Devicetree
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Wangao Wang" <wangao.wang@oss.qualcomm.com>
Cc: devicetree@vger.kernel.org, robh@kernel.org, conor+dt@kernel.org,
	media-ci@linuxtv.org
Subject: Re: [PATCH v3 3/5] arm64: dts: qcom: eliza: Add iris video codec node
Date: Tue, 17 Nov 2026 06:56:00 +0000	[thread overview]
Message-ID: <sashiko-outbox-218283@kernel.org> (raw)
In-Reply-To: <20261117-add_iris_for_eliza-v3-3-02a55a9272ef@oss.qualcomm.com>

Thank you for your contribution! Sashiko AI review found 3 potential
issues to consider:
- [High] arm64: dts: qcom: eliza: missing iommus on iris node breaks
  iris_probe()
- [High] arm64: dts: qcom: eliza: reversed MX and CX required-opps in
  iris_opp_table
- [High] media: iris: missing slice_caps, wd_intr_mask,
  icc_ib_multiplier in eliza_data
--

commit a19bdea4d1c4719e401fe9a32254505afdb19a2d
Author: Wangao Wang <wangao.wang@oss.qualcomm.com>

arm64: dts: qcom: eliza: Add iris video codec node

This commit adds the Iris video codec device tree node and its operating
performance points table for the Qualcomm Eliza SoC.

> diff --git a/arch/arm64/boot/dts/qcom/eliza.dtsi b/arch/arm64/boot/dts/qcom/eliza.dtsi
> index 4a125ef386c02..0c13d3f9b6c17 100644
> --- a/arch/arm64/boot/dts/qcom/eliza.dtsi
> +++ b/arch/arm64/boot/dts/qcom/eliza.dtsi
> @@ -3429,6 +3429,104 @@ gpucc: clock-controller@3d90000 {
>  			#power-domain-cells = <1>;
>  		};
>  
> +		iris: video-codec@aa00000 {
> +			compatible = "qcom,eliza-iris";

[Severity: High]
Does binding to eliza_data (introduced in commit "media: iris: Add support
for Eliza") leave .slice_caps, .wd_intr_mask, and .icc_ib_multiplier
unset?

Unlike the other VPU3x entries in
drivers/media/platform/qcom/iris/iris_platform_vpu3x.c, eliza_data omits
all three fields.  When multi-slice encoding is configured,
iris_set_slice_count() dereferences a NULL slice_caps pointer:

drivers/media/platform/qcom/iris/iris_ctrls.c:iris_set_slice_count() {
	...
	slice_caps = inst->core->iris_platform_data->slice_caps;
	...
	if (fps > slice_caps->max_slice_frame_rate ||
	...
}

Leaving .icc_ib_multiplier at 0 also causes iris_set_icc_bw() to vote 0 peak
bandwidth on video-mem, and leaving .wd_intr_mask at 0 causes
iris_vpu_watchdog() to ignore hardware watchdog interrupts.

> +			reg = <0 0xaa00000 0 0xf0000>;
> +
> +			interrupts = <GIC_SPI 174 IRQ_TYPE_LEVEL_HIGH>;
> +
> +			power-domains = <&videocc VIDEO_CC_MVS0C_GDSC>,
> +					<&videocc VIDEO_CC_MVS0_GDSC>,
> +					<&rpmhpd RPMHPD_MX>,
> +					<&rpmhpd RPMHPD_CX>;
> +			power-domain-names = "venus",
> +					     "vcodec0",
> +					     "mxc",
> +					     "mmcx";

[ ... ]

> +			#address-cells = <1>;
> +			#size-cells = <1>;
> +			dma-ranges = <0x0 0x0 0x0 0xe0000000>;
> +
> +			status = "disabled";
> +
> +			non-pixel {
> +				iommus = <&apps_smmu 0x1940 0x0>;
> +				iommu-ranges = <0x25800000 0xba800000>;
> +			};
> +
> +			pixel {
> +				iommus = <&apps_smmu 0x1947 0x0>;
> +			};

[Severity: High]
Will placing iommus inside the non-pixel and pixel child nodes instead of
on video-codec@aa00000 cause iris_probe() to fail with -ENOMEM?

