On 17.07.2018 09:26, A.s. Dong wrote:

quoted

-----Original Message-----
From: Oleksij Rempel [mailto:o.rempel at pengutronix.de]
Sent: Tuesday, July 17, 2018 3:14 PM
To: A.s. Dong <aisheng.dong@nxp.com>; Shawn Guo
[off-list ref]; Fabio Estevam [off-list ref]; Rob
Herring [off-list ref]; Mark Rutland [off-list ref]
Cc: devicetree at vger.kernel.org; kernel at pengutronix.de; linux-arm-
kernel at lists.infradead.org; dl-linux-imx [off-list ref]
Subject: Re: [PATCH v3 3/3] mailbox: Add support for i.MX7D messaging unit



On 17.07.2018 09:07, A.s. Dong wrote:

quoted

quoted

-----Original Message-----
From: Oleksij Rempel [mailto:o.rempel at pengutronix.de]
Sent: Tuesday, July 17, 2018 2:43 PM
To: A.s. Dong <aisheng.dong@nxp.com>; Shawn Guo
[off-list ref]; Fabio Estevam [off-list ref];

Rob

quoted

quoted

Herring [off-list ref]; Mark Rutland [off-list ref]
Cc: devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org;
kernel at pengutronix.de; dl-linux-imx [off-list ref]
Subject: Re: [PATCH v3 3/3] mailbox: Add support for i.MX7D messaging
unit



On 17.07.2018 08:21, A.s. Dong wrote:

quoted

quoted

-----Original Message-----
From: Oleksij Rempel [mailto:o.rempel at pengutronix.de]
Sent: Monday, July 16, 2018 7:42 PM
To: Shawn Guo <shawnguo@kernel.org>; Fabio Estevam
[off-list ref]; Rob Herring [off-list ref]; Mark
Rutland [off-list ref]; A.s. Dong

[off-list ref]

quoted

quoted

quoted

quoted

Cc: Oleksij Rempel <o.rempel@pengutronix.de>;
kernel at pengutronix.de; linux-arm-kernel at lists.infradead.org;
devicetree at vger.kernel.org;
dl-linux- imx [off-list ref]
Subject: [PATCH v3 3/3] mailbox: Add support for i.MX7D messaging
unit

The Mailbox controller is able to send messages (up to 4 32 bit
words) between the endpoints.

This driver was tested using the mailbox-test driver sending
messages between the Cortex-A7 and the Cortex-M4.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
---
 drivers/mailbox/Kconfig       |   6 +
 drivers/mailbox/Makefile      |   2 +
 drivers/mailbox/imx-mailbox.c | 294
++++++++++++++++++++++++++++++++++
 3 files changed, 302 insertions(+)  create mode 100644
drivers/mailbox/imx-mailbox.c

Generally it looks good to me, a few minor comments:

quoted

diff --git a/drivers/mailbox/Kconfig b/drivers/mailbox/Kconfig
index a2bb27446dce..79060ddc380d 100644
--- a/drivers/mailbox/Kconfig
+++ b/drivers/mailbox/Kconfig

@@ -15,6 +15,12 @@ config ARM_MHU
 	  The controller has 3 mailbox channels, the last of which can be
 	  used in Secure mode only.

+config IMX_MBOX
+	tristate "i.MX Mailbox"
+	depends on ARCH_MXC || COMPILE_TEST
+	help
+	  Mailbox implementation for i.MX Messaging Unit (MU).
+
 config PLATFORM_MHU
 	tristate "Platform MHU Mailbox"
 	depends on OF

diff --git a/drivers/mailbox/Makefile b/drivers/mailbox/Makefile

index
cc23c3a43fcd..ba2fe1b6dd62 100644

--- a/drivers/mailbox/Makefile
+++ b/drivers/mailbox/Makefile

@@ -7,6 +7,8 @@ obj-$(CONFIG_MAILBOX_TEST)	+= mailbox-

test.o

quoted

quoted

quoted

quoted

 obj-$(CONFIG_ARM_MHU)	+= arm_mhu.o

+obj-$(CONFIG_IMX_MBOX)	+= imx-mailbox.o
+
 obj-$(CONFIG_PLATFORM_MHU)	+= platform_mhu.o

 obj-$(CONFIG_PL320_MBOX)	+= pl320-ipc.o

diff --git a/drivers/mailbox/imx-mailbox.c

b/drivers/mailbox/imx-mailbox.c new file mode 100644 index
000000000000..31353366a007

--- /dev/null
+++ b/drivers/mailbox/imx-mailbox.c

@@ -0,0 +1,294 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2018 Pengutronix, Oleksij Rempel
+<o.rempel@pengutronix.de>  */
+
+#include <linux/clk.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/kernel.h>
+#include <linux/mailbox_controller.h> #include <linux/module.h>
+#include <linux/of_device.h>
+
+/* Transmit Register */
+#define IMX_MU_xTRn(x)		(0x00 + 4 * (x))
+/* Receive Register */
+#define IMX_MU_xRRn(x)		(0x10 + 4 * (x))
+/* Status Register */
+#define IMX_MU_xSR		0x20
+#define IMX_MU_xSR_TEn(x)	BIT(20 + (3 - (x)))
+#define IMX_MU_xSR_RFn(x)	BIT(24 + (3 - (x)))
+#define IMX_MU_xSR_BRDIP	BIT(9)
+
+/* Control Register */
+#define IMX_MU_xCR		0x24
+/* Transmit Interrupt Enable */
+#define IMX_MU_xCR_TIEn(x)	BIT(20 + (3 - (x)))
+/* Receive Interrupt Enable */
+#define IMX_MU_xCR_RIEn(x)	BIT(24 + (3 - (x)))
+
+#define IMX_MU_MAX_CHANS	4u
+
+struct imx_mu_priv;
+
+struct imx_mu_cfg {
+	unsigned int		chans;
+	void (*init_hw)(struct imx_mu_priv *priv); };
+
+struct imx_mu_con_priv {
+	int			irq;
+	unsigned int		idx;
+};
+
+struct imx_mu_priv {
+	struct device		*dev;
+	const struct imx_mu_cfg	*dcfg;
+	void __iomem		*base;
+
+	struct mbox_controller	mbox;
+	struct mbox_chan	mbox_chans[IMX_MU_MAX_CHANS];
+
+	struct imx_mu_con_priv  con_priv[IMX_MU_MAX_CHANS];
+	struct clk		*clk;
+
+	bool			side_a;
+};
+
+static struct imx_mu_priv *to_imx_mu_priv(struct mbox_controller
+*mbox) {
+	return container_of(mbox, struct imx_mu_priv, mbox); }
+
+static void imx_mu_write(struct imx_mu_priv *priv, u32 val, u32 offs)

{

quoted

quoted

quoted

quoted

+	iowrite32(val, priv->base + offs); }
+
+static u32 imx_mu_read(struct imx_mu_priv *priv, u32 offs) {
+	return ioread32(priv->base + offs); }
+
+static u32 imx_mu_rmw(struct imx_mu_priv *priv, u32 offs, u32 set,
+u32
+clr) {
+	u32 val;
+
+	val = imx_mu_read(priv, offs);
+	val &= ~clr;
+	val |= set;
+	imx_mu_write(priv, val, offs);
+
+	return val;
+}
+
+static irqreturn_t imx_mu_isr(int irq, void *p) {
+	struct mbox_chan *chan = p;

Better re-order from long to short.

"chis" is used by by following declarations. It make no sense to reorder it.

Yes, you're right.

quoted

quoted

quoted

+	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
+	struct imx_mu_con_priv *cp = chan->con_priv;
+

Unnecessary blank line?

ok,

quoted

quoted

+	u32 val, ctrl, dat;
+
+	ctrl = imx_mu_read(priv, IMX_MU_xCR);
+	val = imx_mu_read(priv, IMX_MU_xSR);
+	val &= IMX_MU_xSR_TEn(cp->idx) | IMX_MU_xSR_RFn(cp->idx);
+	val &= ctrl & (IMX_MU_xCR_TIEn(cp->idx) | IMX_MU_xCR_RIEn(cp-

quoted

idx));

+	if (!val)
+		return IRQ_NONE;
+
+	if (val & IMX_MU_xSR_TEn(cp->idx)) {
+		imx_mu_rmw(priv, IMX_MU_xCR, 0, IMX_MU_xCR_TIEn(cp-

quoted

idx));

+		mbox_chan_txdone(chan, 0);
+	}
+
+	if (val & IMX_MU_xSR_RFn(cp->idx)) {
+		dat = imx_mu_read(priv, IMX_MU_xRRn(cp->idx));
+		mbox_chan_received_data(chan, (void *)&dat);
+	}
+
+	return IRQ_HANDLED;
+}
+
+static bool imx_mu_last_tx_done(struct mbox_chan *chan) {
+	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
+	struct imx_mu_con_priv *cp = chan->con_priv;
+	u32 val;
+
+	val = imx_mu_read(priv, IMX_MU_xSR);
+	/* test if transmit register is empty */
+	return (!!(val & IMX_MU_xSR_TEn(cp->idx))); }
+
+static int imx_mu_send_data(struct mbox_chan *chan, void *data) {
+	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
+	struct imx_mu_con_priv *cp = chan->con_priv;
+	u32 *arg = data;
+
+	if (!imx_mu_last_tx_done(chan))
+		return -EBUSY;
+
+	imx_mu_write(priv, *arg, IMX_MU_xTRn(cp->idx));
+	imx_mu_rmw(priv, IMX_MU_xCR, IMX_MU_xSR_TEn(cp->idx), 0);
+
+	return 0;
+}
+
+static int imx_mu_startup(struct mbox_chan *chan) {
+	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
+	struct imx_mu_con_priv *cp = chan->con_priv;
+	char *irq_desc;
+	int ret;
+
+	irq_desc = devm_kasprintf(priv->dev, GFP_KERNEL,
"imx_mu_chan[%i]",
+				  cp->idx);

