(1) Backport the upstream patches for recognizing the command line option
--key /dev/mapper/VG-LV🔑password
Similarly to the backports for guestfs-tools BZ#2209280 and libguestfs
BZ#2209279, here we need to update the common submodule (thankfully we
need not excise any hunks -- we had to do that for libguestfs).
Unlike those "single-step" submodule updates however, for virt-v2v we
bridge the same submodule commit range 70c10a079a30..b636c3f20a1b in
two steps, stopping at commit 38e6988c1864 in the middle. We do that
simply because that's how upstream virt-v2v moved; i.e., there are two
upstream patches to cherry-pick for advancing our submodule reference.
(2) In dist-git commit ef9a918d7e, there was a typo: the "test" command
was left out. Therefore even our simple test conversion has not been
invoked -- see e.g.
<https://download.eng.bos.redhat.com/brewroot/vol/rhel-9/packages/virt-v2v/2.3.4/2.el9/data/logs/x86_64/build.log>:
> + -s test-data/phony-guests/windows.img
> /var/tmp/rpm-tmp.UMecKA: line 48: -s: command not found
Unfortunately, incorrectly (not) invoking "test -s" has had results
identical to invoking "test -s" correctly and "test -s" failing;
therefore we've been just silently skipping our simple conversion,
assuming "no non-empty guest disk images".
Fix this typo...
(3) ... and then run the sole "test-v2v-fedora-luks-on-lvm-conversion.sh"
test from the test suite, for verifying the backport in the build
environment. (The idea for the future is that we'd run such individual
tests whenever backporting patches.) For this, we also start depending
(at build time) on the sqlite3 command.
resolves: rhbz#2168506
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Laszlo Ersek