Devicetree
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Luca Leonardo Scorcia" <l.scorcia@gmail.com>
Cc: devicetree@vger.kernel.org, robh@kernel.org, conor+dt@kernel.org
Subject: Re: [PATCH] arm64: dts: mediatek: mt8167: Add DRM nodes
Date: Tue, 17 Nov 2026 13:00:03 +0000	[thread overview]
Message-ID: <sashiko-outbox-218823@kernel.org> (raw)
In-Reply-To: <20261117122802.728987-1-l.scorcia@gmail.com>

Thank you for your contribution! Sashiko AI review found 4 potential
issues to consider:
- [High] arm64: dts: mediatek: mt8167: missing reg = <1> on rdma1
  endpoints breaks CRTC_EXT
- [Medium] arm64: dts: mediatek: mt8167: undocumented
  mediatek,mt8167-dpi compatible
- [Medium] arm64: dts: mediatek: mt8167: unconnected rdma0_out breaks
  mtk_drm_probe()
- [Low] arm64: dts: mediatek: mt8167: non-generic display node names
  with indices
--

commit 550fb79e6d290ccbb35fb88f6c76ac7ea3a27e5b
Author: Luca Leonardo Scorcia <l.scorcia@gmail.com>

arm64: dts: mediatek: mt8167: Add DRM nodes

This commit adds the devicetree nodes and aliases for the display subsystem
on the MediaTek MT8167 SoC.

> diff --git a/arch/arm64/boot/dts/mediatek/mt8167.dtsi b/arch/arm64/boot/dts/mediatek/mt8167.dtsi
> index 27cf32d7ae35f..2bcdab2934f4f 100644
> --- a/arch/arm64/boot/dts/mediatek/mt8167.dtsi
> +++ b/arch/arm64/boot/dts/mediatek/mt8167.dtsi
[ ... ]
> @@ -120,10 +136,325 @@ iommu: m4u@10203000 {
[ ... ]
>  		mmsys: syscon@14000000 {
>  			compatible = "mediatek,mt8167-mmsys", "syscon";
>  			reg = <0 0x14000000 0 0x1000>;
> +			power-domains = <&spm MT8167_POWER_DOMAIN_MM>;
>  			#clock-cells = <1>;
> +
> +			port {
> +				#address-cells = <1>;
> +				#size-cells = <0>;
> +
> +				mmsys_main: endpoint@0 {
> +					reg = <0>;
> +					remote-endpoint = <&ovl0_in>;
> +				};
> +
> +				mmsys_ext: endpoint@1 {
> +					reg = <1>;
> +					remote-endpoint = <&rdma1_in>;
> +				};
> +			};
> +		};
> +
> +		ovl0: ovl0@14007000 {

[Severity: Low]
This isn't a bug, but ovl0@14007000, rdma0@14009000, rdma1@1400a000,
wdma0@1400b000, and dpi1@14019000 include instance numbers in their node
names.  Should these use generic node names without trailing indices
(ovl@14007000, rdma@14009000, rdma@1400a000, wdma@1400b000, and
dpi@14019000) to match dpi0 (dpi@14013000), the MediaTek display bindings,
and other MediaTek devicetrees?

[ ... ]

> +		rdma0: rdma0@14009000 {
> +			compatible = "mediatek,mt8167-disp-rdma", "mediatek,mt2701-disp-rdma";
> +			reg = <0 0x14009000 0 0x1000>;
> +			clocks = <&mmsys CLK_MM_DISP_RDMA0>;
> +			interrupts = <GIC_SPI 162 IRQ_TYPE_LEVEL_LOW>;
> +			iommus = <&iommu M4U_PORT_DISP_RDMA0>;
> +			power-domains = <&spm MT8167_POWER_DOMAIN_MM>;
> +
> +			ports {
> +				#address-cells = <1>;
> +				#size-cells = <0>;
> +
> +				port@0 {
> +					reg = <0>;
> +					rdma0_in: endpoint {
> +						remote-endpoint = <&dither_out>;
> +					};
> +				};
> +
> +				port@1 {
> +					reg = <1>;
> +					rdma0_out: endpoint { };
> +				};
> +			};
> +		};

[Severity: Medium]
Does leaving the main pipeline (ovl0 through rdma0) enabled while rdma0_out
has no remote-endpoint cause mtk_drm_probe() to fail on boards that do not
connect rdma0_out?

