When the function consume_dispatch_q() returns true, the dsq lock may
remains held and is not unlocked.

Signed-off-by: Zhang Qiao <zhangqiao22@huawei.com>
---
 kernel/sched/ext.c | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c
index 410a4df8a121..4d80aa3de00e 100644
--- a/kernel/sched/ext.c
+++ b/kernel/sched/ext.c
@@ -2377,7 +2377,8 @@ static inline bool consume_remote_task(struct rq *this_rq, struct task_struct *p
 static bool consume_dispatch_q(struct rq *rq, struct scx_dispatch_q *dsq)
 {
 	struct task_struct *p;
-retry:
+	bool ret = false;
+
 	/*
 	 * The caller can't expect to successfully consume a task if the task's
 	 * addition to @dsq isn't guaranteed to be visible somehow. Test
@@ -2394,19 +2395,20 @@ static bool consume_dispatch_q(struct rq *rq, struct scx_dispatch_q *dsq)
 		if (rq == task_rq) {
 			task_unlink_from_dsq(p, dsq);
 			move_local_task_to_local_dsq(p, 0, dsq, rq);
-			raw_spin_unlock(&dsq->lock);
-			return true;
+			ret = true;
+			break;
 		}
 
 		if (task_can_run_on_remote_rq(p, rq, false)) {
-			if (likely(consume_remote_task(rq, p, dsq, task_rq)))
-				return true;
-			goto retry;
+			if (likely(consume_remote_task(rq, p, dsq, task_rq))) {
+				ret = true;
+				break;
+			}
 		}
 	}
 
 	raw_spin_unlock(&dsq->lock);
-	return false;
+	return ret;
 }
 
 static bool consume_global_dsq(struct rq *rq)
-- 
2.33.0

Hello,

On Tue, Oct 15, 2024 at 10:29:17PM +0800, Zhang Qiao wrote:
> When the function consume_dispatch_q() returns true, the dsq lock may
> remains held and is not unlocked.
> 
> Signed-off-by: Zhang Qiao <zhangqiao22@huawei.com>
> ---
>  kernel/sched/ext.c | 16 +++++++++-------
>  1 file changed, 9 insertions(+), 7 deletions(-)
> 
> diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c
> index 410a4df8a121..4d80aa3de00e 100644
> --- a/kernel/sched/ext.c
> +++ b/kernel/sched/ext.c
> @@ -2377,7 +2377,8 @@ static inline bool consume_remote_task(struct rq *this_rq, struct task_struct *p
>  static bool consume_dispatch_q(struct rq *rq, struct scx_dispatch_q *dsq)
>  {
>  	struct task_struct *p;
> -retry:
> +	bool ret = false;
> +
>  	/*
>  	 * The caller can't expect to successfully consume a task if the task's
>  	 * addition to @dsq isn't guaranteed to be visible somehow. Test
> @@ -2394,19 +2395,20 @@ static bool consume_dispatch_q(struct rq *rq, struct scx_dispatch_q *dsq)
>  		if (rq == task_rq) {
>  			task_unlink_from_dsq(p, dsq);
>  			move_local_task_to_local_dsq(p, 0, dsq, rq);
> -			raw_spin_unlock(&dsq->lock);
> -			return true;
> +			ret = true;
> +			break;
>  		}
>  
>  		if (task_can_run_on_remote_rq(p, rq, false)) {
> -			if (likely(consume_remote_task(rq, p, dsq, task_rq)))
> -				return true;
> -			goto retry;
> +			if (likely(consume_remote_task(rq, p, dsq, task_rq))) {
> +				ret = true;
> +				break;
> +			}
>  		}
>  	}
>  
>  	raw_spin_unlock(&dsq->lock);
> -	return false;
> +	return ret;

Has this change been tested at all? There's quite a bit of lock dancing
happening in the remote consumption path because the task needs to get
hot-migrated to the local CPU - consume_remote_task() calls
unlink_dsq_and_lock_src_rq() which drops the DSQ lock among other things.

Thanks.

-- 
tejun

hello

在 2024/10/16 4:12, Tejun Heo 写道:
> Hello,
> 
> On Tue, Oct 15, 2024 at 10:29:17PM +0800, Zhang Qiao wrote:
>> When the function consume_dispatch_q() returns true, the dsq lock may
>> remains held and is not unlocked.
>>
>> Signed-off-by: Zhang Qiao <zhangqiao22@huawei.com>
>> ---
>>  kernel/sched/ext.c | 16 +++++++++-------
>>  1 file changed, 9 insertions(+), 7 deletions(-)
>>
>> diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c
>> index 410a4df8a121..4d80aa3de00e 100644
>> --- a/kernel/sched/ext.c
>> +++ b/kernel/sched/ext.c
>> @@ -2377,7 +2377,8 @@ static inline bool consume_remote_task(struct rq *this_rq, struct task_struct *p
>>  static bool consume_dispatch_q(struct rq *rq, struct scx_dispatch_q *dsq)
>>  {
>>  	struct task_struct *p;
>> -retry:
>> +	bool ret = false;
>> +
>>  	/*
>>  	 * The caller can't expect to successfully consume a task if the task's
>>  	 * addition to @dsq isn't guaranteed to be visible somehow. Test
>> @@ -2394,19 +2395,20 @@ static bool consume_dispatch_q(struct rq *rq, struct scx_dispatch_q *dsq)
>>  		if (rq == task_rq) {
>>  			task_unlink_from_dsq(p, dsq);
>>  			move_local_task_to_local_dsq(p, 0, dsq, rq);
>> -			raw_spin_unlock(&dsq->lock);
>> -			return true;
>> +			ret = true;
>> +			break;
>>  		}
>>  
>>  		if (task_can_run_on_remote_rq(p, rq, false)) {
>> -			if (likely(consume_remote_task(rq, p, dsq, task_rq)))
>> -				return true;
>> -			goto retry;
>> +			if (likely(consume_remote_task(rq, p, dsq, task_rq))) {
>> +				ret = true;
>> +				break;
>> +			}
>>  		}
>>  	}
>>  
>>  	raw_spin_unlock(&dsq->lock);
>> -	return false;
>> +	return ret;
> 
> Has this change been tested at all? There's quite a bit of lock dancing

I tested this patch using scx_simple scheduler.

> happening in the remote consumption path because the task needs to get
> hot-migrated to the local CPU - consume_remote_task() calls
> unlink_dsq_and_lock_src_rq() which drops the DSQ lock among other things.

Yes, i missed it.

Thanks.

> 
> Thanks.
>