The qcom,sm8550-iris binding requires iommus directly on the video-codec
node and does not define #address-cells, #size-cells, dma-ranges, or the
non-pixel and pixel child nodes.  At runtime, iris_probe() does not parse
those child nodes and instead calls iris_reserve_iova_region() ->
dma_iova_try_alloc() directly on the top-level platform device, which fails
with -ENOMEM because the device has no IOMMU attached.

> +
> +			iris_opp_table: opp-table {
> +				compatible = "operating-points-v2";
> +
> +				opp-192000000 {
> +					opp-hz = /bits/ 64 <192000000>;
> +					required-opps = <&rpmhpd_opp_low_svs_d1>,
> +							<&rpmhpd_opp_low_svs_d1>;
> +				};
> +
> +				opp-240000000 {
> +					opp-hz = /bits/ 64 <240000000>;
> +					required-opps = <&rpmhpd_opp_low_svs>,
> +							<&rpmhpd_opp_svs>;
> +				};
> +
> +				opp-338000000 {
> +					opp-hz = /bits/ 64 <338000000>;
> +					required-opps = <&rpmhpd_opp_svs>,
> +							<&rpmhpd_opp_svs>;
> +				};
> +
> +				opp-366000000 {
> +					opp-hz = /bits/ 64 <366000000>;
> +					required-opps = <&rpmhpd_opp_svs_l1>,
> +							<&rpmhpd_opp_svs>;
> +				};
> +
> +				opp-444000000 {
> +					opp-hz = /bits/ 64 <444000000>;
> +					required-opps = <&rpmhpd_opp_nom>,
> +							<&rpmhpd_opp_svs_l1>;
> +				};
> +
> +				opp-533333333 {
> +					opp-hz = /bits/ 64 <533333333>;
> +					required-opps = <&rpmhpd_opp_turbo>,
> +							<&rpmhpd_opp_svs_l1>;
> +				};
> +			};

[Severity: High]
Are the MX and CX entries in required-opps swapped relative to
power-domains?

Above, power-domains maps mxc to RPMHPD_MX and mmcx to RPMHPD_CX, and
iris_opp_pd_table_vpu3x attaches them in that order ("mxc" at index 0,
"mmcx" at index 1).  In iris_opp_table, however, the first required-opps
entry tracks the CX corner (scaling up to rpmhpd_opp_turbo) while the
second tracks MX (capping at rpmhpd_opp_svs_l1).  This under-volts MX to
low_svs at 240 MHz and under-volts CX to svs at 366 MHz and svs_l1 at
444 MHz and 533.33 MHz.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117-add_iris_for_eliza-v3-0-02a55a9272ef@oss.qualcomm.com?part=3

  reply	other threads:[~2026-11-17  6:56 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-17  6:34 [PATCH v3 0/5] media: iris: Enable Iris video codec on Eliza Wangao Wang
2026-11-17  6:34 ` [PATCH v3 1/5] dt-bindings: media: qcom,sm8550-iris: Document Eliza IRIS accelerator Wangao Wang
2026-11-17  6:34 ` [PATCH v3 2/5] media: iris: Add support for Eliza Wangao Wang
2026-11-17  6:47   ` sashiko-bot
2026-11-17  6:34 ` [PATCH v3 3/5] arm64: dts: qcom: eliza: Add iris video codec node Wangao Wang
2026-11-17  6:56   ` sashiko-bot [this message]
2026-11-17  6:34 ` [PATCH v3 4/5] arm64: dts: qcom: eliza-mtp: Enable iris video codec Wangao Wang
2026-11-17  6:52   ` sashiko-bot
2026-11-17  6:34 ` [PATCH v3 5/5] arm64: dts: qcom: eliza-evk: " Wangao Wang

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-218283@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=conor+dt@kernel.org \
    --cc=devicetree@vger.kernel.org \
    --cc=media-ci@linuxtv.org \
    --cc=robh@kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    --cc=wangao.wang@oss.qualcomm.com \
    /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