openrc: remove duplicate multicall binaries, symlink instead

[mqqc]

There is a bunch of binaries in OpenRC that are essentially multicall binaries, but redundantly compiled multiple times:

https://github.com/OpenRC/openrc/blob/0.45.2/src/einfo/meson.build
https://github.com/OpenRC/openrc/blob/0.45.2/src/service/meson.build
https://github.com/OpenRC/openrc/blob/0.45.2/src/mark_service/meson.build
https://github.com/OpenRC/openrc/blob/0.45.2/src/value/meson.build

By symlinking them to one 'main' multicall binary per group, we can reduce rootfs size. As the minimum ARM binary size is pretty big thanks to default 64 kiB alignment in newer binutils, this can be pretty sizable. This saved about 8MB on my test build.

[jsbronder]

Is this something that's been proposed upstream and rejected? It looks easy enough to patch the build itself.

diff --git a/src/einfo/meson.build b/src/einfo/meson.build
index df11d5fd..b402961c 100644
--- a/src/einfo/meson.build
+++ b/src/einfo/meson.build
@@ -1,6 +1,5 @@
 einfo_execs = [
   'einfon',
-  'einfo',
   'ewarnn',
   'ewarn',
   'eerrorn',
@@ -22,11 +21,15 @@ einfo_execs = [
   'veoutdent',
   ]
 
+executable('einfo',
+  ['einfo.c', version_h],
+  include_directories: [incdir, einfo_incdir, rc_incdir],
+  link_with: [libeinfo, librc],
+  install: true,
+  install_dir: rc_bindir)
+
 foreach exec: einfo_execs
-  executable(exec,
-    ['einfo.c', version_h],
-    include_directories: [incdir, einfo_incdir, rc_incdir],
-    link_with: [libeinfo, librc],
-    install: true,
-    install_dir: rc_bindir)
+  install_symlink(exec,
+    install_dir: rc_bindir,
+    pointing_to: 'einfo')
 endforeach

It seems fine to me, but I would rather carry these as a patchset with the Upstream-Status tag, https://www.openembedded.org/wiki/Commit_Patch_Message_Guidelines#Patch_Header_Recommendations:_Upstream-Status. If you like, I'm happy to submit to upstream on your behalf.

My preference for the patchset is that'll it'll be more likely to work over version updates. If the patches apply, we likely are getting every binary. If not, then we should be looking at the list of executables in meson regardless.

[mqqc]

Sorry for not getting back to you earlier, I got caught up in a few other things.

That's fine and makes sense to me, I'll adjust and submit it upstream.