linux-toradex.git/drivers/power/power_supply_core.c, branch v4.2.2

power_supply: Correct kerneldoc copy paste errors

2015-06-19T12:21:01+00:00

Signed-off-by: Bjorn Andersson 
Reviewed-by: Krzysztof Kozlowski 
Signed-off-by: Sebastian Reichel

power: Add devm_power_supply_get_by_phandle() helper function

2015-06-10T14:15:54+00:00

This commit adds a resource-managed version of the
power_supply_get_by_phandle() function.

Signed-off-by: Hans de Goede 
Signed-off-by: Sebastian Reichel

power_supply: Fix possible NULL pointer dereference on early uevent

2015-05-21T13:41:09+00:00

Don't call the power_supply_changed() from power_supply_register() when
parent is still probing because it may lead to accessing parent too
early.

In bq27x00_battery this caused NULL pointer exception because uevent of
power_supply_changed called back the the get_property() method provided
by the driver. The get_property() method accessed pointer which should
be returned by power_supply_register().

Starting from bq27x00_battery_probe():
  di->bat = power_supply_register()
    power_supply_changed()
      kobject_uevent()
        power_supply_uevent()
          power_supply_show_property()
            power_supply_get_property()
              bq27x00_battery_get_property()
                dereference of di->bat which is NULL here

The dereference of di->bat (value returned by power_supply_register())
is the currently visible problem. However calling back the methods
provided by driver before ending the probe may lead to accessing other
driver-related data which is not yet initialized.

The call to power_supply_changed() is postponed till probing ends -
mutex of parent device is released.

Reported-by: H. Nikolaus Schaller 
Signed-off-by: Krzysztof Kozlowski 
Fixes: 297d716f6260 ("power_supply: Change ownership from driver to core")
Tested-By: Dr. H. Nikolaus Schaller 
Signed-off-by: Sebastian Reichel

power_supply: Fix NULL pointer dereference during bq27x00_battery probe

2015-05-21T13:40:54+00:00

Power supply is often registered during probe of a driver. The
power_supply_register() returns pointer to newly allocated structure as
return value. However before returning the power_supply_register()
calls back the get_property() method provided by the driver through
uevent.

In that time the driver probe is still in progress and driver did not
assigned pointer to power supply to its local variables. This leads to
NULL pointer dereference from get_property() function.
Starting from bq27x00_battery_probe():
  di->bat = power_supply_register()
    device_add()
      kobject_uevent()
        power_supply_uevent()
          power_supply_show_property()
            power_supply_get_property()
              bq27x00_battery_get_property()
                dereference of (di->bat) which is NULL here

The first uevent of power supply (the one coming from device creation)
should not call back to the driver. To prevent that from happening,
increment the atomic use counter at the end of power_supply_register().
This means that power_supply_get_property() will return -ENODEV.

IMPORTANT:
The patch has impact on this first uevent sent from power supply because
it will not contain properties from power supply.

The uevent with properties will be sent later after indicating that
power supply has changed. This also has a race now, but will be fixed in
other patches.

Reported-by: H. Nikolaus Schaller 
Signed-off-by: Krzysztof Kozlowski 
Fixes: 297d716f6260 ("power_supply: Change ownership from driver to core")
Tested-By: Dr. H. Nikolaus Schaller 
Signed-off-by: Sebastian Reichel

power_supply: Increment power supply use counter when obtaining references

2015-03-13T22:15:52+00:00

Increment the power_supply.use_cnt usage counter on:
 - power_supply_get_by_phandle()
 - power_supply_get_by_name()
and decrement it on power_supply_put() call.

This helps tracking of valid usage of power supply instance by
consumers. The usage counter itself also allows safe calling of
power_supply_get_property-like functions even when driver unregisters
this power supply.

Signed-off-by: Krzysztof Kozlowski 
Reviewed-by: Bartlomiej Zolnierkiewicz 
Signed-off-by: Sebastian Reichel

power_supply: Add power_supply_put for decrementing device reference counter

2015-03-13T22:15:52+00:00

The power_supply_get_by_phandle() and power_supply_get_by_name() use
function class_find_device() for obtaining the reference to power
supply. Each use of class_find_device() increases the power supply's
device reference counter.

However the reference counter was not decreased by users of this API.
Thus final device_unregister() call from power_supply_unregister() could
not release the device and clean up its resources. This lead to memory
leak if at least once power_supply_get_by_*() was called between
registering and unregistering the power supply.

Add and document new API power_supply_put() for decrementing the
reference counter.

