Regarding to sched-decrease-number-of-load-balances.patch currently
in -mm tree: I would like to revert the change on adding lb_stopbalance
counter. This count can be calculated by: lb_balanced - lb_nobusyg -
lb_nobusyq. There is no need to create gazillion counters while we
can derive the value. I'm more of against changing sched-stat format
unless it is absolutely necessary as all user land tool parsing
/proc/schedstat needs to be updated and it's a real pain trying to keep
multiple versions of it.
Signed-off-by: Ken Chen <[EMAIL PROTECTED]>
--- ./include/linux/sched.h.orig 2006-12-05 07:56:11.000000000 -0800
+++ ./include/linux/sched.h 2006-12-05 07:57:55.000000000 -0800
@@ -684,7 +684,6 @@ struct sched_domain {
unsigned long lb_hot_gained[MAX_IDLE_TYPES];
unsigned long lb_nobusyg[MAX_IDLE_TYPES];
unsigned long lb_nobusyq[MAX_IDLE_TYPES];
- unsigned long lb_stopbalance[MAX_IDLE_TYPES];
/* Active load balancing */
unsigned long alb_cnt;
--- ./kernel/sched.c.orig 2006-12-05 07:56:31.000000000 -0800
+++ ./kernel/sched.c 2006-12-05 08:02:11.000000000 -0800
@@ -428,7 +428,7 @@ static inline void task_rq_unlock(struct
* bump this up when changing the output format or the meaning of an existing
* format, so that tools can adapt (or abort)
*/
-#define SCHEDSTAT_VERSION 13
+#define SCHEDSTAT_VERSION 12
static int show_schedstat(struct seq_file *seq, void *v)
{
@@ -466,7 +466,7 @@ static int show_schedstat(struct seq_fil
seq_printf(seq, "domain%d %s", dcnt++, mask_str);
for (itype = SCHED_IDLE; itype < MAX_IDLE_TYPES;
itype++) {
- seq_printf(seq, " %lu %lu %lu %lu %lu %lu %lu
%lu %lu",
+ seq_printf(seq, " %lu %lu %lu %lu %lu %lu %lu
%lu",
sd->lb_cnt[itype],
sd->lb_balanced[itype],
sd->lb_failed[itype],
@@ -474,8 +474,7 @@ static int show_schedstat(struct seq_fil
sd->lb_gained[itype],
sd->lb_hot_gained[itype],
sd->lb_nobusyq[itype],
- sd->lb_nobusyg[itype],
- sd->lb_stopbalance[itype]);
+ sd->lb_nobusyg[itype]);
}
seq_printf(seq, " %lu %lu %lu %lu %lu %lu %lu %lu %lu
%lu %lu %lu\n",
sd->alb_cnt, sd->alb_failed, sd->alb_pushed,
@@ -2579,10 +2578,8 @@ redo:
group = find_busiest_group(sd, this_cpu, &imbalance, idle, &sd_idle,
&cpus, balance);
- if (*balance == 0) {
- schedstat_inc(sd, lb_stopbalance[idle]);
+ if (*balance == 0)
goto out_balanced;
- }
if (!group) {
schedstat_inc(sd, lb_nobusyg[idle]);
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
