[PATCH 0/2] irqchip: renesas: Propagate wake-up settings to parent

[PATCH 0/2] irqchip: renesas: Propagate wake-up settings to parent

[Geert Uytterhoeven]

	Hi Thomas, Jason, Marc,

The renesas-intc-irqpin and renesas-irqc interrupt controllers are
cascaded to GICs, but their drivers don't propagate wake-up settings to
their parent interrupt controllers.

Since commit aec89ef72ba6c944 ("irqchip/gic: Enable SKIP_SET_WAKE and
MASK_ON_SUSPEND"), the GIC driver masks interrupts during suspend, and
wake-up through gpio-keys now fails on r8a73a4/ape6evm,
r8a7740/armadillo and sh73a0/kzm9g.

Fix this by propagating wake-up settings to the parent interrupt
controllers. There's no need to handle irq_set_irq_wake() failures, as
the renesas-intc-irqpin and renesas-irqc interrupt controllers are
always cascaded to GICs, and the GIC driver always sets SKIP_SET_WAKE
since the aforementioned commit.

These fixes are meant for v4.3, as commit aec89ef72ba6c944 is already
upstream.

Note that masking GIC interrupts during suspend has another side effect:
before, if no_console_suspend was enabled, the system woke up on console
activity. This is no longer the case, but that's not a bug.

Thanks!

Geert Uytterhoeven (2):
  irqchip: renesas-intc-irqpin: Propagate wake-up settings to parent
  irqchip: renesas-irqc: Propagate wake-up settings to parent

 drivers/irqchip/irq-renesas-intc-irqpin.c | 3 +++
 drivers/irqchip/irq-renesas-irqc.c        | 3 +++
 2 files changed, 6 insertions(+)

-- 
1.9.1

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds

[PATCH 1/2] irqchip: renesas-intc-irqpin: Propagate wake-up settings to parent

[Geert Uytterhoeven]

The renesas-intc-irqpin interrupt controller is cascaded to the GIC, but
its driver doesn't propagate wake-up settings to the parent interrupt
controller.

Since commit aec89ef72ba6c944 ("irqchip/gic: Enable SKIP_SET_WAKE and
MASK_ON_SUSPEND"), the GIC driver masks interrupts during suspend, and
wake-up through gpio-keys now fails on r8a7740/armadillo and
sh73a0/kzm9g.

Fix this by propagating wake-up settings to the parent interrupt
controller. There's no need to handle irq_set_irq_wake() failures, as
the renesas-intc-irqpin interrupt controller is always cascaded to a
GIC, and the GIC driver always sets SKIP_SET_WAKE since the
aforementioned commit.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
---
 drivers/irqchip/irq-renesas-intc-irqpin.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/irqchip/irq-renesas-intc-irqpin.c b/drivers/irqchip/irq-renesas-intc-irqpin.c
index 0670ab4e3897bf61..457d22c09e352d56 100644
--- a/drivers/irqchip/irq-renesas-intc-irqpin.c
+++ b/drivers/irqchip/irq-renesas-intc-irqpin.c
@@ -283,6 +283,9 @@ static int intc_irqpin_irq_set_type(struct irq_data *d, unsigned int type)
 static int intc_irqpin_irq_set_wake(struct irq_data *d, unsigned int on)
 {
 	struct intc_irqpin_priv *p = irq_data_get_irq_chip_data(d);
+	int hw_irq = irqd_to_hwirq(d);
+
+	irq_set_irq_wake(p->irq[hw_irq].requested_irq, on);
 
 	if (!p->clk)
 		return 0;
-- 
1.9.1

[PATCH 2/2] irqchip: renesas-irqc: Propagate wake-up settings to parent

[Geert Uytterhoeven]

The renesas-irqc interrupt controller is cascaded to the GIC, but its
driver doesn't propagate wake-up settings to the parent interrupt
controller.

Since commit aec89ef72ba6c944 ("irqchip/gic: Enable SKIP_SET_WAKE and
MASK_ON_SUSPEND"), the GIC driver masks interrupts during suspend, and
wake-up through gpio-keys now fails on r8a73a4/ape6evm.

