linux-kernel - [RFC/RFT v2 1/2] power_supply: Fix NULL pointer dereference during bq27x00

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Tue, 19 May 2015 16:13:01 +0900
From:	Krzysztof Kozlowski <k.kozlowski@...sung.com>
To:	Sebastian Reichel <sre@...nel.org>,
	Dmitry Eremin-Solenikov <dbaryshkov@...il.com>,
	David Woodhouse <dwmw2@...radead.org>,
	linux-pm@...r.kernel.org, linux-kernel@...r.kernel.org
Cc:	"H. Nikolaus Schaller" <hns@...delico.com>,
	Krzysztof Kozlowski <k.kozlowski@...sung.com>
Subject: [RFC/RFT v2 1/2] power_supply: Fix NULL pointer dereference during
 bq27x00_battery probe

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 <hns@...delico.com>
Signed-off-by: Krzysztof Kozlowski <k.kozlowski@...sung.com>
Fixes: 297d716f6260 ("power_supply: Change ownership from driver to core")
---
 drivers/power/power_supply_core.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/drivers/power/power_supply_core.c b/drivers/power/power_supply_core.c
index 2ed4a4a6b3c5..15da277e0e8d 100644
--- a/drivers/power/power_supply_core.c
+++ b/drivers/power/power_supply_core.c
@@ -659,7 +659,6 @@ __power_supply_register(struct device *parent,
 	dev->release = power_supply_dev_release;
 	dev_set_drvdata(dev, psy);
 	psy->desc = desc;
-	atomic_inc(&psy->use_cnt);
 	if (cfg) {
 		psy->drv_data = cfg->drv_data;
 		psy->of_node = cfg->of_node;
@@ -700,6 +699,16 @@ __power_supply_register(struct device *parent,
 	if (rc)
 		goto create_triggers_failed;
 
+	/*
+	 * Update use_cnt after any uevents (most notably from device_add()).
+	 * We are here still during driver's probe but
+	 * the power_supply_uevent() calls back driver's get_property
+	 * method so:
+	 * 1. Driver did not assigned the returned struct power_supply,
+	 * 2. Driver could not finish initialization (anything in its probe
+	 *    after calling power_supply_register()).
+	 */
+	atomic_inc(&psy->use_cnt);
 	power_supply_changed(psy);
 
 	return psy;
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/