the mini init binary is linked to zlib & lzma, so that it knows how to uncompress these modules when insmoding them at boot time. Also since the init is statically linked, it required _static_

The new run_shell helper is a copy of run_python, and will be used by upcoming changes. Signed-off-by: Olaf Hering <olaf at aepfle.de> --- src/supermin_utils.ml | 9 +++++++++ src/supermin_utils.mli | 5 +++++ 2 files changed, 14 insertions(+) diff --git a/src/supermin_utils.ml b/src/supermin_utils.ml index f98e09a..cb8a27e 100644 --- a/src/supermin_utils.ml +++ b/src/supermin_utils.ml

Check the ID field in /etc/os-release on the current system, before checking for the other old-style release-/version-like files in /etc. Some distributions (openSUSE Thumbleweed) are starting to remove them, breaking the supermin detection. --- src/dpkg.ml | 3 ++- src/pacman.ml | 5 +++-- src/rpm.ml | 15 +++++++++------ 3 files changed, 14 insertions(+), 9 deletions(-) diff --git

Hi, few cleanups in the supermin codebase; no actual functional change. Thanks, -- Pino Toscano (3): ext2: simplify tracking of visited modules utils: remove unused run_python function Add and use an helper error function src/build.ml | 20 +++++----------- src/dpkg.ml | 4 +--- src/ext2_initrd.ml | 10 ++++---- src/kernel.ml | 27

This is just refactoring, with no behaviour changes. --- src/mode_prepare.ml | 40 ++++++++++++++++++++-------------------- 1 file changed, 20 insertions(+), 20 deletions(-) diff --git a/src/mode_prepare.ml b/src/mode_prepare.ml index 7c8221e..7759c58 100644 --- a/src/mode_prepare.ml +++ b/src/mode_prepare.ml @@ -128,28 +128,28 @@ let prepare debug (copy_kernel, format, host_cpu, * be