[PATCH] mmc: sdhci-acpi: Only powered up enabled acpi child devices

[PATCH] mmc: sdhci-acpi: Only powered up enabled acpi child devices

[Hans de Goede]

Commit e5bbf30733f9 ("mmc: sdhci-acpi: Ensure connected devices are
powered when probing") introduced code to powerup any acpi child
nodes listed in the dstd. But some dstd-s list all possible devices
used on some board variants, while reporting if the device is actually
present and enabled in the status field of the device.

So we end up calling the acpi _PS0 (power-on) method for devices which
are not actually present. This does not always end well, e.g. on my
cube iwork8 air tablet, this results in freezing the entire tablet as
soon as the r8723bs module is loaded.

This commit fixes this by checking the child device's status.present
and status.enabled bits and only call acpi_device_fix_up_power()
if both are set.

Fixes: e5bbf30733f9 ("mmc: sdhci-acpi: Ensure connected devices are powered when probing")
BugLink: https://github.com/hadess/rtl8723bs/issues/80
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
---
 drivers/mmc/host/sdhci-acpi.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/mmc/host/sdhci-acpi.c b/drivers/mmc/host/sdhci-acpi.c
index 92096db..34ee237 100644
--- a/drivers/mmc/host/sdhci-acpi.c
+++ b/drivers/mmc/host/sdhci-acpi.c
@@ -396,7 +396,8 @@ static int sdhci_acpi_probe(struct platform_device *pdev)
 	/* Power on the SDHCI controller and its children */
 	acpi_device_fix_up_power(device);
 	list_for_each_entry(child, &device->children, node)
-		acpi_device_fix_up_power(child);
+		if (child->status.present && child->status.enabled)
+			acpi_device_fix_up_power(child);
 
 	if (acpi_bus_get_status(device) || !device->status.present)
 		return -ENODEV;
-- 
2.9.3

Re: [PATCH] mmc: sdhci-acpi: Only powered up enabled acpi child devices

[Ulf Hansson]

On 21 December 2016 at 00:19, Hans de Goede <hdegoede@redhat.com> wrote:
> Commit e5bbf30733f9 ("mmc: sdhci-acpi: Ensure connected devices are
> powered when probing") introduced code to powerup any acpi child
> nodes listed in the dstd. But some dstd-s list all possible devices
> used on some board variants, while reporting if the device is actually
> present and enabled in the status field of the device.
>
> So we end up calling the acpi _PS0 (power-on) method for devices which
> are not actually present. This does not always end well, e.g. on my
> cube iwork8 air tablet, this results in freezing the entire tablet as
> soon as the r8723bs module is loaded.
>
> This commit fixes this by checking the child device's status.present
> and status.enabled bits and only call acpi_device_fix_up_power()
> if both are set.
>
> Fixes: e5bbf30733f9 ("mmc: sdhci-acpi: Ensure connected devices are powered when probing")
> BugLink: https://github.com/hadess/rtl8723bs/issues/80
> Signed-off-by: Hans de Goede <hdegoede@redhat.com>

Seems like I should add stable tag to this as well!?

> ---
>  drivers/mmc/host/sdhci-acpi.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/mmc/host/sdhci-acpi.c b/drivers/mmc/host/sdhci-acpi.c
> index 92096db..34ee237 100644
> --- a/drivers/mmc/host/sdhci-acpi.c
> +++ b/drivers/mmc/host/sdhci-acpi.c
> @@ -396,7 +396,8 @@ static int sdhci_acpi_probe(struct platform_device *pdev)
>         /* Power on the SDHCI controller and its children */
>         acpi_device_fix_up_power(device);
>         list_for_each_entry(child, &device->children, node)
> -               acpi_device_fix_up_power(child);
> +               if (child->status.present && child->status.enabled)
> +                       acpi_device_fix_up_power(child);
>
>         if (acpi_bus_get_status(device) || !device->status.present)
>                 return -ENODEV;
> --
> 2.9.3
>

Adrian, are you happy with this fix?

