On Mon, Mar 31, 2025 at 04:27:23PM -0400, Konrad Rzeszutek Wilk wrote:

Is that the right path forward?

That is replace the MSR write to disable speculative execution with a
non-serialized WRMSR? Doesn't that mean the WRMSRNS is speculative?

Ha, interesting question.

If the WRMSR is non-serializing, when do speculative things like indirect
branches and the like get *actually* cleared and can such a speculation window
be used to leak branch data even if IBRS is actually enabled for example...

Fun.

This change needs to be run by hw folks and I guess until then WRMSRNS should
not get anywhere near mitigation MSRs like SPEC_CTRL or PRED_CMD...

Thx.

-- 
Regards/Gruss,
    Boris.

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