Re: [PATCH] Use back end exceptions for Ada on ia64

[Arnaud Charlet <charlet at ACT-Europe dot FR>]

> Given that sjlj exceptions are completely non-functional (due to a mismatch
> in sizeof(jmp_buf)) and that dwarf2 unwinding is part of the ABI it does

I am unaware of this issue, is there a PR ?
Remember that GNAT is not using the gcc sjlj implementation, it is using
its own implementation, based on builtin_setjmp/longjmp

Also, we have pretty good results of ia64 builds with gcc 3.2.3, so there is
no fundamental issue that I am aware of wrt exception handling with the
current sources and builtin setjmp/longjmp.

> It seems like tasking support is broken (quite a few tests are hanging),
> probably because much of the sysdep part of the RTS is still not ported to
> ia64.

That's not the case, the sysdep part has been ported to ia64.

Tasking tests are very sensitive to exception handling, so it is often the
case that a problem in the exception handling will produce hanging tasking
tests as you are experiencing.

Arno