This patch replace a rwlock and raw notifier by atomic notifier which
protected by spin_lock and rcu.
The first to reason to have this replace is due to a 'scheduling while
atomic' bug of RT kernel on arm/arm64 platform. On arm/arm64, rwlock
cpu_pm_notifier_lock in cpu_pm cause a potential schedule after irq
disable in idle call chain:
cpu_startup_entry
cpu_idle_loop
local_irq_disable()
cpuidle_idle_call
call_cpuidle
cpuidle_enter
cpuidle_enter_state
->enter :arm_enter_idle_state
cpu_pm_enter/exit
CPU_PM_CPU_IDLE_ENTER
read_lock(&cpu_pm_notifier_lock); <-- sleep in idle
__rt_spin_lock();
schedule();
The kernel panic is here:
[ 4.609601] BUG: scheduling while atomic: swapper/1/0/0x00000002
[ 4.609608] [<ffff0000086fae70>] arm_enter_idle_state+0x18/0x70
[ 4.609614] Modules linked in:
[ 4.609615] [<ffff0000086f9298>] cpuidle_enter_state+0xf0/0x218
[ 4.609620] [<ffff0000086f93f8>] cpuidle_enter+0x18/0x20
[ 4.609626] Preemption disabled at:
[ 4.609627] [<ffff0000080fa234>] call_cpuidle+0x24/0x40
[ 4.609635] [<ffff000008882fa4>] schedule_preempt_disabled+0x1c/0x28
[ 4.609639] [<ffff0000080fa49c>] cpu_startup_entry+0x154/0x1f8
[ 4.609645] [<ffff00000808e004>] secondary_start_kernel+0x15c/0x1a0
Daniel Lezcano said this notification is needed on arm/arm64 platforms.
Sebastian suggested using atomic_notifier instead of rwlock, which is not
only removing the sleeping in idle, but also getting better latency
improvement.
This patch passed Fengguang's 0day testing.
Signed-off-by: Alex Shi <[email protected]>
Cc: Sebastian Andrzej Siewior <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: Anders Roxell <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Steven Rostedt <[email protected]>
Cc: Rafael J. Wysocki <[email protected]>
Cc: Daniel Lezcano <[email protected]>
Cc: linux-rt-users <[email protected]>
---
kernel/cpu_pm.c | 43 ++++++-------------------------------------
1 file changed, 6 insertions(+), 37 deletions(-)
diff --git a/kernel/cpu_pm.c b/kernel/cpu_pm.c
index 009cc9a..10f4640 100644
--- a/kernel/cpu_pm.c
+++ b/kernel/cpu_pm.c
@@ -22,14 +22,13 @@
#include <linux/spinlock.h>
#include <linux/syscore_ops.h>
-static DEFINE_RWLOCK(cpu_pm_notifier_lock);
-static RAW_NOTIFIER_HEAD(cpu_pm_notifier_chain);
+static ATOMIC_NOTIFIER_HEAD(cpu_pm_notifier_chain);
static int cpu_pm_notify(enum cpu_pm_event event, int nr_to_call, int *nr_calls)
{
int ret;
- ret = __raw_notifier_call_chain(&cpu_pm_notifier_chain, event, NULL,
+ ret = __atomic_notifier_call_chain(&cpu_pm_notifier_chain, event, NULL,
nr_to_call, nr_calls);
return notifier_to_errno(ret);
@@ -47,14 +46,7 @@ static int cpu_pm_notify(enum cpu_pm_event event, int nr_to_call, int *nr_calls)
*/
int cpu_pm_register_notifier(struct notifier_block *nb)
{
- unsigned long flags;
- int ret;
-
- write_lock_irqsave(&cpu_pm_notifier_lock, flags);
- ret = raw_notifier_chain_register(&cpu_pm_notifier_chain, nb);
- write_unlock_irqrestore(&cpu_pm_notifier_lock, flags);
-
- return ret;
+ return atomic_notifier_chain_register(&cpu_pm_notifier_chain, nb);
}
EXPORT_SYMBOL_GPL(cpu_pm_register_notifier);
@@ -69,14 +61,7 @@ EXPORT_SYMBOL_GPL(cpu_pm_register_notifier);
*/
int cpu_pm_unregister_notifier(struct notifier_block *nb)
{
- unsigned long flags;
- int ret;
-
- write_lock_irqsave(&cpu_pm_notifier_lock, flags);
- ret = raw_notifier_chain_unregister(&cpu_pm_notifier_chain, nb);
- write_unlock_irqrestore(&cpu_pm_notifier_lock, flags);
-
- return ret;
+ return atomic_notifier_chain_unregister(&cpu_pm_notifier_chain, nb);
}
EXPORT_SYMBOL_GPL(cpu_pm_unregister_notifier);
@@ -100,7 +85,6 @@ int cpu_pm_enter(void)
int nr_calls;
int ret = 0;
- read_lock(&cpu_pm_notifier_lock);
ret = cpu_pm_notify(CPU_PM_ENTER, -1, &nr_calls);
if (ret)
/*
@@ -108,7 +92,6 @@ int cpu_pm_enter(void)
* PM entry who are notified earlier to prepare for it.
