On 15/01/2024 19:17, Konrad Dybcio wrote:
> On 15.01.2024 18:59, Krzysztof Kozlowski wrote:
>> On 15/01/2024 17:44, Georgi Djakov wrote:
>>> On 15.01.24 17:34, Krzysztof Kozlowski wrote:
>>>> Revert commit 16862f1b2110 ("interconnect: qcom: sm8450: Enable
>>>> sync_state"), because it causes serial console to corrupt, later freeze
>>>> and become either entirely corrupted or only print without accepting any
>>>> input.
>>>
>>> Sounds like some driver is not requesting bandwidth and is relying on
>>> bandwidth requests made by other drivers. Maybe we are missing some
>>> "interconnects" property in DT?
>>
>> Yes, the debug UART (console) misses the interconnects. They could be
>> added but it does not change the fact that console is broken since v6.6
>> and this was probably never tested on actual hardware :/
>
> This patch? I definitely tested it out on a headless remote board..
OK, then maybe something changed between your tests and when it was
applied? Anyway issue is reproducible 100%. Even older kernel (v6.5)
with your commit cherry-picked fails.
Best regards,
Krzysztof
