[PATCH] power_supply: sysfs: Bring back write to writeable properties

[PATCH] power_supply: sysfs: Bring back write to writeable properties

[Krzysztof Kozlowski]

The fix for NULL pointer exception related to calling uevent for not
finished probe caused to set all writeable properties as non-writeable.
This was caused by checking if property is writeable before the initial
increase of power supply usage counter and in the same time using
wrapper over property_is_writeable(). The wrapper returns ENODEV if the
usage counter is still 0.

The call trace looked like:
  device probe:
    power_supply_register()
      use_cnt = 0;
      device_add()
        create sysfs entries
          power_supply_attr_is_visible()
            power_supply_property_is_writeable()
              if (use_cnt == 0) return -ENODEV;
      use_cnt++;

Replace the usage of wrapper with direct call to property_is_writeable()
from driver. This should be safe call during device probe because
implementations of this callback just return 0/1 for different
properties and they do not access any of the driver's internal data.

Fixes: 8e59c7f23410 ("power_supply: Fix NULL pointer dereference during bq27x00_battery probe")
Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
---
 drivers/power/power_supply_sysfs.c | 2 +-
 include/linux/power_supply.h       | 5 +++++
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/power/power_supply_sysfs.c b/drivers/power/power_supply_sysfs.c
index af026806cba5..ed2d7fd0c734 100644
--- a/drivers/power/power_supply_sysfs.c
+++ b/drivers/power/power_supply_sysfs.c
@@ -223,7 +223,7 @@ static umode_t power_supply_attr_is_visible(struct kobject *kobj,
 
 		if (property == attrno) {
 			if (psy->desc->property_is_writeable &&
-			    power_supply_property_is_writeable(psy, property) > 0)
+			    psy->desc->property_is_writeable(psy, property) > 0)
 				mode |= S_IWUSR;
 
 			return mode;
diff --git a/include/linux/power_supply.h b/include/linux/power_supply.h
index a80f1fd01ddb..0395bcb18ddb 100644
--- a/include/linux/power_supply.h
+++ b/include/linux/power_supply.h
@@ -206,6 +206,11 @@ struct power_supply_desc {
 	int (*set_property)(struct power_supply *psy,
 			    enum power_supply_property psp,
 			    const union power_supply_propval *val);
+	/*
+	 * property_is_writeable() will be called during registration
+	 * of power supply. If this happens during device probe then it must
+	 * not access internal data of device (because probe did not end).
+	 */
 	int (*property_is_writeable)(struct power_supply *psy,
 				     enum power_supply_property psp);
 	void (*external_power_changed)(struct power_supply *psy);
-- 
1.9.1

Re: [PATCH] power_supply: sysfs: Bring back write to writeable properties

[Sebastian Reichel]

[-- Attachment #1: Type: text/plain, Size: 1122 bytes --]

Hi,

On Mon, Jun 08, 2015 at 10:09:48AM +0900, Krzysztof Kozlowski wrote:
> The fix for NULL pointer exception related to calling uevent for not
> finished probe caused to set all writeable properties as non-writeable.
> This was caused by checking if property is writeable before the initial
> increase of power supply usage counter and in the same time using
> wrapper over property_is_writeable(). The wrapper returns ENODEV if the
> usage counter is still 0.
> 
> The call trace looked like:
>   device probe:
>     power_supply_register()
>       use_cnt = 0;
>       device_add()
>         create sysfs entries
>           power_supply_attr_is_visible()
>             power_supply_property_is_writeable()
>               if (use_cnt == 0) return -ENODEV;
>       use_cnt++;
> 
> Replace the usage of wrapper with direct call to property_is_writeable()
> from driver. This should be safe call during device probe because
> implementations of this callback just return 0/1 for different
> properties and they do not access any of the driver's internal data.

Thanks, queued.

-- Sebastian

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]