RFC: riscv64 big endian system attempt

[Olof Johansson]

On Fri, Dec 20, 2024 at 03:57:46PM +0000, Ben Dooks wrote:
> With the latest spec adding configurable endianness, we thought we
> should try putting together a proof of concept riscv64 big endian.
> 
> The full information is documented on our gitlab[1] which includes
> source repositories, build information and project documentation.
> 
> We have a minimal buildroot, qemu and kernel working on QEMU.
> 
> As this is a work in progress any review or help is appreciated.

While this is neat, I wonder if there's any real point in picking this
up broadly and enabling it, with the associated overhead to keep it
maintained?

While big endian ppc64 will always be close to my heart, little endian
really has taken over the world by now, even on ppc64. It used to be
that networking was the area that BE was a (soft) requirement, but with
modern CPUs having advanced faster than memory latencies and speed, doing
endian conversion in software doesn't seem to be a big deal any more.

As far as I know, ARM platforms are in the same boat -- they did some
early enablement, driven by a couple of specific use cases that didn't
significantly grow and are used in very narrow environment and with very
limited userspace.

Is this a parallel situation to that, or are there reasons to support BE
more broadly? It seems like it'd mostly split efforts and add overhead
to make sure it's still supported, even if it's a fun project to prove
that it works.



-Olof