Re: [PATCH] soc: imx8m: Add remove function

[Shawn Guo <shawnguo2@yeah.net>]

On Mon, Dec 09, 2024 at 08:26:48AM +0000, Peng Fan wrote:
> > Subject: Re: [PATCH] soc: imx8m: Add remove function
> > 
> > On 24-12-06, Peng Fan (OSS) wrote:
> > > From: Peng Fan <peng.fan@nxp.com>
> > >
> > > Unregister the cpufreq device and soc device in remove path,
> > otherwise
> > > there will be warning when do removing test:
> > > sysfs: cannot create duplicate filename '/devices/platform/imx-
> > cpufreq-dt'
> > > CPU: 0 UID: 0 PID: 1 Comm: swapper/0 Not tainted
> > > 6.13.0-rc1-next-20241204 Hardware name: NXP i.MX8MPlus EVK
> > board (DT)
> > >
> > > Fixes: 9cc832d37799 ("soc: imx8m: Probe the SoC driver as platform
> > > driver")
> > > Signed-off-by: Peng Fan <peng.fan@nxp.com>
> > > ---
> > >  drivers/soc/imx/soc-imx8m.c | 32 +++++++++++++++++++++++++++----
> > -
> > >  1 file changed, 27 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/drivers/soc/imx/soc-imx8m.c b/drivers/soc/imx/soc-
> > imx8m.c
> > > index 8ac7658e3d52..8c368947d1e5 100644
> > > --- a/drivers/soc/imx/soc-imx8m.c
> > > +++ b/drivers/soc/imx/soc-imx8m.c
> > > @@ -33,6 +33,11 @@ struct imx8_soc_data {
> > >  	int (*soc_revision)(u32 *socrev, u64 *socuid);  };
> > >
> > > +struct imx8m_soc_priv {
> > > +	struct soc_device *soc_dev;
> > > +	struct platform_device *cpufreq_dev; };
> > > +
> > >  #ifdef CONFIG_HAVE_ARM_SMCCC
> > >  static u32 imx8mq_soc_revision_from_atf(void)
> > >  {
> > > @@ -198,7 +203,7 @@ static int imx8m_soc_probe(struct
> > platform_device *pdev)
> > >  	const struct imx8_soc_data *data;
> > >  	struct device *dev = &pdev->dev;
> > >  	const struct of_device_id *id;
> > > -	struct soc_device *soc_dev;
> > > +	struct imx8m_soc_priv *priv;
> > >  	u32 soc_rev = 0;
> > >  	u64 soc_uid = 0;
> > >  	int ret;
> > > @@ -207,6 +212,10 @@ static int imx8m_soc_probe(struct
> > platform_device *pdev)
> > >  	if (!soc_dev_attr)
> > >  		return -ENOMEM;
> > >
> > > +	priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
> > > +	if (!priv)
> > > +		return -ENOMEM;
> > > +
> > >  	soc_dev_attr->family = "Freescale i.MX";
> > >
> > >  	ret = of_property_read_string(of_root, "model",
> > > &soc_dev_attr->machine); @@ -235,21 +244,34 @@ static int
> > imx8m_soc_probe(struct platform_device *pdev)
> > >  	if (!soc_dev_attr->serial_number)
> > >  		return -ENOMEM;
> > >
> > > -	soc_dev = soc_device_register(soc_dev_attr);
> > > -	if (IS_ERR(soc_dev))
> > > -		return PTR_ERR(soc_dev);
> > > +	priv->soc_dev = soc_device_register(soc_dev_attr);
> > > +	if (IS_ERR(priv->soc_dev))
> > > +		return PTR_ERR(priv->soc_dev);
> > >
> > >  	pr_info("SoC: %s revision %s\n", soc_dev_attr->soc_id,
> > >  		soc_dev_attr->revision);
> > >
> > >  	if (IS_ENABLED(CONFIG_ARM_IMX_CPUFREQ_DT))
> > > -		platform_device_register_simple("imx-cpufreq-dt", -1,
> > NULL, 0);
> > > +		priv->cpufreq_dev =
> > > +platform_device_register_simple("imx-cpufreq-dt", -1, NULL, 0);
> > 
> > If CONFIG_ARM_IMX_CPUFREQ_DT is enabled, I asusme that
> > platform_device_register_simple() shouldn't fail else it will be an error,
> > right? Therefore I would like to add the 'if(!IS_ERR())' check here
> > instead of the remove function.
> 
> You mean below?
> dev = platform_device_register_simple("imx-cpufreq-dt", -1, NULL, 0);
> if (!IS_ERR(dev))
>    plat->cpufreq_dev = dev;
> else
>   pr_err("Failed to register imx-cpufreq-dt: %d\n", ERR_PTR(dev))?

Hmm, I'm not sure why we do not have error check on
platform_device_register_simple().  Shouldn't we do the following?

	priv->cpufreq_dev = platform_device_register_simple("imx-cpufreq-dt", -1, NULL, 0);
	if (IS_ERR(priv->cpufreq_dev))
		return PTR_ERR(priv->cpufreq_dev);

Then I'm with Marco that we only need to check 'if (priv->cpufreq_dev)' in
remove function.

Shawn