[PATCH] irqchip/gic-v3: Fix gic_chip specific flag

[PATCH] irqchip/gic-v3: Fix gic_chip specific flag

[fangxiang]

Both IRQCHIP_SKIP_SET_WAKE and .irq_set_wake() is
not set in gic_chip V3. This causes
set_irq_wake_real() return -ENXIO and a
WARN(Unbalanced IRQ ... wake disable ...) in
irq_set_irq_wake().

Signed-off-by: fangxiang <fangxiang3@huawei.com>
---
 drivers/irqchip/irq-gic-v3.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
index b56c3e2..e8104b8 100644
--- a/drivers/irqchip/irq-gic-v3.c
+++ b/drivers/irqchip/irq-gic-v3.c
@@ -800,7 +800,8 @@ static inline void gic_cpu_pm_init(void) { }
 	.irq_set_affinity	= gic_set_affinity,
 	.irq_get_irqchip_state	= gic_irq_get_irqchip_state,
 	.irq_set_irqchip_state	= gic_irq_set_irqchip_state,
-	.flags			= IRQCHIP_SET_TYPE_MASKED,
+	.flags			= IRQCHIP_SET_TYPE_MASKED |
+				  IRQCHIP_SKIP_SET_WAKE,
 };
 
 static struct irq_chip gic_eoimode1_chip = {
-- 
1.7.9.5

Re: [PATCH] irqchip/gic-v3: Fix gic_chip specific flag

[Marc Zyngier]

On 30/01/18 02:56, fangxiang wrote:
> Both IRQCHIP_SKIP_SET_WAKE and .irq_set_wake() is
> not set in gic_chip V3. This causes
> set_irq_wake_real() return -ENXIO and a
> WARN(Unbalanced IRQ ... wake disable ...) in
> irq_set_irq_wake().
> 
> Signed-off-by: fangxiang <fangxiang3@huawei.com>
> ---
>  drivers/irqchip/irq-gic-v3.c |    3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
> index b56c3e2..e8104b8 100644
> --- a/drivers/irqchip/irq-gic-v3.c
> +++ b/drivers/irqchip/irq-gic-v3.c
> @@ -800,7 +800,8 @@ static inline void gic_cpu_pm_init(void) { }
>  	.irq_set_affinity	= gic_set_affinity,
>  	.irq_get_irqchip_state	= gic_irq_get_irqchip_state,
>  	.irq_set_irqchip_state	= gic_irq_set_irqchip_state,
> -	.flags			= IRQCHIP_SET_TYPE_MASKED,
> +	.flags			= IRQCHIP_SET_TYPE_MASKED |
> +				  IRQCHIP_SKIP_SET_WAKE,
>  };
>  
>  static struct irq_chip gic_eoimode1_chip = {
> 

That's sort of expected (the GIC itself doesn't implement any wake-up
mechanism), *unless* you have have a separate, stacked irqchip
implementing irq_set_wake. From your commit message, I cannot deduce
whether that's the case or not.

Please describe what you are fixing, and to what context this applies to.

Thanks,

	M.
-- 
Jazz is not dead. It just smells funny...