Messages in this thread | | | Date | Mon, 18 Jan 2010 10:42:13 +0100 | From | Martin Schwidefsky <> | Subject | Re: [PATCH] sched: cpuacct: Use bigger percpu counter batch values for stats counters |
| |
Hi Anton,
On Mon, 18 Jan 2010 15:41:42 +1100 Anton Blanchard <anton@samba.org> wrote:
> Note: ccing ia64 and s390 who have not yet added code to statically > initialise cputime_one_jiffy at boot. > See a42548a18866e87092db93b771e6c5b060d78401 (cputime: Optimize > jiffies_to_cputime(1) for details). Adding this would help optimise not only > this patch but many other areas of the scheduler when > CONFIG_VIRT_CPU_ACCOUNTING is enabled.
For s390 the jiffies_to_cputime is a compile time constant. No need to initialize it at runtime, no?
> Index: linux.trees.git/kernel/sched.c > =================================================================== > --- linux.trees.git.orig/kernel/sched.c 2010-01-18 14:27:12.000000000 +1100 > +++ linux.trees.git/kernel/sched.c 2010-01-18 15:21:59.000000000 +1100 > @@ -10894,6 +10894,7 @@ static void cpuacct_update_stats(struct > enum cpuacct_stat_index idx, cputime_t val) > { > struct cpuacct *ca; > + int batch; > > if (unlikely(!cpuacct_subsys.active)) > return; > @@ -10901,8 +10902,9 @@ static void cpuacct_update_stats(struct > rcu_read_lock(); > ca = task_ca(tsk); > > + batch = min_t(long, percpu_counter_batch * cputime_one_jiffy, INT_MAX); > do { > - percpu_counter_add(&ca->cpustat[idx], val); > + __percpu_counter_add(&ca->cpustat[idx], val, batch); > ca = ca->parent; > } while (ca); > rcu_read_unlock();
The patch itself trades some accuracy (larger cpu accounting value that are stored per-cpu) against runtime overhead (spinlock to transfer the value to the global variable in __percpu_counter_add). Did you calculate how big the loss in accuracy is?
-- blue skies, Martin.
"Reality continues to ruin my life." - Calvin.
| |