Re: [PATCH] Yielding processor resources during lock contention

[Arnd Bergmann]

On Sonntag, 12. September 2004 12:43, Anton Blanchard wrote:
> cpu_relax doesnt tell us why we are busy looping. In this particular
> case we want to pass to the hypervisor which virtual cpu we are waiting
> on so the hypervisor make better scheduling decisions.

Ah, interesting. I think s390 could do the same with the upcoming 5.1 release
of z/VM. However, we decided to ask for an implementation of a futex hypercall
instead, which lets us implement the linux spinlock as a hypervisor semaphore.

> Did you manage to see any improvement by yielding to the hypervisor
> in cpu_relax?
I'm not sure if this has been measured, maybe Martin or Eberhard knows more.

	Arnd <><

Attachment: pgp00000.pgp
Description: signature