* Re: [PATCH] sched/rt : return accurate release rq lock info
2018-10-02 14:23 [PATCH] sched/rt : return accurate release rq lock info Peng Hao
@ 2018-10-02 8:01 ` Peter Zijlstra
0 siblings, 0 replies; 2+ messages in thread
From: Peter Zijlstra @ 2018-10-02 8:01 UTC (permalink / raw)
To: Peng Hao; +Cc: mingo, linux-kernel, Steven Rostedt, Juri Lelli
On Tue, Oct 02, 2018 at 10:23:41PM +0800, Peng Hao wrote:
> find_lock_lowest_rq may or not releease rq lock, but it is fuzzy.
> If not releasing rq lock, it is unnecessary to re-call
> pick_next_oushable_task.
You forgot to mention how much this matters. That is, why did you take
the effort to write this patch.
> Signed-off-by: Peng Hao <peng.hao2@zte.com.cn>
> ---
> kernel/sched/rt.c | 7 +++++++
> 1 file changed, 7 insertions(+)
>
> diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
> index 2e2955a..4d7d322 100644
> --- a/kernel/sched/rt.c
> +++ b/kernel/sched/rt.c
> @@ -1719,6 +1719,7 @@ static struct rq *find_lock_lowest_rq(struct task_struct *task, struct rq *rq)
> {
> struct rq *lowest_rq = NULL;
> int tries;
> + bool release_lock = false;
> int cpu;
We generally prefer to keep the variable definitions (reverse) ordered
on line length (reverse-xmas-tree).
>
> for (tries = 0; tries < RT_MAX_TRIES; tries++) {
> @@ -1741,6 +1742,7 @@ static struct rq *find_lock_lowest_rq(struct task_struct *task, struct rq *rq)
>
> /* if the prio of this runqueue changed, try again */
> if (double_lock_balance(rq, lowest_rq)) {
> + release_lock = true;
> /*
> * We had to unlock the run queue. In
> * the mean time, task could have
> @@ -1768,6 +1770,8 @@ static struct rq *find_lock_lowest_rq(struct task_struct *task, struct rq *rq)
> lowest_rq = NULL;
> }
>
> + if (!lowest_rq && !release_lock)
> + lowest_rq = (void *) -1;
We have a name for that thing, RETRY_TASK. Funnily it means the exact
opposite of what you did. So maybe use that and invert the logic.
> return lowest_rq;
> }
>
You also forgot to do the same to the deadline code; or explain why it
doesn't need it.
^ permalink raw reply [flat|nested] 2+ messages in thread
* [PATCH] sched/rt : return accurate release rq lock info
@ 2018-10-02 14:23 Peng Hao
2018-10-02 8:01 ` Peter Zijlstra
0 siblings, 1 reply; 2+ messages in thread
From: Peng Hao @ 2018-10-02 14:23 UTC (permalink / raw)
To: mingo, peterz; +Cc: linux-kernel, Peng Hao
find_lock_lowest_rq may or not releease rq lock, but it is fuzzy.
If not releasing rq lock, it is unnecessary to re-call
pick_next_oushable_task.
Signed-off-by: Peng Hao <peng.hao2@zte.com.cn>
---
kernel/sched/rt.c | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
index 2e2955a..4d7d322 100644
--- a/kernel/sched/rt.c
+++ b/kernel/sched/rt.c
@@ -1719,6 +1719,7 @@ static struct rq *find_lock_lowest_rq(struct task_struct *task, struct rq *rq)
{
struct rq *lowest_rq = NULL;
int tries;
+ bool release_lock = false;
int cpu;
for (tries = 0; tries < RT_MAX_TRIES; tries++) {
@@ -1741,6 +1742,7 @@ static struct rq *find_lock_lowest_rq(struct task_struct *task, struct rq *rq)
/* if the prio of this runqueue changed, try again */
if (double_lock_balance(rq, lowest_rq)) {
+ release_lock = true;
/*
* We had to unlock the run queue. In
* the mean time, task could have
@@ -1768,6 +1770,8 @@ static struct rq *find_lock_lowest_rq(struct task_struct *task, struct rq *rq)
lowest_rq = NULL;
}
+ if (!lowest_rq && !release_lock)
+ lowest_rq = (void *) -1;
return lowest_rq;
}
@@ -1863,6 +1867,9 @@ static int push_rt_task(struct rq *rq)
goto retry;
}
+ if (lowest_rq == (void *) -1)
+ goto out;
+
deactivate_task(rq, next_task, 0);
set_task_cpu(next_task, lowest_rq->cpu);
activate_task(lowest_rq, next_task, 0);
--
1.8.3.1
^ permalink raw reply related [flat|nested] 2+ messages in thread
end of thread, other threads:[~2018-10-02 8:01 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-10-02 14:23 [PATCH] sched/rt : return accurate release rq lock info Peng Hao
2018-10-02 8:01 ` Peter Zijlstra
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).