When mtk_drm_of_ddp_path_build_one() walks the main pipeline from
mmsys_main, it reaches rdma0_out, where of_graph_get_remote_port_parent()
returns null and mtk_drm_of_get_ddp_ep_cid() fails with an invalid-argument
error instead of a no-such-device error:

drivers/gpu/drm/mediatek/mtk_drm_drv.c:mtk_drm_of_ddp_path_build_one() {
	...
	if (ret == -ENODEV)
		return ret;

	/* If the last entry is not a final display output, the configuration is wrong */
	switch (temp_path[idx - 1]) {
	case DDP_COMPONENT_DP_INTF0:
	case DDP_COMPONENT_DP_INTF1:
	case DDP_COMPONENT_DPI0:
	case DDP_COMPONENT_DPI1:
	case DDP_COMPONENT_DSI0:
	case DDP_COMPONENT_DSI1:
	case DDP_COMPONENT_DSI2:
	case DDP_COMPONENT_DSI3:
		break;
	default:
		dev_err(dev, "Invalid display hw pipeline. Last component: %d (ret=%d)\n",
			temp_path[idx - 1], ret);
		return -EINVAL;
	}
}

Because rdma0 is not a final display output, mtk_drm_of_ddp_path_build()
aborts mtk_drm_probe() with an error on boards without a main display (such
as mt8167-pumpkin.dts) and before evaluating the external pipeline on
boards that only use rdma1.

> +
> +		rdma1: rdma1@1400a000 {
> +			compatible = "mediatek,mt8167-disp-rdma", "mediatek,mt2701-disp-rdma";
> +			reg = <0 0x1400a000 0 0x1000>;
> +			clocks = <&mmsys CLK_MM_DISP_RDMA1>;
> +			interrupts = <GIC_SPI 163 IRQ_TYPE_LEVEL_LOW>;
> +			iommus = <&iommu M4U_PORT_DISP_RDMA1>;
> +			power-domains = <&spm MT8167_POWER_DOMAIN_MM>;
> +			status = "disabled";
> +
> +			ports {
> +				#address-cells = <1>;
> +				#size-cells = <0>;
> +
> +				port@0 {
> +					reg = <0>;
> +					rdma1_in: endpoint {
> +						remote-endpoint = <&mmsys_ext>;
> +					};
> +				};
> +
> +				port@1 {
> +					reg = <1>;
> +					rdma1_out: endpoint { };
> +				};
> +			};
> +		};

[Severity: High]
Does omitting reg = <1> on the rdma1 endpoints break the external display
pipeline whenever rdma1 is enabled?

mmsys_ext has reg = <1>, so mtk_drm_of_ddp_path_build() builds that path
with crtc_path set to 1.  When mtk_drm_of_ddp_path_build_one() steps from
rdma1 to its output port, mtk_drm_of_get_ddp_ep_cid() looks up the endpoint
whose reg matches crtc_path:

drivers/gpu/drm/mediatek/mtk_drm_drv.c:mtk_drm_of_get_ddp_ep_cid() {
	...
	ep_out = of_graph_get_endpoint_by_regs(node, output_port, crtc_path);
	if (!ep_out)
		return -ENOENT;
}

Because rdma1_out has no reg property, its endpoint id defaults to 0, so
of_graph_get_endpoint_by_regs(rdma1, 1, 1) returns null and
mtk_drm_of_ddp_path_build_one() fails with an invalid pipeline error.
Adding #address-cells = <1> and #size-cells = <0> to port@0 and port@1 and
using endpoint@1 with reg = <1> on rdma1_in and rdma1_out (as in
mt8365.dtsi and mt8183.dtsi) would allow the driver to follow the external
path.

[ ... ]

> +		dpi0: dpi@14013000 {
> +			compatible = "mediatek,mt8167-dpi";
> +			reg = <0 0x14013000 0 0x1000>;
> +			clocks = <&mmsys CLK_MM_DPI0_PXL>, <&mmsys CLK_MM_DPI0_ENGINE>,
> +				 <&apmixedsys CLK_APMIXED_LVDSPLL>;
> +			clock-names = "pixel", "engine", "pll";
> +			interrupts = <GIC_SPI 172 IRQ_TYPE_LEVEL_LOW>;
> +			power-domains = <&spm MT8167_POWER_DOMAIN_MM>;
> +			status = "disabled";

[Severity: Medium]
Is "mediatek,mt8167-dpi" missing from the devicetree bindings and MediaTek
drm driver in this tree, or should dpi0 and dpi1 include a fallback
compatible?

Without a matching entry in Documentation/devicetree/bindings/display/
mediatek/mediatek,dpi.yaml, mtk_dpi_of_ids, and mtk_ddp_comp_dt_ids,
dtbs_check fails on both nodes and mtk_drm_of_get_ddp_comp_type() rejects
the pipeline when either dpi node is enabled.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117122802.728987-1-l.scorcia@gmail.com?part=1

      parent reply	other threads:[~2026-11-17 13:00 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-17 12:26 [PATCH] arm64: dts: mediatek: mt8167: Add DRM nodes Luca Leonardo Scorcia
2026-11-17 12:31 ` Krzysztof Kozlowski
2026-11-17 13:00 ` sashiko-bot [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=sashiko-outbox-218823@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=conor+dt@kernel.org \
    --cc=devicetree@vger.kernel.org \
    --cc=l.scorcia@gmail.com \
    --cc=robh@kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox