[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <adbd07c2-3c01-474d-884f-df9eeb08b580@rock-chips.com>
Date: Fri, 9 Aug 2024 14:11:43 +0800
From: Shawn Lin <shawn.lin@...k-chips.com>
To: Krzysztof Kozlowski <krzk@...nel.org>, Rob Herring <robh+dt@...nel.org>,
"James E . J . Bottomley" <James.Bottomley@...senPartnership.com>,
"Martin K . Petersen" <martin.petersen@...cle.com>,
Krzysztof Kozlowski <krzk+dt@...nel.org>, Conor Dooley <conor+dt@...nel.org>
Cc: shawn.lin@...k-chips.com,
Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>,
Heiko Stuebner <heiko@...ech.de>, Alim Akhtar <alim.akhtar@...sung.com>,
Avri Altman <avri.altman@....com>, Bart Van Assche <bvanassche@....org>,
YiFeng Zhao <zyf@...k-chips.com>, Liang Chen <cl@...k-chips.com>,
linux-scsi@...r.kernel.org, linux-rockchip@...ts.infradead.org,
linux-kernel@...r.kernel.org, devicetree@...r.kernel.org
Subject: Re: [PATCH v2 3/3] scsi: ufs: rockchip: init support for UFS
在 2024/8/9 13:44, Krzysztof Kozlowski 写道:
> On 09/08/2024 02:53, Shawn Lin wrote:
>> 在 2024/8/8 18:36, Krzysztof Kozlowski 写道:
>>> On 08/08/2024 05:52, Shawn Lin wrote:
>>>> RK3576 contains a UFS controller, add init support fot it.
>>>>
>>>> Signed-off-by: Shawn Lin <shawn.lin@...k-chips.com>
>>>>
>>>
>>> ...
>>>
>>>> + err = clk_prepare_enable(host->ref_out_clk);
>>>> + if (err)
>>>> + return dev_err_probe(dev, err, "failed to enable ref out clock\n");
>>>> +
>>>> + host->rst_gpio = devm_gpiod_get(&pdev->dev, "reset", GPIOD_OUT_LOW);
>>>> + if (IS_ERR(host->rst_gpio)) {
>>>> + dev_err_probe(&pdev->dev, PTR_ERR(host->rst_gpio),
>>>> + "invalid reset-gpios property in node\n");
>>>> + err = PTR_ERR(host->rst_gpio);
>>>
>>> No. Look at your code above - you have return dev_err_probe, so logical
>>> is that the syntax is err = dev_err_probe. Don't over-complicate the code.
>>>
>>> Anyway, this is suspicious. You already have resets with four resets
>>> (!!!) and you claim you have fifth reset - GPIO? This looks just wrong,
>>> like you represent some properties which do not belong here.
>>>
>>
>> Thanks for the feadback.
>>
>> Yes, we have 4 resets for controller, and one gpio to reset the
>> device. It happened to be called reset-gpios in DT but can be
>> any name if you like it to be. I added reset-gpios as a required one
>> listed in dt-bindings file, in patch 2.
>
> Then explain in the bindings what this reset-gpios is for.
Got it.
>>
>>
>> > Also, why do you leave the device in the reset state? Logical one
>> means
>> > - reset is asserted. This applies to ufs_rockchip_device_reset() as >
>> well
>> > - that's just wrong code.
>>
>>
>> No, I tested it in linux-net and the output is HIGH, and leave the
>> device in active state.
>>
>> In dts, we add:
>> reset-gpios = <&gpio4 RK_PD0 GPIO_ACTIVE_HIGH>;
>
> ? Don't use random values, what is type of the reset line?
>
>>
>> so gpiod_set_value_cansleep(host->rst_gpio, 1) -> output HIGH for
>> gpio4_d0. Based on your comment, we should change the dts to use
>> GPIO_ACTIVE_LOW and then gpiod_set_value_cansleep(host->rst_gpio, 0)
>>
>>
>> Both can work, however IMO, isn't logical one means HIGH level in line
>> more human readable?
>
> You mix logical level and line level.
>
> If this is reset GPIO, then 1 means asserted so your code is bogus.
>
> Maybe the the line is active low?
>
Yes, I mix the them. Will change to use active-low.
Thanks.
>>
>>
>>
>>> Where is your DTS so we can validate it?
>>>
>>
>> Will add it in the next version, as now the rk3576.dtsi is under
>> reviewed and UFS node was not added. So I was afraid to interfere
>> with that patch and just wanted to add incremental patch once
>> rk3576.dtsi got merged.
>>
>>> Best regards,
>>> Krzysztof
>>>
>
> Best regards,
> Krzysztof
>
Powered by blists - more mailing lists