[PATCH] dt-bindings: net: wireless: qcom,ath10k: add PCI devices

[PATCH] dt-bindings: net: wireless: qcom,ath10k: add PCI devices

[Rosen Penev]

ath10k PCI devices can be described as children of their PCI host
bridge, for example to supply the MAC address and calibration data
through nvmem cells. The only compatible for that was "qcom,ath10k",
but the generic PCI device schema requires a node under a PCI bus to
carry a "pciVVVV,DDDD" or "pciclass,..." compatible, so such nodes fail
validation.

Add the PCI vendor/device IDs handled by the ath10k PCI driver, as the
ath9k and ath11k PCI bindings already do.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
---
 .../devicetree/bindings/net/wireless/qcom,ath10k.yaml    | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/Documentation/devicetree/bindings/net/wireless/qcom,ath10k.yaml b/Documentation/devicetree/bindings/net/wireless/qcom,ath10k.yaml
index 0cb065265df4..992926ce5c5e 100644
--- a/Documentation/devicetree/bindings/net/wireless/qcom,ath10k.yaml
+++ b/Documentation/devicetree/bindings/net/wireless/qcom,ath10k.yaml
@@ -15,6 +15,15 @@ description:
 properties:
   compatible:
     enum:
+      - pci0777,11ac  # QCA988X (Ubiquiti branded)
+      - pci168c,003c  # QCA988X
+      - pci168c,003e  # QCA6174
+      - pci168c,0040  # QCA99X0
+      - pci168c,0041  # QCA6164
+      - pci168c,0042  # QCA9377
+      - pci168c,0046  # QCA9984
+      - pci168c,0050  # QCA9887
+      - pci168c,0056  # QCA9888
       - qcom,ath10k # SDIO-based devices
       - qcom,ipq4019-wifi
       - qcom,wcn3990-wifi # SNoC-based devices
-- 
2.56.0

Re: [PATCH] dt-bindings: net: wireless: qcom,ath10k: add PCI devices

[Krzysztof Kozlowski]

On Tue, Nov 10, 2026 at 07:44:18PM -0800, Rosen Penev wrote:
> ath10k PCI devices can be described as children of their PCI host
> bridge, for example to supply the MAC address and calibration data
> through nvmem cells. The only compatible for that was "qcom,ath10k",
> but the generic PCI device schema requires a node under a PCI bus to
> carry a "pciVVVV,DDDD" or "pciclass,..." compatible, so such nodes fail
> validation.
> 
> Add the PCI vendor/device IDs handled by the ath10k PCI driver, as the
> ath9k and ath11k PCI bindings already do.

We do not need to least all PCI devices here. Were is any actual DTS
user of these?

Best regards,
Krzysztof

Re: [PATCH] dt-bindings: net: wireless: qcom,ath10k: add PCI devices

[Rosen Penev]

On Fri, Nov 13, 2026 at 3:11 AM Krzysztof Kozlowski <krzk@kernel.org> wrote:
>
> On Tue, Nov 10, 2026 at 07:44:18PM -0800, Rosen Penev wrote:
> > ath10k PCI devices can be described as children of their PCI host
> > bridge, for example to supply the MAC address and calibration data
> > through nvmem cells. The only compatible for that was "qcom,ath10k",
> > but the generic PCI device schema requires a node under a PCI bus to
> > carry a "pciVVVV,DDDD" or "pciclass,..." compatible, so such nodes fail
> > validation.
> >
> > Add the PCI vendor/device IDs handled by the ath10k PCI driver, as the
> > ath9k and ath11k PCI bindings already do.
>
> We do not need to least all PCI devices here. Were is any actual DTS
> user of these?
I have some WIP patches that include the QCA988X. Reason for it is
nvram calibration and MAC lookup.

This is basically copy/paste of what's already present in ath9k.yaml.

AFAIK this came up because of dtbs_check
>
> Best regards,
> Krzysztof
>

Re: [PATCH] dt-bindings: net: wireless: qcom,ath10k: add PCI devices

[Krzysztof Kozlowski]

On 13/11/2026 22:34, Rosen Penev wrote:
> On Fri, Nov 13, 2026 at 3:11 AM Krzysztof Kozlowski <krzk@kernel.org> wrote:
>>
>> On Tue, Nov 10, 2026 at 07:44:18PM -0800, Rosen Penev wrote:
>>> ath10k PCI devices can be described as children of their PCI host
>>> bridge, for example to supply the MAC address and calibration data
>>> through nvmem cells. The only compatible for that was "qcom,ath10k",
>>> but the generic PCI device schema requires a node under a PCI bus to
>>> carry a "pciVVVV,DDDD" or "pciclass,..." compatible, so such nodes fail
>>> validation.
>>>
>>> Add the PCI vendor/device IDs handled by the ath10k PCI driver, as the
>>> ath9k and ath11k PCI bindings already do.
>>
>> We do not need to least all PCI devices here. Were is any actual DTS
>> user of these?
> I have some WIP patches that include the QCA988X. Reason for it is
> nvram calibration and MAC lookup.
> 
> This is basically copy/paste of what's already present in ath9k.yaml.
> 
> AFAIK this came up because of dtbs_check

It cannot come as dtbs_check result, because there are no DTS with these.

Best regards,
Krzysztof