I got a minus(very big) dl_b->total_bw during my deadline tests.
# grep dl /proc/sched_debug
dl_rq[0]:
.dl_nr_running : 0
.dl_bw->bw : 996147
.dl_bw->total_bw : -222297900
Something unusual must have happened.
After some digging, I finally noticed that when changing a deadline
task to normal(cfs), and changing it back to deadline immediately,
after it died, we will got the wrong dl_bw->total_bw.
The root cause is in dl_overflow(), it has:
if (new_bw == p->dl.dl_bw)
return 0;
1) When a deadline task is changed to !deadline task, it will start
dl timer in switched_from_dl(), and retain previous deadline parameter
till the timer expires.
2) If we change it back to deadline with the same bandwidth parameter
before the timer expires, as it keeps the old bandwidth although it
is not a deadline task. dl_overflow() simply returns success without
updating the right data, and got the wrong dl_bw->total_bw.
The solution is simple, if @p is not deadline, don't return.
Signed-off-by: Xunlei Pang <[email protected]>
---
kernel/sched/core.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 4a2c79d..5988fee 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -2378,7 +2378,8 @@ static int dl_overflow(struct task_struct *p, int policy,
u64 new_bw = dl_policy(policy) ? to_ratio(period, runtime) : 0;
int cpus, err = -1;
- if (new_bw == p->dl.dl_bw)
+ /* !deadline task may carry old deadline bandwidth */
+ if (new_bw == p->dl.dl_bw && task_has_dl_policy(p))
return 0;
/*
--
1.8.3.1
[+Luca]
Hi,
On 14/04/16 20:19, Xunlei Pang wrote:
> I got a minus(very big) dl_b->total_bw during my deadline tests.
>
> # grep dl /proc/sched_debug
> dl_rq[0]:
> .dl_nr_running : 0
> .dl_bw->bw : 996147
> .dl_bw->total_bw : -222297900
>
> Something unusual must have happened.
>
> After some digging, I finally noticed that when changing a deadline
> task to normal(cfs), and changing it back to deadline immediately,
> after it died, we will got the wrong dl_bw->total_bw.
>
> The root cause is in dl_overflow(), it has:
> if (new_bw == p->dl.dl_bw)
> return 0;
>
> 1) When a deadline task is changed to !deadline task, it will start
> dl timer in switched_from_dl(), and retain previous deadline parameter
> till the timer expires.
> 2) If we change it back to deadline with the same bandwidth parameter
> before the timer expires, as it keeps the old bandwidth although it
> is not a deadline task. dl_overflow() simply returns success without
> updating the right data, and got the wrong dl_bw->total_bw.
>
> The solution is simple, if @p is not deadline, don't return.
>
> Signed-off-by: Xunlei Pang <[email protected]>
> ---
> kernel/sched/core.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 4a2c79d..5988fee 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -2378,7 +2378,8 @@ static int dl_overflow(struct task_struct *p, int policy,
> u64 new_bw = dl_policy(policy) ? to_ratio(period, runtime) : 0;
> int cpus, err = -1;
>
> - if (new_bw == p->dl.dl_bw)
> + /* !deadline task may carry old deadline bandwidth */
> + if (new_bw == p->dl.dl_bw && task_has_dl_policy(p))
Right. I got the same patch that I believe Luca is be already using for
his tests (and he also put together the changelog). I never managed to
send it out, sorry about that. We can take yours, mine follows just in
case we want to take something from the changelog or we want to reverse
the if condition.
Thanks,
- Juri
--->8---
>From 4bf38111bd9383035e03d3dc3d42011aaa9e26e7 Mon Sep 17 00:00:00 2001
From: Juri Lelli <[email protected]>
Date: Thu, 25 Feb 2016 15:50:42 +0100
Subject: [PATCH 2/3] fix a bug in the -deadline utilization tracking mechanism
Currently, a task doing
while(1) {
switch to SCHED_DEADLINE
switch to SCHED_OTHER
}
brings dl_b->total_bw below 0.
This happens because when the task switches back from SCHED_DEADLINE
to SCHED_OTHER, switched_from_dl() does not clear its deadline
parameters (they will be cleared by the deadline timer when it fires).
But dl_overflow() removes its utilization from dl_b->total_bw.
When the task switches back to SCHED_DEADLINE, the
if (new_bw == p->dl.dl_bw)
check in dl_overflow() prevents __dl_add() from being called, and
so when the task switches back to SCHED_OTHER dl_b->total_bw becomes
negative.
This patch changes the check so that if the task is switching from
SCHED_OTHER to SCHED_DEADLINE __dl_add() is correctly invoked.
---
kernel/sched/core.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 9503d59..d59fa20 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -2432,7 +2432,7 @@ static int dl_overflow(struct task_struct *p, int policy,
u64 new_bw = dl_policy(policy) ? to_ratio(period, runtime) : 0;
int cpus, err = -1;
- if (new_bw == p->dl.dl_bw)
+ if (task_has_dl_policy(p) && new_bw == p->dl.dl_bw)
return 0;
/*
--
2.5.0
On 2016/04/15 at 15:07, Juri Lelli wrote:
> [+Luca]
>
> Hi,
>
> On 14/04/16 20:19, Xunlei Pang wrote:
>> I got a minus(very big) dl_b->total_bw during my deadline tests.
>>
>> # grep dl /proc/sched_debug
>> dl_rq[0]:
>> .dl_nr_running : 0
>> .dl_bw->bw : 996147
>> .dl_bw->total_bw : -222297900
>>
>> Something unusual must have happened.
>>
>> After some digging, I finally noticed that when changing a deadline
>> task to normal(cfs), and changing it back to deadline immediately,
>> after it died, we will got the wrong dl_bw->total_bw.
>>
>> The root cause is in dl_overflow(), it has:
>> if (new_bw == p->dl.dl_bw)
>> return 0;
>>
>> 1) When a deadline task is changed to !deadline task, it will start
>> dl timer in switched_from_dl(), and retain previous deadline parameter
>> till the timer expires.
>> 2) If we change it back to deadline with the same bandwidth parameter
>> before the timer expires, as it keeps the old bandwidth although it
>> is not a deadline task. dl_overflow() simply returns success without
>> updating the right data, and got the wrong dl_bw->total_bw.
>>
>> The solution is simple, if @p is not deadline, don't return.
>>
>> Signed-off-by: Xunlei Pang <[email protected]>
>> ---
>> kernel/sched/core.c | 3 ++-
>> 1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
>> index 4a2c79d..5988fee 100644
>> --- a/kernel/sched/core.c
>> +++ b/kernel/sched/core.c
>> @@ -2378,7 +2378,8 @@ static int dl_overflow(struct task_struct *p, int policy,
>> u64 new_bw = dl_policy(policy) ? to_ratio(period, runtime) : 0;
>> int cpus, err = -1;
>>
>> - if (new_bw == p->dl.dl_bw)
>> + /* !deadline task may carry old deadline bandwidth */
>> + if (new_bw == p->dl.dl_bw && task_has_dl_policy(p))
> Right. I got the same patch that I believe Luca is be already using for
> his tests (and he also put together the changelog). I never managed to
> send it out, sorry about that. We can take yours, mine follows just in
> case we want to take something from the changelog or we want to reverse
> the if condition.
Ah, exactly the same issue. Thanks for your feedback :-)
Regards,
Xunlei
>
> Thanks,
>
> - Juri
>
> --->8---
>
> From 4bf38111bd9383035e03d3dc3d42011aaa9e26e7 Mon Sep 17 00:00:00 2001
> From: Juri Lelli <[email protected]>
> Date: Thu, 25 Feb 2016 15:50:42 +0100
> Subject: [PATCH 2/3] fix a bug in the -deadline utilization tracking mechanism
>
> Currently, a task doing
> while(1) {
> switch to SCHED_DEADLINE
> switch to SCHED_OTHER
> }
> brings dl_b->total_bw below 0.
> This happens because when the task switches back from SCHED_DEADLINE
> to SCHED_OTHER, switched_from_dl() does not clear its deadline
> parameters (they will be cleared by the deadline timer when it fires).
> But dl_overflow() removes its utilization from dl_b->total_bw.
> When the task switches back to SCHED_DEADLINE, the
> if (new_bw == p->dl.dl_bw)
> check in dl_overflow() prevents __dl_add() from being called, and
> so when the task switches back to SCHED_OTHER dl_b->total_bw becomes
> negative.
> This patch changes the check so that if the task is switching from
> SCHED_OTHER to SCHED_DEADLINE __dl_add() is correctly invoked.
> ---
> kernel/sched/core.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 9503d59..d59fa20 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -2432,7 +2432,7 @@ static int dl_overflow(struct task_struct *p, int policy,
> u64 new_bw = dl_policy(policy) ? to_ratio(period, runtime) : 0;
> int cpus, err = -1;
>
> - if (new_bw == p->dl.dl_bw)
> + if (task_has_dl_policy(p) && new_bw == p->dl.dl_bw)
> return 0;
>
> /*
Commit-ID: fec148c000d0f9ac21679601722811eb60b4cc52
Gitweb: http://git.kernel.org/tip/fec148c000d0f9ac21679601722811eb60b4cc52
Author: Xunlei Pang <[email protected]>
AuthorDate: Thu, 14 Apr 2016 20:19:28 +0800
Committer: Ingo Molnar <[email protected]>
CommitDate: Sat, 23 Apr 2016 14:20:43 +0200
sched/deadline: Fix a bug in dl_overflow()
I got a minus(very big) dl_b->total_bw during my deadline tests.
# grep dl /proc/sched_debug
dl_rq[0]:
.dl_nr_running : 0
.dl_bw->bw : 996147
.dl_bw->total_bw : -222297900
Something unusual must have happened.
After some digging, I finally noticed that when changing a deadline
task to normal(cfs), and changing it back to deadline immediately,
after it died, we will got the wrong dl_bw->total_bw.
The root cause is in dl_overflow(), it has:
if (new_bw == p->dl.dl_bw)
return 0;
1) When a deadline task is changed to !deadline task, it will start
dl timer in switched_from_dl(), and retain previous deadline parameter
till the timer expires.
2) If we change it back to deadline with the same bandwidth parameter
before the timer expires, as it keeps the old bandwidth although it
is not a deadline task. dl_overflow() simply returns success without
updating the right data, and got the wrong dl_bw->total_bw.
The solution is simple, if @p is not deadline, don't return.
Signed-off-by: Xunlei Pang <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Acked-by: Juri Lelli <[email protected]>
Cc: Mike Galbraith <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Steven Rostedt <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Ingo Molnar <[email protected]>
---
kernel/sched/core.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 71dffbb..9d84d60 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -2378,7 +2378,8 @@ static int dl_overflow(struct task_struct *p, int policy,
u64 new_bw = dl_policy(policy) ? to_ratio(period, runtime) : 0;
int cpus, err = -1;
- if (new_bw == p->dl.dl_bw)
+ /* !deadline task may carry old deadline bandwidth */
+ if (new_bw == p->dl.dl_bw && task_has_dl_policy(p))
return 0;
/*