Re: [PATCH 04/22] arm64: dts: qcom: hamoa: Reserve low IOVA range for Iris

[Krzysztof Kozlowski <krzk@kernel.org>]

On 18/08/2026 10:00, Vikash Garodia wrote:
> 
> On 8/18/2026 1:21 PM, Vikash Garodia wrote:
>>
>> On 8/18/2026 11:53 AM, Dmitry Baryshkov wrote:
>>> On Mon, Aug 17, 2026 at 10:54:47AM +0100, Bryan O'Donoghue wrote:
>>>> On 10/08/2026 14:47, Rob Herring wrote:
>>>>>>> As per the reserved-memory.yaml schema document, the iommu-addresses
>>>>>>> property for IOVA reservation into /reserved-memory is allowed — it's
>>>>>>> documented for IOVA carveout reservations:
>>>>>>> https://github.com/devicetree-org/dt-schema/blob/main/dtschema/
>>>>>>> schemas/reserved-memory/reserved-memory.yaml#L56
>>>>>> I think that is an entirely reasonable reading of the text
>>>>>>
>>>>>> Quote:
>>>>>>
>>>>>> "If no "reg" property is specified, the "iommu-addresses" property
>>>>>>     defines carveout regions in the IOVA space for the given 
>>>>>> device. This
>>>>>>     can be useful if a certain memory region should not be mapped 
>>>>>> through
>>>>>>     the IOMMU."
>>>>>>
>>>>>> Which is exactly what this series does. Excludes a specific IOVA range
>>>>>> from being allocated for this device.
>>>>> But if "iommu-addresses" is the only property of the reserved memory,
>>>>> why put this property in reserved-memory rather than directly in the
>>>>> device's node? I'm only questioning where is the right place to
>>>>> describe a reserved IOVA.
>>>>>
>>>>> Rob
>>>>
>>>> Back from the eclipse, sorry for the delayed response.
>>>>
>>>> Ah I get you.
>>>>
>>>> Yes.
>>>>
>>>> I take your point. This is a property of the node itself; the IOVA < 
>>>> 600MB
>>>> only matters _if_ the venus or iris nodes are enabled - otherwise an 
>>>> IOVA <
>>>> 600MB is perfectly fine.
>>>>
>>>> So it follows its a property specific to this node not specific to the
>>>> overall system.
>>>>
>>>> Yep, I agree - its a device/node property, not a top-level property.
>>>
>>> :-)
>>>
>>> FWIW, I think it's a proper way to describe it too.
>>>
>>
>>  From my readings, a node defined under reserved-memory can have other 
>> properties like "reg" alongwith "iommu-addresses". Are we saying when 
>> *only* "iommu-addresses" is there, then move to the device and when 
>> other property is there, then move to individual device ?
> 
> sorry, typo. Pls read the last few words as ".., then keep under 
> reserve-memory ?"
> 
>>
>> Again the device may have the "reg" property already, in such case, 
>> moving the iommu-addresses + reg in corresponding device might not be 
>> feasible.

Your sub-devices do not have reg property, so there is no such case,
unless you meant the entire iris node?

Anyway, the point is that this does not describe anyhow system memory,
any reserved memory. You want to affect the device virtual memory
mappings and as such this is a property of the device itself, because
this is how we structure DT. Attributes of devices do not go to
top-level but to the devices themselves as properties.

Best regards,
Krzysztof