* [PATCH] KVM: pvclock_clocksource_read while-loop optimization
@ 2011-01-01 20:51 Arjan Koers
2011-01-03 16:50 ` Glauber Costa
0 siblings, 1 reply; 3+ messages in thread
From: Arjan Koers @ 2011-01-01 20:51 UTC (permalink / raw)
To: Avi Kivity, Marcelo Tosatti; +Cc: kvm, Glauber Costa
Reduce the number of while-loop iterations (from two to one in the most
common situation)
Signed-off-by: Arjan Koers <0h61vkll2ly8@xutrox.com>
diff --git a/arch/x86/kernel/pvclock.c b/arch/x86/kernel/pvclock.c
index 42eb330..8f52acb 100644
--- a/arch/x86/kernel/pvclock.c
+++ b/arch/x86/kernel/pvclock.c
@@ -93,7 +93,7 @@ cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src)
struct pvclock_shadow_time shadow;
unsigned version;
cycle_t ret, offset;
- u64 last;
+ u64 last, save;
do {
version = pvclock_get_time_values(&shadow, src);
@@ -125,8 +125,9 @@ cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src)
do {
if (ret < last)
return last;
+ save = last;
last = atomic64_cmpxchg(&last_value, last, ret);
- } while (unlikely(last != ret));
+ } while (unlikely(last != save));
return ret;
}
^ permalink raw reply related [flat|nested] 3+ messages in thread* Re: [PATCH] KVM: pvclock_clocksource_read while-loop optimization
2011-01-01 20:51 [PATCH] KVM: pvclock_clocksource_read while-loop optimization Arjan Koers
@ 2011-01-03 16:50 ` Glauber Costa
0 siblings, 0 replies; 3+ messages in thread
From: Glauber Costa @ 2011-01-03 16:50 UTC (permalink / raw)
To: Arjan Koers; +Cc: Avi Kivity, Marcelo Tosatti, kvm
On Sat, 2011-01-01 at 21:51 +0100, Arjan Koers wrote:
> Reduce the number of while-loop iterations (from two to one in the most
> common situation)
>
> Signed-off-by: Arjan Koers <0h61vkll2ly8@xutrox.com>
>
> diff --git a/arch/x86/kernel/pvclock.c b/arch/x86/kernel/pvclock.c
> index 42eb330..8f52acb 100644
> --- a/arch/x86/kernel/pvclock.c
> +++ b/arch/x86/kernel/pvclock.c
> @@ -93,7 +93,7 @@ cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src)
> struct pvclock_shadow_time shadow;
> unsigned version;
> cycle_t ret, offset;
> - u64 last;
> + u64 last, save;
>
> do {
> version = pvclock_get_time_values(&shadow, src);
> @@ -125,8 +125,9 @@ cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src)
> do {
> if (ret < last)
> return last;
> + save = last;
> last = atomic64_cmpxchg(&last_value, last, ret);
> - } while (unlikely(last != ret));
> + } while (unlikely(last != save));
>
> return ret;
> }
Hey, good catch.
^ permalink raw reply [flat|nested] 3+ messages in thread
* [PATCH] KVM: pvclock_clocksource_read while-loop optimization
@ 2010-10-19 20:05 Arjan Koers
0 siblings, 0 replies; 3+ messages in thread
From: Arjan Koers @ 2010-10-19 20:05 UTC (permalink / raw)
To: kvm
Reduce the number of while-loop iterations (from two to one in the most
common situation)
Signed-off-by: Arjan Koers <0h61vkll2ly8@xutrox.com>
diff --git a/arch/x86/kernel/pvclock.c b/arch/x86/kernel/pvclock.c
index bab3b9e..995d479 100644
--- a/arch/x86/kernel/pvclock.c
+++ b/arch/x86/kernel/pvclock.c
@@ -126,7 +126,7 @@ cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src)
struct pvclock_shadow_time shadow;
unsigned version;
cycle_t ret, offset;
- u64 last;
+ u64 last, save;
do {
version = pvclock_get_time_values(&shadow, src);
@@ -158,8 +158,9 @@ cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src)
do {
if (ret < last)
return last;
+ save = last;
last = atomic64_cmpxchg(&last_value, last, ret);
- } while (unlikely(last != ret));
+ } while (unlikely(last != save));
return ret;
}
^ permalink raw reply related [flat|nested] 3+ messages in thread
end of thread, other threads:[~2011-01-03 16:50 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-01-01 20:51 [PATCH] KVM: pvclock_clocksource_read while-loop optimization Arjan Koers
2011-01-03 16:50 ` Glauber Costa
-- strict thread matches above, loose matches on Subject: below --
2010-10-19 20:05 Arjan Koers
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox