Re: [PATCH v8 10/11] arm64: dts: qcom: sm8350: Add Crypto Engine support

[Krzysztof Kozlowski]

On 17/06/2023 01:33, Konrad Dybcio wrote:
> On 16.06.2023 19:36, Krzysztof Kozlowski wrote:
>> On 26/05/2023 21:22, Bhupesh Sharma wrote:
>>> Add crypto engine (CE) and CE BAM related nodes and definitions to
>>> 'sm8350.dtsi'.
>>>
>>> Tested-by: Anders Roxell <anders.roxell@xxxxxxxxxx>
>>> Tested-by: Linux Kernel Functional Testing <lkft@xxxxxxxxxx>
>>> Co-developed-by and Signed-off-by: Robert Foss <rfoss@xxxxxxxxxx>
>>> [Bhupesh: Switch to '#interconnect-cells = <2>', available since commit 4f287e31ff5f]
>>> Signed-off-by: Bhupesh Sharma <bhupesh.sharma@xxxxxxxxxx>
>>> ---
>>
>> #regzbot introduced: f1040a7fe8f069d2259ab3dab9190210005ceb33
>> #regzbot title: HDK8350 silently crashes early on boot
>>
>> Hi, this landed in the next but unfortunately it causes silent crash
>> (and reboot) of HDK8350. Reverting this commit helps.
> Downstream also references the following SIDs:
> 
> iommus = <&apps_smmu 0x592 0>,
> 	 <&apps_smmu 0x598 0>,
> 	 <&apps_smmu 0x599 0>,
> 	 <&apps_smmu 0x59F 0>;

I already tried iommus from downstream from:
1. qcrypto node (0x584, 0x594)
2. qcedev (0x586, 0x596)
3. qcom_cedev_ns_cb (0x592, 0x598, 0x599, 0x59F), although with 0x0011
last argument.

Same results, but indeed iommu would be nice reason here.

I also double checked the version of block (BAM DMA is v1.7.4) and other
properties. When I disabled crypto but left BAM DMA, the result was the
same, thus it is maybe the BAM who causes abort.


Best regards,
Krzysztof