[PATCH] i2c: ocores: Move system PM hooks to the NOIRQ phase

[PATCH] i2c: ocores: Move system PM hooks to the NOIRQ phase

[Samuel Holland]

When an I2C device contains a wake IRQ subordinate to a regmap-irq chip,
the regmap-irq code must be able to perform I2C transactions during
suspend_device_irqs() and resume_device_irqs(). Therefore, the bus must
be suspended/resumed during the NOIRQ phase.

Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
---

 drivers/i2c/busses/i2c-ocores.c | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

diff --git a/drivers/i2c/busses/i2c-ocores.c b/drivers/i2c/busses/i2c-ocores.c
index 4ac77e57bbbf..b1f621d42910 100644
--- a/drivers/i2c/busses/i2c-ocores.c
+++ b/drivers/i2c/busses/i2c-ocores.c
@@ -743,7 +743,6 @@ static void ocores_i2c_remove(struct platform_device *pdev)
 	i2c_del_adapter(&i2c->adap);
 }
 
-#ifdef CONFIG_PM_SLEEP
 static int ocores_i2c_suspend(struct device *dev)
 {
 	struct ocores_i2c *i2c = dev_get_drvdata(dev);
@@ -772,11 +771,9 @@ static int ocores_i2c_resume(struct device *dev)
 	return ocores_init(dev, i2c);
 }
 
-static SIMPLE_DEV_PM_OPS(ocores_i2c_pm, ocores_i2c_suspend, ocores_i2c_resume);
-#define OCORES_I2C_PM	(&ocores_i2c_pm)
-#else
-#define OCORES_I2C_PM	NULL
-#endif
+static const struct dev_pm_ops ocores_i2c_pm_ops = {
+	NOIRQ_SYSTEM_SLEEP_PM_OPS(ocores_i2c_suspend, ocores_i2c_resume)
+};
 
 static struct platform_driver ocores_i2c_driver = {
 	.probe   = ocores_i2c_probe,
@@ -784,7 +781,7 @@ static struct platform_driver ocores_i2c_driver = {
 	.driver  = {
 		.name = "ocores-i2c",
 		.of_match_table = ocores_i2c_match,
-		.pm = OCORES_I2C_PM,
+		.pm = &ocores_i2c_pm_ops,
 	},
 };
 
-- 
2.40.1

Re: [PATCH] i2c: ocores: Move system PM hooks to the NOIRQ phase

[Andi Shyti]

Hi Samuel,

On Mon, Jul 17, 2023 at 01:38:57PM -0700, Samuel Holland wrote:
> When an I2C device contains a wake IRQ subordinate to a regmap-irq chip,
> the regmap-irq code must be able to perform I2C transactions during
> suspend_device_irqs() and resume_device_irqs(). Therefore, the bus must
> be suspended/resumed during the NOIRQ phase.
> 
> Signed-off-by: Samuel Holland <samuel.holland@sifive.com>

it's OK for me.

Peter, any comment on this?

Andi

> ---
> 
>  drivers/i2c/busses/i2c-ocores.c | 11 ++++-------
>  1 file changed, 4 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/i2c/busses/i2c-ocores.c b/drivers/i2c/busses/i2c-ocores.c
> index 4ac77e57bbbf..b1f621d42910 100644
> --- a/drivers/i2c/busses/i2c-ocores.c
> +++ b/drivers/i2c/busses/i2c-ocores.c
> @@ -743,7 +743,6 @@ static void ocores_i2c_remove(struct platform_device *pdev)
>  	i2c_del_adapter(&i2c->adap);
>  }
>  
> -#ifdef CONFIG_PM_SLEEP
>  static int ocores_i2c_suspend(struct device *dev)
>  {
>  	struct ocores_i2c *i2c = dev_get_drvdata(dev);
> @@ -772,11 +771,9 @@ static int ocores_i2c_resume(struct device *dev)
>  	return ocores_init(dev, i2c);
>  }
>  
> -static SIMPLE_DEV_PM_OPS(ocores_i2c_pm, ocores_i2c_suspend, ocores_i2c_resume);
> -#define OCORES_I2C_PM	(&ocores_i2c_pm)
> -#else
> -#define OCORES_I2C_PM	NULL
> -#endif
> +static const struct dev_pm_ops ocores_i2c_pm_ops = {
> +	NOIRQ_SYSTEM_SLEEP_PM_OPS(ocores_i2c_suspend, ocores_i2c_resume)
> +};
>  
>  static struct platform_driver ocores_i2c_driver = {
>  	.probe   = ocores_i2c_probe,
> @@ -784,7 +781,7 @@ static struct platform_driver ocores_i2c_driver = {
>  	.driver  = {
>  		.name = "ocores-i2c",
>  		.of_match_table = ocores_i2c_match,
> -		.pm = OCORES_I2C_PM,
> +		.pm = &ocores_i2c_pm_ops,
>  	},
>  };
>  
> -- 
> 2.40.1
> 

Re: [PATCH] i2c: ocores: Move system PM hooks to the NOIRQ phase

[Andi Shyti]

Hi,

On Thu, Jul 27, 2023 at 02:35:51AM +0200, Andi Shyti wrote:
> Hi Samuel,
> 
> On Mon, Jul 17, 2023 at 01:38:57PM -0700, Samuel Holland wrote:
> > When an I2C device contains a wake IRQ subordinate to a regmap-irq chip,
> > the regmap-irq code must be able to perform I2C transactions during
> > suspend_device_irqs() and resume_device_irqs(). Therefore, the bus must
> > be suspended/resumed during the NOIRQ phase.
> > 
> > Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
> 
> it's OK for me.
> 
> Peter, any comment on this?

a kind ping...

Andi

Re: [PATCH] i2c: ocores: Move system PM hooks to the NOIRQ phase

[Andi Shyti]

Hi Samuel,

On Mon, Jul 17, 2023 at 01:38:57PM -0700, Samuel Holland wrote:
> When an I2C device contains a wake IRQ subordinate to a regmap-irq chip,
> the regmap-irq code must be able to perform I2C transactions during
> suspend_device_irqs() and resume_device_irqs(). Therefore, the bus must
> be suspended/resumed during the NOIRQ phase.
> 
> Signed-off-by: Samuel Holland <samuel.holland@sifive.com>

I think this patch has failed to receive some comments, I'll go
ahead and give it my blessing:

Reviewed-by: Andi Shyti <andi.shyti@kernel.org>

Thanks,
Andi

Re: [PATCH] i2c: ocores: Move system PM hooks to the NOIRQ phase

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 761 bytes --]


> > When an I2C device contains a wake IRQ subordinate to a regmap-irq chip,
> > the regmap-irq code must be able to perform I2C transactions during
> > suspend_device_irqs() and resume_device_irqs(). Therefore, the bus must
> > be suspended/resumed during the NOIRQ phase.
> > 
> > Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
> 
> I think this patch has failed to receive some comments, I'll go
> ahead and give it my blessing:
> 
> Reviewed-by: Andi Shyti <andi.shyti@kernel.org>

It needs to be rebased on top of 0ad93449b043 ("i2c: ocores: Remove
#ifdef guards for PM related functions"), though. This series did clean
this up for the whole subsystem. It also introduced using pm_sleep_ptr
which probably was missing here?


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]