*/
cpu_pm_notify(CPU_PM_ENTER_FAILED, nr_calls - 1, NULL);
- read_unlock(&cpu_pm_notifier_lock);
return ret;
}
@@ -128,13 +111,7 @@ EXPORT_SYMBOL_GPL(cpu_pm_enter);
*/
int cpu_pm_exit(void)
{
- int ret;
-
- read_lock(&cpu_pm_notifier_lock);
- ret = cpu_pm_notify(CPU_PM_EXIT, -1, NULL);
- read_unlock(&cpu_pm_notifier_lock);
-
- return ret;
+ return cpu_pm_notify(CPU_PM_EXIT, -1, NULL);
}
EXPORT_SYMBOL_GPL(cpu_pm_exit);
@@ -159,7 +136,6 @@ int cpu_cluster_pm_enter(void)
int nr_calls;
int ret = 0;
- read_lock(&cpu_pm_notifier_lock);
ret = cpu_pm_notify(CPU_CLUSTER_PM_ENTER, -1, &nr_calls);
if (ret)
/*
@@ -167,7 +143,6 @@ int cpu_cluster_pm_enter(void)
* PM entry who are notified earlier to prepare for it.
*/
cpu_pm_notify(CPU_CLUSTER_PM_ENTER_FAILED, nr_calls - 1, NULL);
- read_unlock(&cpu_pm_notifier_lock);
return ret;
}
@@ -190,13 +165,7 @@ EXPORT_SYMBOL_GPL(cpu_cluster_pm_enter);
*/
int cpu_cluster_pm_exit(void)
{
- int ret;
-
- read_lock(&cpu_pm_notifier_lock);
- ret = cpu_pm_notify(CPU_CLUSTER_PM_EXIT, -1, NULL);
- read_unlock(&cpu_pm_notifier_lock);
-
- return ret;
+ return cpu_pm_notify(CPU_CLUSTER_PM_EXIT, -1, NULL);
}
EXPORT_SYMBOL_GPL(cpu_cluster_pm_exit);
--
2.7.4
On 2017-07-06 16:47:46 [+0800], Alex Shi wrote:
> This patch replace a rwlock and raw notifier by atomic notifier which
+ is
> protected by spin_lock and rcu.
…
> Sebastian suggested using atomic_notifier instead of rwlock, which is not
> only removing the sleeping in idle, but also getting better latency
> improvement.
Did you measure this / have numbers or did you write this because this
is what RCU does in general?
Acked-by: Sebastian Andrzej Siewior <[email protected]>
Sebastian
On 07/07/2017 06:25 PM, Sebastian Andrzej Siewior wrote:
> On 2017-07-06 16:47:46 [+0800], Alex Shi wrote:
>> This patch replace a rwlock and raw notifier by atomic notifier which
> + is
>> protected by spin_lock and rcu.
> …
>> Sebastian suggested using atomic_notifier instead of rwlock, which is not
>> only removing the sleeping in idle, but also getting better latency
>> improvement.
>
> Did you measure this / have numbers or did you write this because this
> is what RCU does in general?
Uh, I have no number on the performance. The benefit is in the theory.
https://lwn.net/Articles/263130/
But above article give some comparison data of rwlock and RCU.
>
> Acked-by: Sebastian Andrzej Siewior <[email protected]>
>
Thanks a lot!
Regards
Alex
It is a serious bug: add a waiting lock in idle and cause boot failure
in arm/arm64 RT.
Any more comments for this change?
On 07/06/2017 04:47 PM, Alex Shi wrote:
> This patch replace a rwlock and raw notifier by atomic notifier which
> protected by spin_lock and rcu.
