Olaf Hering

This was rebased, take the chance to remove the `v2` from the commit message and also the the MR text.

I see nothing particular wrong with it. Since all systems are (most likely) headless, this does not seem to change behavior.

I think you have a point here. For `olh-autoinst` I introduced a separation of firmware and kernel interface. ```` firmware_terminal='serial|graphical|native' operating_system_terminal='serial|graphical' serial_port_zerobased='0' serial_speed='115200' ```` With Open Firmware the kernel is...

Could the mentioned `separation of firmware and kernel interface` above cover all cases? It is probably too x86 centric. Using `native` as firmware mode could mean "reuse firmware UI", and...

For me a simple `if d.kernel|d.initrd:`fixed `cobbler sync`

What is the expected outcome anyway? For me it fails like this: ```` [ 129s] $ (cd _build/.sandbox/f948a16328af0270471659460c901e07/default && /usr/bin/diff -u tests/test-dirs/locate/context-detection/cd-mod_constr.t/run.t tests/test-dirs/locate/context-detection/cd-mod_constr.t/run.t.corrected) [ 129s] --- tests/test-dirs/locate/context-detection/cd-mod_constr.t/run.t 2021-04-14 12:16:29.943666650 +0000...

Well, I added this change to `ocaml-merlin.spec` to fix build. ````patch %check +sed -i~ ' +s@"lib/ocaml/string.ml"@"%{ocaml_standard_library}/string.mli"@ +' tests/test-dirs/locate/context-detection/cd-mod_constr.t/run.t +diff -u "$_"~ "$_" && exit 1 %ocaml_dune_test ````

The change above fails when applied on a 32bit host, such as `i586/armv7l/armv6l`. Not sure why the testcase makes a difference between 64bit and 32bit builds. The question above still...

One more review required to get enable the `Merge` knob. I think this change is not controversial.

I have no resources to do further work in this specific bug.