-fno-weak, typeinfo, and AIX

[Jamie Lokier]

Mark Mitchell wrote:
>     Jamie> Oh dear.  We have an address in unmapped memory.  Maybe
>     Jamie> something will get mapped in it later, such as typeinfo C
>     Jamie> from libC.so.
> 
> Good point.  
> 
> I do not believe the ABI committee considered this situation.
> 
> I suspect that at this point, we're out of luck, but I'm willing to
> entertain suggestions.

Something to do with malloc() at dynamic load time?  For each typeinfo
you should have one dynamically allocated structure that won't go away
when the module referencing it is unloaded.  The structure can contain
just a pointer to the real typeinfo, or the real typeinfo itself.

-- Jamie