[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Fri, 1 Jan 2021 12:27:41 +0100
From: Bert Vermeulen <bert@...t.com>
To: Lukas Wunner <lukas@...ner.de>
Cc: Mark Brown <broonie@...nel.org>, Rob Herring <robh+dt@...nel.org>,
Birger Koblitz <mail@...ger-koblitz.de>,
linux-spi@...r.kernel.org, devicetree@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH RESEND v2 2/2] Add support for Realtek RTL838x/RTL839x SoC
SPI controllers
On 12/31/20 3:23 PM, Lukas Wunner wrote:
> On Wed, Dec 30, 2020 at 12:19:04AM +0100, Bert Vermeulen wrote:
>> +static inline void wait_ready(struct rtspi *rtspi)
>> +{
>> + while (!(readl(REG(RTL8380_SPI_SFCSR)) & RTL8380_SPI_SFCSR_RDY))
>> + ;
>> +}
>
> I'd suggest calling cpu_relax() in the loop's body.
>
>
>> + err = devm_spi_register_controller(&pdev->dev, ctrl);
>
> Since you're invoking devm_spi_register_controller() on probe,
> the controller must not be unregistered explicitly on remove.
> So the ->remove hook can be dropped altogether:
>
>> +static int realtek_spi_remove(struct platform_device *pdev)
>> +{
>> + struct spi_controller *ctrl = platform_get_drvdata(pdev);
>> +
>> + spi_unregister_controller(ctrl);
>> +
>> + return 0;
>> +}
> [...]
>> + .remove = realtek_spi_remove,
>
> The ->probe hook otherwise LGTM.
Thanks Lukas, will fix those issues.
--
Bert Vermeulen
bert@...t.com
Powered by blists - more mailing lists