diff mbox

[1/2] : pkt_sched: Convert qdisc_watchdog to tasklet_hrtimer

Message ID 20090821.170330.217935645.davem@davemloft.net
State Accepted, archived
Delegated to: David Miller
Headers show

Commit Message

David Miller Aug. 22, 2009, 12:03 a.m. UTC
None of this stuff should execute in hw IRQ context, therefore
use a tasklet_hrtimer so that it runs in softirq context.

Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/pkt_sched.h |    4 ++--
 net/sched/sch_api.c     |    8 ++++----
 2 files changed, 6 insertions(+), 6 deletions(-)

Comments

Thomas Gleixner Aug. 22, 2009, 8:41 a.m. UTC | #1
On Fri, 21 Aug 2009, David Miller wrote:

> 
> None of this stuff should execute in hw IRQ context, therefore
> use a tasklet_hrtimer so that it runs in softirq context.
> 
> Signed-off-by: David S. Miller <davem@davemloft.net>

Acked-by: Thomas Gleixner <tglx@linutronix.de>

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Jarek Poplawski Aug. 22, 2009, 12:04 p.m. UTC | #2
David Miller wrote, On 08/22/2009 02:03 AM:

> None of this stuff should execute in hw IRQ context, therefore
> use a tasklet_hrtimer so that it runs in softirq context.


Do you mean __netif_schedule()? Could you explain it more?

Thanks,
Jarek P.

> 
> Signed-off-by: David S. Miller <davem@davemloft.net>
> ---
>  include/net/pkt_sched.h |    4 ++--
>  net/sched/sch_api.c     |    8 ++++----
>  2 files changed, 6 insertions(+), 6 deletions(-)
> 
> diff --git a/include/net/pkt_sched.h b/include/net/pkt_sched.h
> index 82a3191..7eafb8d 100644
> --- a/include/net/pkt_sched.h
> +++ b/include/net/pkt_sched.h
> @@ -61,8 +61,8 @@ psched_tdiff_bounded(psched_time_t tv1, psched_time_t tv2, psched_time_t bound)
>  }
>  
>  struct qdisc_watchdog {
> -	struct hrtimer	timer;
> -	struct Qdisc	*qdisc;
> +	struct tasklet_hrtimer	timer;
> +	struct Qdisc		*qdisc;
>  };
>  
>  extern void qdisc_watchdog_init(struct qdisc_watchdog *wd, struct Qdisc *qdisc);
> diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
> index 24d17ce..e1c2bf7 100644
> --- a/net/sched/sch_api.c
> +++ b/net/sched/sch_api.c
> @@ -468,8 +468,8 @@ static enum hrtimer_restart qdisc_watchdog(struct hrtimer *timer)
>  
>  void qdisc_watchdog_init(struct qdisc_watchdog *wd, struct Qdisc *qdisc)
>  {
> -	hrtimer_init(&wd->timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
> -	wd->timer.function = qdisc_watchdog;
> +	tasklet_hrtimer_init(&wd->timer, qdisc_watchdog,
> +			     CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
>  	wd->qdisc = qdisc;
>  }
>  EXPORT_SYMBOL(qdisc_watchdog_init);
> @@ -485,13 +485,13 @@ void qdisc_watchdog_schedule(struct qdisc_watchdog *wd, psched_time_t expires)
>  	wd->qdisc->flags |= TCQ_F_THROTTLED;
>  	time = ktime_set(0, 0);
>  	time = ktime_add_ns(time, PSCHED_TICKS2NS(expires));
> -	hrtimer_start(&wd->timer, time, HRTIMER_MODE_ABS);
> +	tasklet_hrtimer_start(&wd->timer, time, HRTIMER_MODE_ABS);
>  }
>  EXPORT_SYMBOL(qdisc_watchdog_schedule);
>  
>  void qdisc_watchdog_cancel(struct qdisc_watchdog *wd)
>  {
> -	hrtimer_cancel(&wd->timer);
> +	tasklet_hrtimer_cancel(&wd->timer);
>  	wd->qdisc->flags &= ~TCQ_F_THROTTLED;
>  }
>  EXPORT_SYMBOL(qdisc_watchdog_cancel);


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller Aug. 23, 2009, 12:54 a.m. UTC | #3
From: Jarek Poplawski <jarkao2@gmail.com>
Date: Sat, 22 Aug 2009 14:04:31 +0200

> David Miller wrote, On 08/22/2009 02:03 AM:
> 
>> None of this stuff should execute in hw IRQ context, therefore
>> use a tasklet_hrtimer so that it runs in softirq context.
> 
> Do you mean __netif_schedule()?

No, that works fine from HW irq context.

> Could you explain it more?

