RTEMS C++ Issues

[Robert S. Grimes]

Ian Caddy wrote:
> Hi Bob,
>
> I realise I am coming a bit late into this conversation, which I have 
> been loosely following, but...
Better late than never - thanks!
>
> Have you checked the section on C++ on the wiki:
Yes, and while I understand it in principle, the problem I'm having is 
reproducing it in practice, especially wrt the automake/autoconf tools.
> We had to get C++ working for our applications people, and these 
> applications run under RTEMS but are not directly linked with RTEMS, 
> but linked independantly using the RTEMS compiler, then our main RTEMS 
> system starts them up at startup.
>
> I also had problems with the global CTORS which also causes problems 
> with exceptions, as these are lumped in with CTORS.
I haven't even begun to consider exceptions yet!  ;-)
>
> Somewhere in your code (after kernel start, but before init) call a 
> function that either the compiler/linker creates or your own one to 
> init the CTOR function handlers.
This is what I understand is the process - again, it's the mechanics I'm 
unsure of...
>
> In our case we wrote our own global CTOR init, as it was being run 
> outside of RTEMS directly.
>
> I hope this helps.
While your approach is different from mine, I think the info you've 
provided is useful in demonstrating it is not only feasible, but also 
perhaps not quite as daunting as I first thought.  To be honest, I've 
been on vacation for 1.5 weeks, so I'm not exactly sure where I left off 
in my quest, but I'll be resuming it tomorrow.  While it seems I should 
be able to cobble something together that would work for my apps, I'm 
hoping to have it all incorporated into the Virtex PowerPC BSP - to make 
it easier for others!

Thanks!
-Bob
>
> regards,
>
> Ian Caddy
>
>