Re: [PATCH] staging: octeon: release the port device node in cvm_oct_probe

[Krzysztof Kozlowski <krzk@kernel.org>]

On 04/11/2026 08:45, Dan Carpenter wrote:
> On Wed, Nov 04, 2026 at 02:14:50AM +0000, Wentao Liang wrote:
>> cvm_oct_probe() looks up the device node of every port with
>> cvm_oct_node_for_port(), which walks the children with
>> for_each_child_of_node() and therefore returns the node with an
>> incremented reference count.  The reference is stored in
>> priv->of_node and used for the lifetime of the net_device, but it is
>> never dropped: both probe error paths free the net_device without
>> releasing the node, and cvm_oct_remove_device() does the same on
>> unbind.  Each failed or torn down port therefore keeps its device node
>> referenced forever.
>>
>> Add the missing of_node_put() in the two probe error paths and in
>> cvm_oct_remove_device(), which also covers the error cleanup done after
>> the ports have been registered.  of_node_put() accepts the NULL node of
>> the POW device.
>>
>> Fixes: 422d97b8b05e ("Revert "staging: octeon: delete driver"")
>> Cc: stable@vger.kernel.org
>> Assisted-by: DeepSeek-V4.1-Flash and OpenAI Codex
>> Signed-off-by: Wentao Liang <vulab@iscas.ac.cn>
> 
> Normally with firmware, we just take a reference to the firmware node,
> read the information we need to probe, and then release the reference.
> 
> Is there any need to hold the reference here until basically the system
> shuts down?

Just FYI, you are discussing with a bot. :)

Best regards,
Krzysztof