4.9-stable review patch. If anyone has any objections, please let me know.
------------------
[ Upstream commit 5ddb0869bfc1bca6cfc592c74c64a026f936638c ]
I've stumbled upon a kernel crash and the logs pointed me towards the lp5562 driver:
<4>[306013.841294] lp5562 0-0030: Direct firmware load for lp5562 failed with error -2 <4>[306013.894990] lp5562 0-0030: Falling back to user helper ... <3>[306073.924886] lp5562 0-0030: firmware request failed <1>[306073.939456] Unable to handle kernel NULL pointer dereference at virtual address 00000000 <4>[306074.251011] PC is at _raw_spin_lock+0x1c/0x58 <4>[306074.255539] LR is at release_firmware+0x6c/0x138 ...
After taking a look I noticed firmware_release() could be called with either NULL or a dangling pointer.
Fixes: 10c06d178df11 ("leds-lp55xx: support firmware interface") Signed-off-by: Michal Kazior michal@plume.com Signed-off-by: Jacek Anaszewski jacek.anaszewski@gmail.com Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/leds/leds-lp55xx-common.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/leds/leds-lp55xx-common.c b/drivers/leds/leds-lp55xx-common.c index 5377f22ff994..e2655953667c 100644 --- a/drivers/leds/leds-lp55xx-common.c +++ b/drivers/leds/leds-lp55xx-common.c @@ -201,7 +201,7 @@ static void lp55xx_firmware_loaded(const struct firmware *fw, void *context)
if (!fw) { dev_err(dev, "firmware request failed\n"); - goto out; + return; }
/* handling firmware data is chip dependent */ @@ -214,9 +214,9 @@ static void lp55xx_firmware_loaded(const struct firmware *fw, void *context)
mutex_unlock(&chip->lock);
-out: /* firmware should be released for other channel use */ release_firmware(chip->fw); + chip->fw = NULL; }
static int lp55xx_request_firmware(struct lp55xx_chip *chip)