[PATCH v4 0/2] make arm_cpuidle_suspend() a bit more efficient

There's one corner case need to be fixed: !cpuidle_ops[cpu].init.
patch1 tries to address this corner case.

patch2 tries to improve arm_cpuidle_suspend() a bit by moving .suspend
check into arm_cpuidle_read_ops().

Since v3:
 - rebased on 4.7-rc1

Since v2:
 - add back patch1.

Since v1:
 - drop patch1, I.E "ARM: cpuidle: fix !cpuidle_ops[cpu].init case
   during init", because there may be platforms which doesn't need the
   init member at all.
 - adopt Daniel's suggestion to move the cpuidle_ops.suspend check to
   arm_cpuidle_read_ops().


Jisheng Zhang (2):
  ARM: cpuidle: fix !cpuidle_ops[cpu].init case during init
  ARM: cpuidle: make arm_cpuidle_suspend() a bit more efficient

 arch/arm/kernel/cpuidle.c | 23 +++++++++++++----------
 1 file changed, 13 insertions(+), 10 deletions(-)

-- 
2.8.1

Currently, we check cpuidle_ops.suspend every time when entering a
low-power idle state. But this check could be avoided in this hot path
by moving it into arm_cpuidle_read_ops() to reduce arm_cpuidle_suspend
overhead a bit.

Signed-off-by: Jisheng Zhang <redacted>
---
 arch/arm/kernel/cpuidle.c | 17 ++++++-----------
 1 file changed, 6 insertions(+), 11 deletions(-)

diff --git a/arch/arm/kernel/cpuidle.c b/arch/arm/kernel/cpuidle.c
index 2129b29..7dccc96 100644
--- a/arch/arm/kernel/cpuidle.c
+++ b/arch/arm/kernel/cpuidle.c

@@ -47,18 +47,13 @@ int arm_cpuidle_simple_enter(struct cpuidle_device *dev,
  * This function calls the underlying arch specific low level PM code as
  * registered at the init time.
  *
- * Returns -EOPNOTSUPP if no suspend callback is defined, the result of the
- * callback otherwise.
+ * Returns the result of the suspend callback.
  */
 int arm_cpuidle_suspend(int index)
 {
-	int ret = -EOPNOTSUPP;
 	int cpu = smp_processor_id();
 
-	if (cpuidle_ops[cpu].suspend)
-		ret = cpuidle_ops[cpu].suspend(index);
-
-	return ret;
+	return cpuidle_ops[cpu].suspend(index);
 }
 
 /**

@@ -92,8 +87,8 @@ static const struct cpuidle_ops *__init arm_cpuidle_get_ops(const char *method)
  * process.
  *
  * Return 0 on sucess, -ENOENT if no 'enable-method' is defined, -EOPNOTSUPP if
- * no cpuidle_ops is registered for the 'enable-method', or if no init callback
- * is defined.
+ * no cpuidle_ops is registered for the 'enable-method', or if either init or
+ * suspend callback isn't defined.
  */
 static int __init arm_cpuidle_read_ops(struct device_node *dn, int cpu)
 {

@@ -111,8 +106,8 @@ static int __init arm_cpuidle_read_ops(struct device_node *dn, int cpu)
 		return -EOPNOTSUPP;
 	}
 
-	if (!ops->init) {
-		pr_warn("cpuidle_ops '%s': no init callback\n",
+	if (!ops->init || !ops->suspend) {
+		pr_warn("cpuidle_ops '%s': no init or suspend callback\n",
 			enable_method);
 		return -EOPNOTSUPP;
 	}

-- 
2.8.1

Let's assume cpuidle_ops exists but it doesn't implement the according
init callback, current arm_cpuidle_init() will return success to its
caller, but in fact it should return -EOPNOTSUPP.