It's that throttled flag bit setting.  It's still unsafe but at least
we're in a position now to use softirq locking or atomic bitops to
maintain it properly.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller Aug. 23, 2009, 1:08 a.m. UTC | #4
From: Thomas Gleixner <tglx@linutronix.de>
Date: Sat, 22 Aug 2009 10:41:18 +0200 (CEST)

> On Fri, 21 Aug 2009, David Miller wrote:
> 
>> 
>> None of this stuff should execute in hw IRQ context, therefore
>> use a tasklet_hrtimer so that it runs in softirq context.
>> 
>> Signed-off-by: David S. Miller <davem@davemloft.net>
> 
> Acked-by: Thomas Gleixner <tglx@linutronix.de>

Thanks for reviewing.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Jarek Poplawski Aug. 27, 2009, 8:36 a.m. UTC | #5
On 22-08-2009 02:03, David Miller wrote:
> None of this stuff should execute in hw IRQ context, therefore
> use a tasklet_hrtimer so that it runs in softirq context.
> 
> Signed-off-by: David S. Miller <davem@davemloft.net>
> ---
...
>  void qdisc_watchdog_cancel(struct qdisc_watchdog *wd)
>  {
> -	hrtimer_cancel(&wd->timer);
> +	tasklet_hrtimer_cancel(&wd->timer);
>  	wd->qdisc->flags &= ~TCQ_F_THROTTLED;
>  }
>  EXPORT_SYMBOL(qdisc_watchdog_cancel);

I've had a second look at it and I wonder why it's not reported, but
since qdisc_watchdog_cancel is run in qdisc_reset with BHs disabled
it seems there should be (at least) a warning triggered in
tasklet_kill on positive in_interrupt test (unless I miss something,
but can't test it now).

Jarek P.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller Sept. 1, 2009, 10:49 p.m. UTC | #6
From: Jarek Poplawski <jarkao2@gmail.com>
Date: Thu, 27 Aug 2009 08:36:41 +0000

> On 22-08-2009 02:03, David Miller wrote:
>> None of this stuff should execute in hw IRQ context, therefore
>> use a tasklet_hrtimer so that it runs in softirq context.
>> 
>> Signed-off-by: David S. Miller <davem@davemloft.net>
>> ---
> ...
>>  void qdisc_watchdog_cancel(struct qdisc_watchdog *wd)
>>  {
>> -	hrtimer_cancel(&wd->timer);
>> +	tasklet_hrtimer_cancel(&wd->timer);
>>  	wd->qdisc->flags &= ~TCQ_F_THROTTLED;
>>  }
>>  EXPORT_SYMBOL(qdisc_watchdog_cancel);
> 
> I've had a second look at it and I wonder why it's not reported, but
> since qdisc_watchdog_cancel is run in qdisc_reset with BHs disabled
> it seems there should be (at least) a warning triggered in
> tasklet_kill on positive in_interrupt test (unless I miss something,
> but can't test it now).

There are a lot of unknowns about this stuff.

It's very likely I'm going to simply revert these two patches
and try to work on these issues in net-next-2.6 instead.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/include/net/pkt_sched.h b/include/net/pkt_sched.h
index 82a3191..7eafb8d 100644
--- a/include/net/pkt_sched.h
+++ b/include/net/pkt_sched.h
@@ -61,8 +61,8 @@  psched_tdiff_bounded(psched_time_t tv1, psched_time_t tv2, psched_time_t bound)
 }
 
 struct qdisc_watchdog {
-	struct hrtimer	timer;
-	struct Qdisc	*qdisc;
+	struct tasklet_hrtimer	timer;
+	struct Qdisc		*qdisc;
 };
 
 extern void qdisc_watchdog_init(struct qdisc_watchdog *wd, struct Qdisc *qdisc);
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index 24d17ce..e1c2bf7 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -468,8 +468,8 @@  static enum hrtimer_restart qdisc_watchdog(struct hrtimer *timer)
 
 void qdisc_watchdog_init(struct qdisc_watchdog *wd, struct Qdisc *qdisc)
 {
-	hrtimer_init(&wd->timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
-	wd->timer.function = qdisc_watchdog;
+	tasklet_hrtimer_init(&wd->timer, qdisc_watchdog,
+			     CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
 	wd->qdisc = qdisc;
 }
 EXPORT_SYMBOL(qdisc_watchdog_init);
@@ -485,13 +485,13 @@  void qdisc_watchdog_schedule(struct qdisc_watchdog *wd, psched_time_t expires)
 	wd->qdisc->flags |= TCQ_F_THROTTLED;
 	time = ktime_set(0, 0);
 	time = ktime_add_ns(time, PSCHED_TICKS2NS(expires));
-	hrtimer_start(&wd->timer, time, HRTIMER_MODE_ABS);
+	tasklet_hrtimer_start(&wd->timer, time, HRTIMER_MODE_ABS);
 }
 EXPORT_SYMBOL(qdisc_watchdog_schedule);
 
 void qdisc_watchdog_cancel(struct qdisc_watchdog *wd)
 {
-	hrtimer_cancel(&wd->timer);
+	tasklet_hrtimer_cancel(&wd->timer);
 	wd->qdisc->flags &= ~TCQ_F_THROTTLED;
 }
 EXPORT_SYMBOL(qdisc_watchdog_cancel);