I like the name differentiation, just wondering whether this could
cause memory leak if users repeatly open/close MU channels due to I
see no free.

good point.

quoted

quoted

+	if (!irq_desc)
+		return -ENOMEM;
+
+	ret = devm_request_irq(priv->dev, cp->irq, imx_mu_isr,
+			       IRQF_SHARED, irq_desc, chan);
+	if (ret) {
+		dev_err(priv->dev,
+			"Unable to acquire IRQ %d\n", cp->irq);
+		return ret;
+	}
+
+	imx_mu_rmw(priv, IMX_MU_xCR, IMX_MU_xCR_RIEn(cp->idx), 0);
+
+	return 0;
+}
+
+static void imx_mu_shutdown(struct mbox_chan *chan) {
+	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
+	struct imx_mu_con_priv *cp = chan->con_priv;
+
+	imx_mu_rmw(priv, IMX_MU_xCR, 0,
+		   IMX_MU_xCR_TIEn(cp->idx) | IMX_MU_xCR_RIEn(cp-

quoted

idx));

+
+	devm_free_irq(priv->dev, cp->irq, chan); }
+
+static const struct mbox_chan_ops imx_mu_ops = {
+	.send_data = imx_mu_send_data,
+	.startup = imx_mu_startup,
+	.shutdown = imx_mu_shutdown,
+};
+
+static int imx_mu_probe(struct platform_device *pdev) {
+	struct device *dev = &pdev->dev;
+	struct device_node *np = dev->of_node;
+	struct resource *iomem;
+	struct imx_mu_priv *priv;
+	const struct imx_mu_cfg *dcfg;
+	unsigned int i, chans;
+	int irq, ret;
+
+	priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	dcfg = of_device_get_match_data(dev);
+	if (!dcfg)
+		return -EINVAL;
+
+	priv->dcfg = dcfg;
+	priv->dev = dev;
+
+	iomem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	priv->base = devm_ioremap_resource(&pdev->dev, iomem);
+	if (IS_ERR(priv->base))
+		return PTR_ERR(priv->base);
+
+	irq = platform_get_irq(pdev, 0);
+	if (irq <= 0)
+		return irq < 0 ? irq : -EINVAL;
+
+	priv->clk = devm_clk_get(dev, NULL);
+	if (IS_ERR(priv->clk)) {
+		if (PTR_ERR(priv->clk) != -ENOENT)
+			return PTR_ERR(priv->clk);
+
+		priv->clk = NULL;
+	}
+
+	ret = clk_prepare_enable(priv->clk);
+	if (ret) {
+		dev_err(dev, "Failed to enable clock\n");
+		return ret;
+	}
+
+	chans = min(dcfg->chans, IMX_MU_MAX_CHANS);
+	/* Initialize channel identifiers */
+	for (i = 0; i < chans; i++) {
+		struct imx_mu_con_priv *cp = &priv->con_priv[i];
+
+		cp->idx = i;
+		cp->irq = irq;
+		priv->mbox_chans[i].con_priv = cp;
+	}
+
+	if (of_property_read_bool(np, "fsl,mu-side-a"))
+		priv->side_a = true;

See property comments in former emails.

ok.

quoted

quoted

+
+	priv->mbox.dev = dev;
+	priv->mbox.ops = &imx_mu_ops;
+	priv->mbox.chans = priv->mbox_chans;
+	priv->mbox.num_chans = chans;
+	priv->mbox.txdone_irq = true;
+
+	platform_set_drvdata(pdev, priv);
+
+	if (priv->dcfg->init_hw)
+		priv->dcfg->init_hw(priv);
+
+	return mbox_controller_register(&priv->mbox);
+}
+
+static int imx_mu_remove(struct platform_device *pdev) {
+	struct imx_mu_priv *priv = platform_get_drvdata(pdev);
+
+	mbox_controller_unregister(&priv->mbox);
+	clk_disable_unprepare(priv->clk);
+
+	return 0;
+}
+
+
+static void imx_mu_init_imx7d(struct imx_mu_priv *priv) {

I guess we could remove the soc postfix now.

ok.

quoted

quoted

+	/* Set default config */
+	if (priv->side_a)
+		imx_mu_write(priv, 0, IMX_MU_xCR); }
+
+static const struct imx_mu_cfg imx_mu_cfg_imx7d = {
+	.chans = IMX_MU_MAX_CHANS,

What's point of another chans here?
This can also be controlled by IMX_MU_MAX_CHANS.

SCU has one channel configuration.

I guess the SCU one channel(4 rx/tx register) actually is not the one
channel specified here. e.g. For M4 case, users can still specify only
using one chan
(1 rx/tx register). They're slightly different.
So for SCU case, we actually did not use the struct imx_mu_cfg.

quoted

It is possible to exctend this driver with existing MU to provide 8
channels: 4 channels with FIFO (can be used stand alone) + 4 channes
based on General purpose interrupt and should be used with shared

memory.

quoted

Yes, but that still seems common to me. Not per SoC specific.

Correct, it is not SoC specific, it is end-product specific. All depends on what
exactly is on opposite side of MU. Currently I don't see any other way as
providing product specific compatible with needed configuration. And
current driver version allows it without ugly patching.

I think the point is your current implementation of .hw_init() for MX7D
is quite general to other SoCs as well, not much end-product specific.

Let's see it's only a simple initialization of MU and reset both side.
+static void imx_mu_init_imx7d(struct imx_mu_priv *priv) {
+	/* Set default config */
+	if (priv->side_a)
+		imx_mu_write(priv, 0, IMX_MU_xCR);
+}
That's why I thought it could be moved into general part instead of being
provided as SoC specific .hw_init() callback.

Here is probably some misunderstanding. I'm not against
imx_mu_cfg_generic. I just prefer to keep "struct imx_mu_cfg", because i
already see me patching it back in some months .

Because I'm afraid we may write similar code for other SoCs as well:
e.g.
static const struct imx_mu_cfg imx_mu_cfg_imx6sx = {
        .chans = IMX_MU_MAX_CHANS,
        .init_hw = imx_mu_init_imx7d,
};

static const struct imx_mu_cfg imx_mu_cfg_imx7d = {
        .chans = IMX_MU_MAX_CHANS,
        .init_hw = imx_mu_init_imx7d,
};

static const struct imx_mu_cfg imx_mu_cfg_imx7ulp = {
        .chans = IMX_MU_MAX_CHANS,
        .init_hw = imx_mu_init_imx7d,
};

static const struct imx_mu_cfg imx_mu_cfg_imx8qxp = {
        .chans = IMX_MU_MAX_CHANS,
        .init_hw = imx_mu_init_imx7d,
};

static const struct imx_mu_cfg imx_mu_cfg_imx8qm = {
        .chans = IMX_MU_MAX_CHANS,
        .init_hw = imx_mu_init_imx7d,
};

static const struct imx_mu_cfg imx_mu_cfg_imx8mq = {
        .chans = IMX_MU_MAX_CHANS,
        .init_hw = imx_mu_init_imx7d,
};

Anyway, I'm not strongly against it. Maybe we can also refine
it later when we see the real problem.

Regards
Dong Aisheng

quoted

quoted

quoted

quoted

quoted

+	.init_hw = imx_mu_init_imx7d,

Can we imagine a diferent .init_hw callback for another SoC?

yes, for example SCU case, or any other case where A side is used on
slave system.

SCU does not use this as we still did not see SoC specific init_hw

requirement.

quoted

quoted

quoted

If no, how about make it default as I see the implementation is
quite simple and seems not SoC specific. Then we probably can
totally remove struct imx_mu_cfg.

i prefer not to do it.

quoted

quoted

+};
+
+static const struct of_device_id imx_mu_dt_ids[] = {
+	{ .compatible = "fsl,imx7s-mu", .data = &imx_mu_cfg_imx7d },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, imx_mu_dt_ids);
+
+static struct platform_driver imx_mu_driver = {
+	.probe		= imx_mu_probe,
+	.remove		= imx_mu_remove,
+	.driver = {
+		.name	= "imx_mu",
+		.of_match_table = imx_mu_dt_ids,
+	},
+};
+module_platform_driver(imx_mu_driver);

Do you think if we can escalated it a bit earlier as it's used by SCU?
e.g. core_initcall ?

Some more testing shoukld be done. Same MU driver is on master side a
clock consumer on slave side it is clock provider. It is valid for my
R&D project as well.
Let us concentrate on generic part first and then extend it as needed.

Okay to me.

Regards
Dong Aisheng

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20180717/08f0e58e/attachment-0001.sig>