[PATCH v2 0/5] rtc: nintendo: Add a RTC driver for the GameCube, Wii and Wii U

[PATCH v2 0/5] rtc: nintendo: Add a RTC driver for the GameCube, Wii and Wii U

[Emmanuel Gil Peyrot]

These three consoles share a device, the MX23L4005, which contains a
clock and 64 bytes of SRAM storage, and is exposed on the EXI bus
(similar to SPI) on channel 0, device 1.  This driver allows it to be
used as a Linux RTC device, where time can be read and set.

The hardware also exposes two timers, one which shuts down the console
and one which powers it on, but these aren’t supported currently.

On the Wii U, the counter bias is stored in a XML file, /config/rtc.xml,
encrypted in the SLC (eMMC storage), using a proprietary filesystem.  In
order to avoid having to implement all that, this driver assumes a
bootloader will parse this XML file and write the bias into the SRAM, at
the same location the other two consoles have it.

Changes since v1:
- Rename the driver to rtc-gamecube.
- Switch to the regmap API for debugfs support.
- Report low battery and unstable power as invalid data.
- Remove Wii U support in Kconfig, nothing specific to this console
  needs to be changed in the code.
- Don’t attempt to change HW_SRNPROT on the GameCube, this register
  doesn’t exist so we can use SRAM just fine without doing anything.
- Add needed changes to the wii device tree.
- Enable this driver on the gamecube and wii platforms.

Emmanuel Gil Peyrot (5):
  rtc: gamecube: Add a RTC driver for the GameCube, Wii and Wii U
  rtc: gamecube: Report low battery as invalid data
  powerpc: wii.dts: Expose HW_SRNPROT on this platform
  powerpc: gamecube_defconfig: Enable the RTC driver
  powerpc: wii_defconfig: Enable the RTC driver

 arch/powerpc/boot/dts/wii.dts           |   5 +
 arch/powerpc/configs/gamecube_defconfig |   2 +-
 arch/powerpc/configs/wii_defconfig      |   2 +-
 drivers/rtc/Kconfig                     |  11 +
 drivers/rtc/Makefile                    |   1 +
 drivers/rtc/rtc-gamecube.c              | 377 ++++++++++++++++++++++++
 6 files changed, 396 insertions(+), 2 deletions(-)
 create mode 100644 drivers/rtc/rtc-gamecube.c

-- 
2.33.1

[PATCH v2 1/5] rtc: gamecube: Add a RTC driver for the GameCube, Wii and Wii U

[Emmanuel Gil Peyrot]

These three consoles share a device, the MX23L4005, which contains a
clock and 64 bytes of SRAM storage, and is exposed on the EXI bus
(similar to SPI) on channel 0, device 1.  This driver allows it to be
used as a Linux RTC device, where time can be read and set.

The hardware also exposes two timers, one which shuts down the console
and one which powers it on, but these aren’t supported currently.

On the Wii U, the counter bias is stored in a XML file, /config/rtc.xml,
encrypted in the SLC (eMMC storage), using a proprietary filesystem.  In
order to avoid having to implement all that, this driver assumes a
bootloader will parse this XML file and write the bias into the SRAM, at
the same location the other two consoles have it.

Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
---
 drivers/rtc/Kconfig        |  11 ++
 drivers/rtc/Makefile       |   1 +
 drivers/rtc/rtc-gamecube.c | 347 +++++++++++++++++++++++++++++++++++++
 3 files changed, 359 insertions(+)
 create mode 100644 drivers/rtc/rtc-gamecube.c

diff --git a/drivers/rtc/Kconfig b/drivers/rtc/Kconfig
index 914497abeef9..503089ca370d 100644
--- a/drivers/rtc/Kconfig
+++ b/drivers/rtc/Kconfig
@@ -1214,6 +1214,17 @@ config RTC_DRV_V3020
 	  This driver can also be built as a module. If so, the module
 	  will be called rtc-v3020.
 
+config RTC_DRV_GAMECUBE
+	tristate "Nintendo GameCube, Wii and Wii U RTC"
+	depends on GAMECUBE || WII || COMPILE_TEST
+	select REGMAP
+	help
+	  If you say yes here you will get support for the RTC subsystem
+	  of the Nintendo GameCube, Wii and Wii U.
+
+	  This driver can also be built as a module. If so, the module
+	  will be called "rtc-gamecube".
+
 config RTC_DRV_WM831X
 	tristate "Wolfson Microelectronics WM831x RTC"
 	depends on MFD_WM831X
diff --git a/drivers/rtc/Makefile b/drivers/rtc/Makefile
index 2dd0dd956b0e..d781aaf0909c 100644
--- a/drivers/rtc/Makefile
+++ b/drivers/rtc/Makefile
@@ -108,6 +108,7 @@ obj-$(CONFIG_RTC_DRV_MT7622)	+= rtc-mt7622.o
 obj-$(CONFIG_RTC_DRV_MV)	+= rtc-mv.o
 obj-$(CONFIG_RTC_DRV_MXC)	+= rtc-mxc.o
 obj-$(CONFIG_RTC_DRV_MXC_V2)	+= rtc-mxc_v2.o
+obj-$(CONFIG_RTC_DRV_GAMECUBE)	+= rtc-gamecube.o
 obj-$(CONFIG_RTC_DRV_NTXEC)	+= rtc-ntxec.o
 obj-$(CONFIG_RTC_DRV_OMAP)	+= rtc-omap.o
 obj-$(CONFIG_RTC_DRV_OPAL)	+= rtc-opal.o
diff --git a/drivers/rtc/rtc-gamecube.c b/drivers/rtc/rtc-gamecube.c
new file mode 100644
index 000000000000..e8260c82c07d
--- /dev/null
+++ b/drivers/rtc/rtc-gamecube.c
@@ -0,0 +1,347 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Nintendo GameCube, Wii and Wii U RTC driver
+ *
+ * This driver is for the MX23L4005, more specifically its real-time clock and
+ * SRAM storage.  The value returned by the RTC counter must be added with the
+ * offset stored in a bias register in SRAM (on the GameCube and Wii) or in
+ * /config/rtc.xml (on the Wii U).  The latter being very impractical to access
+ * from Linux, this driver assumes the bootloader has read it and stored it in
+ * SRAM like for the other two consoles.
+ *
+ * This device sits on a bus named EXI (which is similar to SPI), channel 0,
+ * device 1.  This driver assumes no other user of the EXI bus, which is
+ * currently the case but would have to be reworked to add support for other
+ * GameCube hardware exposed on this bus.
+ *
+ * References:
+ * - https://wiiubrew.org/wiki/Hardware/RTC
+ * - https://wiibrew.org/wiki/MX23L4005
+ *
+ * Copyright (C) 2018 rw-r-r-0644
+ * Copyright (C) 2021 Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
+ *
+ * Based on rtc-gcn.c
+ * Copyright (C) 2004-2009 The GameCube Linux Team
+ * Copyright (C) 2005,2008,2009 Albert Herranz
+ * Based on gamecube_time.c from Torben Nielsen.
+ */
+
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/platform_device.h>
+#include <linux/regmap.h>
+#include <linux/rtc.h>
+#include <linux/time.h>
+
+/* EXI registers */
+#define EXICSR	0
+#define EXICR	12
+#define EXIDATA	16
+
+/* EXI register values */
+#define EXICSR_DEV		0x380
+	#define EXICSR_DEV1	0x100
+#define EXICSR_CLK		0x070
+	#define EXICSR_CLK_1MHZ	0x000
+	#define EXICSR_CLK_2MHZ	0x010
+	#define EXICSR_CLK_4MHZ	0x020
+	#define EXICSR_CLK_8MHZ	0x030
+	#define EXICSR_CLK_16MHZ 0x040
+	#define EXICSR_CLK_32MHZ 0x050
+#define EXICSR_INT		0x008
+	#define EXICSR_INTSET	0x008
+
+#define EXICR_TSTART		0x001
+#define EXICR_TRSMODE		0x002
+	#define EXICR_TRSMODE_IMM 0x000
+#define EXICR_TRSTYPE		0x00C
+	#define EXICR_TRSTYPE_R	0x000
+	#define EXICR_TRSTYPE_W	0x004
+#define EXICR_TLEN		0x030
+	#define EXICR_TLEN32	0x030
+
+/* EXI registers values to access the RTC */
+#define RTC_EXICSR	(EXICSR_DEV1 | EXICSR_CLK_8MHZ | EXICSR_INTSET)
+#define RTC_EXICR_W	(EXICR_TSTART | EXICR_TRSMODE_IMM | EXICR_TRSTYPE_W | EXICR_TLEN32)
+#define RTC_EXICR_R	(EXICR_TSTART | EXICR_TRSMODE_IMM | EXICR_TRSTYPE_R | EXICR_TLEN32)
+#define RTC_EXIDATA_W	0x80000000
+
+/* RTC registers */
+#define RTC_COUNTER	0x200000
+#define RTC_SRAM	0x200001
+#define RTC_SRAM_BIAS	0x200004
+#define RTC_SNAPSHOT	0x204000
+#define RTC_ONTMR	0x210000
+#define RTC_OFFTMR	0x210001
+#define RTC_TEST0	0x210004
+#define RTC_TEST1	0x210005
+#define RTC_TEST2	0x210006
+#define RTC_TEST3	0x210007
+#define RTC_CONTROL0	0x21000c
+#define RTC_CONTROL1	0x21000d
+
+struct priv {
+	struct regmap *regmap;
+	void __iomem *iob;
+	u32 rtc_bias;
+};
+
+static int exi_read(void *context, u32 reg, u32 *data)
+{
+	struct priv *d = (struct priv *)context;
+	void __iomem *iob = d->iob;
+
+	/* The spin loops here loop about 15~16 times each, so there is no need
+	 * to use a more expensive sleep method.
+	 */
+
+	/* Write register offset */
+	iowrite32be(RTC_EXICSR, iob + EXICSR);
+	iowrite32be(reg << 8, iob + EXIDATA);
+	iowrite32be(RTC_EXICR_W, iob + EXICR);
+	while (!(ioread32be(iob + EXICSR) & EXICSR_INTSET))
+		cpu_relax();
+
+	/* Read data */
+	iowrite32be(RTC_EXICSR, iob + EXICSR);
+	iowrite32be(RTC_EXICR_R, iob + EXICR);
+	while (!(ioread32be(iob + EXICSR) & EXICSR_INTSET))
+		cpu_relax();
+	*data = ioread32be(iob + EXIDATA);
+
+	/* Clear channel parameters */
+	iowrite32be(0, iob + EXICSR);
+
+	return 0;
+}
+
+static int exi_write(void *context, u32 reg, u32 data)
+{
+	struct priv *d = (struct priv *)context;
+	void __iomem *iob = d->iob;
+
+	/* The spin loops here loop about 15~16 times each, so there is no need
+	 * to use a more expensive sleep method.
+	 */
+
+	/* Write register offset */
+	iowrite32be(RTC_EXICSR, iob + EXICSR);
+	iowrite32be(RTC_EXIDATA_W | (reg << 8), iob + EXIDATA);
+	iowrite32be(RTC_EXICR_W, iob + EXICR);
+	while (!(ioread32be(iob + EXICSR) & EXICSR_INTSET))
+		cpu_relax();
+
+	/* Write data */
+	iowrite32be(RTC_EXICSR, iob + EXICSR);
+	iowrite32be(data, iob + EXIDATA);
+	iowrite32be(RTC_EXICR_W, iob + EXICR);
+	while (!(ioread32be(iob + EXICSR) & EXICSR_INTSET))
+		cpu_relax();
+
+	/* Clear channel parameters */
+	iowrite32be(0, iob + EXICSR);
+
+	return 0;
+}
+
+static const struct regmap_bus exi_bus = {
+	/* TODO: is that true?  Not that it matters here, but still. */
+	.fast_io = true,
+	.reg_read = exi_read,
+	.reg_write = exi_write,
+};
+
+static int gamecube_rtc_read_time(struct device *dev, struct rtc_time *t)
+{
+	struct priv *d = dev_get_drvdata(dev);
+	int ret;
+	u32 counter;
+	time64_t timestamp;
+
+	ret = regmap_read(d->regmap, RTC_COUNTER, &counter);
+	if (ret)
+		return ret;
+
+	/* Add the counter and the bias to obtain the timestamp */
+	timestamp = (time64_t)d->rtc_bias + counter;
+	rtc_time64_to_tm(timestamp, t);
+
+	return 0;
+}
+
+static int gamecube_rtc_set_time(struct device *dev, struct rtc_time *t)
+{
+	struct priv *d = dev_get_drvdata(dev);
+	time64_t timestamp;
+
+	/* Subtract the timestamp and the bias to obtain the counter value */
+	timestamp = rtc_tm_to_time64(t);
+	return regmap_write(d->regmap, RTC_COUNTER, timestamp - d->rtc_bias);
+}
+
+static const struct rtc_class_ops gamecube_rtc_ops = {
+	.read_time	= gamecube_rtc_read_time,
+	.set_time	= gamecube_rtc_set_time,
+};
+
+static int gamecube_rtc_read_offset_from_sram(struct priv *d)
+{
+	struct device_node *np;
+	int ret;
+	struct resource res;
+	void __iomem *hw_srnprot;
+	u32 old;
+
+	np = of_find_compatible_node(NULL, NULL, "nintendo,latte-srnprot");
+	if (!np)
+		np = of_find_compatible_node(NULL, NULL,
+					     "nintendo,hollywood-srnprot");
+	if (!np) {
+		pr_info("HW_SRNPROT not found, assuming a GameCube\n");
+		return regmap_read(d->regmap, RTC_SRAM_BIAS, &d->rtc_bias);
+	}
+
+	ret = of_address_to_resource(np, 0, &res);
+	if (ret) {
+		pr_err("no io memory range found\n");
+		return -1;
+	}
+
+	hw_srnprot = ioremap(res.start, resource_size(&res));
+	old = ioread32be(hw_srnprot);
+
+	/* TODO: figure out why we use this magic constant.  I obtained it by
+	 * reading the leftover value after boot, after IOSU already ran.
+	 *
+	 * On my Wii U, setting this register to 1 prevents the console from
+	 * rebooting properly, so wiiubrew.org must be missing something.
+	 *
+	 * See https://wiiubrew.org/wiki/Hardware/Latte_registers
+	 */
+	if (old != 0x7bf)
+		iowrite32be(0x7bf, hw_srnprot);
+
+	/* Get the offset from RTC SRAM.
+	 *
+	 * Its default location on the GameCube and on the Wii is in the SRAM,
+	 * while on the Wii U the bootloader needs to fill it with the contents
+	 * of /config/rtc.xml on the SLC (the eMMC).  We don’t do that from
+	 * Linux since it requires implementing a proprietary filesystem and do
+	 * file decryption, instead we require the bootloader to fill the same
+	 * SRAM address as on previous consoles.
+	 */
+	ret = regmap_read(d->regmap, RTC_SRAM_BIAS, &d->rtc_bias);
+	if (ret) {
+		pr_err("failed to get the RTC bias\n");
+		return -1;
+	}
+
+	/* Reset SRAM access to how it was before, our job here is done. */
+	if (old != 0x7bf)
+		iowrite32be(old, hw_srnprot);
+	iounmap(hw_srnprot);
+
+	return 0;
+}
+
+static const struct regmap_range rtc_rd_ranges[] = {
+	regmap_reg_range(0x200000, 0x200010),
+	regmap_reg_range(0x204000, 0x204000),
+	regmap_reg_range(0x210000, 0x210001),
+	regmap_reg_range(0x210004, 0x210007),
+	regmap_reg_range(0x21000c, 0x21000d),
+};
+
+static const struct regmap_access_table rtc_rd_regs = {
+	.yes_ranges =	rtc_rd_ranges,
+	.n_yes_ranges =	ARRAY_SIZE(rtc_rd_ranges),
+};
+
+static const struct regmap_range rtc_wr_ranges[] = {
+	regmap_reg_range(0x200000, 0x200010),
+	regmap_reg_range(0x204000, 0x204000),
+	regmap_reg_range(0x210000, 0x210001),
+	regmap_reg_range(0x21000d, 0x21000d),
+};
+
+static const struct regmap_access_table rtc_wr_regs = {
+	.yes_ranges =	rtc_wr_ranges,
+	.n_yes_ranges =	ARRAY_SIZE(rtc_wr_ranges),
+};
+
+static const struct regmap_config gamecube_rtc_regmap_config = {
+	.reg_bits = 24,
+	.val_bits = 32,
+	.rd_table = &rtc_rd_regs,
+	.wr_table = &rtc_wr_regs,
+	.max_register = 0x21000d,
+	.name = "gamecube-rtc",
+};
+
+static int gamecube_rtc_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct rtc_device *rtc;
+	struct priv *d;
+	int ret;
+
+	d = devm_kzalloc(dev, sizeof(struct priv), GFP_KERNEL);
+	if (IS_ERR(d))
+		return PTR_ERR(d);
+
+	d->iob = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(d->iob))
+		return PTR_ERR(d->iob);
+
+	d->regmap = devm_regmap_init(dev, &exi_bus, d,
+				     &gamecube_rtc_regmap_config);
+	if (IS_ERR(d->regmap))
+		return PTR_ERR(d->regmap);
+
+	ret = gamecube_rtc_read_offset_from_sram(d);
+	if (ret)
+		return ret;
+	dev_dbg(dev, "SRAM bias: 0x%x", d->rtc_bias);
+
+	dev_set_drvdata(dev, d);
+
+	rtc = devm_rtc_allocate_device(dev);
+	if (IS_ERR(rtc))
+		return PTR_ERR(rtc);
+
+	/* We can represent further than that, but it depends on the stored
+	 * bias and we can’t modify it persistently on all supported consoles,
+	 * so here we pretend to be limited to 2106.
+	 */
+	rtc->range_min = 0;
+	rtc->range_max = U32_MAX;
+	rtc->ops = &gamecube_rtc_ops;
+
+	devm_rtc_register_device(rtc);
+
+	return 0;
+}
+
+static const struct of_device_id gamecube_rtc_of_match[] = {
+	{.compatible = "nintendo,latte-exi" },
+	{.compatible = "nintendo,hollywood-exi" },
+	{.compatible = "nintendo,flipper-exi" },
+	{ }
+};
+MODULE_DEVICE_TABLE(of, gamecube_rtc_of_match);
+
+static struct platform_driver gamecube_rtc_driver = {
+	.probe		= gamecube_rtc_probe,
+	.driver		= {
+		.name	= "rtc-gamecube",
+		.of_match_table	= gamecube_rtc_of_match,
+	},
+};
+module_platform_driver(gamecube_rtc_driver);
+
+MODULE_AUTHOR("Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>");
+MODULE_DESCRIPTION("Nintendo GameCube, Wii and Wii U RTC driver");
+MODULE_LICENSE("GPL");
-- 
2.33.1

