Bug#939563: pulseaudio: core-util.c: Failed to create secure directory (/nonexistent/.config/pulse)

[mike at mike-burns.com]

I looked at the logs more closely and identified a minimal way to reproduce this: switch user to `nobody` as if we were systemd.

```
sudo systemd-run su nobody -c "/bin/true"
```

When I run that, I get this in the logs (abbreviated):

```
Sep 07 21:07:07 fling sudo[654179]:     mike : TTY=pts/3 ; PWD=/etc/cron.daily ; USER=root ; COMMAND=/usr/bin/systemd-run su nobody -c /bin/true
Sep 07 21:07:07 fling sudo[654179]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=1000)
Sep 07 21:07:07 fling systemd[1]: Started run-r4a62261dbedd4a1cbcbab7e0f38f16db.service - /usr/bin/su nobody -c /bin/true.
Sep 07 21:07:07 fling sudo[654179]: pam_unix(sudo:session): session closed for user root
Sep 07 21:07:07 fling su[654182]: (to nobody) root on none
Sep 07 21:07:07 fling su[654182]: pam_unix(su:session): session opened for user nobody(uid=65534) by (uid=0)
Sep 07 21:07:07 fling systemd[1]: Created slice user-65534.slice - User Slice of UID 65534.
Sep 07 21:07:07 fling systemd[1]: Starting user-runtime-dir at 65534.service - User Runtime Directory /run/user/65534...
Sep 07 21:07:07 fling systemd[1]: Finished user-runtime-dir at 65534.service - User Runtime Directory /run/user/65534.
Sep 07 21:07:07 fling systemd[1]: Starting user at 65534.service - User Manager for UID 65534...
Sep 07 21:07:07 fling (systemd)[654184]: pam_unix(systemd-user:session): session opened for user nobody(uid=65534) by (uid=0)
Sep 07 21:07:07 fling systemd[654184]: Queued start job for default target default.target.
Sep 07 21:07:07 fling systemd[654184]: Created slice app.slice - User Application Slice.
Sep 07 21:07:07 fling systemd[654184]: Created slice session.slice - User Core Session Slice.
Sep 07 21:07:07 fling systemd[654184]: Reached target paths.target - Paths.
Sep 07 21:07:07 fling systemd[654184]: Reached target timers.target - Timers.
Sep 07 21:07:07 fling systemd[654184]: Starting dbus.socket - D-Bus User Message Bus Socket...
[...]
Sep 07 21:07:07 fling systemd[654184]: Listening on pulseaudio.socket - Sound System.
[...]
Sep 07 21:07:07 fling systemd[654184]: Starting pulseaudio.service - Sound Service...
[...]
Sep 07 21:07:07 fling pulseaudio[654201]: Failed to create secure directory (/nonexistent/.config/pulse): No such file or directory
```

What's happening is systemd is switching from root to nobody, and in the process it's activating the `default.target` for a user.

OK, why is this relevant? If you look at the journal around the daily pulseaudio error and scroll up a bit, you can see that the errors occur at the same time. It starts like this:

```
Sep 07 06:25:01 fling CRON[631236]: pam_unix(cron:session): session opened for user root(uid=0) by (uid=0)
Sep 07 06:25:01 fling CRON[631237]: (root) CMD (test -x /usr/sbin/anacron || { cd / && run-parts --report /etc/cron.daily; })
Sep 07 06:25:01 fling su[631260]: (to nobody) root on none
Sep 07 06:25:01 fling su[631260]: pam_unix(su:session): session opened for user nobody(uid=65534) by (uid=0)
```

I only have one functioning daily cronjob, `locate`. It eventually calls `updatedb.findutils` with `LOCALUSER=nobody`, and then calls:

```
su $LOCALUSER ...
```

And that's how all our trouble begins.

So a less-minimal way to reproduce this is:

```
sudo systemd-run /etc/cron.daily/locate
```

Anyway, I don't know how to solve this, but that's the cause of the noise. Moving the `locate` daily to a timer would (potentially) cause us to no longer see the errors, but it'd mask that the `nobody` user probably shouldn't cause all of (user) `default.target` to load. I'll leave the solution to all this to someone smarter.

-Mike