Signed-off-by: Jisheng Zhang <redacted>
---
 arch/arm/kernel/cpuidle.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/arch/arm/kernel/cpuidle.c b/arch/arm/kernel/cpuidle.c
index a44b268e..2129b29 100644
--- a/arch/arm/kernel/cpuidle.c
+++ b/arch/arm/kernel/cpuidle.c

@@ -92,7 +92,8 @@ static const struct cpuidle_ops *__init arm_cpuidle_get_ops(const char *method)
  * process.
  *
  * Return 0 on sucess, -ENOENT if no 'enable-method' is defined, -EOPNOTSUPP if
- * no cpuidle_ops is registered for the 'enable-method'.
+ * no cpuidle_ops is registered for the 'enable-method', or if no init callback
+ * is defined.
  */
 static int __init arm_cpuidle_read_ops(struct device_node *dn, int cpu)
 {

@@ -110,6 +111,12 @@ static int __init arm_cpuidle_read_ops(struct device_node *dn, int cpu)
 		return -EOPNOTSUPP;
 	}
 
+	if (!ops->init) {
+		pr_warn("cpuidle_ops '%s': no init callback\n",
+			enable_method);
+		return -EOPNOTSUPP;
+	}
+
 	cpuidle_ops[cpu] = *ops; /* structure copy */
 
 	pr_notice("cpuidle: enable-method property '%s'"

@@ -129,7 +136,8 @@ static int __init arm_cpuidle_read_ops(struct device_node *dn, int cpu)
  * Returns:
  *  0 on success,
  *  -ENODEV if it fails to find the cpu node in the device tree,
- *  -EOPNOTSUPP if it does not find a registered cpuidle_ops for this cpu,
+ *  -EOPNOTSUPP if it does not find a registered and valid cpuidle_ops for
+ *  this cpu,
  *  -ENOENT if it fails to find an 'enable-method' property,
  *  -ENXIO if the HW reports a failure or a misconfiguration,
  *  -ENOMEM if the HW report an memory allocation failure 

@@ -143,7 +151,7 @@ int __init arm_cpuidle_init(int cpu)
 		return -ENODEV;
 
 	ret = arm_cpuidle_read_ops(cpu_node, cpu);
-	if (!ret && cpuidle_ops[cpu].init)
+	if (!ret)
 		ret = cpuidle_ops[cpu].init(cpu_node, cpu);
 
 	of_node_put(cpu_node);

-- 
2.8.1

On Wed, Jul 06, 2016 at 02:15:54PM +0800, Jisheng Zhang wrote:

Let's assume cpuidle_ops exists but it doesn't implement the according
init callback, current arm_cpuidle_init() will return success to its
caller, but in fact it should return -EOPNOTSUPP.

Signed-off-by: Jisheng Zhang <redacted>
---

Acked-by: Daniel Lezcano <redacted>

On Wed, Jul 06, 2016 at 02:15:55PM +0800, Jisheng Zhang wrote:

Currently, we check cpuidle_ops.suspend every time when entering a
low-power idle state. But this check could be avoided in this hot path
by moving it into arm_cpuidle_read_ops() to reduce arm_cpuidle_suspend
overhead a bit.

Signed-off-by: Jisheng Zhang <redacted>
---

Acked-by: Daniel Lezcano <redacted>

Dear Daniel,

On Thu, 7 Jul 2016 15:46:28 +0200 Daniel Lezcano wrote:

On Wed, Jul 06, 2016 at 02:15:55PM +0800, Jisheng Zhang wrote:

quoted

Currently, we check cpuidle_ops.suspend every time when entering a
low-power idle state. But this check could be avoided in this hot path
by moving it into arm_cpuidle_read_ops() to reduce arm_cpuidle_suspend
overhead a bit.

Signed-off-by: Jisheng Zhang <redacted>
---  

Acked-by: Daniel Lezcano <redacted>

I'm not sure the upstream merging path this patch should follow, Per my
understanding, I need to put it into Russell's PATCH system.

Thanks in advance,
Jisheng

On Fri, Jul 08, 2016 at 02:17:29PM +0800, Jisheng Zhang wrote:

Dear Daniel,

On Thu, 7 Jul 2016 15:46:28 +0200 Daniel Lezcano wrote:

quoted

On Wed, Jul 06, 2016 at 02:15:55PM +0800, Jisheng Zhang wrote:

quoted

Currently, we check cpuidle_ops.suspend every time when entering a
low-power idle state. But this check could be avoided in this hot path
by moving it into arm_cpuidle_read_ops() to reduce arm_cpuidle_suspend
overhead a bit.

Signed-off-by: Jisheng Zhang <redacted>
---  

Acked-by: Daniel Lezcano <redacted>

I'm not sure the upstream merging path this patch should follow, Per my
understanding, I need to put it into Russell's PATCH system.

Or alternatively through arm-soc

  -- Daniel

On Fri, 8 Jul 2016 12:50:12 +0200 Daniel Lezcano  wrote:

On Fri, Jul 08, 2016 at 02:17:29PM +0800, Jisheng Zhang wrote:

quoted

Dear Daniel,

On Thu, 7 Jul 2016 15:46:28 +0200 Daniel Lezcano wrote:
  

quoted

On Wed, Jul 06, 2016 at 02:15:55PM +0800, Jisheng Zhang wrote:  

quoted

Currently, we check cpuidle_ops.suspend every time when entering a
low-power idle state. But this check could be avoided in this hot path
by moving it into arm_cpuidle_read_ops() to reduce arm_cpuidle_suspend
overhead a bit.

Signed-off-by: Jisheng Zhang <redacted>
---    

Acked-by: Daniel Lezcano <redacted>  

I'm not sure the upstream merging path this patch should follow, Per my
understanding, I need to put it into Russell's PATCH system.  

Or alternatively through arm-soc

Got it. thanks.

Dear Arnd, Olof,

I have no pull request permission. what's your preference? Could you please
advise?

Thanks

On Fri, Jul 08, 2016 at 06:58:54PM +0800, Jisheng Zhang wrote:

On Fri, 8 Jul 2016 12:50:12 +0200 Daniel Lezcano  wrote:

quoted

On Fri, Jul 08, 2016 at 02:17:29PM +0800, Jisheng Zhang wrote:

quoted

Dear Daniel,

On Thu, 7 Jul 2016 15:46:28 +0200 Daniel Lezcano wrote:
  

quoted

On Wed, Jul 06, 2016 at 02:15:55PM +0800, Jisheng Zhang wrote:  

quoted

Currently, we check cpuidle_ops.suspend every time when entering a
low-power idle state. But this check could be avoided in this hot path
by moving it into arm_cpuidle_read_ops() to reduce arm_cpuidle_suspend
overhead a bit.

Signed-off-by: Jisheng Zhang <redacted>
---    

Acked-by: Daniel Lezcano <redacted>  

I'm not sure the upstream merging path this patch should follow, Per my
understanding, I need to put it into Russell's PATCH system.  

Or alternatively through arm-soc

Got it. thanks.

Dear Arnd, Olof,

I have no pull request permission. what's your preference? Could you please
advise?

... and because arm-soc people haven't responded, they've now ended up
in the patch system... So I've applied them to my tree in a separate
branch.

Thanks!

-- 
RMK's Patch system: http://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line: currently at 9.6Mbps down 400kbps up
according to speedtest.net.

On Thu, 14 Jul 2016 16:31:13 +0100 Russell King - ARM Linux wrote:

On Fri, Jul 08, 2016 at 06:58:54PM +0800, Jisheng Zhang wrote:

quoted

On Fri, 8 Jul 2016 12:50:12 +0200 Daniel Lezcano  wrote:
  

quoted

On Fri, Jul 08, 2016 at 02:17:29PM +0800, Jisheng Zhang wrote:  

quoted

Dear Daniel,

On Thu, 7 Jul 2016 15:46:28 +0200 Daniel Lezcano wrote:
    

quoted

On Wed, Jul 06, 2016 at 02:15:55PM +0800, Jisheng Zhang wrote:    

quoted

Currently, we check cpuidle_ops.suspend every time when entering a
low-power idle state. But this check could be avoided in this hot path
by moving it into arm_cpuidle_read_ops() to reduce arm_cpuidle_suspend
overhead a bit.

Signed-off-by: Jisheng Zhang <redacted>
---      

Acked-by: Daniel Lezcano <redacted>    

I'm not sure the upstream merging path this patch should follow, Per my
understanding, I need to put it into Russell's PATCH system.    

Or alternatively through arm-soc
  

Got it. thanks.

Dear Arnd, Olof,

I have no pull request permission. what's your preference? Could you please
advise?  

... and because arm-soc people haven't responded, they've now ended up
in the patch system... So I've applied them to my tree in a separate
branch.

Thank you so much!