syslogd: Read configuration outside of init loop

[jfree]

Summary

Move all configuration-file code outside of the initialization loop and
into its own set of functions. Create a readconfigfile() to open the
config and call parseconfigfile() to parse it.

The init() function no longer returns if there was a failure to open the
configuration file. Initialization will be finished, using the backup
logging rules: *.ERR to /dev/console and *.PANIC to all logged in users.

[jfree]

Update diff with context

[markj]

usr.sbin/syslogd/syslogd.c

2468

This inverts the flag, but the commit message makes this seem unintentional.

2662

We no longer return early in this case. I think that's probably ok though?

[jfree]

usr.sbin/syslogd/syslogd.c

2468

This inverts the flag, but the commit message makes this seem unintentional.

Ah, I ended up removing the allow_includes parameter later because of this unintentional flip.

I don't see why recursive includes would be an issue, unless you had an inclusion loop. Maybe this is there to mitigate that?

2662

We no longer return early in this case. I think that's probably ok though?

Yeah, I went through the logic when I made this change. There shouldn't be an error in removing the early return.

[markj]

usr.sbin/syslogd/syslogd.c

2468

Yes, I'm fairly sure this is a low-tech mechanism to avoid infinite loops.

2662

Ok. In general I would suggest including this kind of information in the review/commit description, since otherwise it's hard to tell if this is intentional or not.

[jfree]

Address Mark's suggestions

[markj]

usr.sbin/syslogd/syslogd.c

2549

I would drop this comment too, in the spirit of "comments should explain why, not what."

[jfree]

Remove comment in readconfigfile()