However since I'm technically a "Ports" committer (and far from a kernel expert, though hopefully getting there), I'd recommend waiting for approval from @hselasky and/or @manu.

Use (1U << 2) as per @hselasky's request.

I chose (1U << 1) mainly for compatibility with Linux, but it could also be (1U << 2).

In D30747#691142, @hselasky wrote:

What does this new flag do? Should it be (1U << 2) perhaps?

Using function macro as per @hselasky's suggestion.

Technically, from what I understand, these are macros to locks in the kernel for scheduling purposes, but as with other macros we aren't doing anything with them so we just use "do {} while (0)".

This is used for the addresses in mutexes and log messages.

On Linux, spin_trylock_irqsave() is used when you want to save the IRQ before doing a trylock, however we aren't doing this as we do in spin_trylock_irq(). This is called in the i915 part of the drm-kmod 5.7 update.

I believe linuxkpi's list_for_each_entry is lockless (but am not sure), so I updated the message.

Use original OpenBSD file with headers.

LGTM.

LGTM.