[PATCH v2 2/5] rtc: gamecube: Report low battery as invalid data

[Emmanuel Gil Peyrot]

I haven’t been able to test this patch as all of my consoles have a
working RTC battery, but according to the documentation it should work
like that.

Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
---
 drivers/rtc/rtc-gamecube.c | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

diff --git a/drivers/rtc/rtc-gamecube.c b/drivers/rtc/rtc-gamecube.c
index e8260c82c07d..1932c6fe1301 100644
--- a/drivers/rtc/rtc-gamecube.c
+++ b/drivers/rtc/rtc-gamecube.c
@@ -83,6 +83,10 @@
 #define RTC_CONTROL0	0x21000c
 #define RTC_CONTROL1	0x21000d
 
+/* RTC flags */
+#define RTC_CONTROL0_UNSTABLE_POWER	0x00000800
+#define RTC_CONTROL0_LOW_BATTERY	0x00000200
+
 struct priv {
 	struct regmap *regmap;
 	void __iomem *iob;
@@ -182,9 +186,35 @@ static int gamecube_rtc_set_time(struct device *dev, struct rtc_time *t)
 	return regmap_write(d->regmap, RTC_COUNTER, timestamp - d->rtc_bias);
 }
 
+static int gamecube_rtc_ioctl(struct device *dev, unsigned int cmd, unsigned long arg)
+{
+	struct priv *d = dev_get_drvdata(dev);
+	int value;
+	int control0;
+	int ret;
+
+	switch (cmd) {
+	case RTC_VL_READ:
+		ret = regmap_read(d->regmap, RTC_CONTROL0, &control0);
+		if (ret)
+			return ret;
+
+		value = 0;
+		if (control0 & RTC_CONTROL0_UNSTABLE_POWER)
+			value |= RTC_VL_DATA_INVALID;
+		if (control0 & RTC_CONTROL0_LOW_BATTERY)
+			value |= RTC_VL_DATA_INVALID;
+		return put_user(value, (unsigned int __user *)arg);
+
+	default:
+		return -ENOIOCTLCMD;
+	}
+}
+
 static const struct rtc_class_ops gamecube_rtc_ops = {
 	.read_time	= gamecube_rtc_read_time,
 	.set_time	= gamecube_rtc_set_time,
+	.ioctl		= gamecube_rtc_ioctl,
 };
 
 static int gamecube_rtc_read_offset_from_sram(struct priv *d)
-- 
2.33.1

Re: [PATCH v2 2/5] rtc: gamecube: Report low battery as invalid data

[Alexandre Belloni]

Hello,

On 28/10/2021 00:35:12+0200, Emmanuel Gil Peyrot wrote:
> I haven’t been able to test this patch as all of my consoles have a
> working RTC battery, but according to the documentation it should work
> like that.
> 
> Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
> ---
>  drivers/rtc/rtc-gamecube.c | 30 ++++++++++++++++++++++++++++++
>  1 file changed, 30 insertions(+)
> 
> diff --git a/drivers/rtc/rtc-gamecube.c b/drivers/rtc/rtc-gamecube.c
> index e8260c82c07d..1932c6fe1301 100644
> --- a/drivers/rtc/rtc-gamecube.c
> +++ b/drivers/rtc/rtc-gamecube.c
> @@ -83,6 +83,10 @@
>  #define RTC_CONTROL0	0x21000c
>  #define RTC_CONTROL1	0x21000d
>  
> +/* RTC flags */
> +#define RTC_CONTROL0_UNSTABLE_POWER	0x00000800
> +#define RTC_CONTROL0_LOW_BATTERY	0x00000200
> +
>  struct priv {
>  	struct regmap *regmap;
>  	void __iomem *iob;
> @@ -182,9 +186,35 @@ static int gamecube_rtc_set_time(struct device *dev, struct rtc_time *t)
>  	return regmap_write(d->regmap, RTC_COUNTER, timestamp - d->rtc_bias);
>  }
>  
> +static int gamecube_rtc_ioctl(struct device *dev, unsigned int cmd, unsigned long arg)
> +{
> +	struct priv *d = dev_get_drvdata(dev);
> +	int value;
> +	int control0;
> +	int ret;
> +
> +	switch (cmd) {
> +	case RTC_VL_READ:
> +		ret = regmap_read(d->regmap, RTC_CONTROL0, &control0);
> +		if (ret)
> +			return ret;
> +
> +		value = 0;
> +		if (control0 & RTC_CONTROL0_UNSTABLE_POWER)
> +			value |= RTC_VL_DATA_INVALID;
> +		if (control0 & RTC_CONTROL0_LOW_BATTERY)
> +			value |= RTC_VL_DATA_INVALID;

Shouldn't that one be RTC_VL_BACKUP_LOW?

