Re: Partial linking interacts badly with N_WARNING and N_INDIR

[Ian Dall <ian at beware dot dropbear dot id dot au>]

Alan Modra writes:
 > Hi Ian,
 >   I think this will fix the problem for you.

Thanks, it seems to.

Alan Modra writes:
 > On Tue, Aug 06, 2002 at 02:15:50AM +0930, Ian Dall wrote:
 > > How should indirect symbols behave with the "ld -x -r" option (a.out)?
 > > 
 > > With the following code:
 > > 
 > > 	.stabs "_daylight",10,0,0,0
 > > 	.stabs "__daylight",1,0,0,0
 > > 
 > > [...]
 > Well, they are local syms, so I suppose you get what you asked for
 > with -x.

Well, I suppose the question is whether they should be considered local
symbols.  Certainly the 01 bit is clear in the encoding, but I don't think the required semantics are that of a local symbol (*). Local symbols do not take
part in further linking. Indirect symbols clearly do impact linking, because
it matters when they are deleted. Ergo, they are not local.

* I'm not really clear what the semantics of indirect symbols are. It
looks like weak symbols are a better alternative, but it would still
be nice if it worked.

Ian