Fix this by propagating wake-up settings to the parent interrupt
controller. There's no need to handle irq_set_irq_wake() failures, as
the renesas-irqc interrupt controller is always cascaded to a GIC, and
the GIC driver always sets SKIP_SET_WAKE since the aforementioned
commit.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
---
 drivers/irqchip/irq-renesas-irqc.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/irqchip/irq-renesas-irqc.c b/drivers/irqchip/irq-renesas-irqc.c
index 2aa3add711a6612a..49f6c41d94f91d4d 100644
--- a/drivers/irqchip/irq-renesas-irqc.c
+++ b/drivers/irqchip/irq-renesas-irqc.c
@@ -121,6 +121,9 @@ static int irqc_irq_set_type(struct irq_data *d, unsigned int type)
 static int irqc_irq_set_wake(struct irq_data *d, unsigned int on)
 {
 	struct irqc_priv *p = irq_data_get_irq_chip_data(d);
+	int hw_irq = irqd_to_hwirq(d);
+
+	irq_set_irq_wake(p->irq[hw_irq].requested_irq, on);
 
 	if (!p->clk)
 		return 0;
-- 
1.9.1

Re: [PATCH 0/2] irqchip: renesas: Propagate wake-up settings to parent

[Sudeep Holla]

On 08/09/15 18:00, Geert Uytterhoeven wrote:
> 	Hi Thomas, Jason, Marc,
>
> The renesas-intc-irqpin and renesas-irqc interrupt controllers are
> cascaded to GICs, but their drivers don't propagate wake-up settings to
> their parent interrupt controllers.
>
> Since commit aec89ef72ba6c944 ("irqchip/gic: Enable SKIP_SET_WAKE and
> MASK_ON_SUSPEND"), the GIC driver masks interrupts during suspend, and
> wake-up through gpio-keys now fails on r8a73a4/ape6evm,
> r8a7740/armadillo and sh73a0/kzm9g.
>
> Fix this by propagating wake-up settings to the parent interrupt
> controllers. There's no need to handle irq_set_irq_wake() failures, as
> the renesas-intc-irqpin and renesas-irqc interrupt controllers are
> always cascaded to GICs, and the GIC driver always sets SKIP_SET_WAKE
> since the aforementioned commit.
>

Ah so, we have the first culprit. I expect many more to come, as I had
seen many drivers didn't bother to enable wakeup source correctly.

> These fixes are meant for v4.3, as commit aec89ef72ba6c944 is already
> upstream.
>

Looks fine to me, so for the series:
Acked-by: Sudeep Holla <sudeep.holla@arm.com>

> Note that masking GIC interrupts during suspend has another side effect:
> before, if no_console_suspend was enabled, the system woke up on console
> activity. This is no longer the case, but that's not a bug.
>

It's very much similar to IRQF_NO_SUSPEND flag which keeps the IRQ
enabled but doesn't guarantee wakeup. no_console_suspend doesn't suspend
console but doesn't mean it can wake up. I expected this to come up :)
IIUC, it was added to help in debugging the suspend path.

Regards,
Sudeep

Re: [PATCH 1/2] irqchip: renesas-intc-irqpin: Propagate wake-up settings to parent

[Thomas Gleixner]

On Tue, 8 Sep 2015, Geert Uytterhoeven wrote:
> The renesas-intc-irqpin interrupt controller is cascaded to the GIC, but
> its driver doesn't propagate wake-up settings to the parent interrupt
> controller.
> 
> Since commit aec89ef72ba6c944 ("irqchip/gic: Enable SKIP_SET_WAKE and
> MASK_ON_SUSPEND"), the GIC driver masks interrupts during suspend, and
> wake-up through gpio-keys now fails on r8a7740/armadillo and
> sh73a0/kzm9g.
> 
> Fix this by propagating wake-up settings to the parent interrupt
> controller. There's no need to handle irq_set_irq_wake() failures, as
> the renesas-intc-irqpin interrupt controller is always cascaded to a
> GIC, and the GIC driver always sets SKIP_SET_WAKE since the
> aforementioned commit.
> 
> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
> ---
>  drivers/irqchip/irq-renesas-intc-irqpin.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/irqchip/irq-renesas-intc-irqpin.c b/drivers/irqchip/irq-renesas-intc-irqpin.c
> index 0670ab4e3897bf61..457d22c09e352d56 100644
> --- a/drivers/irqchip/irq-renesas-intc-irqpin.c
> +++ b/drivers/irqchip/irq-renesas-intc-irqpin.c
> @@ -283,6 +283,9 @@ static int intc_irqpin_irq_set_type(struct irq_data *d, unsigned int type)
>  static int intc_irqpin_irq_set_wake(struct irq_data *d, unsigned int on)
>  {
>  	struct intc_irqpin_priv *p = irq_data_get_irq_chip_data(d);
> +	int hw_irq = irqd_to_hwirq(d);
> +
> +	irq_set_irq_wake(p->irq[hw_irq].requested_irq, on);

Are you sure that this does not make lockdep unhappy due to lock
nesting?

