[PATCH] ata: libahci: properly propagate return value of platform_get_irq()

[PATCH] ata: libahci: properly propagate return value of platform_get_irq()

[Thomas Petazzoni]

When platform_get_irq() fails, it returns an error code, which
libahci_platform and replaces it by -EINVAL. This commit fixes that by
propagating the error code. It fixes the situation where
platform_get_irq() returns -EPROBE_DEFER because the interrupt
controller is not available yet, and generally looks like the right
thing to do.

We pay attention to not show the "no irq" message when we are in an
EPROBE_DEFER situation, because the driver probing will be retried
later on, once the interrupt controller becomes available to provide
the interrupt.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
---
 drivers/ata/libahci_platform.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/ata/libahci_platform.c b/drivers/ata/libahci_platform.c
index aaa761b..cd2eab6 100644
--- a/drivers/ata/libahci_platform.c
+++ b/drivers/ata/libahci_platform.c
@@ -514,8 +514,9 @@ int ahci_platform_init_host(struct platform_device *pdev,
 
 	irq = platform_get_irq(pdev, 0);
 	if (irq <= 0) {
-		dev_err(dev, "no irq\n");
-		return -EINVAL;
+		if (irq != -EPROBE_DEFER)
+			dev_err(dev, "no irq\n");
+		return irq;
 	}
 
 	hpriv->irq = irq;
-- 
2.7.4

Re: [PATCH] ata: libahci: properly propagate return value of platform_get_irq()

[Hans de Goede]

Hi,

On 16-05-17 14:06, Thomas Petazzoni wrote:
> When platform_get_irq() fails, it returns an error code, which
> libahci_platform and replaces it by -EINVAL. This commit fixes that by
> propagating the error code. It fixes the situation where
> platform_get_irq() returns -EPROBE_DEFER because the interrupt
> controller is not available yet, and generally looks like the right
> thing to do.
> 
> We pay attention to not show the "no irq" message when we are in an
> EPROBE_DEFER situation, because the driver probing will be retried
> later on, once the interrupt controller becomes available to provide
> the interrupt.
> 
> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Looks good to me:

Reviewed-by: Hans de Goede <hdegoede@redhat.com>

Regards,

Hans


> ---
>   drivers/ata/libahci_platform.c | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/ata/libahci_platform.c b/drivers/ata/libahci_platform.c
> index aaa761b..cd2eab6 100644
> --- a/drivers/ata/libahci_platform.c
> +++ b/drivers/ata/libahci_platform.c
> @@ -514,8 +514,9 @@ int ahci_platform_init_host(struct platform_device *pdev,
>   
>   	irq = platform_get_irq(pdev, 0);
>   	if (irq <= 0) {
> -		dev_err(dev, "no irq\n");
> -		return -EINVAL;
> +		if (irq != -EPROBE_DEFER)
> +			dev_err(dev, "no irq\n");
> +		return irq;
>   	}
>   
>   	hpriv->irq = irq;
> 

Re: [PATCH] ata: libahci: properly propagate return value of platform_get_irq()

[Tejun Heo]

On Tue, May 16, 2017 at 02:06:12PM +0200, Thomas Petazzoni wrote:
> When platform_get_irq() fails, it returns an error code, which
> libahci_platform and replaces it by -EINVAL. This commit fixes that by
> propagating the error code. It fixes the situation where
> platform_get_irq() returns -EPROBE_DEFER because the interrupt
> controller is not available yet, and generally looks like the right
> thing to do.
> 
> We pay attention to not show the "no irq" message when we are in an
> EPROBE_DEFER situation, because the driver probing will be retried
> later on, once the interrupt controller becomes available to provide
> the interrupt.
> 
> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Applied to libata/for-4.12-fixes.

Thanks.

-- 
tejun

Re: [PATCH] ata: libahci: properly propagate return value of platform_get_irq()

[Sergei Shtylyov]

Hello!

On 5/16/2017 3:06 PM, Thomas Petazzoni wrote:

> When platform_get_irq() fails, it returns an error code, which
> libahci_platform and replaces it by -EINVAL.

    Couldn't parse this sentence...

> This commit fixes that by
> propagating the error code. It fixes the situation where
> platform_get_irq() returns -EPROBE_DEFER because the interrupt
> controller is not available yet, and generally looks like the right
> thing to do.
>
> We pay attention to not show the "no irq" message when we are in an
> EPROBE_DEFER situation, because the driver probing will be retried
> later on, once the interrupt controller becomes available to provide
> the interrupt.
>
> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
> ---
>  drivers/ata/libahci_platform.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/ata/libahci_platform.c b/drivers/ata/libahci_platform.c
> index aaa761b..cd2eab6 100644
> --- a/drivers/ata/libahci_platform.c
> +++ b/drivers/ata/libahci_platform.c
> @@ -514,8 +514,9 @@ int ahci_platform_init_host(struct platform_device *pdev,
>
>  	irq = platform_get_irq(pdev, 0);
>  	if (irq <= 0) {
> -		dev_err(dev, "no irq\n");
> -		return -EINVAL;
> +		if (irq != -EPROBE_DEFER)
> +			dev_err(dev, "no irq\n");
> +		return irq;

    Iff 'irq' is 0, you'll return 0 here, indicating success.
    Actually, I've fixed platform_get_irq() not to return 0 on error.

[...]

MBR, Sergei