ld 2.11: unaligned memory access for unions

[Ian Lance Taylor]

Nick Clifton <nickc@cambridge.redhat.com> writes:

> > The internal structure should be aligned appropriately any place
> > where BFD uses it.
> 
> But presumably it is not.

Do you know that or are you just assuming it?

> > If it is not, where is it being used?
> 
> I do not know.  I was not able to repeat the failure mode myself.  I
> assume that something in Eberhard's environment is generating
> unaligned internal structures.

Well, personally, I think this needs further investigation.  I don't
think the hack of adding a function to call memcpy is appropriate
until we understand why it is needed.  Perhaps there is a bug in gcc.
Perhaps BFD is failing to align structures when it should.  Perhaps
the real problem has been completely misdiagnosed.  Perhaps I am wrong
and BFD is really doing something wrong; perhaps we can fix that
instead.

Ian