dlclose()

[Jim Kingdon]

> Am I mistaken in the observation that people rarely post test suite
> results when they submit patches?

Well, solib.exp only runs on HPUX and if there are other shared
library tests, I didn't see them.

I'd probably argue that the distributed testing resulting from getting
patches checked in is more effective QA than getting obsessed with
testsuites but that isn't really the point in the sense that I'd like
to see both.  On the testsuite side, I'm looking into the weird.exp
failures - what I've seen so far as that it has nothing to do with
relocations (at least not on my Red Hat Linux 6.1 system), that it has
nothing to do with how the stabs are laid out (they are correct), but
it has to do with C++ code which applies the baseclass offset twice,
thus pointing the wrong place.  The two places in question are:

cp-valprint.c:572
		  if (target_read_memory (address + boffset, base_valaddr,
					  TYPE_LENGTH (baseclass)) != 0)

cp-valprint.c:599

	cp_print_value_fields (baseclass, real_type, base_valaddr, offset + boffset, address,
			       stream, format, recurse, pretty,
		     (struct type **) obstack_base (&dont_print_vb_obstack),
			       0);