[PATCH v2 0/3] Add initial devicetree for Turing RK1

Sam Edwards posted 3 patches 2 years, 2 months ago
.../devicetree/bindings/arm/rockchip.yaml     |   5 +
.../devicetree/bindings/vendor-prefixes.yaml  |   2 +
arch/arm64/boot/dts/rockchip/Makefile         |   1 +
.../boot/dts/rockchip/rk3588-turing-rk1.dts   |  21 +
.../boot/dts/rockchip/rk3588-turing-rk1.dtsi  | 623 ++++++++++++++++++
5 files changed, 652 insertions(+)
create mode 100644 arch/arm64/boot/dts/rockchip/rk3588-turing-rk1.dts
create mode 100644 arch/arm64/boot/dts/rockchip/rk3588-turing-rk1.dtsi
[PATCH v2 0/3] Add initial devicetree for Turing RK1
Posted by Sam Edwards 2 years, 2 months ago
Hi again list,

This is the second version of my patch to bring in support for the RK3588-based
Turing RK1 SoM. In my previous cover letter, I perhaps should have specified
that the RK1 is a little bit unusual in that, though it *is* a true SoM, it is
targeted toward home-hosting/edge users directly as a compute node, and as a
result the vast majority of users will be seeing it more like a
micro-bladeserver, rather than an off-the-shelf part meant to power a larger
system. This was my rationale for previously sending this as a single .dts,
targeting that use case.

However, Heiko previously made a good point that it still depends on a carrier
board to be "complete," and then I reminded myself that not all users will be
treating the RK1 like a mere "node" -- some will be incorporating them into
larger carriers, which the RK1 is meant to enable (not the other way around).

This version tries to strike a compromise, by moving the SoM-specific stuff to
a .dtsi, introducing a .dts specifically for the "opinionated" view of the RK1
as a carrier-agnostic node, and adding another paragraph to the PATCH 3/3
changelog explaining that.

I am wondering if there will be an objection about the .dts having the exact
same base filename and `compatible` as the .dtsi. I am not sure what word/term
to use to mean "RK1 but thought of as the system itself and not as a piece of a
system," but am open to suggestions. :)

Cheers all,
Sam

Sam Edwards (3):
  dt-bindings: vendor-prefixes: add turing
  dt-bindings: arm: rockchip: Add Turing RK1
  arm64: dts: rockchip: Add Turing RK1 SoM support

 .../devicetree/bindings/arm/rockchip.yaml     |   5 +
 .../devicetree/bindings/vendor-prefixes.yaml  |   2 +
 arch/arm64/boot/dts/rockchip/Makefile         |   1 +
 .../boot/dts/rockchip/rk3588-turing-rk1.dts   |  21 +
 .../boot/dts/rockchip/rk3588-turing-rk1.dtsi  | 623 ++++++++++++++++++
 5 files changed, 652 insertions(+)
 create mode 100644 arch/arm64/boot/dts/rockchip/rk3588-turing-rk1.dts
 create mode 100644 arch/arm64/boot/dts/rockchip/rk3588-turing-rk1.dtsi

-- 
2.41.0
Re: [PATCH v2 0/3] Add initial devicetree for Turing RK1
Posted by Heiko Stuebner 2 years, 2 months ago
On Wed, 11 Oct 2023 16:58:20 -0600, Sam Edwards wrote:
> This is the second version of my patch to bring in support for the RK3588-based
> Turing RK1 SoM. In my previous cover letter, I perhaps should have specified
> that the RK1 is a little bit unusual in that, though it *is* a true SoM, it is
> targeted toward home-hosting/edge users directly as a compute node, and as a
> result the vast majority of users will be seeing it more like a
> micro-bladeserver, rather than an off-the-shelf part meant to power a larger
> system. This was my rationale for previously sending this as a single .dts,
> targeting that use case.
> 
> [...]

Applied, thanks!

[1/3] dt-bindings: vendor-prefixes: add turing
      commit: 817bacc3a648cc55a0b07a699c03ecc70309ae50
[2/3] dt-bindings: arm: rockchip: Add Turing RK1
      commit: e30ecfcbe4ed3706af67dff5aa1418fba6ba2c29
[3/3] arm64: dts: rockchip: Add Turing RK1 SoM support
      commit: 2806a69f3fef61d7353ea8206add8ffb15064b51

I've dropped the mem-supply references from the cpu nodes.
I think some at Collabora is working on upstreaming the
cpufreq driver that will utilize those.

Until then they are not part of the binding, so please add
them via a new patch once the cpufreq support has landed.


Best regards,
-- 
Heiko Stuebner <heiko@sntech.de>