Hi Gabi,

Sorry for the delay. It was a hectic week last week.

As promised:

This patch adds ST Keyscan driver to use the keypad hw a subset
of ST boards provide. Specific board setup will be put in the
given dt.

Signed-off-by: Giuseppe Condorelli <redacted>
Signed-off-by: Gabriel Fernandez <redacted>

Are you sure these are in the correct order?

What is the history of this commit?

---
 .../devicetree/bindings/input/st-keypad.txt        |  50 ++++

This should be submitted as a seperate patch.

s/device tree/Device Tree

+
+The ST keypad controller device tree binding is based on the

As above.

+matrix-keymap.
+
+Required properties:
+
+- compatible: "st,keypad"

I think there will be subsequent st,keypad drivers? Is there any way
we can make this compatible string more specific to _this_ device?

st,stih4xx-keypad?

+- reg: Register base address of st-keypad controler.

s/address/address and size AND s/controler/controller

+- interrupts: Interrupt numberof st-keypad controler.

s/numberof/number for the

+- clocks: Must contain one entry, for the module clock.
+  See ../clocks/clock-bindings.txt for details.

Great!

+- pinctrl-0: Should specify pin control groups used for this controller.
+
+- pinctrl-names: Should contain only one value - "default".

Like to ../pinctrl/pinctrl-bindings.txt

+- linux,keymap: The keymap for keys as described in the binding document
+  devicetree/bindings/input/matrix-keymap.txt.
+
+- keypad,num-rows: Number of row lines connected to the keypad controller.
+
+- keypad,num-columns: Number of column lines connected to the keypad
+  controller.
+
+- st,debounce_us: Debouncing interval time in microseconds

I'm sure there will be a shared binding for de-bounce.

If not, there certainly should be.

+
+

Superfluous new lines.

+Example:
+
+keyscan: keyscan at fe4b0000 {
+	compatible = "st,keypad";

Is there any way we can make this more specific to _this_ IP?

+	reg = <0xfe4b0000 0x2000>;
+	interrupts = <GIC_SPI 212 IRQ_TYPE_NONE>;
+	clocks	= <&CLK_SYSIN>;
+	pinctrl-names = "default";
+	pinctrl-0 = <&pinctrl_keyscan>;
+
+	keypad,num-rows = <4>;
+	keypad,num-columns = <4>;
+	st,debounce_us = <5000>;
+	linux,keymap = < /* in0	in1	in2	in3 */

Do these line up in the file? I know Git can be a bit funny about this
sometimes.

Might be worth mentioning which ones.

Did sh_keysc.c ever exist in Mainline?

+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/module.h>
+#include <linux/interrupt.h>
+#include <linux/platform_device.h>
+#include <linux/clk.h>
+#include <linux/io.h>
+#include <linux/input/matrix_keypad.h>
+
+#define ST_KEYSCAN_MAXKEYS 16
+
+#define KEYSCAN_CONFIG_OFF		0x000
+#define KEYSCAN_CONFIG_ENABLE		1

0x001?

+#define KEYSCAN_DEBOUNCE_TIME_OFF	0x004
+#define KEYSCAN_MATRIX_STATE_OFF	0x008
+#define KEYSCAN_MATRIX_DIM_OFF		0x00c

Odd that these are 3 digit padded? Is there a reason for that?

+#define KEYSCAN_MATRIX_DIM_X_SHIFT	0
+#define KEYSCAN_MATRIX_DIM_Y_SHIFT	2

For all 'ST_KEYSCAN_...'?

+struct keypad_platform_data {
+	const struct matrix_keymap_data *keymap_data;
+	unsigned int num_out_pads;
+	unsigned int num_in_pads;
+	unsigned int debounce_us;
+};
+
+struct keyscan_priv {
+	void __iomem *base;
+	int irq;
+	struct clk *clk;
+	struct input_dev *input_dev;
+	struct keypad_platform_data *config;
+	unsigned int last_state;
+	u32 keycodes[ST_KEYSCAN_MAXKEYS];

Seems odd to limit this. Can't the information come from DT
i.e. keypad,num-rows and keypad,num-columns?

