arm64: add tests for swp/swpb emulation

[kevans]

Summary

One test is suitable to be hooked up to the build, so I've done this
here. The other test lives in tools/regression because failure is a
bit more subjective -- generally, one runs it for some unbounded amount
of time and observe if it eventually exits because two threads acquired
the same mutex.

Sponsored by: Stormshield
Sponsored by: Klara, Inc.

[andrew]

tests/sys/compat32/Makefile

3–4

Should this be MACHINE or MACHINE_CPUARCH? I've preferred the latter in userspace, but given this is for a kernel test the former might be better.

[imp]

tests/sys/compat32/Makefile

3–4

It should be MACHINE_ARCH I'd think, with the corresponding renames below.
I think MACHINE_ARCH because it's userland (and it's about binaries executing, which are tied to MACHINE_ARCH, not MACHINE).
And MACHINE_ARCH instead of MACHINE_CPUARCH because we'd want to do this possibly for powerpc64, but not powerpc64le, since we don't have a powerpcle.

[kevans]

MACHINE -> MACHINE_ARCH

[kevans]

tests/sys/compat32/Makefile

3–4

I also changed the dir name in tools/regression/compat32 to be consistent, although we won't be hooking that up with any make infrastructure at the compat32 level most likely.

[mmel]

Hmm, the SWP/SWPB instruction is obsolete in ARMv7 ISA (in favor of atomic operations). It not supported/implemented on ARMv7 SMP cpus and arm32 kernel doesn't support it at all. IMHO we should be consistent at least between arm32 and compat32 on aarch64.

https://developer.arm.com/documentation/ddi0406/c/System-Level-Architecture/System-Control-Registers-in-a-VMSA-implementation/VMSA-System-control-registers-descriptions--in-register-order/ID-ISAR4--Instruction-Set-Attribute-Register-4--VMSA?lang=en

[kevans]

In D39668#904485, @mmel wrote:

Hmm, the SWP/SWPB instruction is obsolete in ARMv7 ISA (in favor of atomic operations). It not supported/implemented on ARMv7 SMP cpus and arm32 kernel doesn't support it at all. IMHO we should be consistent at least between arm32 and compat32 on aarch64.

https://developer.arm.com/documentation/ddi0406/c/System-Level-Architecture/System-Control-Registers-in-a-VMSA-implementation/VMSA-System-control-registers-descriptions--in-register-order/ID-ISAR4--Instruction-Set-Attribute-Register-4--VMSA?lang=en

This is the only missing piece these folks need to build/run some specific software they have w/ compat32 on aarch64, though... it'd be really nice not to block that just because of this relatively minor detail.

[mmel]

I see. So in this case can we enable swp emulation using sysctl, disabled by default? I'm worried that this may cause bad behavior for ports build for arm32 - the port may detect swp during build. A bit paranoid, I know.

Also, can you put comment briefly on why we need to emulate deprecated, unimplemented (not enabled) on real 32bit hw ? It doesn't make sense without context. Thanks.

[kevans]

• Also check that the kernel was built with COMPAT_FREEBSD32
• Use the new sysctl to turn swp emulation on as needed, make sure we can't run these tests in parallel in normal kyua runs.

[kevans]

In D39668#905298, @mmel wrote:

I see. So in this case can we enable swp emulation using sysctl, disabled by default? I'm worried that this may cause bad behavior for ports build for arm32 - the port may detect swp during build. A bit paranoid, I know.

I agree; I added a compat.arm.emul_swp sysctl.

Also, can you put comment briefly on why we need to emulate deprecated, unimplemented (not enabled) on real 32bit hw ? It doesn't make sense without context. Thanks.

I put a mention in the commit message of D39667 (the implementation), let me know if you'd prefer to see it elsewhere as well.

[mmel]

More than enough, thanks. And sorry for starting this in the wrong review.

[imp]

for the makefiles