>
> The first to reason to have this replace is due to a 'scheduling while
> atomic' bug of RT kernel on arm/arm64 platform. On arm/arm64, rwlock
> cpu_pm_notifier_lock in cpu_pm cause a potential schedule after irq
> disable in idle call chain:
>
> cpu_startup_entry
> cpu_idle_loop
> local_irq_disable()
> cpuidle_idle_call
> call_cpuidle
> cpuidle_enter
> cpuidle_enter_state
> ->enter :arm_enter_idle_state
> cpu_pm_enter/exit
> CPU_PM_CPU_IDLE_ENTER
> read_lock(&cpu_pm_notifier_lock); <-- sleep in idle
> __rt_spin_lock();
> schedule();
>
> The kernel panic is here:
> [ 4.609601] BUG: scheduling while atomic: swapper/1/0/0x00000002
> [ 4.609608] [<ffff0000086fae70>] arm_enter_idle_state+0x18/0x70
> [ 4.609614] Modules linked in:
> [ 4.609615] [<ffff0000086f9298>] cpuidle_enter_state+0xf0/0x218
> [ 4.609620] [<ffff0000086f93f8>] cpuidle_enter+0x18/0x20
> [ 4.609626] Preemption disabled at:
> [ 4.609627] [<ffff0000080fa234>] call_cpuidle+0x24/0x40
> [ 4.609635] [<ffff000008882fa4>] schedule_preempt_disabled+0x1c/0x28
> [ 4.609639] [<ffff0000080fa49c>] cpu_startup_entry+0x154/0x1f8
> [ 4.609645] [<ffff00000808e004>] secondary_start_kernel+0x15c/0x1a0
>
> Daniel Lezcano said this notification is needed on arm/arm64 platforms.
> Sebastian suggested using atomic_notifier instead of rwlock, which is not
> only removing the sleeping in idle, but also getting better latency
> improvement.
>
> This patch passed Fengguang's 0day testing.
>
> Signed-off-by: Alex Shi <[email protected]>
> Cc: Sebastian Andrzej Siewior <[email protected]>
> Cc: Thomas Gleixner <[email protected]>
> Cc: Anders Roxell <[email protected]>
> Cc: Rik van Riel <[email protected]>
> Cc: Steven Rostedt <[email protected]>
> Cc: Rafael J. Wysocki <[email protected]>
> Cc: Daniel Lezcano <[email protected]>
> Cc: linux-rt-users <[email protected]>
> ---
> kernel/cpu_pm.c | 43 ++++++-------------------------------------
> 1 file changed, 6 insertions(+), 37 deletions(-)
>
> diff --git a/kernel/cpu_pm.c b/kernel/cpu_pm.c
> index 009cc9a..10f4640 100644
> --- a/kernel/cpu_pm.c
> +++ b/kernel/cpu_pm.c
> @@ -22,14 +22,13 @@
> #include <linux/spinlock.h>
> #include <linux/syscore_ops.h>
>
> -static DEFINE_RWLOCK(cpu_pm_notifier_lock);
> -static RAW_NOTIFIER_HEAD(cpu_pm_notifier_chain);
> +static ATOMIC_NOTIFIER_HEAD(cpu_pm_notifier_chain);
>
> static int cpu_pm_notify(enum cpu_pm_event event, int nr_to_call, int *nr_calls)
> {
> int ret;
>
> - ret = __raw_notifier_call_chain(&cpu_pm_notifier_chain, event, NULL,
> + ret = __atomic_notifier_call_chain(&cpu_pm_notifier_chain, event, NULL,
> nr_to_call, nr_calls);
>
> return notifier_to_errno(ret);
> @@ -47,14 +46,7 @@ static int cpu_pm_notify(enum cpu_pm_event event, int nr_to_call, int *nr_calls)
> */
> int cpu_pm_register_notifier(struct notifier_block *nb)
> {
> - unsigned long flags;
> - int ret;
> -
> - write_lock_irqsave(&cpu_pm_notifier_lock, flags);
> - ret = raw_notifier_chain_register(&cpu_pm_notifier_chain, nb);
> - write_unlock_irqrestore(&cpu_pm_notifier_lock, flags);
> -
> - return ret;
> + return atomic_notifier_chain_register(&cpu_pm_notifier_chain, nb);
> }
> EXPORT_SYMBOL_GPL(cpu_pm_register_notifier);
>
> @@ -69,14 +61,7 @@ EXPORT_SYMBOL_GPL(cpu_pm_register_notifier);
> */
> int cpu_pm_unregister_notifier(struct notifier_block *nb)
> {
> - unsigned long flags;
> - int ret;
> -
> - write_lock_irqsave(&cpu_pm_notifier_lock, flags);
> - ret = raw_notifier_chain_unregister(&cpu_pm_notifier_chain, nb);
> - write_unlock_irqrestore(&cpu_pm_notifier_lock, flags);
> -
> - return ret;
> + return atomic_notifier_chain_unregister(&cpu_pm_notifier_chain, nb);
> }
> EXPORT_SYMBOL_GPL(cpu_pm_unregister_notifier);
>
> @@ -100,7 +85,6 @@ int cpu_pm_enter(void)
> int nr_calls;
> int ret = 0;
>
> - read_lock(&cpu_pm_notifier_lock);
> ret = cpu_pm_notify(CPU_PM_ENTER, -1, &nr_calls);
> if (ret)
> /*
> @@ -108,7 +92,6 @@ int cpu_pm_enter(void)
> * PM entry who are notified earlier to prepare for it.
