Re: [PATCH 1/5] mmc: sdhci: make sdhci-esdhc-imx driver self registered

On Wed, Mar 23, 2011 at 10:28:58PM -0600, Grant Likely wrote:

On Mon, Mar 21, 2011 at 04:06:59PM +0800, Shawn Guo wrote:

quoted

The patch turns the common stuff to in sdhci-pltfm.c into functions,
and add sdhci-esdhc-imx its own .probe and .remove which in turn call
into the common functions, so that sdhci-esdhc-imx driver registers
itself and keep all sdhci-esdhc-imx specific things like
sdhci_esdhc_imx_pdata away from sdhci-pltfm.c which is common.

Signed-off-by: Shawn Guo <redacted>

Hi Shawn,

Hi Grant,

Thanks for all this work, I think it is the right thing to do.  A few
relatively minor comments below, but otherwise I like it.

g.

quoted

---
 drivers/mmc/host/sdhci-esdhc-imx.c |   98 +++++++++++++++++++++++++++++-------
 drivers/mmc/host/sdhci-pltfm.c     |   84 +++++++++++++++++++++++++++++--
 drivers/mmc/host/sdhci-pltfm.h     |   11 ++++-

I think this patch should be split in 2.  One patch to refactor
edhci-pltfm* to create the common utility functions, and one patch to
convert the imx driver.

I squashed this whole series into one patch in a relevant patch set
I just posted out minutes ago, primarily to reduce the patch quantity
and ease the bisect, since it's logically integral.

[...]

quoted

+static int __init sdhci_esdhc_imx_init(void)
+{
+	return platform_driver_register(&sdhci_esdhc_imx_driver);
+}
+
+static void __exit sdhci_esdhc_imx_exit(void)
+{
+	platform_driver_unregister(&sdhci_esdhc_imx_driver);
+}
+
+module_init(sdhci_esdhc_imx_init);
+module_exit(sdhci_esdhc_imx_exit);

Typically, I like to see module_init() and module_exit() immediately
adjacent to the functions they register.

Incorporated in the new patch set just posted ...

quoted

+
+MODULE_DESCRIPTION("SDHCI driver for i.MX eSDHC");
+MODULE_AUTHOR("Wolfram Sang [off-list ref]");
+MODULE_LICENSE("GPL v2");

diff --git a/drivers/mmc/host/sdhci-pltfm.c b/drivers/mmc/host/sdhci-pltfm.c
index ccc04ac..38b8cb4 100644
--- a/drivers/mmc/host/sdhci-pltfm.c
+++ b/drivers/mmc/host/sdhci-pltfm.c

@@ -44,6 +44,83 @@
 static struct sdhci_ops sdhci_pltfm_ops = {
 };
 
+struct sdhci_host *sdhci_pltfm_init(struct platform_device *pdev,
+				    struct sdhci_pltfm_data *pdata)

Since there is no chance of *pdata being passed via the
pdev->dev.platform_data pointer (there aren't any users in mainline of
that feature) then I would take the opportunity to rename this
parameter and structure to eliminate any confusion.  'struct
sdhci_pltfm' would be sufficient I think.

The chance comes along with the new patch set (function
sdhci_esdhc_probe in sdhci-esdhc.c).  And I would not rename 'struct
sdhci_pltfmi_data' to 'struct sdhci_pltfm' in terms of that we have
another name of 'struct sdhci_pltfm_host'.  But yes, pdata does
confuse.  Any suggestion?  I can fix all of them with a follow-up
patch against the new series.

quoted

+{
+	struct sdhci_host *host;
+	struct sdhci_pltfm_host *pltfm_host;
+	struct resource *iomem;
+	int ret;
+
+	iomem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (!iomem) {
+		ret = -ENOMEM;
+		goto err;
+	}
+
+	if (resource_size(iomem) < 0x100)
+		dev_err(&pdev->dev, "Invalid iomem size!\n");
+
+	/* Some PCI-based MFD need the parent here */
+	if (pdev->dev.parent != &platform_bus)
+		host = sdhci_alloc_host(pdev->dev.parent, sizeof(*pltfm_host));
+	else
+		host = sdhci_alloc_host(&pdev->dev, sizeof(*pltfm_host));
+
+	if (IS_ERR(host)) {
+		ret = PTR_ERR(host);
+		goto err;
+	}
+
+	pltfm_host = sdhci_priv(host);
+
+	host->hw_name = "SDHCI";
+	if (pdata && pdata->ops)
+		host->ops = pdata->ops;
+	else
+		host->ops = &sdhci_pltfm_ops;
+	if (pdata)
+		host->quirks = pdata->quirks;
+	host->irq = platform_get_irq(pdev, 0);
+
+	if (!request_mem_region(iomem->start, resource_size(iomem),
+		mmc_hostname(host->mmc))) {
+		dev_err(&pdev->dev, "cannot request region\n");
+		ret = -EBUSY;
+		goto err_request;
+	}
+
+	host->ioaddr = ioremap(iomem->start, resource_size(iomem));
+	if (!host->ioaddr) {
+		dev_err(&pdev->dev, "failed to remap registers\n");
+		ret = -ENOMEM;
+		goto err_remap;
+	}
+
+	platform_set_drvdata(pdev, host);
+
+	return host;
+
+err_remap:
+	release_mem_region(iomem->start, resource_size(iomem));
+err_request:
+	sdhci_free_host(host);
+err:
+	pr_err("%s failed %d\n", __func__, ret);
+	return NULL;
+}

Since the bulk of this function is a direct clone of the
body of sdhci_pltfm_probe(), this patch should also modify
sdhci_pltfm_probe() to use the new utility function.  That will also
make it clearer to readers that this new block is simply a refactor of
the old.

Good suggestion.  Will take it next time where applies ...

-- 
Regards,
Shawn