Re: [tip: x86/cpu] x86/cpu: Use INVPCID mnemonic in invpcid.h

From: Borislav Petkov
Date: Tue May 12 2020 - 12:28:56 EST


On Tue, May 12, 2020 at 05:54:49PM +0200, Uros Bizjak wrote:
> Symbolic operands are agnostic to the position in the asm clause, so
> it really doesn't matter much. It just doesn't feel right, when other
> cases follow different order.
>
> > $ diff -suprN /tmp/before /tmp/after
> > Files /tmp/before and /tmp/after are identical
>
> Sure, otherwise assembler would complain.
>
> > Makes sense?
>
> Well, I don't want to bikeshed around this anymore, so any way is good.

Look at it this way: the symbolic operand names feature has made inline
assembly *orders* of magnitude more readable than what it was before.
Kernel folks, including myself, have stumbled upon the question which
operand is which, on a regular basis and having the operand names there
makes reading the inline asm almost trivial.

So while we should not convert wholesale, I think we should aim to
gradually convert those other cases to the a-lot-more readable variant
with symbolic operand names.

Thx.

--
Regards/Gruss,
Boris.

https://people.kernel.org/tglx/notes-about-netiquette