Signed-off-by: Krzysztof Kozlowski 
Acked-by: Pavel Machek 
Reviewed-by: Bartlomiej Zolnierkiewicz 
Reviewed-by: Sebastian Reichel 
Signed-off-by: Sebastian Reichel

power_supply: Change ownership from driver to core

2015-03-13T22:15:51+00:00

Change the ownership of power_supply structure from each driver
implementing the class to the power supply core.

The patch changes power_supply_register() function thus all drivers
implementing power supply class are adjusted.

Each driver provides the implementation of power supply. However it
should not be the owner of power supply class instance because it is
exposed by core to other subsystems with power_supply_get_by_name().
These other subsystems have no knowledge when the driver will unregister
the power supply. This leads to several issues when driver is unbound -
mostly because user of power supply accesses freed memory.

Instead let the core own the instance of struct 'power_supply'.  Other
users of this power supply will still access valid memory because it
will be freed when device reference count reaches 0. Currently this
means "it will leak" but power_supply_put() call in next patches will
solve it.

This solves invalid memory references in following race condition
scenario:

Thread 1: charger manager
Thread 2: power supply driver, used by charger manager

THREAD 1 (charger manager)         THREAD 2 (power supply driver)
==========================         ==============================
psy = power_supply_get_by_name()
                                   Driver unbind, .remove
                                     power_supply_unregister()
                                     Device fully removed
psy->get_property()

The 'get_property' call is executed in invalid context because the driver was
unbound and struct 'power_supply' memory was freed.

This could be observed easily with charger manager driver (here compiled
with max17040 fuel gauge):

$ cat /sys/devices/virtual/power_supply/cm-battery/capacity &
$ echo "1-0036" > /sys/bus/i2c/drivers/max17040/unbind
[   55.725123] Unable to handle kernel NULL pointer dereference at virtual address 00000000
[   55.732584] pgd = d98d4000
[   55.734060] [00000000] *pgd=5afa2831, *pte=00000000, *ppte=00000000
[   55.740318] Internal error: Oops: 80000007 [#1] PREEMPT SMP ARM
[   55.746210] Modules linked in:
[   55.749259] CPU: 1 PID: 2936 Comm: cat Tainted: G        W       3.19.0-rc1-next-20141226-00048-gf79f475f3c44-dirty #1496
[   55.760190] Hardware name: SAMSUNG EXYNOS (Flattened Device Tree)
[   55.766270] task: d9b76f00 ti: daf54000 task.ti: daf54000
[   55.771647] PC is at 0x0
[   55.774182] LR is at charger_get_property+0x2f4/0x36c
[   55.779201] pc : [<00000000>]    lr : []    psr: 60000013
[   55.779201] sp : daf55e90  ip : 00000003  fp : 00000000
[   55.790657] r10: 00000000  r9 : c06e2878  r8 : d9b26c68
[   55.795865] r7 : dad81610  r6 : daec7410  r5 : daf55ebc  r4 : 00000000
[   55.802367] r3 : 00000000  r2 : daf55ebc  r1 : 0000002a  r0 : d9b26c68
[   55.808879] Flags: nZCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment user
[   55.815994] Control: 10c5387d  Table: 598d406a  DAC: 00000015
[   55.821723] Process cat (pid: 2936, stack limit = 0xdaf54210)
[   55.827451] Stack: (0xdaf55e90 to 0xdaf56000)
[   55.831795] 5e80:                                     60000013 c01459c4 0000002a c06f8ef8
[   55.839956] 5ea0: db651000 c06f8ef8 daebac00 c04cb668 daebac08 c0346864 00000000 c01459c4
[   55.848115] 5ec0: d99eaa80 c06f8ef8 00000fff 00001000 db651000 c027f25c c027f240 d99eaa80
[   55.856274] 5ee0: d9a06c00 c0146218 daf55f18 00001000 d99eaa80 db4c18c0 00000001 00000001
[   55.864468] 5f00: daf55f80 c0144c78 c0144c54 c0107f90 00015000 d99eaab0 00000000 00000000
[   55.872603] 5f20: 000051c7 00000000 db4c18c0 c04a9370 00015000 00001000 daf55f80 00001000
[   55.880763] 5f40: daf54000 00015000 00000000 c00e53dc db4c18c0 c00e548c 0000000d 00008124
[   55.888937] 5f60: 00000001 00000000 00000000 db4c18c0 db4c18c0 00001000 00015000 c00e5550
[   55.897099] 5f80: 00000000 00000000 00001000 00001000 00015000 00000003 00000003 c000f364
[   55.905239] 5fa0: 00000000 c000f1a0 00001000 00015000 00000003 00015000 00001000 0001333c
[   55.913399] 5fc0: 00001000 00015000 00000003 00000003 00000002 00000000 00000000 00000000
[   55.921560] 5fe0: 7fffe000 be999850 0000a225 b6f3c19c 60000010 00000003 00000000 00000000
[   55.929744] [] (charger_get_property) from [] (power_supply_show_property+0x48/0x20c)
[   55.939286] [] (power_supply_show_property) from [] (dev_attr_show+0x1c/0x48)
[   55.948130] [] (dev_attr_show) from [] (sysfs_kf_seq_show+0x84/0x104)
[   55.956298] [] (sysfs_kf_seq_show) from [] (kernfs_seq_show+0x24/0x28)
[   55.964536] [] (kernfs_seq_show) from [] (seq_read+0x1b0/0x484)
[   55.972172] [] (seq_read) from [] (__vfs_read+0x18/0x4c)
[   55.979188] [] (__vfs_read) from [] (vfs_read+0x7c/0x100)
[   55.986304] [] (vfs_read) from [] (SyS_read+0x40/0x8c)
[   55.993164] [] (SyS_read) from [] (ret_fast_syscall+0x0/0x48)
[   56.000626] Code: bad PC value
[   56.011652] ---[ end trace 7b64343fbdae8ef1 ]---

