[PATCH v2] sched/deadline: reduce rq lock contention by eliminating locking of non-feasible target

[PATCH v2] sched/deadline: reduce rq lock contention by eliminating locking of non-feasible target

[Wanpeng Li]

This patch adds checks that prevents futile attempts to move dl tasks to
a CPU with active tasks of equal or earlier deadline. The same behavior as
commit 80e3d87b2c55 ("sched/rt: Reduce rq lock contention by eliminating
locking of non-feasible target") for rt class.

Signed-off-by: Wanpeng Li <wanpeng.li@linux.intel.com>
---
v1 -> v2:
 * use dl_time_before macro

 kernel/sched/deadline.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index 19d716e..f48d319 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -955,7 +955,9 @@ select_task_rq_dl(struct task_struct *p, int cpu, int sd_flag, int flags)
 	    (p->nr_cpus_allowed > 1)) {
 		int target = find_later_rq(p);
 
-		if (target != -1)
+		if (target != -1 &&
+			dl_time_before(p->dl.deadline,
+				cpu_rq(target)->dl.earliest_dl.curr))
 			cpu = target;
 	}
 	rcu_read_unlock();
-- 
1.9.1

Re: [PATCH v2] sched/deadline: reduce rq lock contention by eliminating locking of non-feasible target

[Wanpeng Li]

Ping Juri, ;)
On Mon, Mar 23, 2015 at 05:02:13PM +0800, Wanpeng Li wrote:
>This patch adds checks that prevents futile attempts to move dl tasks to
>a CPU with active tasks of equal or earlier deadline. The same behavior as
>commit 80e3d87b2c55 ("sched/rt: Reduce rq lock contention by eliminating
>locking of non-feasible target") for rt class.
>
>Signed-off-by: Wanpeng Li <wanpeng.li@linux.intel.com>
>---
>v1 -> v2:
> * use dl_time_before macro
>
> kernel/sched/deadline.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
>diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
>index 19d716e..f48d319 100644
>--- a/kernel/sched/deadline.c
>+++ b/kernel/sched/deadline.c
>@@ -955,7 +955,9 @@ select_task_rq_dl(struct task_struct *p, int cpu, int sd_flag, int flags)
> 	    (p->nr_cpus_allowed > 1)) {
> 		int target = find_later_rq(p);
> 
>-		if (target != -1)
>+		if (target != -1 &&
>+			dl_time_before(p->dl.deadline,
>+				cpu_rq(target)->dl.earliest_dl.curr))
> 			cpu = target;
> 	}
> 	rcu_read_unlock();
>-- 
>1.9.1

Re: [PATCH v2] sched/deadline: reduce rq lock contention by eliminating locking of non-feasible target

[Wanpeng Li]

Ping Peter, this patch is collected into this patchset https://lkml.org/lkml/2015/4/6/44
On Fri, Mar 27, 2015 at 07:03:00AM +0800, Wanpeng Li wrote:
>Ping Juri, ;)
>On Mon, Mar 23, 2015 at 05:02:13PM +0800, Wanpeng Li wrote:
>>This patch adds checks that prevents futile attempts to move dl tasks to
>>a CPU with active tasks of equal or earlier deadline. The same behavior as
>>commit 80e3d87b2c55 ("sched/rt: Reduce rq lock contention by eliminating
>>locking of non-feasible target") for rt class.
>>
>>Signed-off-by: Wanpeng Li <wanpeng.li@linux.intel.com>
>>---
>>v1 -> v2:
>> * use dl_time_before macro
>>
>> kernel/sched/deadline.c | 4 +++-
>> 1 file changed, 3 insertions(+), 1 deletion(-)
>>
>>diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
>>index 19d716e..f48d319 100644
>>--- a/kernel/sched/deadline.c
>>+++ b/kernel/sched/deadline.c
>>@@ -955,7 +955,9 @@ select_task_rq_dl(struct task_struct *p, int cpu, int sd_flag, int flags)
>> 	    (p->nr_cpus_allowed > 1)) {
>> 		int target = find_later_rq(p);
>> 
>>-		if (target != -1)
>>+		if (target != -1 &&
>>+			dl_time_before(p->dl.deadline,
>>+				cpu_rq(target)->dl.earliest_dl.curr))
>> 			cpu = target;
>> 	}
>> 	rcu_read_unlock();
>>-- 
>>1.9.1