Else, the driver is great, I'm ready to apply it.

> +		return put_user(value, (unsigned int __user *)arg);
> +
> +	default:
> +		return -ENOIOCTLCMD;
> +	}
> +}
> +
>  static const struct rtc_class_ops gamecube_rtc_ops = {
>  	.read_time	= gamecube_rtc_read_time,
>  	.set_time	= gamecube_rtc_set_time,
> +	.ioctl		= gamecube_rtc_ioctl,
>  };
>  
>  static int gamecube_rtc_read_offset_from_sram(struct priv *d)
> -- 
> 2.33.1
> 

-- 
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

Re: [PATCH v2 2/5] rtc: gamecube: Report low battery as invalid data

[Emmanuel Gil Peyrot]

[-- Attachment #1: Type: text/plain, Size: 2567 bytes --]

On Tue, Nov 30, 2021 at 11:45:18PM +0100, Alexandre Belloni wrote:
> Hello,

Hi,

> 
> On 28/10/2021 00:35:12+0200, Emmanuel Gil Peyrot wrote:
> > I haven’t been able to test this patch as all of my consoles have a
> > working RTC battery, but according to the documentation it should work
> > like that.
> > 
> > Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
> > ---
> >  drivers/rtc/rtc-gamecube.c | 30 ++++++++++++++++++++++++++++++
> >  1 file changed, 30 insertions(+)
> > 
> > diff --git a/drivers/rtc/rtc-gamecube.c b/drivers/rtc/rtc-gamecube.c
> > index e8260c82c07d..1932c6fe1301 100644
> > --- a/drivers/rtc/rtc-gamecube.c
> > +++ b/drivers/rtc/rtc-gamecube.c
> > @@ -83,6 +83,10 @@
> >  #define RTC_CONTROL0	0x21000c
> >  #define RTC_CONTROL1	0x21000d
> >  
> > +/* RTC flags */
> > +#define RTC_CONTROL0_UNSTABLE_POWER	0x00000800
> > +#define RTC_CONTROL0_LOW_BATTERY	0x00000200
> > +
> >  struct priv {
> >  	struct regmap *regmap;
> >  	void __iomem *iob;
> > @@ -182,9 +186,35 @@ static int gamecube_rtc_set_time(struct device *dev, struct rtc_time *t)
> >  	return regmap_write(d->regmap, RTC_COUNTER, timestamp - d->rtc_bias);
> >  }
> >  
> > +static int gamecube_rtc_ioctl(struct device *dev, unsigned int cmd, unsigned long arg)
> > +{
> > +	struct priv *d = dev_get_drvdata(dev);
> > +	int value;
> > +	int control0;
> > +	int ret;
> > +
> > +	switch (cmd) {
> > +	case RTC_VL_READ:
> > +		ret = regmap_read(d->regmap, RTC_CONTROL0, &control0);
> > +		if (ret)
> > +			return ret;
> > +
> > +		value = 0;
> > +		if (control0 & RTC_CONTROL0_UNSTABLE_POWER)
> > +			value |= RTC_VL_DATA_INVALID;
> > +		if (control0 & RTC_CONTROL0_LOW_BATTERY)
> > +			value |= RTC_VL_DATA_INVALID;
> 
> Shouldn't that one be RTC_VL_BACKUP_LOW?

Oops indeed, that seems like a better report, I’ll send a v3 with this
change only.

> 
> Else, the driver is great, I'm ready to apply it.

Perfect!

> 
> > +		return put_user(value, (unsigned int __user *)arg);
> > +
> > +	default:
> > +		return -ENOIOCTLCMD;
> > +	}
> > +}
> > +
> >  static const struct rtc_class_ops gamecube_rtc_ops = {
> >  	.read_time	= gamecube_rtc_read_time,
> >  	.set_time	= gamecube_rtc_set_time,
> > +	.ioctl		= gamecube_rtc_ioctl,
> >  };
> >  
> >  static int gamecube_rtc_read_offset_from_sram(struct priv *d)
> > -- 
> > 2.33.1
> > 
> 
> -- 
> Alexandre Belloni, co-owner and COO, Bootlin
> Embedded Linux and Kernel engineering
> https://bootlin.com

-- 
Emmanuel Gil Peyrot

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

[PATCH v2 3/5] powerpc: wii.dts: Expose HW_SRNPROT on this platform

[Emmanuel Gil Peyrot]

This Hollywood register isn’t properly understood, but can allow or
reject access to the SRAM, which we need to set for RTC usage if it
isn’t previously set correctly beforehand.

See https://wiibrew.org/wiki/Hardware/Hollywood_Registers#HW_SRNPROT

Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
---
 arch/powerpc/boot/dts/wii.dts | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/arch/powerpc/boot/dts/wii.dts b/arch/powerpc/boot/dts/wii.dts
index c5720fdd0686..34d9732d5910 100644
--- a/arch/powerpc/boot/dts/wii.dts
+++ b/arch/powerpc/boot/dts/wii.dts
@@ -175,6 +175,11 @@ PIC1: pic1@d800030 {
 			interrupts = <14>;
 		};
 
+		srnprot@d800060 {
+			compatible = "nintendo,hollywood-srnprot";
+			reg = <0x0d800060 0x4>;
+		};
+
 		GPIO: gpio@d8000c0 {
 			#gpio-cells = <2>;
 			compatible = "nintendo,hollywood-gpio";
-- 
2.33.1

[PATCH v2 4/5] powerpc: gamecube_defconfig: Enable the RTC driver

[Emmanuel Gil Peyrot]

This selects the rtc-gamecube driver, which provides a real-time clock
on this platform.

Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
---
 arch/powerpc/configs/gamecube_defconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/powerpc/configs/gamecube_defconfig b/arch/powerpc/configs/gamecube_defconfig
index 24c0e0ea5aeb..91a1b99f4e8f 100644
--- a/arch/powerpc/configs/gamecube_defconfig
+++ b/arch/powerpc/configs/gamecube_defconfig
@@ -68,7 +68,7 @@ CONFIG_SND_SEQUENCER=y
 CONFIG_SND_SEQUENCER_OSS=y
 # CONFIG_USB_SUPPORT is not set
 CONFIG_RTC_CLASS=y
-CONFIG_RTC_DRV_GENERIC=y
+CONFIG_RTC_DRV_GAMECUBE=y
 CONFIG_EXT2_FS=y
 CONFIG_EXT4_FS=y
 CONFIG_ISO9660_FS=y
-- 
2.33.1

[PATCH v2 5/5] powerpc: wii_defconfig: Enable the RTC driver

[Emmanuel Gil Peyrot]

This selects the rtc-gamecube driver, which provides a real-time clock
on this platform.

Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
---
 arch/powerpc/configs/wii_defconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/powerpc/configs/wii_defconfig b/arch/powerpc/configs/wii_defconfig
index 752e081d28d0..ad4302a12fd7 100644
--- a/arch/powerpc/configs/wii_defconfig
+++ b/arch/powerpc/configs/wii_defconfig
@@ -98,7 +98,7 @@ CONFIG_LEDS_TRIGGERS=y
 CONFIG_LEDS_TRIGGER_HEARTBEAT=y
 CONFIG_LEDS_TRIGGER_PANIC=y
 CONFIG_RTC_CLASS=y
-CONFIG_RTC_DRV_GENERIC=y
+CONFIG_RTC_DRV_GAMECUBE=y
 CONFIG_EXT2_FS=y
 CONFIG_EXT4_FS=y
 CONFIG_FUSE_FS=m
-- 
2.33.1

[PATCH v3 0/5] rtc: nintendo: Add a RTC driver for the GameCube, Wii and Wii U

[Emmanuel Gil Peyrot]

These three consoles share a device, the MX23L4005, which contains a
clock and 64 bytes of SRAM storage, and is exposed on the EXI bus
(similar to SPI) on channel 0, device 1.  This driver allows it to be
used as a Linux RTC device, where time can be read and set.

The hardware also exposes two timers, one which shuts down the console
and one which powers it on, but these aren’t supported currently.

On the Wii U, the counter bias is stored in a XML file, /config/rtc.xml,
encrypted in the SLC (eMMC storage), using a proprietary filesystem.  In
order to avoid having to implement all that, this driver assumes a
bootloader will parse this XML file and write the bias into the SRAM, at
the same location the other two consoles have it.

Changes since v1:
- Rename the driver to rtc-gamecube.
- Switch to the regmap API for debugfs support.
- Report low battery and unstable power as invalid data.
- Remove Wii U support in Kconfig, nothing specific to this console
  needs to be changed in the code.
- Don’t attempt to change HW_SRNPROT on the GameCube, this register
  doesn’t exist so we can use SRAM just fine without doing anything.
- Add needed changes to the wii device tree.
- Enable this driver on the gamecube and wii platforms.

Changes since v2:
- Report low battery correctly.

Emmanuel Gil Peyrot (5):
  rtc: gamecube: Add a RTC driver for the GameCube, Wii and Wii U
  rtc: gamecube: Report low battery as invalid data
  powerpc: wii.dts: Expose HW_SRNPROT on this platform
  powerpc: gamecube_defconfig: Enable the RTC driver
  powerpc: wii_defconfig: Enable the RTC driver

 arch/powerpc/boot/dts/wii.dts           |   5 +
 arch/powerpc/configs/gamecube_defconfig |   2 +-
 arch/powerpc/configs/wii_defconfig      |   2 +-
 drivers/rtc/Kconfig                     |  11 +
 drivers/rtc/Makefile                    |   1 +
 drivers/rtc/rtc-gamecube.c              | 377 ++++++++++++++++++++++++
 6 files changed, 396 insertions(+), 2 deletions(-)
 create mode 100644 drivers/rtc/rtc-gamecube.c

-- 
2.34.1

[PATCH v3 1/5] rtc: gamecube: Add a RTC driver for the GameCube, Wii and Wii U

[Emmanuel Gil Peyrot]

These three consoles share a device, the MX23L4005, which contains a
clock and 64 bytes of SRAM storage, and is exposed on the EXI bus
(similar to SPI) on channel 0, device 1.  This driver allows it to be
used as a Linux RTC device, where time can be read and set.

The hardware also exposes two timers, one which shuts down the console
and one which powers it on, but these aren’t supported currently.

On the Wii U, the counter bias is stored in a XML file, /config/rtc.xml,
encrypted in the SLC (eMMC storage), using a proprietary filesystem.  In
order to avoid having to implement all that, this driver assumes a
bootloader will parse this XML file and write the bias into the SRAM, at
the same location the other two consoles have it.

Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
---
 drivers/rtc/Kconfig        |  11 ++
 drivers/rtc/Makefile       |   1 +
 drivers/rtc/rtc-gamecube.c | 347 +++++++++++++++++++++++++++++++++++++
 3 files changed, 359 insertions(+)
 create mode 100644 drivers/rtc/rtc-gamecube.c

diff --git a/drivers/rtc/Kconfig b/drivers/rtc/Kconfig
index 914497abeef9..503089ca370d 100644
--- a/drivers/rtc/Kconfig
+++ b/drivers/rtc/Kconfig
@@ -1214,6 +1214,17 @@ config RTC_DRV_V3020
 	  This driver can also be built as a module. If so, the module
 	  will be called rtc-v3020.
 
