On 07/01/2025 12:30, Sakari Ailus wrote:
> Hi Lubomir,
>
> Thanks for the patch.
>
> On Tue, Dec 31, 2024 at 08:04:34PM +0100, Lubomir Rintel wrote:
>> In commit 4af65141e38e ("media: marvell: cafe: Register V4L2 device
>> earlier"), a call to v4l2_device_register() was moved away from
>> mccic_register() into its caller, marvell/cafe's cafe_pci_probe().
>> This is not the only caller though -- there's also marvell/mmp.
>>
>> Add v4l2_device_register() into mmpcam_probe() to unbreak the MMP camera
>> driver, in a fashion analogous to what's been done to the Cafe driver.
>> Same for the teardown path.
>>
>> Fixes: 4af65141e38e ("media: marvell: cafe: Register V4L2 device earlier")
>> Signed-off-by: Lubomir Rintel <lkundrak@xxxxx>
>> Cc: stable@xxxxxxxxxxxxxxx # v6.6+
>> ---
>> drivers/media/platform/marvell/mmp-driver.c | 21 +++++++++++++++++----
>> 1 file changed, 17 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/media/platform/marvell/mmp-driver.c b/drivers/media/platform/marvell/mmp-driver.c
>> index 3fd4fc1b9c48..d3da7ebb4a2b 100644
>> --- a/drivers/media/platform/marvell/mmp-driver.c
>> +++ b/drivers/media/platform/marvell/mmp-driver.c
>> @@ -231,13 +231,23 @@ static int mmpcam_probe(struct platform_device *pdev)
>>
>> mcam_init_clk(mcam);
>>
>> + /*
>> + * Register with V4L.
>> + */
>> +
>> + ret = v4l2_device_register(mcam->dev, &mcam->v4l2_dev);
>
> I'd do this just before initialising the notifier (as in the patch in
> Fixes: tag): registering the V4L2 device requires probably severe memory
> pressure while it's entirely plausible there's no endpoint for the device.
Actually, v4l2_device_register does very little. No memory allocation is involved.
I was planning to take this patch as-is.
Regards,
Hans
>
>> + if (ret)
>> + return ret;
>> +
>> /*
>> * Create a match of the sensor against its OF node.
>> */
>> ep = fwnode_graph_get_next_endpoint(of_fwnode_handle(pdev->dev.of_node),
>> NULL);
>> - if (!ep)
>> - return -ENODEV;
>> + if (!ep) {
>> + ret = -ENODEV;
>> + goto out_v4l2_device_unregister;
>> + }
>>
>> v4l2_async_nf_init(&mcam->notifier, &mcam->v4l2_dev);
>>
>> @@ -246,7 +256,7 @@ static int mmpcam_probe(struct platform_device *pdev)
>> fwnode_handle_put(ep);
>> if (IS_ERR(asd)) {
>> ret = PTR_ERR(asd);
>> - goto out;
>> + goto out_v4l2_device_unregister;
>> }
>>
>> /*
>> @@ -254,7 +264,7 @@ static int mmpcam_probe(struct platform_device *pdev)
>> */
>> ret = mccic_register(mcam);
>> if (ret)
>> - goto out;
>> + goto out_v4l2_device_unregister;
>>
>> /*
>> * Add OF clock provider.
>> @@ -283,6 +293,8 @@ static int mmpcam_probe(struct platform_device *pdev)
>> return 0;
>> out:
>
> For clarity, it'd be best to rename the out label to something more
> specific, i.e. out_mccic_shutdown. Either way,
>
> Reviewed-by: Sakari Ailus <sakari.ailus@xxxxxxxxxxxxxxx>
>
>> mccic_shutdown(mcam);
>> +out_v4l2_device_unregister:
>> + v4l2_device_unregister(&mcam->v4l2_dev);
>>
>> return ret;
>> }
>> @@ -293,6 +305,7 @@ static void mmpcam_remove(struct platform_device *pdev)
>> struct mcam_camera *mcam = &cam->mcam;
>>
>> mccic_shutdown(mcam);
>> + v4l2_device_unregister(&mcam->v4l2_dev);
>> pm_runtime_force_suspend(mcam->dev);
>> }
>>
>