Thanks,

	tglx

Re: [PATCH 1/2] irqchip: renesas-intc-irqpin: Propagate wake-up settings to parent

[Geert Uytterhoeven]

Hi Thomas,

On Tue, Sep 8, 2015 at 9:05 PM, Thomas Gleixner <tglx@linutronix.de> wrote:
> On Tue, 8 Sep 2015, Geert Uytterhoeven wrote:
>> The renesas-intc-irqpin interrupt controller is cascaded to the GIC, but
>> its driver doesn't propagate wake-up settings to the parent interrupt
>> controller.
>>
>> Since commit aec89ef72ba6c944 ("irqchip/gic: Enable SKIP_SET_WAKE and
>> MASK_ON_SUSPEND"), the GIC driver masks interrupts during suspend, and
>> wake-up through gpio-keys now fails on r8a7740/armadillo and
>> sh73a0/kzm9g.
>>
>> Fix this by propagating wake-up settings to the parent interrupt
>> controller. There's no need to handle irq_set_irq_wake() failures, as
>> the renesas-intc-irqpin interrupt controller is always cascaded to a
>> GIC, and the GIC driver always sets SKIP_SET_WAKE since the
>> aforementioned commit.
>>
>> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
>> ---
>>  drivers/irqchip/irq-renesas-intc-irqpin.c | 3 +++
>>  1 file changed, 3 insertions(+)
>>
>> diff --git a/drivers/irqchip/irq-renesas-intc-irqpin.c b/drivers/irqchip/irq-renesas-intc-irqpin.c
>> index 0670ab4e3897bf61..457d22c09e352d56 100644
>> --- a/drivers/irqchip/irq-renesas-intc-irqpin.c
>> +++ b/drivers/irqchip/irq-renesas-intc-irqpin.c
>> @@ -283,6 +283,9 @@ static int intc_irqpin_irq_set_type(struct irq_data *d, unsigned int type)
>>  static int intc_irqpin_irq_set_wake(struct irq_data *d, unsigned int on)
>>  {
>>       struct intc_irqpin_priv *p = irq_data_get_irq_chip_data(d);
>> +     int hw_irq = irqd_to_hwirq(d);
>> +
>> +     irq_set_irq_wake(p->irq[hw_irq].requested_irq, on);
>
> Are you sure that this does not make lockdep unhappy due to lock
> nesting?

Actually I did see one lockdep warning, so I'm aware we're probably gonna
need a similar solution like commit a0a8bcf4670c2c69 ("gpiolib: irqchip:
use different lockdep class for each gpio irqchip")?

To be honest, these lockdep warnings aren't helping much here: on embedded
we typical have several stacked interrupt controllers, so wake-up settings
have to propagate from the bottom to the top of the stack.

E.g. on sh73a0/kzm9g gpio-keys wake-up goes through 3 interrupt controllers:
pcf875x -> renesas-intc-irqpin -> gic.

Thanks!

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH 1/2] irqchip: renesas-intc-irqpin: Propagate wake-up settings to parent

[Thomas Gleixner]

On Tue, 8 Sep 2015, Geert Uytterhoeven wrote:
> >> --- a/drivers/irqchip/irq-renesas-intc-irqpin.c
> >> +++ b/drivers/irqchip/irq-renesas-intc-irqpin.c
> >> @@ -283,6 +283,9 @@ static int intc_irqpin_irq_set_type(struct irq_data *d, unsigned int type)
> >>  static int intc_irqpin_irq_set_wake(struct irq_data *d, unsigned int on)
> >>  {
> >>       struct intc_irqpin_priv *p = irq_data_get_irq_chip_data(d);
> >> +     int hw_irq = irqd_to_hwirq(d);
> >> +
> >> +     irq_set_irq_wake(p->irq[hw_irq].requested_irq, on);
> >
> > Are you sure that this does not make lockdep unhappy due to lock
> > nesting?
> 
> Actually I did see one lockdep warning, so I'm aware we're probably gonna
> need a similar solution like commit a0a8bcf4670c2c69 ("gpiolib: irqchip:
> use different lockdep class for each gpio irqchip")?
> 
> To be honest, these lockdep warnings aren't helping much here: on embedded
> we typical have several stacked interrupt controllers, so wake-up settings
> have to propagate from the bottom to the top of the stack.

