[PATCH RFC 0/3] Allow DT boards to disable WCN7850 hardware rfkill

[PATCH RFC 0/3] Allow DT boards to disable WCN7850 hardware rfkill

[Franz]

On the Microsoft Surface Pro 11 (X1E80100, board "denali"), the WCN7850
hardware rfkill input reads as "radio off". ath12k turns on hardware
rfkill handling, so Wi-Fi stays hard-blocked:

  iwd: Error bringing interface 4 up: Operation not possible due to RF-kill

ACPI platforms can already turn this off (ath12k_acpi_get_disable_rfkill()),
but DT-based boards have no way to. This series adds a boolean
"disable-rfkill" property to the ath12k binding, makes ath12k honour
it, and sets it on the Surface Pro 11. Out-of-tree Surface Pro 11
kernels already ship an equivalent change.

This is an RFC mainly because of the property itself. Is a generic
"disable-rfkill" acceptable, or would you prefer a vendor-prefixed
name, or a different way to describe a board whose rfkill line is not
usable?

Testing: next-20260929 on a Surface Pro 11, with the ath12k change and
the property added to the installed DTB with fdtput. The interface
comes up and connects (5 GHz, 160 MHz). The DTS change builds, and the
built DTB carries the property. dt_binding_check / dtbs_check were not
run, because dtschema is not installed here.

On this machine the endpoint is only enumerated with "power: sequencing:
qcom-wcn: power off WLAN at probe", sent separately as a fix for
b921aa3f8dec.

This series was prepared with the help of an AI coding assistant
(Claude): finding the cause, writing the code and the changelogs. I
reviewed it and tested it on the hardware.

Franz (3):
  dt-bindings: net: wireless: qcom,ath12k: add disable-rfkill
  wifi: ath12k: honour disable-rfkill from the device tree
  arm64: dts: qcom: x1-microsoft-denali: disable WCN7850 hardware rfkill

 .../devicetree/bindings/net/wireless/qcom,ath12k.yaml      | 7 +++++++
 arch/arm64/boot/dts/qcom/x1-microsoft-denali.dtsi          | 2 ++
 drivers/net/wireless/ath/ath12k/core.c                     | 3 +++
 3 files changed, 12 insertions(+)


base-commit: 6474fa070f2b8013b4b87350b775b8c3be6e8aac
prerequisite-patch-id: aafe3fac38f4020441979665509928f8cc832095
-- 
2.56.0

[PATCH RFC 1/3] dt-bindings: net: wireless: qcom,ath12k: add disable-rfkill

[Franz]

Some boards do not wire the WCN7850 hardware rfkill input, or wire it so
that it always reads as "radio off". On the Microsoft Surface Pro 11
(X1E80100), ath12k enables hardware rfkill handling and the interface
ends up hard-blocked for good.

ACPI platforms can already turn this off through a firmware setting
that ath12k reads (ath12k_acpi_get_disable_rfkill()). Add a boolean
property so that DT-based boards can describe the same thing.

Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Franz <franzelfranzel@gmail.com>
---
 .../devicetree/bindings/net/wireless/qcom,ath12k.yaml      | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/Documentation/devicetree/bindings/net/wireless/qcom,ath12k.yaml b/Documentation/devicetree/bindings/net/wireless/qcom,ath12k.yaml
index dc68dd599..d56050e77 100644
--- a/Documentation/devicetree/bindings/net/wireless/qcom,ath12k.yaml
+++ b/Documentation/devicetree/bindings/net/wireless/qcom,ath12k.yaml
@@ -27,6 +27,13 @@ properties:
       If present, a board or platform specific string used to lookup
       usecase-specific firmware files for the device.
 
+  disable-rfkill:
+    type: boolean
+    description:
+      The device's hardware rfkill input is not usable on this board (for
+      example, it is not wired and reads as "radio off"). The driver must
+      not enable hardware rfkill handling.
+
   vddaon-supply:
     description: VDD_AON supply regulator handle
 
-- 
2.56.0

[PATCH RFC 2/3] wifi: ath12k: honour disable-rfkill from the device tree

[Franz]

ath12k_core_rfkill_config() skips the hardware rfkill setup only when
ACPI says so. DT-based boards have no equivalent. On the Microsoft
Surface Pro 11 (X1E80100), the WCN7850 rfkill input reads as "radio
off", so the interface is hard-blocked:

  iwd: Error bringing interface 4 up: Operation not possible due to RF-kill

Also skip the setup when the device's DT node has "disable-rfkill". The
PCI core attaches the node to the pci_dev, so ab->dev->of_node sees it.

