The call to of_parse_phandle() returns a node pointer with refcount
incremented thus it must be explicitly decremented here after the last
usage.
The of_find_device_by_node() takes a reference to the underlying device
structure, we also should release that reference.
This patch fixes those two issues.
Hans Verkuil says:
The cec driver should never take a reference of the hdmi device.
It never accesses the HDMI device, it only needs the HDMI device pointer as
a key in the notifier list.
The real problem is that several CEC drivers take a reference of the HDMI
device and never release it. So those drivers need to be fixed.
Fixes: a93d429b51fb ("[media] s5p-cec: add cec-notifier support, move out of staging")
Suggested-by: Hans Verkuil (hansverk) <[email protected]>
Signed-off-by: Wen Yang <[email protected]>
Cc: Hans Verkuil (hansverk) <[email protected]>
Cc: Hans Verkuil <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Cc: Mauro Carvalho Chehab <[email protected]>
Cc: Wen Yang <[email protected]>
Cc: [email protected]
Cc: [email protected]
---
v2->v1:
- move of_node_put() to just after the 'hdmi_dev = of_find_device_by_node(np)'.
- put_device() can be done before the cec = devm_kzalloc line.
drivers/media/platform/s5p-cec/s5p_cec.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/media/platform/s5p-cec/s5p_cec.c b/drivers/media/platform/s5p-cec/s5p_cec.c
index 8837e26..1f5c355 100644
--- a/drivers/media/platform/s5p-cec/s5p_cec.c
+++ b/drivers/media/platform/s5p-cec/s5p_cec.c
@@ -192,9 +192,11 @@ static int s5p_cec_probe(struct platform_device *pdev)
return -ENODEV;
}
hdmi_dev = of_find_device_by_node(np);
+ of_node_put(np);
if (hdmi_dev == NULL)
return -EPROBE_DEFER;
+ put_device(&hdmi_dev->dev);
cec = devm_kzalloc(&pdev->dev, sizeof(*cec), GFP_KERNEL);
if (!cec)
return -ENOMEM;
--
2.9.5
Hi
On 2019-02-13 00:47, Wen Yang wrote:
> The call to of_parse_phandle() returns a node pointer with refcount
> incremented thus it must be explicitly decremented here after the last
> usage.
> The of_find_device_by_node() takes a reference to the underlying device
> structure, we also should release that reference.
> This patch fixes those two issues.
>
> Hans Verkuil says:
> The cec driver should never take a reference of the hdmi device.
> It never accesses the HDMI device, it only needs the HDMI device pointer as
> a key in the notifier list.
> The real problem is that several CEC drivers take a reference of the HDMI
> device and never release it. So those drivers need to be fixed.
>
> Fixes: a93d429b51fb ("[media] s5p-cec: add cec-notifier support, move out of staging")
> Suggested-by: Hans Verkuil (hansverk) <[email protected]>
> Signed-off-by: Wen Yang <[email protected]>
> Cc: Hans Verkuil (hansverk) <[email protected]>
> Cc: Hans Verkuil <[email protected]>
> Cc: Marek Szyprowski <[email protected]>
> Cc: Mauro Carvalho Chehab <[email protected]>
> Cc: Wen Yang <[email protected]>
> Cc: [email protected]
> Cc: [email protected]
Acked-by: Marek Szyprowski <[email protected]>
> ---
> v2->v1:
> - move of_node_put() to just after the 'hdmi_dev = of_find_device_by_node(np)'.
> - put_device() can be done before the cec = devm_kzalloc line.
>
> drivers/media/platform/s5p-cec/s5p_cec.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/drivers/media/platform/s5p-cec/s5p_cec.c b/drivers/media/platform/s5p-cec/s5p_cec.c
> index 8837e26..1f5c355 100644
> --- a/drivers/media/platform/s5p-cec/s5p_cec.c
> +++ b/drivers/media/platform/s5p-cec/s5p_cec.c
> @@ -192,9 +192,11 @@ static int s5p_cec_probe(struct platform_device *pdev)
> return -ENODEV;
> }
> hdmi_dev = of_find_device_by_node(np);
> + of_node_put(np);
> if (hdmi_dev == NULL)
> return -EPROBE_DEFER;
>
> + put_device(&hdmi_dev->dev);
> cec = devm_kzalloc(&pdev->dev, sizeof(*cec), GFP_KERNEL);
> if (!cec)
> return -ENOMEM;
Best regards
--
Marek Szyprowski, PhD
Samsung R&D Institute Poland