On Tue, 2014-06-17 at 22:09 +0300, Mihai Caraman wrote:
> On vcpu schedule, the condition checked for tlb pollution is too loose.
> The tlb entries of a vcpu become polluted (vs stale) only when a different
> vcpu within the same logical partition runs in-between. Optimize the tlb
> invalidation condition keeping last_vcpu_on_cpu per logical partition id.
>
> With the new invalidation condition, a guest shows 4% performance improvement
> on P5020DS while running a memory stress application with the cpu oversubscribed,
> the other guest running a cpu intensive workload.
>
> Guest - old invalidation condition
> real 3.89
> user 3.87
> sys 0.01
>
> Guest - enhanced invalidation condition
> real 3.75
> user 3.73
> sys 0.01
>
> Host
> real 3.70
> user 1.85
> sys 0.00
>
> The memory stress application accesses 4KB pages backed by 75% of available
> TLB0 entries:
>
> char foo[ENTRIES][4096] __attribute__ ((aligned (4096)));
>
> int main()
> {
> char bar;
> int i, j;
>
> for (i = 0; i < ITERATIONS; i++)
> for (j = 0; j < ENTRIES; j++)
> bar = foo[j][0];
>
> return 0;
> }
>
> Signed-off-by: Mihai Caraman <mihai.caraman@xxxxxxxxxxxxx>
> Cc: Scott Wood <scottwood@xxxxxxxxxxxxx>
> ---
> v3:
> - use existing logic while keeping last_vcpu_per_cpu per lpid
>
> v2:
> - improve patch name and description
> - add performance results
>
>
> arch/powerpc/kvm/e500mc.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/arch/powerpc/kvm/e500mc.c b/arch/powerpc/kvm/e500mc.c
> index 17e4562..95e33e3 100644
> --- a/arch/powerpc/kvm/e500mc.c
> +++ b/arch/powerpc/kvm/e500mc.c
> @@ -110,7 +110,7 @@ void kvmppc_mmu_msr_notify(struct kvm_vcpu *vcpu, u32 old_msr)
> {
> }
>
> -static DEFINE_PER_CPU(struct kvm_vcpu *, last_vcpu_on_cpu);
> +static DEFINE_PER_CPU(struct kvm_vcpu * [KVMPPC_NR_LPIDS], last_vcpu_on_cpu);
Hmm, I didn't know you could express types like that. Is this special
syntax that only works for typeof?
No space after *
Name should be adjusted to match, something like last_vcpu_of_lpid (with
the _on_cpu being implied by the fact that it's PER_CPU).
-Scott
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