Tested on a Surface Pro 11 with next-20260929 and the matching DT
change: the interface comes up and connects (5 GHz, 160 MHz).

Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Franz <franzelfranzel@gmail.com>
---
 drivers/net/wireless/ath/ath12k/core.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/net/wireless/ath/ath12k/core.c b/drivers/net/wireless/ath/ath12k/core.c
index 0c72948f3..8cac84917 100644
--- a/drivers/net/wireless/ath/ath12k/core.c
+++ b/drivers/net/wireless/ath/ath12k/core.c
@@ -83,6 +83,9 @@ static int ath12k_core_rfkill_config(struct ath12k_base *ab)
 	if (ath12k_acpi_get_disable_rfkill(ab))
 		return 0;
 
+	if (of_property_read_bool(ab->dev->of_node, "disable-rfkill"))
+		return 0;
+
 	for (i = 0; i < ab->num_radios; i++) {
 		ar = ab->pdevs[i].ar;
 
-- 
2.56.0

[PATCH RFC 3/3] arm64: dts: qcom: x1-microsoft-denali: disable WCN7850 hardware rfkill

[Franz]

The WCN7850 hardware rfkill input reads as "radio off" on the Surface
Pro 11, which keeps Wi-Fi hard-blocked. Mark it as unusable.

Tested by adding the property to the installed DTB with fdtput, on
next-20260929 with the matching ath12k change.

Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Franz <franzelfranzel@gmail.com>
---
 arch/arm64/boot/dts/qcom/x1-microsoft-denali.dtsi | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/arch/arm64/boot/dts/qcom/x1-microsoft-denali.dtsi b/arch/arm64/boot/dts/qcom/x1-microsoft-denali.dtsi
index a71c29762..36592b8ed 100644
--- a/arch/arm64/boot/dts/qcom/x1-microsoft-denali.dtsi
+++ b/arch/arm64/boot/dts/qcom/x1-microsoft-denali.dtsi
@@ -1009,6 +1009,8 @@ wifi@0 {
 		vddrfa1p8-supply = <&vreg_pmu_rfa_1p8>;
 		vddpcie0p9-supply = <&vreg_pmu_pcie_0p9>;
 		vddpcie1p8-supply = <&vreg_pmu_pcie_1p8>;
+
+		disable-rfkill;
 	};
 };
 
-- 
2.56.0

Re: [PATCH RFC 1/3] dt-bindings: net: wireless: qcom,ath12k: add disable-rfkill

[Krzysztof Kozlowski]

On 03/10/2026 11:15, Franz wrote:
> Some boards do not wire the WCN7850 hardware rfkill input, or wire it so
> that it always reads as "radio off". On the Microsoft Surface Pro 11
> (X1E80100), ath12k enables hardware rfkill handling and the interface
> ends up hard-blocked for good.
> 
> ACPI platforms can already turn this off through a firmware setting
> that ath12k reads (ath12k_acpi_get_disable_rfkill()). Add a boolean
> property so that DT-based boards can describe the same thing.
> 
> Assisted-by: Claude:claude-opus-5-5
> Signed-off-by: Franz <franzelfranzel@gmail.com>

Please do not duplicate work but build on top or reference previous
discussions.

https://lore.kernel.org/all/20251220-surface-sp11-for-next-v5-3-16065bef8ef3@gmail.com/
https://lore.kernel.org/all/20251201011457.17422-7-daleyo@gmail.com/
and some more

Best regards,
Krzysztof

Re: [PATCH RFC 2/3] wifi: ath12k: honour disable-rfkill from the device tree

[Krzysztof Kozlowski]

On 03/10/2026 11:15, Franz wrote:
> ath12k_core_rfkill_config() skips the hardware rfkill setup only when
> ACPI says so. DT-based boards have no equivalent. On the Microsoft
> Surface Pro 11 (X1E80100), the WCN7850 rfkill input reads as "radio
> off", so the interface is hard-blocked:
> 
>   iwd: Error bringing interface 4 up: Operation not possible due to RF-kill
> 
> Also skip the setup when the device's DT node has "disable-rfkill". The
> PCI core attaches the node to the pci_dev, so ab->dev->of_node sees it.
> 
> Tested on a Surface Pro 11 with next-20260929 and the matching DT
> change: the interface comes up and connects (5 GHz, 160 MHz).
> 
> Assisted-by: Claude:claude-opus-5-5

LLM, see docs.


> Signed-off-by: Franz <franzelfranzel@gmail.com>

Please read submitting patches document carefully to understand what is
needed when sending patch, e.g. what you need to certify with your full
name.

Your LLM cannot certify anything.

Best regards,
Krzysztof