Re: [PATCH RFC 0/6] x86/msr: Rename MSR access functions

[Jürgen Groß]

On 20.04.26 13:35, Peter Zijlstra wrote:
> On Mon, Apr 20, 2026 at 11:16:28AM +0200, Juergen Gross wrote:
>
> > - Use functions instead of macros for accessing MSRs, which will drop
> >    modifying variables passed as a parameter.
> >
> > - Eliminate multiple accessors doing exactly the same thing (e.g.
> >    rdmsrl() and rdmsrq()).
>
> So far so sane.
>
> > - Instead of having function names based on the underlying instruction
> >    mnemonics, have functions of a common name space (msr_*()).
>
> Not sure on this one. The whole msr_{read,write}_{safe,noser}() thing is
> a royal pain. Also 'noser' reads to me as the noun that goes with 'to
> nose' [he that noses (around), like baker: he that bakes].

Naming is hard. :-)

What about s/ser/sync/ then?

> I would much rather we just stick to the mnemonics here. All of this
> really is about wrapping single instructions, no need to make it an
> unreadable mess.

I'm pretty sure most of the wrmsr*() use cases could switch to the non
serializing variants. The problem not making the serializing aspect visible
in the function name will probably result in most new instances still using
the serializing variant instead of the probably possible non serializing one.

Many of those use cases will even suffer more, as they won't use the
immediate form of WRMSRNS then, which would waste the additional benefits of
that instruction.


Juergen

Attachment: OpenPGP_0xB0DE9DD628BF132F.asc
Description: OpenPGP public key

Attachment: OpenPGP_signature.asc
Description: OpenPGP digital signature