Re: [RFC/RFT v2 0/2] power_supply: Fix NULL pointer dereference from uevent

Hi,

Am 19.05.2015 um 09:13 schrieb Krzysztof Kozlowski [off-list ref]:

Hi,


Changes since v1:
=================
1. Patch 2: fix invalid member used for container_of().
2. Patch 2: Replace WARN with pr_warn() in __power_supply_register()
  if parent is missing.


Description:
============
This is an idea to fix issue in bq27x00 driver (and probably in others)
reported by H. Nikolaus Schaller [0].

The fixes are marked RFC/RFT because:
1. I do not have bq27x00-like device. I confirmed this and tested on
  modified drivers (max77693, ACPI AC). These drivers are not
  impacted by the issue but one can easily adjust them to reproduce
  the problem.
2. The first uevent coming from power_supply_register() is now
  different because it won't contain device properties. I am
  not sure how this impacts userspace.

Comments are welcomed.

appears to work! bq27000 is coming up now without hickups.

What I can’t test is if the uevent is reasonable because we do not
have a specific rule triggered by it in our system.

So from my side we are now happy with the solution.

Thanks and BR,
Nikolaus

[0] https://lkml.org/lkml/2015/5/18/152


Best regards,
Krzysztof

Krzysztof Kozlowski (2):
 power_supply: Fix NULL pointer dereference during bq27x00_battery
   probe
 power_supply: Fix possible NULL pointer dereference on early uevent

drivers/power/power_supply_core.c | 61 +++++++++++++++++++++++++++++++++++----
include/linux/power_supply.h      |  1 +
2 files changed, 56 insertions(+), 6 deletions(-)

-- 
1.9.1