There's some confusion around if an irq that's disabled with
disable_irq() can still wake the system from sleep states such as
"suspend to RAM". Let's clarify this in the kernel documentation for
irq_set_irq_wake() so that it's clear that an irq can be disabled and
still wake the system if it has been marked for wakeup.
Cc: Marc Zyngier <[email protected]>
Cc: Douglas Anderson <[email protected]>
Cc: Lina Iyer <[email protected]>
Cc: Maulik Shah <[email protected]>
Signed-off-by: Stephen Boyd <[email protected]>
---
kernel/irq/manage.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c
index 818b2802d3e7..fa8db98c8699 100644
--- a/kernel/irq/manage.c
+++ b/kernel/irq/manage.c
@@ -731,6 +731,11 @@ static int set_irq_wake_real(unsigned int irq, unsigned int on)
*
* Wakeup mode lets this IRQ wake the system from sleep
* states like "suspend to RAM".
+ *
+ * Note: irq enable/disable state is completely orthogonal
+ * to the enable/disable state of irq wake. An irq can be
+ * disabled with disable_irq() and still wake the system as
+ * long as the irq has wake enabled.
*/
int irq_set_irq_wake(unsigned int irq, unsigned int on)
{
--
Sent by a computer, using git, on the internet
Stephen Boyd <[email protected]> writes:
> There's some confusion around if an irq that's disabled with
> disable_irq() can still wake the system from sleep states such as
> "suspend to RAM". Let's clarify this in the kernel documentation for
> irq_set_irq_wake() so that it's clear that an irq can be disabled and
> still wake the system if it has been marked for wakeup.
>
> Cc: Marc Zyngier <[email protected]>
> Cc: Douglas Anderson <[email protected]>
> Cc: Lina Iyer <[email protected]>
> Cc: Maulik Shah <[email protected]>
> Signed-off-by: Stephen Boyd <[email protected]>
> ---
> kernel/irq/manage.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c
> index 818b2802d3e7..fa8db98c8699 100644
> --- a/kernel/irq/manage.c
> +++ b/kernel/irq/manage.c
> @@ -731,6 +731,11 @@ static int set_irq_wake_real(unsigned int irq, unsigned int on)
> *
> * Wakeup mode lets this IRQ wake the system from sleep
> * states like "suspend to RAM".
> + *
> + * Note: irq enable/disable state is completely orthogonal
> + * to the enable/disable state of irq wake. An irq can be
> + * disabled with disable_irq() and still wake the system as
> + * long as the irq has wake enabled.
It clearly should say that this is really depending on the hardware
implementation of the particual interrupt chip whether disabled + wake
mode is supported.
Thanks,
tglx
On Wed, Feb 05 2020 at 05:27 -0700, Thomas Gleixner wrote:
>Stephen Boyd <[email protected]> writes:
>> There's some confusion around if an irq that's disabled with
>> disable_irq() can still wake the system from sleep states such as
>> "suspend to RAM". Let's clarify this in the kernel documentation for
>> irq_set_irq_wake() so that it's clear that an irq can be disabled and
>> still wake the system if it has been marked for wakeup.
>>
>> Cc: Marc Zyngier <[email protected]>
>> Cc: Douglas Anderson <[email protected]>
>> Cc: Lina Iyer <[email protected]>
>> Cc: Maulik Shah <[email protected]>
>> Signed-off-by: Stephen Boyd <[email protected]>
>> ---
>> kernel/irq/manage.c | 5 +++++
>> 1 file changed, 5 insertions(+)
>>
>> diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c
>> index 818b2802d3e7..fa8db98c8699 100644
>> --- a/kernel/irq/manage.c
>> +++ b/kernel/irq/manage.c
>> @@ -731,6 +731,11 @@ static int set_irq_wake_real(unsigned int irq, unsigned int on)
>> *
>> * Wakeup mode lets this IRQ wake the system from sleep
>> * states like "suspend to RAM".
>> + *
>> + * Note: irq enable/disable state is completely orthogonal
>> + * to the enable/disable state of irq wake. An irq can be
>> + * disabled with disable_irq() and still wake the system as
>> + * long as the irq has wake enabled.
>
>It clearly should say that this is really depending on the hardware
>implementation of the particual interrupt chip whether disabled + wake
>mode is supported.
>
Could an irqchip flag be used to warn users that we may not wakeup from
suspend/resume if the interrupt if the hardware does not support wakeup
when disabled ?
