[RFC][BZ #16046] dl_iterate_phdr should not expose internal stub.

[Roland McGrath]

macro should participate in the fix, since it was his change that
introduced the regression.

Firstly, there should be a regression test for this (i.e. a test that
passes under 2.17 and breaks under 2.18).

I don't think that's the best way to implement it.  Look at the !SHARED
version of dl_iterate_phdr.  It's doing a special case for the main program
because it wasn't in the list.  After macro's change, it is in the list.
So it seems cleanest to drop the !SHARED special-case code altogether and
instead fill out _dl_main_map sufficiently so the general-case code dtrt
for it too.


Thanks,
Roland