ns32k status

ns32k status

[Gilbert Coville]

I see that ns32k is on the chopping block once again.

As I understand it, ns32k isn’t the last arch to use a.out, but it is the last arch to use only a.out.  With its removal, it would then be possible to remove a.out completely.

However, I do think that dropping ns32k from binutils would be a shame.  Although only a handful of people use it in its current form, it does actually work quite well.  Obviously, it’s no longer being used for netbsd, but it is still useful for hobby embedded or bare metal-type environments.

I have started investigating what it would take to create an ns32k-elf target.  My initial thought was something like “There’s already a working example of ns32k a.out, plus plenty of other elf targets.  It should be straightforward.”  Perhaps it is to someone with years of binutils experience.  However, the more I dig into it, the more daunting it becomes.  Adding to the problem is that there isn’t a ns32k ELF addendum.  We’ve looked pretty hard and only found a reference to it in National’s documentation.  Most likely the document was vaporware.  All we’re left with is a value for e_machine.  I had hoped to have some code to show by the upcoming branch date, but that is clearly not going to happen.

My primary request is that the removal of ns32k targets be postponed for a release (or two).  I don’t mind if ns32k-*-netbsd remains in the obsolete list, but if the ns32k code and files are removed, it will make the task significantly more difficult.

I am open to any advice, or ridicule if necessary.

Thoughts?

Gilbert

Re: ns32k status

[Alan Modra]

On Mon, Jan 17, 2022 at 08:02:49PM -0800, Gilbert Coville wrote:
> I see that ns32k is on the chopping block once again.

Just ns32k-netbsd.  The other ns32k aout targets are still available
(and even ns32k-netbsd with --enable-obsolete).  Nobody is planning to
remove ns32k code until all ns32k targets have been marked obsolete.

-- 
Alan Modra
Australia Development Lab, IBM