From mboxrd@z Thu Jan  1 00:00:00 1970
From: grygorii.strashko@ti.com (Grygorii Strashko)
Date: Thu, 13 Aug 2015 13:31:04 +0300
Subject: [PATCH v2 2/6] genirq: fix irqchip_set_wake_parent if
 IRQCHIP_SKIP_SET_WAKE
In-Reply-To: <55CC6AFF.8060400@arm.com>
References: <1439401562-28874-1-git-send-email-grygorii.strashko@ti.com>
 <1439401562-28874-3-git-send-email-grygorii.strashko@ti.com>
 <55CC6AFF.8060400@arm.com>
Message-ID: <55CC71E8.80604@ti.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

On 08/13/2015 01:01 PM, Marc Zyngier wrote:
> On 12/08/15 18:45, Grygorii Strashko wrote:
>> The irqchip_set_wake_parent should not fail if IRQ chip
>> specifies IRQCHIP_SKIP_SET_WAKE. Otherwise, IRQ wakeup
>> configuration can't be propagated properly through IRQ
>> domains hierarchy.
>>
>> In case of TI OMAP DRA7 the issue reproduced with following
>> configuration:
>> ARM GIC<-OMAP wakeupgen<-TI CBAR<-GPIO<-GPIO pcf857x<-gpio_key
>>
>> gpio_key is wakeup source
>>
>> Failure is reproduced during suspend/resume to RAM:
>> suspend:
>>   - gpio_keys_suspend
>>     enable_irq_wake
>>       + pcf857x_irq_set_wake
>>         + omap_gpio_wake_enable
>>           + TI CBAR irq_chip_set_wake_parent
>>             + OMAP wakeupgen has no .irq_set_wake()
>
> Most importantly, wakeupgen has IRQCHIP_SKIP_SET_WAKE set.
>
>>             and -ENOSYS will be returned
>>
>> resume:
>>   - gpio_keys_resume
>>     + disable_irq_wake
>>       + irq_set_irq_wake
>>         + WARN(1, "Unbalanced IRQ %d wake disable\n", irq);
>>
>> Fixes: 08b55e2a9208 ('genirq: Add irqchip_set_wake_parent')
>> Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com>
>> ---
>>   kernel/irq/chip.c | 4 ++++
>>   1 file changed, 4 insertions(+)
>>
>> diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c
>> index 6de638b..bdb1b9d 100644
>> --- a/kernel/irq/chip.c
>> +++ b/kernel/irq/chip.c
>> @@ -1024,6 +1024,10 @@ int irq_chip_set_vcpu_affinity_parent(struct irq_data *data, void *vcpu_info)
>>   int irq_chip_set_wake_parent(struct irq_data *data, unsigned int on)
>>   {
>>   	data = data->parent_data;
>> +
>> +	if (irq_data_get_irq_chip(data)->flags & IRQCHIP_SKIP_SET_WAKE)
>> +		return 0;
>> +
>>   	if (data->chip->irq_set_wake)
>>   		return data->chip->irq_set_wake(data, on);
>>
>>
>
> We have a more general issue with chip flags, and how they combine
> within a stack of irqchips.

Indeed. Problem looks similar to IRQCHIP_MASK_ON_SUSPEND flag usage.

>
> What if you remove the irq_chip_set_wake_parent from the crossbar
> driver, and instead set IRQCHIP_SKIP_SET_WAKE?

I've thought about this and it should work for me.
One question - what if crossbar will be not the last one in
IRQ domains hierarchy?


-- 
regards,
-grygorii