Re: [PATCH v3 1/3] dt-bindings: clock: qcom,camcc-sm8250: add interconnect provider

[Atanas Filipov]

On 10/20/2026 3:44 PM, Atanas Filipov wrote:
> On Tue, Oct 20, 2026 at 09:12:03AM +0200, Krzysztof Kozlowski wrote:
>> If you only need to solve "last writer" problem, why not fixing the
>> clock framework to accept the votes? Otherwise this should have
>> description why it is an interconnect.

A follow-up on the second part, since the problem turned out to be
more than the last writer.

The last-writer problem between the CAMSS VFEs is now handled in CAMSS
itself, in the direction you and Bryan suggested: each VFE sets a
clk_set_rate_range() floor, and on power off drops it and sets the
lowest rate explicitly, so the clock no longer stays high:

https://lore.kernel.org/all/20261028132403.3505037-1-atanas.filipov@xxxxxxxxxxxxxxxx/

What clock votes cannot express is why I think the camera NoC is an
interconnect:

 - The camera NoC is a bus with several independent masters: the VFE
   write ports, JPEG and OPE. They share it on the way to the MMSS NoC,
   and the camnoc AXI clock sets its throughput.
 - The consumers know their bandwidth, not the NoC rate that serves it.
   Converting one to the other depends on the NoC width, which is a
   property of the bus, not of each consumer.
 - Masters running at the same time load the NoC by the sum of their
   traffic. Clock rate requests can only be aggregated by maximum; a
   sum of rates has no meaning, a sum of bandwidth does.
 - The rest of the camera DMA path, from the MMSS NoC to DDR, is
   already described with interconnects, and CAMSS votes bandwidth on
   it. The camera NoC is its first hop.

The interconnect core already sums the average bandwidth of all
requests. icc-clk currently derives the rate from the aggregated peak
only, so it does not use that sum yet; that is a limitation of the
icc-clk translation, not of describing the camera NoC as an
interconnect. It can be addressed in icc-clk by taking the average
into account together with the bus width.

Is this a valid reason to describe the camera NoC as an interconnect?

--
Best regards,
Atanas