Re: [PATCH v2 4/4] scsi: ufs: probe and init of variant driver from the platform device

[Paul Bolle <pebolle@tiscali.nl>]

On Wed, 2015-06-03 at 12:37 +0300, Yaniv Gardi wrote:
> --- a/drivers/scsi/ufs/ufs-qcom.c
> +++ b/drivers/scsi/ufs/ufs-qcom.c

>  EXPORT_SYMBOL(ufs_hba_qcom_vops);

Nothing uses this export. It's still a (static) symbol that is not
included in any header. I think this export serves no purpose. Am I
missing something subtle here?

> +/**
> + * ufs_qcom_probe - probe routine of the driver
> + * @pdev: pointer to Platform device handle
> + *
> + * Always return 0
> + */
> +static int ufs_qcom_probe(struct platform_device *pdev)
> +{
> +	dev_set_drvdata(&pdev->dev, (void *)&ufs_hba_qcom_vops);

(Cast to void * should not be needed.)

> +	return 0;
> +}
> +
> +/**
> + * ufs_qcom_remove - set driver_data of the device to NULL
> + * @pdev: pointer to platform device handle
> + *
> + * Always return 0
> + */
> +static int ufs_qcom_remove(struct platform_device *pdev)
> +{
> +	dev_set_drvdata(&pdev->dev, NULL);
> +	return 0;
> +}
> +
> +static const struct of_device_id ufs_qcom_of_match[] = {
> +	{ .compatible = "qcom,ufs_variant"},
> +	{},
> +};
> +
> +static struct platform_driver ufs_qcom_pltform = {
> +	.probe	= ufs_qcom_probe,
> +	.remove	= ufs_qcom_remove,
> +	.driver	= {
> +		.name	= "ufs_qcom",
> +		.owner	= THIS_MODULE,
> +		.of_match_table = of_match_ptr(ufs_qcom_of_match),
> +	},
> +};
> +module_platform_driver(ufs_qcom_pltform);

> --- a/drivers/scsi/ufs/ufshcd-pltfrm.c
> +++ b/drivers/scsi/ufs/ufshcd-pltfrm.c

> +	struct device_node *node = pdev->dev.of_node;
> +	struct device_node *ufs_variant_node;
> +	struct platform_device *ufs_variant_pdev;
 
> -	hba->vops = get_variant_ops(&pdev->dev);
> +	err = of_platform_populate(node, NULL, NULL, &pdev->dev);
> +	if (err)
> +		dev_err(&pdev->dev,
> +			"%s: of_platform_populate() failed\n", __func__);
> +
> +	ufs_variant_node = of_get_next_available_child(node, NULL);
> +
> +	if (!ufs_variant_node) {
> +		dev_dbg(&pdev->dev, "failed to find ufs_variant_node child\n");
> +	} else {
> +		ufs_variant_pdev = of_find_device_by_node(ufs_variant_node);
> +
> +		if (ufs_variant_pdev)
> +			hba->vops = (struct ufs_hba_variant_ops *)
> +				dev_get_drvdata(&ufs_variant_pdev->dev);

(Another cast that I think is not needed.)

> +	}

If I scanned this correctly, the dev_set_drvdata() and dev_get_drvdata()
pair adds an actual user of ufs_hba_qcom_vops. So that ends the obvious
issue I think the code currently has. And I gladly defer to the scsi
people to determine whether that is done the right way.

Thanks,


Paul Bolle