lightdm: allow dm-tool lock to work properly

[philandstuff]

Fixes #5038.

The dm-tool lock command fails to work for reasons described in #5038:
it tries passing arguments to the xserver command, but the wrapper
throws the arguments away and replaces them with different arguments.

I fixed this by specifying the xserver args in the lightdm.conf file and
having the wrapper pass them through. This works for lightdm on
startup, and also works for dm-tool lock when it later passes a
different set of arguments to start a separate X server on another
display.

Ping @ocharles @wkennington, who are listed as maintainers of the lightdm package.

[mention-bot]

By analyzing the blame information on this pull request, we identified @ocharles, @viric and @wkennington to be potential reviewers

[philandstuff]

It's probably worth saying that dm-tool lock doesn't, by itself, lock the screen so there's still more work to be done after this PR.

[ocharles]

This looks OK from me, @viric, @wkennington - any thoughts?

[zimbatm]

@philandstuff do you mind rebasing on top of master ?

There is just one thing that needs to be fixed: $@ needs to be quoted for shell expansion to work properly. So "$@" instead of just $@

[philandstuff]

hmm, it looks like 9be012f by @obadz (merged in #11338) has already added the "$@" pass-through. It certainly looks like it was solving a related problem.

The only thing remaining in this PR is whether ${dmcfg.xserverArgs} should remain hardcoded in the xserver-wrapper script, or if it should be put in to lightdm.conf. I kind of feel that putting the xserverArgs in lightdm.conf makes some of the logic in the xserver-wrapper unnecessary, but since I opened this PR I've moved to gdm instead of lightdm and so I don't have the setup to test this change any more. #11338 changed the semantics of xserverArgs so I don't have confidence in this.

I've rebased the commit and I'll leave this open for a few days if someone wants to take it up, but if there's no further activity within a week I suggest closing this PR.

As an aside, I wonder if #5038 was fixed by #11338?

[obadz]

@philandstuff I wonder that too

[obadz]

@philandstuff actually I'm pretty sure it's fixed

[philandstuff]

Closing (re #11686 (comment) )