--Lina
Lina Iyer <[email protected]> writes:
> On Wed, Feb 05 2020 at 05:27 -0700, Thomas Gleixner wrote:
>>> @@ -731,6 +731,11 @@ static int set_irq_wake_real(unsigned int irq, unsigned int on)
>>> *
>>> * Wakeup mode lets this IRQ wake the system from sleep
>>> * states like "suspend to RAM".
>>> + *
>>> + * Note: irq enable/disable state is completely orthogonal
>>> + * to the enable/disable state of irq wake. An irq can be
>>> + * disabled with disable_irq() and still wake the system as
>>> + * long as the irq has wake enabled.
>>
>>It clearly should say that this is really depending on the hardware
>>implementation of the particual interrupt chip whether disabled + wake
>>mode is supported.
>>
> Could an irqchip flag be used to warn users that we may not wakeup from
> suspend/resume if the interrupt if the hardware does not support wakeup
> when disabled ?
There are also magic ways of wakeup for irqchips which do not have wake
setup functions and still wake the system up when the interrupt line is
disabled by the kernel on suspend. :)
Thanks,
tglx
Quoting Thomas Gleixner (2020-02-05 04:27:06)
> Stephen Boyd <[email protected]> writes:
> > There's some confusion around if an irq that's disabled with
> > disable_irq() can still wake the system from sleep states such as
> > "suspend to RAM". Let's clarify this in the kernel documentation for
> > irq_set_irq_wake() so that it's clear that an irq can be disabled and
> > still wake the system if it has been marked for wakeup.
> >
> > Cc: Marc Zyngier <[email protected]>
> > Cc: Douglas Anderson <[email protected]>
> > Cc: Lina Iyer <[email protected]>
> > Cc: Maulik Shah <[email protected]>
> > Signed-off-by: Stephen Boyd <[email protected]>
> > ---
> > kernel/irq/manage.c | 5 +++++
> > 1 file changed, 5 insertions(+)
> >
> > diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c
> > index 818b2802d3e7..fa8db98c8699 100644
> > --- a/kernel/irq/manage.c
> > +++ b/kernel/irq/manage.c
> > @@ -731,6 +731,11 @@ static int set_irq_wake_real(unsigned int irq, unsigned int on)
> > *
> > * Wakeup mode lets this IRQ wake the system from sleep
> > * states like "suspend to RAM".
> > + *
> > + * Note: irq enable/disable state is completely orthogonal
> > + * to the enable/disable state of irq wake. An irq can be
> > + * disabled with disable_irq() and still wake the system as
> > + * long as the irq has wake enabled.
>
> It clearly should say that this is really depending on the hardware
> implementation of the particual interrupt chip whether disabled + wake
> mode is supported.
>
Ok. I'm having trouble parsing this. Is there a consistent wording that
can be put here?
The API seems fraught with peril if an implementation of an irqchip is
allowed to ignore wakeup on interrupts that are marked for wakeup while
the irq is disabled. Driver writers won't be able to write drivers that
work across implementations if the irq can't wake the system reliably.
Stephen Boyd <[email protected]> writes:
> Quoting Thomas Gleixner (2020-02-05 04:27:06)
>> > * Wakeup mode lets this IRQ wake the system from sleep
>> > * states like "suspend to RAM".
>> > + *
>> > + * Note: irq enable/disable state is completely orthogonal
>> > + * to the enable/disable state of irq wake. An irq can be
>> > + * disabled with disable_irq() and still wake the system as
>> > + * long as the irq has wake enabled.
>>
>> It clearly should say that this is really depending on the hardware
>> implementation of the particual interrupt chip whether disabled + wake
>> mode is supported.
>>
>
> Ok. I'm having trouble parsing this. Is there a consistent wording that
> can be put here?
See below.
> The API seems fraught with peril if an implementation of an irqchip is
> allowed to ignore wakeup on interrupts that are marked for wakeup while
> the irq is disabled. Driver writers won't be able to write drivers that
> work across implementations if the irq can't wake the system reliably.
It's not really well defined but thats a result of the gazillion
variants of irq chips which all have their own quirks. The wakeup
mechansims are also widely different, some of them are built into the
SOC, others require external logic. And a large part of these things is
completely undocumented. Welcome to my wonderful world.
So versus consistent wording. I'm fine with the paragraph you suggested,
but please amend it with something like this:
If this does not hold, then either the underlying irq chip and the
related driver need to be investigated.
Thanks,
tglx