Jorgen Lundman

It's ready, the two failures is probably unrelated.

> Please change the commit message to say "assembly files". We do not call our C files "compiler files", so saying "assembler files" is inconsistent. Somewhat interesting linguistically, when C64...

> This seems to have not given any attention to our ARM and POWER assembly. Was that intentional? Yep - just doing Intel, with the idea that we will just...

rebased.

Does the test `Test: /usr/share/zfs/zfs-tests/tests/functional/pyzfs/pyzfs_unittest (run as root) [18:46] [KILLED]` have anything to do with crypto, is it something to look into?

Yeah, this is a good point. We could take align out of ENTRY_NP() and keep it separate, or as you suggest, add it to macro. But since it's an illumos...

macOS has the 2 argument align, you specify the value you want in there, 0x90 is NOP or something. https://github.com/openzfsonosx/openzfs/blob/development/module/icp/asm-x86_64/os/macos/aes/aes_amd64.S#L696 but I see I've left the align problem alone for...

OK so, I don't see anything using `MCOUNT`, whatever that is. I think maybe it was something `mdb` related for illumos kernels? Either way, we are not using it. (ok,...

> try .balign first. Will do. @behlendorf I believe we are ready.

Might not be clear, but #14228 takes care of all the changes in here and expands it to make all assembler files sharable with windows and macOS. Moving the header...