+config RTC_DRV_GAMECUBE
+	tristate "Nintendo GameCube, Wii and Wii U RTC"
+	depends on GAMECUBE || WII || COMPILE_TEST
+	select REGMAP
+	help
+	  If you say yes here you will get support for the RTC subsystem
+	  of the Nintendo GameCube, Wii and Wii U.
+
+	  This driver can also be built as a module. If so, the module
+	  will be called "rtc-gamecube".
+
 config RTC_DRV_WM831X
 	tristate "Wolfson Microelectronics WM831x RTC"
 	depends on MFD_WM831X
diff --git a/drivers/rtc/Makefile b/drivers/rtc/Makefile
index 2dd0dd956b0e..d781aaf0909c 100644
--- a/drivers/rtc/Makefile
+++ b/drivers/rtc/Makefile
@@ -108,6 +108,7 @@ obj-$(CONFIG_RTC_DRV_MT7622)	+= rtc-mt7622.o
 obj-$(CONFIG_RTC_DRV_MV)	+= rtc-mv.o
 obj-$(CONFIG_RTC_DRV_MXC)	+= rtc-mxc.o
 obj-$(CONFIG_RTC_DRV_MXC_V2)	+= rtc-mxc_v2.o
+obj-$(CONFIG_RTC_DRV_GAMECUBE)	+= rtc-gamecube.o
 obj-$(CONFIG_RTC_DRV_NTXEC)	+= rtc-ntxec.o
 obj-$(CONFIG_RTC_DRV_OMAP)	+= rtc-omap.o
 obj-$(CONFIG_RTC_DRV_OPAL)	+= rtc-opal.o
diff --git a/drivers/rtc/rtc-gamecube.c b/drivers/rtc/rtc-gamecube.c
new file mode 100644
index 000000000000..e8260c82c07d
--- /dev/null
+++ b/drivers/rtc/rtc-gamecube.c
@@ -0,0 +1,347 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Nintendo GameCube, Wii and Wii U RTC driver
+ *
+ * This driver is for the MX23L4005, more specifically its real-time clock and
+ * SRAM storage.  The value returned by the RTC counter must be added with the
+ * offset stored in a bias register in SRAM (on the GameCube and Wii) or in
+ * /config/rtc.xml (on the Wii U).  The latter being very impractical to access
+ * from Linux, this driver assumes the bootloader has read it and stored it in
+ * SRAM like for the other two consoles.
+ *
+ * This device sits on a bus named EXI (which is similar to SPI), channel 0,
+ * device 1.  This driver assumes no other user of the EXI bus, which is
+ * currently the case but would have to be reworked to add support for other
+ * GameCube hardware exposed on this bus.
+ *
+ * References:
+ * - https://wiiubrew.org/wiki/Hardware/RTC
+ * - https://wiibrew.org/wiki/MX23L4005
+ *
+ * Copyright (C) 2018 rw-r-r-0644
+ * Copyright (C) 2021 Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
+ *
+ * Based on rtc-gcn.c
+ * Copyright (C) 2004-2009 The GameCube Linux Team
+ * Copyright (C) 2005,2008,2009 Albert Herranz
+ * Based on gamecube_time.c from Torben Nielsen.
+ */
+
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/platform_device.h>
+#include <linux/regmap.h>
+#include <linux/rtc.h>
+#include <linux/time.h>
+
+/* EXI registers */
+#define EXICSR	0
+#define EXICR	12
+#define EXIDATA	16
+
+/* EXI register values */
+#define EXICSR_DEV		0x380
+	#define EXICSR_DEV1	0x100
+#define EXICSR_CLK		0x070
+	#define EXICSR_CLK_1MHZ	0x000
+	#define EXICSR_CLK_2MHZ	0x010
+	#define EXICSR_CLK_4MHZ	0x020
+	#define EXICSR_CLK_8MHZ	0x030
+	#define EXICSR_CLK_16MHZ 0x040
+	#define EXICSR_CLK_32MHZ 0x050
+#define EXICSR_INT		0x008
+	#define EXICSR_INTSET	0x008
+
+#define EXICR_TSTART		0x001
+#define EXICR_TRSMODE		0x002
+	#define EXICR_TRSMODE_IMM 0x000
+#define EXICR_TRSTYPE		0x00C
+	#define EXICR_TRSTYPE_R	0x000
+	#define EXICR_TRSTYPE_W	0x004
+#define EXICR_TLEN		0x030
+	#define EXICR_TLEN32	0x030
+
+/* EXI registers values to access the RTC */
+#define RTC_EXICSR	(EXICSR_DEV1 | EXICSR_CLK_8MHZ | EXICSR_INTSET)
+#define RTC_EXICR_W	(EXICR_TSTART | EXICR_TRSMODE_IMM | EXICR_TRSTYPE_W | EXICR_TLEN32)
+#define RTC_EXICR_R	(EXICR_TSTART | EXICR_TRSMODE_IMM | EXICR_TRSTYPE_R | EXICR_TLEN32)
+#define RTC_EXIDATA_W	0x80000000
+
+/* RTC registers */
+#define RTC_COUNTER	0x200000
+#define RTC_SRAM	0x200001
+#define RTC_SRAM_BIAS	0x200004
+#define RTC_SNAPSHOT	0x204000
+#define RTC_ONTMR	0x210000
+#define RTC_OFFTMR	0x210001
+#define RTC_TEST0	0x210004
+#define RTC_TEST1	0x210005
+#define RTC_TEST2	0x210006
+#define RTC_TEST3	0x210007
+#define RTC_CONTROL0	0x21000c
+#define RTC_CONTROL1	0x21000d
+
+struct priv {
+	struct regmap *regmap;
+	void __iomem *iob;
+	u32 rtc_bias;
+};
+
+static int exi_read(void *context, u32 reg, u32 *data)
+{
+	struct priv *d = (struct priv *)context;
+	void __iomem *iob = d->iob;
+
+	/* The spin loops here loop about 15~16 times each, so there is no need
+	 * to use a more expensive sleep method.
+	 */
+
+	/* Write register offset */
+	iowrite32be(RTC_EXICSR, iob + EXICSR);
+	iowrite32be(reg << 8, iob + EXIDATA);
+	iowrite32be(RTC_EXICR_W, iob + EXICR);
+	while (!(ioread32be(iob + EXICSR) & EXICSR_INTSET))
+		cpu_relax();
+
+	/* Read data */
+	iowrite32be(RTC_EXICSR, iob + EXICSR);
+	iowrite32be(RTC_EXICR_R, iob + EXICR);
+	while (!(ioread32be(iob + EXICSR) & EXICSR_INTSET))
+		cpu_relax();
+	*data = ioread32be(iob + EXIDATA);
+
+	/* Clear channel parameters */
+	iowrite32be(0, iob + EXICSR);
+
+	return 0;
+}
+
+static int exi_write(void *context, u32 reg, u32 data)
+{
+	struct priv *d = (struct priv *)context;
+	void __iomem *iob = d->iob;
+
+	/* The spin loops here loop about 15~16 times each, so there is no need
+	 * to use a more expensive sleep method.
+	 */
+
+	/* Write register offset */
+	iowrite32be(RTC_EXICSR, iob + EXICSR);
+	iowrite32be(RTC_EXIDATA_W | (reg << 8), iob + EXIDATA);
+	iowrite32be(RTC_EXICR_W, iob + EXICR);
+	while (!(ioread32be(iob + EXICSR) & EXICSR_INTSET))
+		cpu_relax();
+
+	/* Write data */
+	iowrite32be(RTC_EXICSR, iob + EXICSR);
+	iowrite32be(data, iob + EXIDATA);
+	iowrite32be(RTC_EXICR_W, iob + EXICR);
+	while (!(ioread32be(iob + EXICSR) & EXICSR_INTSET))
+		cpu_relax();
+
+	/* Clear channel parameters */
+	iowrite32be(0, iob + EXICSR);
+
+	return 0;
+}
+
+static const struct regmap_bus exi_bus = {
+	/* TODO: is that true?  Not that it matters here, but still. */
+	.fast_io = true,
+	.reg_read = exi_read,
+	.reg_write = exi_write,
+};
+
+static int gamecube_rtc_read_time(struct device *dev, struct rtc_time *t)
+{
+	struct priv *d = dev_get_drvdata(dev);
+	int ret;
+	u32 counter;
+	time64_t timestamp;
+
+	ret = regmap_read(d->regmap, RTC_COUNTER, &counter);
+	if (ret)
+		return ret;
+
+	/* Add the counter and the bias to obtain the timestamp */
+	timestamp = (time64_t)d->rtc_bias + counter;
+	rtc_time64_to_tm(timestamp, t);
+
+	return 0;
+}
+
+static int gamecube_rtc_set_time(struct device *dev, struct rtc_time *t)
+{
+	struct priv *d = dev_get_drvdata(dev);
+	time64_t timestamp;
+
+	/* Subtract the timestamp and the bias to obtain the counter value */
+	timestamp = rtc_tm_to_time64(t);
+	return regmap_write(d->regmap, RTC_COUNTER, timestamp - d->rtc_bias);
+}
+
+static const struct rtc_class_ops gamecube_rtc_ops = {
+	.read_time	= gamecube_rtc_read_time,
+	.set_time	= gamecube_rtc_set_time,
+};
+
+static int gamecube_rtc_read_offset_from_sram(struct priv *d)
+{
+	struct device_node *np;
+	int ret;
+	struct resource res;
+	void __iomem *hw_srnprot;
+	u32 old;
+
+	np = of_find_compatible_node(NULL, NULL, "nintendo,latte-srnprot");
+	if (!np)
+		np = of_find_compatible_node(NULL, NULL,
+					     "nintendo,hollywood-srnprot");
+	if (!np) {
+		pr_info("HW_SRNPROT not found, assuming a GameCube\n");
+		return regmap_read(d->regmap, RTC_SRAM_BIAS, &d->rtc_bias);
+	}
+
+	ret = of_address_to_resource(np, 0, &res);
+	if (ret) {
+		pr_err("no io memory range found\n");
+		return -1;
+	}
+
+	hw_srnprot = ioremap(res.start, resource_size(&res));
+	old = ioread32be(hw_srnprot);
+
+	/* TODO: figure out why we use this magic constant.  I obtained it by
+	 * reading the leftover value after boot, after IOSU already ran.
+	 *
+	 * On my Wii U, setting this register to 1 prevents the console from
+	 * rebooting properly, so wiiubrew.org must be missing something.
+	 *
+	 * See https://wiiubrew.org/wiki/Hardware/Latte_registers
+	 */
+	if (old != 0x7bf)
+		iowrite32be(0x7bf, hw_srnprot);
+
+	/* Get the offset from RTC SRAM.
+	 *
+	 * Its default location on the GameCube and on the Wii is in the SRAM,
+	 * while on the Wii U the bootloader needs to fill it with the contents
+	 * of /config/rtc.xml on the SLC (the eMMC).  We don’t do that from
+	 * Linux since it requires implementing a proprietary filesystem and do
+	 * file decryption, instead we require the bootloader to fill the same
+	 * SRAM address as on previous consoles.
+	 */
+	ret = regmap_read(d->regmap, RTC_SRAM_BIAS, &d->rtc_bias);
+	if (ret) {
+		pr_err("failed to get the RTC bias\n");
+		return -1;
+	}
+
+	/* Reset SRAM access to how it was before, our job here is done. */
+	if (old != 0x7bf)
+		iowrite32be(old, hw_srnprot);
+	iounmap(hw_srnprot);
+
+	return 0;
+}
+
+static const struct regmap_range rtc_rd_ranges[] = {
+	regmap_reg_range(0x200000, 0x200010),
+	regmap_reg_range(0x204000, 0x204000),
+	regmap_reg_range(0x210000, 0x210001),
+	regmap_reg_range(0x210004, 0x210007),
+	regmap_reg_range(0x21000c, 0x21000d),
+};
+
+static const struct regmap_access_table rtc_rd_regs = {
+	.yes_ranges =	rtc_rd_ranges,
+	.n_yes_ranges =	ARRAY_SIZE(rtc_rd_ranges),
+};
+
+static const struct regmap_range rtc_wr_ranges[] = {
+	regmap_reg_range(0x200000, 0x200010),
+	regmap_reg_range(0x204000, 0x204000),
+	regmap_reg_range(0x210000, 0x210001),
+	regmap_reg_range(0x21000d, 0x21000d),
+};
+
+static const struct regmap_access_table rtc_wr_regs = {
+	.yes_ranges =	rtc_wr_ranges,
+	.n_yes_ranges =	ARRAY_SIZE(rtc_wr_ranges),
+};
+
+static const struct regmap_config gamecube_rtc_regmap_config = {
+	.reg_bits = 24,
+	.val_bits = 32,
+	.rd_table = &rtc_rd_regs,
+	.wr_table = &rtc_wr_regs,
+	.max_register = 0x21000d,
+	.name = "gamecube-rtc",
+};
+
+static int gamecube_rtc_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct rtc_device *rtc;
+	struct priv *d;
+	int ret;
+
+	d = devm_kzalloc(dev, sizeof(struct priv), GFP_KERNEL);
+	if (IS_ERR(d))
+		return PTR_ERR(d);
+
+	d->iob = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(d->iob))
+		return PTR_ERR(d->iob);
+
+	d->regmap = devm_regmap_init(dev, &exi_bus, d,
+				     &gamecube_rtc_regmap_config);
+	if (IS_ERR(d->regmap))
+		return PTR_ERR(d->regmap);
+
+	ret = gamecube_rtc_read_offset_from_sram(d);
+	if (ret)
+		return ret;
+	dev_dbg(dev, "SRAM bias: 0x%x", d->rtc_bias);
+
+	dev_set_drvdata(dev, d);
+
+	rtc = devm_rtc_allocate_device(dev);
+	if (IS_ERR(rtc))
+		return PTR_ERR(rtc);
+
+	/* We can represent further than that, but it depends on the stored
+	 * bias and we can’t modify it persistently on all supported consoles,
+	 * so here we pretend to be limited to 2106.
+	 */
+	rtc->range_min = 0;
+	rtc->range_max = U32_MAX;
+	rtc->ops = &gamecube_rtc_ops;
+
+	devm_rtc_register_device(rtc);
+
+	return 0;
+}
+
+static const struct of_device_id gamecube_rtc_of_match[] = {
+	{.compatible = "nintendo,latte-exi" },
+	{.compatible = "nintendo,hollywood-exi" },
+	{.compatible = "nintendo,flipper-exi" },
+	{ }
+};
+MODULE_DEVICE_TABLE(of, gamecube_rtc_of_match);
+
+static struct platform_driver gamecube_rtc_driver = {
+	.probe		= gamecube_rtc_probe,
+	.driver		= {
+		.name	= "rtc-gamecube",
+		.of_match_table	= gamecube_rtc_of_match,
+	},
+};
+module_platform_driver(gamecube_rtc_driver);
+
+MODULE_AUTHOR("Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>");
+MODULE_DESCRIPTION("Nintendo GameCube, Wii and Wii U RTC driver");
+MODULE_LICENSE("GPL");
-- 
2.34.1

