* [PATCH 0/1] sched: defer idle accounting till after load update period
@ 2010-03-21 22:45 Chase Douglas
2010-03-21 22:45 ` [PATCH 1/1] " Chase Douglas
0 siblings, 1 reply; 2+ messages in thread
From: Chase Douglas @ 2010-03-21 22:45 UTC (permalink / raw)
To: linux-kernel; +Cc: Ingo Molnar, Peter Zijlstra, Thomas Gleixner
The following patch fixes a load avg calculation bug. Details are included with
the patch. Essentially, a task that often runs for less than 10 ticks at a time
is likely to be left out of the load avg calculation. A test case is provided
below. If you run the test case on a near zero-load system you will find top
report 90% cpu usage while the load avg stays at or near 0.00. With the patch,
the load avg is calculated correctly to be at least 0.90.
--
#include <asm/param.h>
#include <sys/time.h>
#include <time.h>
int main() {
struct timespec ts;
ts.tv_sec = 0;
ts.tv_nsec = 1000000000 / HZ;
/*
* Run gettimeofday in a tight loop 9 ticks, then sleep for 1 tick
*/
while (1) {
struct timeval tv;
do {
gettimeofday(&tv, NULL);
} while ((tv.tv_usec * HZ / 1000000) % 10 != 0);
nanosleep(&ts, NULL);
}
return 0;
}
^ permalink raw reply [flat|nested] 2+ messages in thread
* [PATCH 1/1] sched: defer idle accounting till after load update period
2010-03-21 22:45 [PATCH 0/1] sched: defer idle accounting till after load update period Chase Douglas
@ 2010-03-21 22:45 ` Chase Douglas
0 siblings, 0 replies; 2+ messages in thread
From: Chase Douglas @ 2010-03-21 22:45 UTC (permalink / raw)
To: linux-kernel; +Cc: Ingo Molnar, Peter Zijlstra, Thomas Gleixner
There's a period of 10 ticks where calc_load_tasks is updated by all the
cpus for the load avg. Usually all the cpus do this during the first
tick. If any cpus go idle, calc_load_tasks is decremented accordingly.
However, if they wake up calc_load_tasks is not incremented. Thus, if
cpus go idle during the 10 tick period, calc_load_tasks may be
decremented to a non-representative value. This issue can lead to
systems having a load avg of exactly 0, even though the real load avg
could theoretically be up to NR_CPUS.
This change defers calc_load_tasks accounting after each cpu updates the
count until after the 10 tick period.
BugLink: http://bugs.launchpad.net/bugs/513848
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
---
kernel/sched.c | 16 ++++++++++++++--
1 files changed, 14 insertions(+), 2 deletions(-)
diff --git a/kernel/sched.c b/kernel/sched.c
index 9ab3cd7..c0aedac 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -3064,7 +3064,8 @@ void calc_global_load(void)
*/
static void calc_load_account_active(struct rq *this_rq)
{
- long nr_active, delta;
+ static atomic_long_t deferred;
+ long nr_active, delta, deferred_delta;
nr_active = this_rq->nr_running;
nr_active += (long) this_rq->nr_uninterruptible;
@@ -3072,6 +3073,17 @@ static void calc_load_account_active(struct rq *this_rq)
if (nr_active != this_rq->calc_load_active) {
delta = nr_active - this_rq->calc_load_active;
this_rq->calc_load_active = nr_active;
+
+ /* Need to defer idle accounting during load update period: */
+ if (unlikely(time_before(jiffies, this_rq->calc_load_update) &&
+ time_after_eq(jiffies, calc_load_update))) {
+ atomic_long_add(delta, &deferred);
+ return;
+ }
+
+ deferred_delta = atomic_long_xchg(&deferred, 0);
+ delta += deferred_delta;
+
atomic_long_add(delta, &calc_load_tasks);
}
}
@@ -3106,8 +3118,8 @@ static void update_cpu_load(struct rq *this_rq)
}
if (time_after_eq(jiffies, this_rq->calc_load_update)) {
- this_rq->calc_load_update += LOAD_FREQ;
calc_load_account_active(this_rq);
+ this_rq->calc_load_update += LOAD_FREQ;
}
}
--
1.6.3.3
^ permalink raw reply related [flat|nested] 2+ messages in thread
end of thread, other threads:[~2010-03-21 22:45 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-03-21 22:45 [PATCH 0/1] sched: defer idle accounting till after load update period Chase Douglas
2010-03-21 22:45 ` [PATCH 1/1] " Chase Douglas
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.