Re: [PATCH v4 1/2] dt-bindings: interconnect: qcom: Add SM7250 RPMh NoC

[Ishu Brar]

On Tue, 27 Oct 2026 19:38:34 +0100, Krzysztof Kozlowski wrote:
To: krzk@xxxxxxxxxx
Cc: djakov@xxxxxxxxxx, robh@xxxxxxxxxx, krzk+dt@xxxxxxxxxx, conor+dt@xxxxxxxxxx, linux-arm-msm@xxxxxxxxxxxxxxx, linux-pm@xxxxxxxxxxxxxxx, konradybcio@xxxxxxxxxx, devicetree@xxxxxxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx
Subject: Re: [PATCH v4 1/2] dt-bindings: interconnect: qcom: Add SM7250 RPMh NoC

On Tue, 27 Oct 2026 19:38:34 +0100, Krzysztof Kozlowski wrote:
> Cover letter is missing,

Sorry, my mail setup rewrote the Message-IDs, so the patches lost the
link to the cover. I will fix that and resend.

> but you also never responded what is the actual hardware here - why
> sometimes you have children, sometimes not. How is it supposed to work?

compute-noc has no register block of its own: its QoS registers sit
inside the aggre2 NoC region (0x1700000, at offset 0x9000). mc-virt
and qup-virt are virtual providers with no registers at all, only BCM
votes; they belong to gem-noc and system-noc. So all three depend on
the parent's register space, which is why they are child nodes.
v1 and v2 had them as children; v3 flattened them because I misread
your comment, and v4 restores that.

Ishu Brar