power: charger-manager: Avoid recursive thermal get_temp call
authorKrzysztof Kozlowski <k.kozlowski@samsung.com>
Tue, 7 Oct 2014 15:47:37 +0000 (17:47 +0200)
committerSebastian Reichel <sre@kernel.org>
Tue, 28 Oct 2014 02:30:19 +0000 (03:30 +0100)
The charger manager supports POWER_SUPPLY_PROP_TEMP property and acts
as a thermal zone if any of these conditions match:
1. Fuel gauge used by charger manager supports POWER_SUPPLY_PROP_TEMP.
2. 'cm-thermal-zone' property is present in DTS (then it will supersede
   the fuel gauge temperature property).

However in case 1 (fuel gauge reports temperature and 'cm-thermal-zone'
is not set) the charger manager forwards its get_temp calls to fuel
gauge thermal zone.

This leads to reporting by lockdep a false positive deadlock for thermal
zone's mutex because of nested calls to thermal_zone_get_temp(). This is
false positive because these are different mutexes: one for charger
manager thermal zone and second for fuel gauge thermal zone.

Get rid of false lockdep alert and recursive call by setting
'no_thermal' property for this power supply class. The thermal zone for
charger manager won't be created (user space does not use it anyway).

The lockdep report:
[    2.540339] charger-manager charger-manager@0: Ignoring full-battery voltage threshold as it is not supplied
[    2.540351] charger-manager charger-manager@0: Ignoring full-battery full capacity threshold as it is not supplied
[    2.546296]
[    2.546302] =============================================
[    2.546305] [ INFO: possible recursive locking detected ]
[    2.546312] 3.17.0-rc6-next-20140926-00012-gbb13895e46af-dirty #39 Not tainted
[    2.546316] ---------------------------------------------
[    2.546321] swapper/0/1 is trying to acquire lock:
[    2.546348]  (&tz->lock){+.+...}, at: [<c0321d24>] thermal_zone_get_temp+0x38/0x68
[    2.546352]
[    2.546352] but task is already holding lock:
[    2.546369]  (&tz->lock){+.+...}, at: [<c0321d24>] thermal_zone_get_temp+0x38/0x68
[    2.546373]
[    2.546373] other info that might help us debug this:
[    2.546376]  Possible unsafe locking scenario:
[    2.546376]
[    2.546378]        CPU0
[    2.546380]        ----
[    2.546386]   lock(&tz->lock);
[    2.546392]   lock(&tz->lock);
[    2.546394]
[    2.546394]  *** DEADLOCK ***
[    2.546394]
[    2.546397]  May be due to missing lock nesting notation
[    2.546397]
[    2.546401] 2 locks held by swapper/0/1:
[    2.546430]  #0:  (&dev->mutex){......}, at: [<c02720c4>] __driver_attach+0x58/0x98
[    2.546448]  #1:  (&tz->lock){+.+...}, at: [<c0321d24>] thermal_zone_get_temp+0x38/0x68
[    2.546451]
[    2.546451] stack backtrace:
[    2.546460] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 3.17.0-rc6-next-20140926-00012-gbb13895e46af-dirty #39
[    2.546497] [<c00140f0>] (unwind_backtrace) from [<c0011228>] (show_stack+0x10/0x14)
[    2.546526] [<c0011228>] (show_stack) from [<c046158c>] (dump_stack+0x70/0xbc)
[    2.546554] [<c046158c>] (dump_stack) from [<c005e32c>] (validate_chain.isra.24+0x718/0x890)
[    2.546569] [<c005e32c>] (validate_chain.isra.24) from [<c005f0a0>] (__lock_acquire+0x498/0xa78)
[    2.546581] [<c005f0a0>] (__lock_acquire) from [<c005fb50>] (lock_acquire+0x78/0xb8)
[    2.546594] [<c005fb50>] (lock_acquire) from [<c0464260>] (mutex_lock_nested+0x64/0x458)
[    2.546605] [<c0464260>] (mutex_lock_nested) from [<c0321d24>] (thermal_zone_get_temp+0x38/0x68)
[    2.546634] [<c0321d24>] (thermal_zone_get_temp) from [<c031f1e0>] (charger_get_property+0x10c/0x348)
[    2.546649] [<c031f1e0>] (charger_get_property) from [<c031af18>] (power_supply_read_temp+0x28/0x58)
[    2.546662] [<c031af18>] (power_supply_read_temp) from [<c0321d38>] (thermal_zone_get_temp+0x4c/0x68)
[    2.546676] [<c0321d38>] (thermal_zone_get_temp) from [<c03233d8>] (thermal_zone_device_update+0x24/0x9c)
[    2.546687] [<c03233d8>] (thermal_zone_device_update) from [<c0323874>] (thermal_zone_device_register+0x424/0x550)
[    2.546701] [<c0323874>] (thermal_zone_device_register) from [<c031b3c0>] (__power_supply_register+0x2a4/0x348)
[    2.546714] [<c031b3c0>] (__power_supply_register) from [<c031ff64>] (charger_manager_probe+0x600/0xe5c)
[    2.546727] [<c031ff64>] (charger_manager_probe) from [<c0273384>] (platform_drv_probe+0x48/0xa4)
[    2.546746] [<c0273384>] (platform_drv_probe) from [<c0271f54>] (driver_probe_device+0x10c/0x224)
[    2.546760] [<c0271f54>] (driver_probe_device) from [<c0272100>] (__driver_attach+0x94/0x98)
[    2.546772] [<c0272100>] (__driver_attach) from [<c0270780>] (bus_for_each_dev+0x54/0x88)
[    2.546784] [<c0270780>] (bus_for_each_dev) from [<c027173c>] (bus_add_driver+0xd4/0x1d0)
[    2.546797] [<c027173c>] (bus_add_driver) from [<c027271c>] (driver_register+0x78/0xf4)
[    2.546809] [<c027271c>] (driver_register) from [<c0008984>] (do_one_initcall+0x80/0x1d4)
[    2.546829] [<c0008984>] (do_one_initcall) from [<c0612d60>] (kernel_init_freeable+0x10c/0x1d8)
[    2.546847] [<c0612d60>] (kernel_init_freeable) from [<c045c238>] (kernel_init+0x8/0xec)
[    2.546863] [<c045c238>] (kernel_init) from [<c000e828>] (ret_from_fork+0x14/0x2c)
[    2.551396] charger-manager charger-manager@0: 'chg-reg' regulator's externally_control is 0

Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Signed-off-by: Sebastian Reichel <sre@kernel.org>
drivers/power/charger-manager.c

index 7098a1ce2d3c1068e2ef6cc5fbbe4d185ece9850..7a1177ea238dd5686df5e020438c1b661ccbb14b 100644 (file)
@@ -970,6 +970,7 @@ static struct power_supply psy_default = {
        .properties = default_charger_props,
        .num_properties = ARRAY_SIZE(default_charger_props),
        .get_property = charger_get_property,
+       .no_thermal = true,
 };
 
 /**