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: <32e9a512-fd74-b2f6-6b8a-fefb9ad5912d@quicinc.com>
Date:   Mon, 10 Jul 2023 20:52:48 -0700
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:91.0) Gecko/20100101
 Thunderbird/91.5.0
Subject: Re: [PATCH 1/7] dt-bindings: soc: qcom: Add qcom-pbs bindings
Content-Language: en-US
To:     Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>,
        Rob Herring <robh@kernel.org>
CC:     <pavel@ucw.cz>, <lee@kernel.org>, <thierry.reding@gmail.com>,
        <krzysztof.kozlowski+dt@linaro.org>, <conor+dt@kernel.org>,
        <agross@kernel.org>, <andersson@kernel.org>,
        <konrad.dybcio@linaro.org>, <u.kleine-koenig@pengutronix.de>,
        <linux-leds@vger.kernel.org>, <devicetree@vger.kernel.org>,
        <linux-kernel@vger.kernel.org>, <linux-arm-msm@vger.kernel.org>,
        <linux-pwm@vger.kernel.org>
References: <20230621185949.2068-1-quic_amelende@quicinc.com>
 <20230621185949.2068-2-quic_amelende@quicinc.com>
 <20230626135857.GA3118929-robh@kernel.org>
 <2e871e21-a81d-0d7d-993b-9a9d7bd9d962@quicinc.com>
 <e7298704-5a03-0961-90a3-dab4af60c326@linaro.org>
From:   Anjelique Melendez <quic_amelende@quicinc.com>
In-Reply-To: <e7298704-5a03-0961-90a3-dab4af60c326@linaro.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 7/1/2023 4:03 AM, Krzysztof Kozlowski wrote:
> On 29/06/2023 03:19, Anjelique Melendez wrote:
> 
>>>> +examples:
>>>> +  - |
>>>> +    pmic {
>>>> +      #address-cells = <1>;
>>>> +      #size-cells = <0>;
>>>> +
>>>> +      qcom,pbs@7400 {
>>>> +        compatible = "qcom,pbs";
>>>> +        reg = <0x7400>;
>>>> +      };
>>>
>>> Why do you need a child node for this? Is there more than 1 instance in 
>>> a PMIC? Every sub-function of a PMIC doesn't have to have a DT node.
>>>
>>
>> We currently have another downstream driver (which is planned to get upstreamed)
>> which also needs a handle to a pbs device in order to properly trigger events. 
> 
> I don't see how does it answer Rob's concerns. Neither mine about
> incomplete binding. You don't need pbs node here for that.
> 
> Anyway, whatever you have downstream also does not justify any changes.
> Either upstream these so we can see it or drop this binding.
> 
> Best regards,
> Krzysztof
> 

On PMI632, peripherals are partitioned over 2 different SIDs
(https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/arm64/boot/dts/qcom/pmi632.dtsi?h=v6.5-rc1#n42
and https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/arm64/boot/dts/qcom/pmi632.dtsi?h=v6.5-rc1#n149).
Unfortunately, the pbs peripheral and the lpg peripherals are on different
PMI632 devices and therefore have different regmaps.
 
If we get rid of the pbs node we need to get a handle to the proper regmap.
I see two possible options, we could either introduce a new client property
which points to a peripheral on the same device as pbs.

i.e.
	led-controller {
		compatible = "qcom,pmi632-lpg";
      		#address-cells = <1>;
      		#size-cells = <0>;
      		#pwm-cells = <2>;
     		nvmem-names = "lpg_chan_sdam";
      		nvmem = <&pmi632_sdam7>;
      		qcom,pbs-phandle = <&pmi632_gpios>;
      		..... 
	};
Then when client is probing could do something like the following to get the regmap

	dn = of_parse_phandle(node, "qcom,pbs-phandle", 0);
	pdev = of_find_device_by_node(dn);
	pbs_regmap = dev_get_regmap(&pdev->dev->parent, NULL);


Or we could use the nvmem phandle and just have something like this in client's probe

	dn = of_parse_phandle(node, "nvmem", 0);
	pdev = of_find_device_by_node(dn);
	pbs_regmap = dev_get_regmap(&pdev->dev->parent, NULL);


Let me know what your thoughts are on this.

Thanks,
Anjelique