Re: [PATCH 2/2] arm64: dts: mediatek: mt8365: Describe infracfg-nao as a pure syscon
From: David Lechner <dlechner@baylibre.com>
Date: 2026-02-02 17:20:32
Also in:
linux-devicetree, linux-mediatek, lkml
On 5/2/25 11:43 AM, Nícolas F. R. A. Prado wrote:
quoted hunk ↗ jump to hunk
The infracfg-nao register space at 0x1020e000 has different registers than the infracfg space at 0x10001000, and most importantly, doesn't contain any clock controls. Therefore it shouldn't use the same compatible used for the mt8365 infracfg clocks driver: mediatek,mt8365-infracfg. Since it currently does, probe errors are reported in the kernel logs: [ 0.245959] Failed to register clk ifr_pmic_tmr: -EEXIST [ 0.245998] clk-mt8365 1020e000.infracfg: probe with driver clk-mt8365 failed with error -17 This register space is used only as a syscon for bus control by the power domain controller, so in order to properly describe it and fix the errors, set its compatible to a distinct compatible used exclusively as a syscon, drop the clock-cells, and while at it rename the node to 'syscon' following the naming convention. Fixes: 6ff945376556 ("arm64: dts: mediatek: Initial mt8365-evk support") Signed-off-by: Nícolas F. R. A. Prado <redacted> --- arch/arm64/boot/dts/mediatek/mt8365.dtsi | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)diff --git a/arch/arm64/boot/dts/mediatek/mt8365.dtsi b/arch/arm64/boot/dts/mediatek/mt8365.dtsi index e6d2b3221a3b7a855129258b379ae4bc2fd05449..49ad4dee9c4cf563743dc55d5e0b055cfb69986a 100644 --- a/arch/arm64/boot/dts/mediatek/mt8365.dtsi +++ b/arch/arm64/boot/dts/mediatek/mt8365.dtsi@@ -495,10 +495,9 @@ iommu: iommu@10205000 { #iommu-cells = <1>; }; - infracfg_nao: infracfg@1020e000 { - compatible = "mediatek,mt8365-infracfg", "syscon"; + infracfg_nao: syscon@1020e000 { + compatible = "mediatek,mt8365-infracfg-nao", "syscon"; reg = <0 0x1020e000 0 0x1000>; - #clock-cells = <1>; }; rng: rng@1020f000 {
Reviewed-by: David Lechner <dlechner@baylibre.com> It looks like this never got picked up. I noticed this was a problem in U-Boot because it was registering this as a clock provider. And I sent a similar patch [1] recently that has also not been acted on yet. I prefer this patch since it also fixes the node name to use a standard name. Who should be responsible for actually picking up the patch? [1]: https://lore.kernel.org/linux-mediatek/20251216-mtk-fix-infracfg_nao-compatibile-v1-1-d339b151ac81@baylibre.com/ (local)