> */
> cpu_pm_notify(CPU_PM_ENTER_FAILED, nr_calls - 1, NULL);
> - read_unlock(&cpu_pm_notifier_lock);
>
> return ret;
> }
> @@ -128,13 +111,7 @@ EXPORT_SYMBOL_GPL(cpu_pm_enter);
> */
> int cpu_pm_exit(void)
> {
> - int ret;
> -
> - read_lock(&cpu_pm_notifier_lock);
> - ret = cpu_pm_notify(CPU_PM_EXIT, -1, NULL);
> - read_unlock(&cpu_pm_notifier_lock);
> -
> - return ret;
> + return cpu_pm_notify(CPU_PM_EXIT, -1, NULL);
> }
> EXPORT_SYMBOL_GPL(cpu_pm_exit);
>
> @@ -159,7 +136,6 @@ int cpu_cluster_pm_enter(void)
> int nr_calls;
> int ret = 0;
>
> - read_lock(&cpu_pm_notifier_lock);
> ret = cpu_pm_notify(CPU_CLUSTER_PM_ENTER, -1, &nr_calls);
> if (ret)
> /*
> @@ -167,7 +143,6 @@ int cpu_cluster_pm_enter(void)
> * PM entry who are notified earlier to prepare for it.
> */
> cpu_pm_notify(CPU_CLUSTER_PM_ENTER_FAILED, nr_calls - 1, NULL);
> - read_unlock(&cpu_pm_notifier_lock);
>
> return ret;
> }
> @@ -190,13 +165,7 @@ EXPORT_SYMBOL_GPL(cpu_cluster_pm_enter);
> */
> int cpu_cluster_pm_exit(void)
> {
> - int ret;
> -
> - read_lock(&cpu_pm_notifier_lock);
> - ret = cpu_pm_notify(CPU_CLUSTER_PM_EXIT, -1, NULL);
> - read_unlock(&cpu_pm_notifier_lock);
> -
> - return ret;
> + return cpu_pm_notify(CPU_CLUSTER_PM_EXIT, -1, NULL);
> }
> EXPORT_SYMBOL_GPL(cpu_cluster_pm_exit);
>
>
On 2017-07-11 22:42:04 [+0800], Alex Shi wrote:
> It is a serious bug: add a waiting lock in idle and cause boot failure
> in arm/arm64 RT.
>
> Any more comments for this change?
As far as RT is concerned, I am taking this for the next v4.11 release.
I would appreciate if upstream would apply this as well.
Rafael do you feel responsible for this?
Sebastian
On Tuesday, July 11, 2017 05:06:35 PM Sebastian Andrzej Siewior wrote:
> On 2017-07-11 22:42:04 [+0800], Alex Shi wrote:
> > It is a serious bug: add a waiting lock in idle and cause boot failure
> > in arm/arm64 RT.
> >
> > Any more comments for this change?
>
> As far as RT is concerned, I am taking this for the next v4.11 release.
> I would appreciate if upstream would apply this as well.
> Rafael do you feel responsible for this?
I can apply this if no one else wants to. :-)
However, I'd appreciate a resend with a CC to [email protected] for
easier processing.
Thanks,
Rafael
On 2017-07-11 17:01:09 [+0200], Rafael J. Wysocki wrote:
> > As far as RT is concerned, I am taking this for the next v4.11 release.
> > I would appreciate if upstream would apply this as well.
> > Rafael do you feel responsible for this?
>
> I can apply this if no one else wants to. :-)
prosze.
> However, I'd appreciate a resend with a CC to [email protected] for
> easier processing.
Alex: please repost.
> Thanks,
> Rafael
Sebastian
On 07/11/2017 11:16 PM, Sebastian Andrzej Siewior wrote:
> On 2017-07-11 17:01:09 [+0200], Rafael J. Wysocki wrote:
>>> As far as RT is concerned, I am taking this for the next v4.11 release.
>>> I would appreciate if upstream would apply this as well.
>>> Rafael do you feel responsible for this?
>>
>> I can apply this if no one else wants to. :-)
>
> prosze.
>
>> However, I'd appreciate a resend with a CC to [email protected] for
>> easier processing.
>
> Alex: please repost.
Reposted. Thanks for you all! :)