Re: [patch][rfc] Complex Relocations gas and bfd

[fche at redhat dot com (Frank Ch. Eigler)]

Nick Clifton <nickc@redhat.com> writes:

>    It is great to see this work being contributed at last.

Yes.  I had been dreading the submission of this code, since it was so
ambitious, and we designed it without outside (upstream) expertise,
and thus expected controversy.

> One point though - with a new feature like this there really ought
> to be some new testsuite entries to check its behaviour.  Ideally
> they should be generic tests (rather than MeP specific).  Is there
> any chance of doing this ?

If this feature is accepted within binutils, then possibly gas could
be extended in a target-neutral way to generate such complex
relocations from sufficiently complex expressions.  Then a collection
of target-independent "label: .word (expression)" source lines could
constitute the test case.  These could culminate in a linker test that
asserts correct evaluation of those expressions, perhaps via "nm".

- FChE