Re: [RFC PATCH 10/11] arm64: dts: qcom: monaco: Move Iris IOMMUs to child nodes

[Vikash Garodia <vikash.garodia@oss.qualcomm.com>]

On 7/9/2026 8:33 PM, sashiko-bot@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [High] Misuse of `iommu-addresses` in a `reserved-memory` node creates an `IOMMU_RESV_RESERVED` region that actively blocks IOVA allocations in the 0-600MB range, rather than restricting allocations to it.
> --
> 
> --- Patch [10]: [RFC PATCH 10/11] arm64: dts: qcom: monaco: Move Iris IOMMUs to child nodes ---
> commit 2ee0f65b225528a8b439c85350380c9a56ad101f
> Author: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
> 
> arm64: dts: qcom: monaco: Move Iris IOMMUs to child nodes
> 
> This commit splits the Iris VPU device into non-pixel and pixel child nodes
> to accommodate different IOVA constraints. It adds a reserved IOVA range
> intending to limit the non-pixel stream to the 0-600 MB range.
> 
>> diff --git a/arch/arm64/boot/dts/qcom/monaco.dtsi b/arch/arm64/boot/dts/qcom/monaco.dtsi
>> --- a/arch/arm64/boot/dts/qcom/monaco.dtsi
>> +++ b/arch/arm64/boot/dts/qcom/monaco.dtsi
>> @@ -831,6 +831,10 @@ video_mem: video-region@9be82000 {
>>   			reg = <0x0 0x9be82000 0x0 0x700000>;
>>   			no-map;
>>   		};
>> +
>> +		iris_resv: reservation-iris {
>> +			iommu-addresses = <&iris_non_pixel 0x0 0x0 0x0 0x25800000>;
>> +		};
> 
> [Severity: High]
> Does this iommu-addresses property block allocations in the 0-600MB range
> instead of restricting allocations to it?
> 
> During device probe, of_iommu_get_resv_regions() parses this memory-region.
> Because iris_resv has an iommu-addresses property but lacks a physical reg
> property, it is classified as IOMMU_RESV_RESERVED.
> 
> The IOMMU DMA allocator then passes this range to reserve_iova(), which
> explicitly excludes the 0-600MB range from the available IOVA pool.
> 
> Will this force subsequent allocations for the non-pixel stream into addresses
> above 600MB, resulting in the exact IOMMU faults and device crashes this
> patch intends to fix?
> 

Yes, thats the intended behavior too. The "non-pixel" sub node is 
suppose to skip the region 0-600MB entirely, and return IOVAs above and 
outside the range 0-600MB.

Regards,
Vikash