2018-11-21 03:57:37

by Ravi Bangoria

Subject: [PATCH v2] Powerpc/perf: Wire up PMI throttling

Commit 14c63f17b1fde ("perf: Drop sample rate when sampling is too
slow") introduced a way to throttle PMU interrupts if we're spending
too much time just processing those. Wire up powerpc PMI handler to
use this infrastructure.

We have throttling of the *rate* of interrupts, but this adds
throttling based on the *time taken* to process the interrupts.

Signed-off-by: Ravi Bangoria <[email protected]>
---
v1: https://lkml.org/lkml/2018/11/14/627

arch/powerpc/perf/core-book3s.c | 11 ++++++++++-
1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/arch/powerpc/perf/core-book3s.c b/arch/powerpc/perf/core-book3s.c
index 81f8a0c838ae..3cd2ec9c9732 100644
--- a/arch/powerpc/perf/core-book3s.c
+++ b/arch/powerpc/perf/core-book3s.c
@@ -10,6 +10,7 @@
*/
#include <linux/kernel.h>
#include <linux/sched.h>
+#include <linux/sched/clock.h>
#include <linux/perf_event.h>
#include <linux/percpu.h>
#include <linux/hardirq.h>
@@ -2148,7 +2149,7 @@ static bool pmc_overflow(unsigned long val)
/*
* Performance monitor interrupt stuff
*/
-static void perf_event_interrupt(struct pt_regs *regs)
+static void __perf_event_interrupt(struct pt_regs *regs)
{
int i, j;
struct cpu_hw_events *cpuhw = this_cpu_ptr(&cpu_hw_events);
@@ -2232,6 +2233,14 @@ static void perf_event_interrupt(struct pt_regs *regs)
irq_exit();
}

+static void perf_event_interrupt(struct pt_regs *regs)
+{
+ u64 start_clock = sched_clock();
+
+ __perf_event_interrupt(regs);
+ perf_sample_event_took(sched_clock() - start_clock);
+}
+
static int power_pmu_prepare_cpu(unsigned int cpu)
{
struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu);
--
2.17.1

2018-12-23 16:44:31

by Michael Ellerman

Subject: Re: [v2] Powerpc/perf: Wire up PMI throttling

On Wed, 2018-11-21 at 03:56:37 UTC, Ravi Bangoria wrote:
> Commit 14c63f17b1fde ("perf: Drop sample rate when sampling is too
> slow") introduced a way to throttle PMU interrupts if we're spending
> too much time just processing those. Wire up powerpc PMI handler to
> use this infrastructure.
>
> We have throttling of the *rate* of interrupts, but this adds
> throttling based on the *time taken* to process the interrupts.
>
> Signed-off-by: Ravi Bangoria <[email protected]>

Applied to powerpc next, thanks.

https://git.kernel.org/powerpc/c/0c9108b083706330cd5484d121fbb0

cheers