linux-kernel - Re: [RFC PATCH 2/6] dt-bindings: net: wireless: ath12k: describe WSI property for QCN9274

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <88489191-2dbd-4018-b35b-84f43f8eb55e@quicinc.com>
Date: Fri, 25 Oct 2024 15:36:55 +0530
From: Raj Kumar Bhagat <quic_rajkbhag@...cinc.com>
To: Krzysztof Kozlowski <krzk@...nel.org>, <ath12k@...ts.infradead.org>
CC: <linux-wireless@...r.kernel.org>, Kalle Valo <kvalo@...nel.org>,
        Rob
 Herring <robh@...nel.org>,
        Krzysztof Kozlowski <krzk+dt@...nel.org>,
        Conor
 Dooley <conor+dt@...nel.org>,
        Jeff Johnson <jjohnson@...nel.org>,
        Bjorn
 Andersson <andersson@...nel.org>,
        Konrad Dybcio <konradybcio@...nel.org>, <devicetree@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>, <linux-arm-msm@...r.kernel.org>
Subject: Re: [RFC PATCH 2/6] dt-bindings: net: wireless: ath12k: describe WSI
 property for QCN9274

On 10/23/2024 11:27 PM, Krzysztof Kozlowski wrote:
> On 23/10/2024 14:22, Raj Kumar Bhagat wrote:
>> The above three blocks represent the QCN9274 WiFi devices connected to their
>> respective PCI slots. The dotted line represents the WSI connection that connects
>> these three devices together. Hence, the WSI interface is part of the QCN9274 device.
>>
>> To describe this WSI hardware connection in the device tree, we are adding three
>> properties inside the WSI object:
>>
>> 1. qcom,wsi-group-id:
>>    In the above diagram, we have one WSI connection connecting all three devices.
>>    Hence, “qcom,wsi-group-id” for all three devices can be 0.
>>
>>    This cannot be implied by the compatible property, as explained below:
>>    Let’s take the case of a platform that can have four QCN9274 WiFi devices. Below
>>    is one possibility of a WSI connection:
>>
>>          +-------+       +-------+          +-------+      +-------+
>>          | pcie2 |       | pcie3 |          | pcie1 |      | pcie0 |
>>          |       |       |       |          |       |      |       |
>>    +---->|  wsi  |------>|  wsi  |--+   +-->|  wsi  |----->|  wsi  |----+
>>    |     | idx 0 |       | idx 1 |  |   |   | idx 0 |      | idx 1 |    |
>>    |     +-------+       +-------+  |   |   +-------+      +-------+    |
>>    +--------------------------------+   +-------------------------------+
>>
>>    In this case, QCN9274 devices connected in PCIe2 and PCIe3 will have the same
>>    “qcom,wsi-group-id”. This group-id will be different from the “qcom,wsi-group-id”
>>    of QCN9274 devices connected at PCIe1 and PCIe0.
> Thanks, this explains why group-id cannot be same...
> 
>> 2. qcom,wsi-index:
>>    This is a unique identifier of the device within the same group. The value of
>>    wsi-idx is represented in both the above cases (RDP433 and the 4 WiFi device
>>    platform) in the diagram itself.
> But still any device-indexing is in general not accepted (and was
> mentioned during reviews multiple times).
> 
> This looks like circular list, so phandle will be enough. You only need
> to mark devices being part of the same chain.
> 
> Actually graph with endpoints would be more suitable, assuming above
> diagram represents connections.
> 

Thanks for suggesting "graph will endpoints" approach for representing WSI
connections.

I will check on this and come back.

> Please include that diagram in binding description.
> 

Sure will include the above diagram in next version.

>> 3. qcom,wsi-num-devices:
>>    Represents the number of devices connected through WSI within the same WSI group to
>>    which the device belongs.
>>    
>>    In the case of RDP433, all devices will have this number as 3.
>>    For the second example with four WiFi devices but with two WSI connections, the
>>    value of “qcom,wsi-num-devices” for each device will be 2.
> Not needed, just iterate over the graph children.

Thanks, based on "graph with endpoints" implementation will have this as well.