[PATCH] ld: Fix segfault in populate_publics_stream

[Mark Harmstone]

 > it ... should actually be accompanied by a code comment.

Not really, as it was a straightforward mistake in my original patch. If I'd
included a NULL check to begin with, I doubt anyone would have batted an
eyelid.

I could have easily worked around this in my test, either by adding in a dummy
section or by using the preprocessor rather than .equ, but figured that any
way of causing a segfault is a bug that needs to be fixed.

Mark