i386, arm, arm64, riscv: move ABI code from machdep.c to exec_machdep.c

[kib]

Summary

and ptrace bits to ptrace_machdep.c.

[imp]

Looks good to me, though I didn't verify that the moved code was identical to the original.
Might want to do the ptrace move and the exec move as separate commits, but it's not terrible if not.

sys/arm/arm/exec_machdep.c

43–45

Any reason to retain this information?

sys/conf/files.arm64

51

This duplicates the entry below, so can be removed.

[kib]

In D32310#729364, @imp wrote:

Might want to do the ptrace move and the exec move as separate commits, but it's not terrible if not.

The diff is split per-arch, I do not plan to split it inside arches. Per commit view is available at https://kib.kiev.ua/git/gitweb.cgi?p=deviant3.git;a=shortlog;h=refs/heads/alloca

sys/arm/arm/exec_machdep.c

43–45

Removed. I just copied the whole blocks.

[emaste]

Same as @imp, I did not compare all of the removed/added blocks.

[jrtc27]

Couple of minor comments (and some positive observations) from reviewing the diff locally with --color-moved

sys/arm/arm/machdep.c

526

More K&R C gone :)

sys/i386/i386/exec_machdep.c

1400

This modified line is using tabs, not spaces like the rest of the function (which all your other modified returns in the function adhere to)

sys/i386/i386/machdep.c

737

These got lost, I assume deliberate because of where our SMP support is these days :)

sys/riscv/riscv/exec_machdep.c

193

You fixed the arm64 typo for this but not the riscv copy of it (as if we needed more reasons to point out how much of sys/riscv is a copy of sys/arm64 :P)

sys/riscv/riscv/machdep.c

548

This disappeared from the arm64 and riscv copies. I assume it's legacy crap that's no longer needed?

[emaste]

sys/arm/arm/machdep.c

526

\o/

sys/i386/i386/exec_machdep.c

541–547

I guess if we're doing other little cleanups like deleting MPSAFE we might as well rewrap this at a reasonable width?

[kib]

sys/i386/i386/exec_machdep.c

1400

Emacs whitespace-mode only reported spaces to me.

Anyway, I re-indented the function with tabs, and looked at other issues in this file highlighted by emacs.

sys/i386/i386/machdep.c

737

Yes, MPSAFE annotations are pointless today.

sys/riscv/riscv/machdep.c

548

Yes, I believe its perceived usefulness long time go. It is in fact detriment, since whatever changes syscalls.master and gen script take, SYSPROTO_H reminders are not updated.

[kib]

Handle notes

[imp]

Per arch is fine...