Re: [PATCH v10 03/14] media: iris: Add helper to create a context bank device

On 7/26/2026 11:33 AM, [email protected] wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [High] The newly created context bank device is never attached to an IOMMU domain because `of_dma_configure()` is not called.
> --
>
> --- Patch [3]: [PATCH v10 03/14] media: iris: Add helper to create a context bank device ---
> commit 461ae3ede65da976a81591c71089c7621d163869
> Author: Vikash Garodia <[email protected]>
>
> media: iris: Add helper to create a context bank device
>     
> This commit adds a helper function to create a context bank device for the
> iris driver. The subnode in the device tree is registered as a separate
> device so that it gets its own IOMMU context for mapping pixel/non-pixel
> buffers.
>
>> diff --git a/drivers/media/platform/qcom/iris/iris_resources.c b/drivers/media/platform/qcom/iris/iris_resources.c
>> index 773f6548370a2..a685915a19e9b 100644
>> --- a/drivers/media/platform/qcom/iris/iris_resources.c
>> +++ b/drivers/media/platform/qcom/iris/iris_resources.c
> [ ... ]
>
>> @@ -141,3 +142,30 @@ int iris_disable_unprepare_clock(struct iris_core *core, enum platform_clk_type
>>  
>>  	return 0;
>>  }
> [ ... ]
>
>> +	pdev = platform_device_register_full(&plat_dev_info);
>> +	of_node_put(child_of_node);
>> +	if (IS_ERR(pdev))
>> +		return ERR_CAST(pdev);
>> +
>> +	dma_set_max_seg_size(&pdev->dev, DMA_BIT_MASK(32));
>> +	dma_set_seg_boundary(&pdev->dev, DMA_BIT_MASK(32));
>> +
>> +	return &pdev->dev;
>> +}
> [Severity: High]
> Since this subnode device won't be probed by a driver (due to lacking a
> compatible string), it won't go through the device driver core's
> really_probe() which would automatically call platform_dma_configure().
>
> Will this newly created context bank device fail to attach to an IOMMU domain
> without an explicit call to of_dma_configure() here? If it isn't attached,
> could subsequent dma_map_* calls using this device bypass the IOMMU or fail?

This is not true, please refer the other path

device_add() (called by platform_device_add(), which
platform_device_register_full() invokes) unconditionally calls
bus_notify(dev, BUS_NOTIFY_ADD_DEVICE) in core.c. For any device on a
bus in iommu_buses[] (platform bus included), that fires
iommu_bus_notifier() → iommu_probe_device() → __iommu_probe_device() →
iommu_init_device(), all at device-registration time, independent of any
driver binding. And iommu_init_device() explicitly handles exactly this case
dev->bus->dma_configure(dev);