Re: Premature removal of aarch64 ilp32

Radek,

any chance to revert the ilp32 case?

On Jul 21 11:25, Joel Sherrill wrote:
> Hi
> 
> A recent commit broke aarch64 ilp32. While I agree that it has been
> deprecated in GCC, it is not scheduled for removal until GCC 16 which
> has not been released yet. It is deprecated in GCC 15 but there it is
> still possible to build a tool chain with current released versions. The
> commit was:
> 
> commit de479a54e22e8fcb6262639a8e67fe8b00a27c37
> Author: Radek Bartoň <[email protected]>
> Date:   Thu Jun 5 11:32:08 2025 +0200
> 
> The full commit message is below my signature.
> 
> Newlib has never been aggressive about removing support for
> architectures. I point to the nios2 as a recent example. It is last
> supported by GCC 14 and that requires enabling obsolete.
> 
> RTEMS is planning on a release with GCC 15 as the primary
> compiler. nios2 will be locked onto gcc 14. Both ilp32 and nios2
> will have to be removed as we move past GCC 15.
> 
> Why was ilp32 support removed while still supported in the current
> gcc release?
> 
> What was special about this particular case that breaks a few
> decades of newlib being OK with leaving old targets in the tree?

It was discussed in this very mailing list and the commit message
says it all.  If this ILP32 code is still required, it will surely be
possible to send patches?


Corinna