[PATCH] rpmsg: fix possible refcount leak in rpmsg_register_device_override()

[PATCH] rpmsg: fix possible refcount leak in rpmsg_register_device_override()

[Hangyu Hua]

[1] commit 1680939e9ecf ("rpmsg: virtio: Fix possible double free in
rpmsg_virtio_add_ctrl_dev()")
[2] commit c2eecefec5df ("rpmsg: virtio: Fix possible double free in
rpmsg_probe()")
[3] commit bb17d110cbf2 ("rpmsg: Fix calling device_lock() on
non-initialized device")

The above three patches merged at the same time introduced a new bug.
[1] and [2] make rpmsg_ns_register_device and rpmsg_ctrldev_register_device
need to call the callback function internally to free vch when it fails.
[3] has an error return path not handled vch.

Fix this by adding a put_device() to the error path.

Fixes: bb17d110cbf2 ("rpmsg: Fix calling device_lock() on non-initialized device")
Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
---
 drivers/rpmsg/rpmsg_core.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/rpmsg/rpmsg_core.c b/drivers/rpmsg/rpmsg_core.c
index 290c1f02da10..5a47cad89fdc 100644
--- a/drivers/rpmsg/rpmsg_core.c
+++ b/drivers/rpmsg/rpmsg_core.c
@@ -618,6 +618,7 @@ int rpmsg_register_device_override(struct rpmsg_device *rpdev,
 					  strlen(driver_override));
 		if (ret) {
 			dev_err(dev, "device_set_override failed: %d\n", ret);
+			put_device(dev);
 			return ret;
 		}
 	}
-- 
2.25.1

Re: [PATCH] rpmsg: fix possible refcount leak in rpmsg_register_device_override()

[Krzysztof Kozlowski]

On 23/06/2022 09:36, Hangyu Hua wrote:
> [1] commit 1680939e9ecf ("rpmsg: virtio: Fix possible double free in
> rpmsg_virtio_add_ctrl_dev()")
> [2] commit c2eecefec5df ("rpmsg: virtio: Fix possible double free in
> rpmsg_probe()")
> [3] commit bb17d110cbf2 ("rpmsg: Fix calling device_lock() on
> non-initialized device")

I think only the last [3] introduced it, because it's the commit missing
put_device in first error path.

> 
> The above three patches merged at the same time introduced a new bug.
> [1] and [2] make rpmsg_ns_register_device and rpmsg_ctrldev_register_device
> need to call the callback function internally to free vch when it fails.
> [3] has an error return path not handled vch.
> 
> Fix this by adding a put_device() to the error path.
> 
> Fixes: bb17d110cbf2 ("rpmsg: Fix calling device_lock() on non-initialized device")
> Signed-off-by: Hangyu Hua <hbh25y@gmail.com>


Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>

Best regards,
Krzysztof

Re: [PATCH] rpmsg: fix possible refcount leak in rpmsg_register_device_override()

[Hangyu Hua]

On 2022/6/23 16:15, Krzysztof Kozlowski wrote:
> On 23/06/2022 09:36, Hangyu Hua wrote:
>> [1] commit 1680939e9ecf ("rpmsg: virtio: Fix possible double free in
>> rpmsg_virtio_add_ctrl_dev()")
>> [2] commit c2eecefec5df ("rpmsg: virtio: Fix possible double free in
>> rpmsg_probe()")
>> [3] commit bb17d110cbf2 ("rpmsg: Fix calling device_lock() on
>> non-initialized device")
> 
> I think only the last [3] introduced it, because it's the commit missing
> put_device in first error path.
> 

I see. Do i need to change the commit log and then send a v2?

>>
>> The above three patches merged at the same time introduced a new bug.
>> [1] and [2] make rpmsg_ns_register_device and rpmsg_ctrldev_register_device
>> need to call the callback function internally to free vch when it fails.
>> [3] has an error return path not handled vch.
>>
>> Fix this by adding a put_device() to the error path.
>>
>> Fixes: bb17d110cbf2 ("rpmsg: Fix calling device_lock() on non-initialized device")
>> Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
> 
> 
> Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
> 
> Best regards,
> Krzysztof

Re: [PATCH] rpmsg: fix possible refcount leak in rpmsg_register_device_override()

[Krzysztof Kozlowski]

On 23/06/2022 12:53, Hangyu Hua wrote:
> On 2022/6/23 16:15, Krzysztof Kozlowski wrote:
>> On 23/06/2022 09:36, Hangyu Hua wrote:
>>> [1] commit 1680939e9ecf ("rpmsg: virtio: Fix possible double free in
>>> rpmsg_virtio_add_ctrl_dev()")
>>> [2] commit c2eecefec5df ("rpmsg: virtio: Fix possible double free in
>>> rpmsg_probe()")
>>> [3] commit bb17d110cbf2 ("rpmsg: Fix calling device_lock() on
>>> non-initialized device")
>>
>> I think only the last [3] introduced it, because it's the commit missing
>> put_device in first error path.
>>
> 
> I see. Do i need to change the commit log and then send a v2?

Yes, please. With my Reviewed-by tag.


Best regards,
Krzysztof