+};
+
+static irqreturn_t keyscan_isr(int irq, void *dev_id)
+{
+	struct platform_device *pdev = dev_id;
+	struct keyscan_priv *priv = platform_get_drvdata(pdev);
+	int state;
+	int change;
+
+	state = readl(priv->base + KEYSCAN_MATRIX_STATE_OFF) & 0xffff;
+	change = priv->last_state ^ state;
+
+	while (change) {
+		int scancode = __ffs(change);
+		int down = state & (1 << scancode);

int down = state & BIT(scancode);

+		input_report_key(priv->input_dev, priv->keycodes[scancode],
+				 down);
+		change ^= 1 << scancode;

change ^= BIT(scancode);

+	};
+
+	input_sync(priv->input_dev);
+
+	priv->last_state = state;
+
+	return IRQ_HANDLED;
+}
+
+static void keyscan_start(struct keyscan_priv *priv)
+{
+	clk_enable(priv->clk);
+
+	writel(priv->config->debounce_us * (clk_get_rate(priv->clk) / 1000000),
+	       priv->base + KEYSCAN_DEBOUNCE_TIME_OFF);
+
+	writel(((priv->config->num_in_pads - 1) << KEYSCAN_MATRIX_DIM_X_SHIFT) |
+	       ((priv->config->num_out_pads - 1) << KEYSCAN_MATRIX_DIM_Y_SHIFT),
+	       priv->base + KEYSCAN_MATRIX_DIM_OFF);
+
+	writel(KEYSCAN_CONFIG_ENABLE, priv->base + KEYSCAN_CONFIG_OFF);
+}
+
+static void keyscan_stop(struct keyscan_priv *priv)
+{
+	writel(0, priv->base + KEYSCAN_CONFIG_OFF);
+
+	clk_disable(priv->clk);
+}
+
+static int keypad_matrix_key_parse_dt(struct keyscan_priv *st_kp)
+{
+	struct device *dev = st_kp->input_dev->dev.parent;
+	struct device_node *np = dev->of_node;
+	struct keypad_platform_data *pdata;
+	int error;
+
+	pdata = devm_kzalloc(dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata) {
+		dev_err(dev, "failed to allocate driver pdata\n");

If the system is OOM, this sort of error message will be pretty
redundant. Just return -ENOMEM instead.

+		return -ENOMEM;
+	}
+
+	error = matrix_keypad_parse_of_params(dev, &pdata->num_out_pads,
+			&pdata->num_in_pads);
+	if (error) {
+		dev_err(dev, "failed to parse keypad params\n");
+		return error;

Nit: It's pretty unusual to use this for a standard error handling
variable. Consider 'ret' or 'err' as a replacement.

+	}
+
+	of_property_read_u32(np, "st,debounce_us", &pdata->debounce_us);

Isn't this a required property? Might be worth checking the return
value if so.

+	st_kp->config = pdata;
+
+	dev_info(dev, "rows=%d col=%d debounce=%d\n",
+			pdata->num_out_pads,
+			pdata->num_in_pads,
+			pdata->debounce_us);

Might be worth moving this down passed the final point of failure.

+	error = of_property_read_u32_array(np, "linux,keymap",
+					st_kp->keycodes, ST_KEYSCAN_MAXKEYS);
+	if (error) {
+		dev_err(dev, "failed to parse keymap\n");
+		return error;
+	}
+
+	return 0;
+}
+
+static int __init keyscan_probe(struct platform_device *pdev)
+{
+	struct keypad_platform_data *pdata =
+		dev_get_platdata(&pdev->dev);

Do we really support platform data still?

+	struct keyscan_priv *st_kp;
+	struct input_dev *input_dev;
+	struct device *dev = &pdev->dev;

dev and pdev are used randomly in probe(). Just remove the dev
de-reference and use &pdev->dev exclusively.

It's pretty common to de-reference 'np = pdev->dev.of_node' though.

+	struct resource *res;
+	int len;
+	int error;
+	int i;
+
+	if (!pdata && !pdev->dev.of_node) {
+		dev_err(&pdev->dev, "no keymap defined\n");
+		return -EINVAL;
+	}
+
+	st_kp = devm_kzalloc(dev, sizeof(*st_kp), GFP_KERNEL);
+	if (!st_kp) {
+		dev_err(dev, "failed to allocate driver data\n");
+		return -ENOMEM;

I wouldn't print any messages on -ENOMEM personally. It's not a driver
failure per se, but a system one where the user will soon be notified.

+	}
+	st_kp->config = pdata;

You only want to do this in the !np case.

+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (!res) {
+		dev_err(dev, "no I/O memory specified\n");
+		return -ENXIO;
+	}
+
+	len = resource_size(res);
+	if (!devm_request_mem_region(dev, res->start, len, pdev->name)) {
+		dev_err(dev, "failed to reserve I/O memory\n");
+		return -EBUSY;
+	}
+
+	st_kp->base = devm_ioremap_nocache(dev, res->start, len);
+	if (st_kp->base == NULL) {
+		dev_err(dev, "failed to remap I/O memory\n");
+		return -ENXIO;
+	}

Replace the two above with devm_ioremap_resource().

Make sure the IORESOURCE_CACHEABLE is set.

+	st_kp->irq = platform_get_irq(pdev, 0);
+	if (st_kp->irq < 0) {
+		dev_err(dev, "no IRQ specified\n");
+		return -ENXIO;

No such device or address, are you sure?

Perhaps -EINVAL would be better, as one should be specified.

+	}
+
+	error = devm_request_irq(dev, st_kp->irq, keyscan_isr, 0,
+				 pdev->name, pdev);
+	if (error) {
+		dev_err(dev, "failed to request IRQ\n");
+		return error;
+	}
+
+	input_dev = devm_input_allocate_device(&pdev->dev);
+	if (!input_dev) {
+		dev_err(&pdev->dev, "failed to allocate the input device\n");
+		return -ENOMEM;
+	}
+
+	st_kp->clk = devm_clk_get(dev, NULL);
+	if (IS_ERR(st_kp->clk)) {
+		dev_err(dev, "cannot get clock");
+		return PTR_ERR(st_kp->clk);
+	}
+
+	input_dev = input_allocate_device();
+	if (!input_dev) {
+		dev_err(dev, "failed to allocate input device\n");
+		return -ENOMEM;
+	}

Wasn't this done already?

+	st_kp->input_dev = input_dev;
+
+	input_dev->name = pdev->name;
+	input_dev->phys = "keyscan-keys/input0";
+	input_dev->dev.parent = dev;
+
+	input_dev->id.bustype = BUS_HOST;
+	input_dev->id.vendor = 0x0001;
+	input_dev->id.product = 0x0001;
+	input_dev->id.version = 0x0100;

Any chance we can #define these?

+	if (!pdata) {
+		error = keypad_matrix_key_parse_dt(st_kp);
+		if (error)
+			return error;
+		pdata = st_kp->config;

Is pdata used after this?

+	}
+
+	input_dev->keycode = st_kp->keycodes;
+	input_dev->keycodesize = sizeof(st_kp->keycodes[0]);
+	input_dev->keycodemax = ARRAY_SIZE(st_kp->keycodes);
+
+	for (i = 0; i < ST_KEYSCAN_MAXKEYS; i++)
+		input_set_capability(input_dev, EV_KEY, st_kp->keycodes[i]);
+	__clear_bit(KEY_RESERVED, input_dev->keybit);
+
+	error = input_register_device(input_dev);
+	if (error) {
+		dev_err(dev, "failed to register input device\n");
+		input_free_device(input_dev);
+		platform_set_drvdata(pdev, NULL);

You don't need to do this anymore. It's taken care of elsewhere.

+		return error;
+	}
+
+	platform_set_drvdata(pdev, st_kp);
+
+	keyscan_start(st_kp);
+
+	device_set_wakeup_capable(&pdev->dev, 1);
+
+	return 0;
+}
+
+static int __exit keyscan_remove(struct platform_device *pdev)
+{
+	struct keyscan_priv *priv = platform_get_drvdata(pdev);
+
+	keyscan_stop(priv);
+
+	input_unregister_device(priv->input_dev);
+	platform_set_drvdata(pdev, NULL);
+
+	return 0;

I already saw that Dmitry commented on the rest of the file.

<snip>

-- 
Lee Jones
Linaro STMicroelectronics Landing Team Lead
Linaro.org ? Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog