Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <9581ba8b-1d28-3292-872b-026c60e20424@linaro.org>
Date:   Thu, 5 May 2022 10:18:09 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.8.1
Subject: Re: [PATCH v2 1/3] dt-bindings: misc: add bindings for Intel HPS Copy
 Engine
Content-Language: en-US
To:     matthew.gerlach@linux.intel.com
Cc:     dinguyen@kernel.org, robh+dt@kernel.org,
        linux-kernel@vger.kernel.org, devicetree@vger.kernel.org,
        krzysztof.kozlowski+dt@linaro.org
References: <20220503194546.1287679-1-matthew.gerlach@linux.intel.com>
 <20220503194546.1287679-2-matthew.gerlach@linux.intel.com>
 <17407fe7-b11d-2ba8-acca-3e71cf1a3b2f@linaro.org>
 <alpine.DEB.2.22.394.2205041425230.2669897@rhweight-WRK1>
From:   Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
In-Reply-To: <alpine.DEB.2.22.394.2205041425230.2669897@rhweight-WRK1>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 05/05/2022 00:41, matthew.gerlach@linux.intel.com wrote:
>> Please find appropriate directory matching this hardware, not "misc". As
>> a fallback SoC related bindings end up in "soc".
> 
> I thought misc seemed appropriate because it is a very specific IP block 
> in the FPGA connected to the HPS.  It does perform a simple DMA function; 
> so I considered putting it in the dma directory, but it also has some
> hand-shaking registers between the HPS and a host processor connected to the
> FPGA via PCIe; so I thought misc.  Since the HPS "soc" accesses the 
> component, I can put it in the "soc" directory, unless there is a better 
> suggestion.

So let it be "soc".

>>
>> $ git grep address-cell
>> Do not use inconsistent coding. The same applies to your DTS.
> 
> Is the inconsistency the use of '0x' in the values of #address-cells and 
> #size-cells, or is the consistency having different values for 
> #address-cells and #size-cells or both?

It's about "0x".

> 
>>
>>> +        #size-cells = <0x1>;
>>> +        ranges = <0x00000000 0x00000000 0xf9000000 0x00001000>;
>>
>> Why do you even need the simple-bus above and cannot put the device
>> directly on the bus?
> 
> On an Agilex chip, the HPS is connected to the FPGA via two bridges, 
> referred as the "HPS to FPGA bridge" and the "Lightweight HPS to FPGA 
> bridge".  An IP block in the FPGA could be connected to one or both of 
> these bridges.  I am anticipating device tree overlays being applied for 
> other IP blocks instantiated in the FPGA.

OK

> 
>>
>>> +
>>> +        hps_cp_eng@0 {
>>
>> No underscores in node names. Generic node name.
> 
> I understand.  I am considering dma@0 for the generic node name.

Let's keep the same as in DTS.


Best regards,
Krzysztof