But ignoring them does not help much either, right?
 
> E.g. on sh73a0/kzm9g gpio-keys wake-up goes through 3 interrupt controllers:
> pcf875x -> renesas-intc-irqpin -> gic.

So, yes a seperate locking class for that intc trainwreck is probably
required.

Thanks,

	tglx

Re: [PATCH 1/2] irqchip: renesas-intc-irqpin: Propagate wake-up settings to parent

[Grygorii Strashko]

Hi,

On 09/08/2015 11:03 PM, Thomas Gleixner wrote:
> On Tue, 8 Sep 2015, Geert Uytterhoeven wrote:
>>>> --- a/drivers/irqchip/irq-renesas-intc-irqpin.c
>>>> +++ b/drivers/irqchip/irq-renesas-intc-irqpin.c
>>>> @@ -283,6 +283,9 @@ static int intc_irqpin_irq_set_type(struct irq_data *d, unsigned int type)
>>>>   static int intc_irqpin_irq_set_wake(struct irq_data *d, unsigned int on)
>>>>   {
>>>>        struct intc_irqpin_priv *p = irq_data_get_irq_chip_data(d);
>>>> +     int hw_irq = irqd_to_hwirq(d);
>>>> +
>>>> +     irq_set_irq_wake(p->irq[hw_irq].requested_irq, on);
>>>
>>> Are you sure that this does not make lockdep unhappy due to lock
>>> nesting?
>>
>> Actually I did see one lockdep warning, so I'm aware we're probably gonna
>> need a similar solution like commit a0a8bcf4670c2c69 ("gpiolib: irqchip:
>> use different lockdep class for each gpio irqchip")?
>>
>> To be honest, these lockdep warnings aren't helping much here: on embedded
>> we typical have several stacked interrupt controllers, so wake-up settings
>> have to propagate from the bottom to the top of the stack.
> 
> But ignoring them does not help much either, right?
>   
>> E.g. on sh73a0/kzm9g gpio-keys wake-up goes through 3 interrupt controllers:
>> pcf875x -> renesas-intc-irqpin -> gic.
> 
> So, yes a seperate locking class for that intc trainwreck is probably
> required.
> 

Just as an option, May be we can proceed with patch:
[PATCH v2 2/6] genirq: fix irqchip_set_wake_parent if IRQCHIP_SKIP_SET_WAKE
http://www.spinics.net/lists/linux-omap/msg121262.html

As result, irq_chip_set_wake_parent() can be used here and no lockdep issues.

Marc, what do you think? renesas-intc-irqpin has own .irq_set_wake()
implementation and IRQCHIP_SKIP_SET_WAKE can't be used here as W/A.


-- 
regards,
-grygorii

Re: [PATCH 1/2] irqchip: renesas-intc-irqpin: Propagate wake-up settings to parent

[Thomas Gleixner]

On Wed, 9 Sep 2015, Grygorii Strashko wrote:
> On 09/08/2015 11:03 PM, Thomas Gleixner wrote:
> > So, yes a seperate locking class for that intc trainwreck is probably
> > required.
> > 
> 
> Just as an option, May be we can proceed with patch:
> [PATCH v2 2/6] genirq: fix irqchip_set_wake_parent if IRQCHIP_SKIP_SET_WAKE
> http://www.spinics.net/lists/linux-omap/msg121262.html
> 
> As result, irq_chip_set_wake_parent() can be used here and no lockdep issues.

The driver in question is not using hierarchical irq domains. So what
would that patch solve?

Thanks,

	tglx

Re: [PATCH 1/2] irqchip: renesas-intc-irqpin: Propagate wake-up settings to parent

[Grygorii Strashko]

On 09/09/2015 11:53 AM, Thomas Gleixner wrote:
> On Wed, 9 Sep 2015, Grygorii Strashko wrote:
>> On 09/08/2015 11:03 PM, Thomas Gleixner wrote:
>>> So, yes a seperate locking class for that intc trainwreck is probably
>>> required.
>>>
>>
>> Just as an option, May be we can proceed with patch:
>> [PATCH v2 2/6] genirq: fix irqchip_set_wake_parent if IRQCHIP_SKIP_SET_WAKE
>> http://www.spinics.net/lists/linux-omap/msg121262.html
>>
>> As result, irq_chip_set_wake_parent() can be used here and no lockdep issues.
>
> The driver in question is not using hierarchical irq domains. So what
> would that patch solve?
>

You are right.
I misunderstood it. Sorry for the noise.

-- 
regards,
-grygorii