[systemd-devel] Antw: Re: Antw: Re: Antw: Re: Arbitrary restrictions (e.g. for RuntimeDirectory)

[Lennart Poettering]

On Mi, 15.05.19 07:48, Ulrich Windl (Ulrich.Windl at rz.uni-regensburg.de) wrote:

> >> Or more like this (in the user directory):?
> >> > systemd‑analyze verify systemd/iotwatch.target.in
> >> Failed to open /dev/tty0: Permission denied
> >> Failed to load systemd/iotwatch.target.in: Invalid argument
> >
> > That's not a valid unit file name. On systemd unit files are named in
> > a very strict fashion, and this is documented. A service unit file
> > name must be suffixed with ".service" and a target unit file name must
> > be suffixed with ".target". The suffix ".target.in" is not valid for a
> > unit file.
>
> So the type to check is deduced from the name? my "target.in" is the template
> used by the generator to create the real target....

Yes, the unit type is derived from the unit file name. If a unit file
name ends in ".service" it means it may have a [Service] section in
it, with an ExecStart= line. otoh if it ends in ".mount" it instead
may have [Mount] section with an Option= line.

> Usually I wouldn't associate "Inavalid argument" with "syntax error". I'd
> prefer "unknown suffix" over "Invalid argument". There's less room for
> speculation then...

Please file an RFE issue, or even better prepare a patch!

Lennart

--
Lennart Poettering, Berlin