Kind regards
Uffe

Re: [PATCH] mmc: sdhci-acpi: Only powered up enabled acpi child devices

[Adrian Hunter]

On 12/01/17 12:03, Ulf Hansson wrote:
> On 21 December 2016 at 00:19, Hans de Goede <hdegoede@redhat.com> wrote:
>> Commit e5bbf30733f9 ("mmc: sdhci-acpi: Ensure connected devices are
>> powered when probing") introduced code to powerup any acpi child
>> nodes listed in the dstd. But some dstd-s list all possible devices
>> used on some board variants, while reporting if the device is actually
>> present and enabled in the status field of the device.
>>
>> So we end up calling the acpi _PS0 (power-on) method for devices which
>> are not actually present. This does not always end well, e.g. on my
>> cube iwork8 air tablet, this results in freezing the entire tablet as
>> soon as the r8723bs module is loaded.
>>
>> This commit fixes this by checking the child device's status.present
>> and status.enabled bits and only call acpi_device_fix_up_power()
>> if both are set.
>>
>> Fixes: e5bbf30733f9 ("mmc: sdhci-acpi: Ensure connected devices are powered when probing")
>> BugLink: https://github.com/hadess/rtl8723bs/issues/80
>> Signed-off-by: Hans de Goede <hdegoede@redhat.com>
> 
> Seems like I should add stable tag to this as well!?

Yes I think so.

> 
>> ---
>>  drivers/mmc/host/sdhci-acpi.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/mmc/host/sdhci-acpi.c b/drivers/mmc/host/sdhci-acpi.c
>> index 92096db..34ee237 100644
>> --- a/drivers/mmc/host/sdhci-acpi.c
>> +++ b/drivers/mmc/host/sdhci-acpi.c
>> @@ -396,7 +396,8 @@ static int sdhci_acpi_probe(struct platform_device *pdev)
>>         /* Power on the SDHCI controller and its children */
>>         acpi_device_fix_up_power(device);
>>         list_for_each_entry(child, &device->children, node)
>> -               acpi_device_fix_up_power(child);
>> +               if (child->status.present && child->status.enabled)
>> +                       acpi_device_fix_up_power(child);
>>
>>         if (acpi_bus_get_status(device) || !device->status.present)
>>                 return -ENODEV;
>> --
>> 2.9.3
>>
> 
> Adrian, are you happy with this fix?

Yes.

Acked-by: Adrian Hunter <adrian.hunter@intel.com>

Re: [PATCH] mmc: sdhci-acpi: Only powered up enabled acpi child devices

[Ulf Hansson]

+stable

On 21 December 2016 at 00:19, Hans de Goede <hdegoede@redhat.com> wrote:
> Commit e5bbf30733f9 ("mmc: sdhci-acpi: Ensure connected devices are
> powered when probing") introduced code to powerup any acpi child
> nodes listed in the dstd. But some dstd-s list all possible devices
> used on some board variants, while reporting if the device is actually
> present and enabled in the status field of the device.
>
> So we end up calling the acpi _PS0 (power-on) method for devices which
> are not actually present. This does not always end well, e.g. on my
> cube iwork8 air tablet, this results in freezing the entire tablet as
> soon as the r8723bs module is loaded.
>
> This commit fixes this by checking the child device's status.present
> and status.enabled bits and only call acpi_device_fix_up_power()
> if both are set.
>
> Fixes: e5bbf30733f9 ("mmc: sdhci-acpi: Ensure connected devices are powered when probing")
> BugLink: https://github.com/hadess/rtl8723bs/issues/80
> Signed-off-by: Hans de Goede <hdegoede@redhat.com>

Thanks, applied for fixes and added a stable tag.

Kind regards
Uffe

> ---
>  drivers/mmc/host/sdhci-acpi.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/mmc/host/sdhci-acpi.c b/drivers/mmc/host/sdhci-acpi.c
> index 92096db..34ee237 100644
> --- a/drivers/mmc/host/sdhci-acpi.c
> +++ b/drivers/mmc/host/sdhci-acpi.c
> @@ -396,7 +396,8 @@ static int sdhci_acpi_probe(struct platform_device *pdev)
>         /* Power on the SDHCI controller and its children */
>         acpi_device_fix_up_power(device);
>         list_for_each_entry(child, &device->children, node)
> -               acpi_device_fix_up_power(child);
> +               if (child->status.present && child->status.enabled)
> +                       acpi_device_fix_up_power(child);
>
>         if (acpi_bus_get_status(device) || !device->status.present)
>                 return -ENODEV;
> --
> 2.9.3
>