Signed-off-by: Krzysztof Kozlowski 
Reviewed-by: Bartlomiej Zolnierkiewicz 

[for the nvec part]
Reviewed-by: Marc Dietrich 

[for compal-laptop.c]
Acked-by: Darren Hart 

[for the mfd part]
Acked-by: Lee Jones 

[for the hid part]
Acked-by: Jiri Kosina 

[for the acpi part]
Acked-by: Rafael J. Wysocki 

Signed-off-by: Sebastian Reichel

power_supply: Add API for safe access of power supply function attrs

2015-03-13T22:15:48+00:00

Add simple wrappers for accessing power supply's function attributes:
 - get_property -> power_supply_get_property
 - set_property -> power_supply_set_property
 - property_is_writeable -> power_supply_property_is_writeable
 - external_power_changed -> power_supply_external_power_changed

This API along with atomic usage counter adds a safe way of accessing a
power supply from another driver. If power supply is unregistered after
obtaining reference to it by some driver, then the API wrappers won't be
executed in invalid (freed) context.

Next patch changing the ownership of power supply class is still needed
to fully fix race conditions in accessing freed power supply.

Signed-off-by: Krzysztof Kozlowski 
Reviewed-by: Bartlomiej Zolnierkiewicz 
Reviewed-by: Sebastian Reichel 
Acked-by: Pavel Machek 
Signed-off-by: Sebastian Reichel

power_supply: Move run-time configuration to separate structure

2015-03-13T22:15:12+00:00

Add new structure 'power_supply_config' for holding run-time
initialization data like of_node, supplies and private driver data.

The power_supply_register() function is changed so all power supply
drivers need updating.

When registering the power supply this new 'power_supply_config' should be
used instead of directly initializing 'struct power_supply'. This allows
changing the ownership of power_supply structure from driver to the
power supply core in next patches.

When a driver does not use of_node or supplies then it should use NULL
as config. If driver uses of_node or supplies then it should allocate
config on stack and initialize it with proper values.

Signed-off-by: Krzysztof Kozlowski 
Reviewed-by: Bartlomiej Zolnierkiewicz 
Acked-by: Pavel Machek 

[for the nvec part]
Reviewed-by: Marc Dietrich 

[for drivers/platform/x86/compal-laptop.c]
Reviewed-by: Darren Hart 

[for drivers/hid/*]
Reviewed-by: Jiri Kosina 

Signed-off-by: Sebastian Reichel

power_supply: Add driver private data

2015-03-13T21:52:52+00:00

Allow drivers to store private data inside power_supply structure for
later usage in power supply operations.

Usage of driver private data is necessary to access driver's state
container object from power supply calls (like get_property()) if struct
'power_supply' is a stored there as a pointer, for example:

struct some_driver_info {
	struct i2c_client       *client;
	struct power_supply     *power_supply;
	...
}

In such case one cannot use container_of() and must store pointer to
state container as private data.

Signed-off-by: Krzysztof Kozlowski 
Reviewed-by: Bartlomiej Zolnierkiewicz 
Reviewed-by: Sebastian Reichel 
Acked-by: Pavel Machek 
Signed-off-by: Sebastian Reichel