[PATCH v3 2/5] rtc: gamecube: Report low battery as invalid data

[Emmanuel Gil Peyrot]

I haven’t been able to test this patch as all of my consoles have a
working RTC battery, but according to the documentation it should work
like that.

Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
---
 drivers/rtc/rtc-gamecube.c | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

diff --git a/drivers/rtc/rtc-gamecube.c b/drivers/rtc/rtc-gamecube.c
index e8260c82c07d..98128746171e 100644
--- a/drivers/rtc/rtc-gamecube.c
+++ b/drivers/rtc/rtc-gamecube.c
@@ -83,6 +83,10 @@
 #define RTC_CONTROL0	0x21000c
 #define RTC_CONTROL1	0x21000d
 
+/* RTC flags */
+#define RTC_CONTROL0_UNSTABLE_POWER	0x00000800
+#define RTC_CONTROL0_LOW_BATTERY	0x00000200
+
 struct priv {
 	struct regmap *regmap;
 	void __iomem *iob;
@@ -182,9 +186,35 @@ static int gamecube_rtc_set_time(struct device *dev, struct rtc_time *t)
 	return regmap_write(d->regmap, RTC_COUNTER, timestamp - d->rtc_bias);
 }
 
+static int gamecube_rtc_ioctl(struct device *dev, unsigned int cmd, unsigned long arg)
+{
+	struct priv *d = dev_get_drvdata(dev);
+	int value;
+	int control0;
+	int ret;
+
+	switch (cmd) {
+	case RTC_VL_READ:
+		ret = regmap_read(d->regmap, RTC_CONTROL0, &control0);
+		if (ret)
+			return ret;
+
+		value = 0;
+		if (control0 & RTC_CONTROL0_UNSTABLE_POWER)
+			value |= RTC_VL_DATA_INVALID;
+		if (control0 & RTC_CONTROL0_LOW_BATTERY)
+			value |= RTC_VL_BACKUP_LOW;
+		return put_user(value, (unsigned int __user *)arg);
+
+	default:
+		return -ENOIOCTLCMD;
+	}
+}
+
 static const struct rtc_class_ops gamecube_rtc_ops = {
 	.read_time	= gamecube_rtc_read_time,
 	.set_time	= gamecube_rtc_set_time,
+	.ioctl		= gamecube_rtc_ioctl,
 };
 
 static int gamecube_rtc_read_offset_from_sram(struct priv *d)
-- 
2.34.1

[PATCH v3 3/5] powerpc: wii.dts: Expose HW_SRNPROT on this platform

[Emmanuel Gil Peyrot]

This Hollywood register isn’t properly understood, but can allow or
reject access to the SRAM, which we need to set for RTC usage if it
isn’t previously set correctly beforehand.

See https://wiibrew.org/wiki/Hardware/Hollywood_Registers#HW_SRNPROT

Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
---
 arch/powerpc/boot/dts/wii.dts | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/arch/powerpc/boot/dts/wii.dts b/arch/powerpc/boot/dts/wii.dts
index c5720fdd0686..34d9732d5910 100644
--- a/arch/powerpc/boot/dts/wii.dts
+++ b/arch/powerpc/boot/dts/wii.dts
@@ -175,6 +175,11 @@ PIC1: pic1@d800030 {
 			interrupts = <14>;
 		};
 
+		srnprot@d800060 {
+			compatible = "nintendo,hollywood-srnprot";
+			reg = <0x0d800060 0x4>;
+		};
+
 		GPIO: gpio@d8000c0 {
 			#gpio-cells = <2>;
 			compatible = "nintendo,hollywood-gpio";
-- 
2.34.1

[PATCH v3 4/5] powerpc: gamecube_defconfig: Enable the RTC driver

[Emmanuel Gil Peyrot]

This selects the rtc-gamecube driver, which provides a real-time clock
on this platform.

Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
---
 arch/powerpc/configs/gamecube_defconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/powerpc/configs/gamecube_defconfig b/arch/powerpc/configs/gamecube_defconfig
index 24c0e0ea5aeb..91a1b99f4e8f 100644
--- a/arch/powerpc/configs/gamecube_defconfig
+++ b/arch/powerpc/configs/gamecube_defconfig
@@ -68,7 +68,7 @@ CONFIG_SND_SEQUENCER=y
 CONFIG_SND_SEQUENCER_OSS=y
 # CONFIG_USB_SUPPORT is not set
 CONFIG_RTC_CLASS=y
-CONFIG_RTC_DRV_GENERIC=y
+CONFIG_RTC_DRV_GAMECUBE=y
 CONFIG_EXT2_FS=y
 CONFIG_EXT4_FS=y
 CONFIG_ISO9660_FS=y
-- 
2.34.1

[PATCH v3 5/5] powerpc: wii_defconfig: Enable the RTC driver

[Emmanuel Gil Peyrot]

This selects the rtc-gamecube driver, which provides a real-time clock
on this platform.

Signed-off-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
---
 arch/powerpc/configs/wii_defconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/powerpc/configs/wii_defconfig b/arch/powerpc/configs/wii_defconfig
index 752e081d28d0..ad4302a12fd7 100644
--- a/arch/powerpc/configs/wii_defconfig
+++ b/arch/powerpc/configs/wii_defconfig
@@ -98,7 +98,7 @@ CONFIG_LEDS_TRIGGERS=y
 CONFIG_LEDS_TRIGGER_HEARTBEAT=y
 CONFIG_LEDS_TRIGGER_PANIC=y
 CONFIG_RTC_CLASS=y
-CONFIG_RTC_DRV_GENERIC=y
+CONFIG_RTC_DRV_GAMECUBE=y
 CONFIG_EXT2_FS=y
 CONFIG_EXT4_FS=y
 CONFIG_FUSE_FS=m
-- 
2.34.1

Re: [PATCH v3 0/5] rtc: nintendo: Add a RTC driver for the GameCube, Wii and Wii U

[Michael Ellerman]

Emmanuel Gil Peyrot <linkmauve@linkmauve.fr> writes:
> These three consoles share a device, the MX23L4005, which contains a
> clock and 64 bytes of SRAM storage, and is exposed on the EXI bus
> (similar to SPI) on channel 0, device 1.  This driver allows it to be
> used as a Linux RTC device, where time can be read and set.
>
> The hardware also exposes two timers, one which shuts down the console
> and one which powers it on, but these aren’t supported currently.
>
> On the Wii U, the counter bias is stored in a XML file, /config/rtc.xml,
> encrypted in the SLC (eMMC storage), using a proprietary filesystem.  In
> order to avoid having to implement all that, this driver assumes a
> bootloader will parse this XML file and write the bias into the SRAM, at
> the same location the other two consoles have it.
>
> Changes since v1:
> - Rename the driver to rtc-gamecube.
> - Switch to the regmap API for debugfs support.
> - Report low battery and unstable power as invalid data.
> - Remove Wii U support in Kconfig, nothing specific to this console
>   needs to be changed in the code.
> - Don’t attempt to change HW_SRNPROT on the GameCube, this register
>   doesn’t exist so we can use SRAM just fine without doing anything.
> - Add needed changes to the wii device tree.
> - Enable this driver on the gamecube and wii platforms.
>
> Changes since v2:
> - Report low battery correctly.
>
> Emmanuel Gil Peyrot (5):
>   rtc: gamecube: Add a RTC driver for the GameCube, Wii and Wii U
>   rtc: gamecube: Report low battery as invalid data
>   powerpc: wii.dts: Expose HW_SRNPROT on this platform
>   powerpc: gamecube_defconfig: Enable the RTC driver
>   powerpc: wii_defconfig: Enable the RTC driver
>
>  drivers/rtc/Kconfig                     |  11 +
>  drivers/rtc/Makefile                    |   1 +
>  drivers/rtc/rtc-gamecube.c              | 377 ++++++++++++++++++++++++

This is basically an rtc series as far as I'm concerned.

>  arch/powerpc/boot/dts/wii.dts           |   5 +
>  arch/powerpc/configs/gamecube_defconfig |   2 +-
>  arch/powerpc/configs/wii_defconfig      |   2 +-

I have nothing queued in the powerpc tree that touches any of those
files, so conflicts are unlikely.

So I'm happy for this to go via the rtc tree whenever it's ready.

Acked-by: Michael Ellerman <mpe@ellerman.id.au> (powerpc)

cheers

Re: [PATCH v3 0/5] rtc: nintendo: Add a RTC driver for the GameCube, Wii and Wii U

[Alexandre Belloni]

Hello,

On 16/12/2021 15:52:59+1100, Michael Ellerman wrote:
> > Emmanuel Gil Peyrot (5):
> >   rtc: gamecube: Add a RTC driver for the GameCube, Wii and Wii U
> >   rtc: gamecube: Report low battery as invalid data
> >   powerpc: wii.dts: Expose HW_SRNPROT on this platform
> >   powerpc: gamecube_defconfig: Enable the RTC driver
> >   powerpc: wii_defconfig: Enable the RTC driver
> >
> >  drivers/rtc/Kconfig                     |  11 +
> >  drivers/rtc/Makefile                    |   1 +
> >  drivers/rtc/rtc-gamecube.c              | 377 ++++++++++++++++++++++++
> 
> This is basically an rtc series as far as I'm concerned.
> 
> >  arch/powerpc/boot/dts/wii.dts           |   5 +
> >  arch/powerpc/configs/gamecube_defconfig |   2 +-
> >  arch/powerpc/configs/wii_defconfig      |   2 +-
> 
> I have nothing queued in the powerpc tree that touches any of those
> files, so conflicts are unlikely.
> 
> So I'm happy for this to go via the rtc tree whenever it's ready.
> 
> Acked-by: Michael Ellerman <mpe@ellerman.id.au> (powerpc)

That's done, thanks.


-- 
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

Re: [PATCH v3 0/5] rtc: nintendo: Add a RTC driver for the GameCube, Wii and Wii U

[Alexandre Belloni]

On Wed, 15 Dec 2021 18:54:56 +0100, Emmanuel Gil Peyrot wrote:
> These three consoles share a device, the MX23L4005, which contains a
> clock and 64 bytes of SRAM storage, and is exposed on the EXI bus
> (similar to SPI) on channel 0, device 1.  This driver allows it to be
> used as a Linux RTC device, where time can be read and set.
> 
> The hardware also exposes two timers, one which shuts down the console
> and one which powers it on, but these aren’t supported currently.
> 
> [...]

Applied, thanks!

[1/5] rtc: gamecube: Add a RTC driver for the GameCube, Wii and Wii U
      commit: 86559400b3ef9de93ba50523cffe767c35cd531a
[2/5] rtc: gamecube: Report low battery as invalid data
      commit: 322539a014bcd24cbb9281832c09b24e07912237
[3/5] powerpc: wii.dts: Expose HW_SRNPROT on this platform
      commit: 5479618e1e2641dd57352a73b7b7b2f6908fbeee
[4/5] powerpc: gamecube_defconfig: Enable the RTC driver
      commit: 57bd7d356506b713d0df8d8e42da7810a18864df
[5/5] powerpc: wii_defconfig: Enable the RTC driver
      commit: 69e8ba80ddda4db31e59facbf2db19773ad3785b

This one didn't apply ceanly but I believe I did the right thing. Can you check?


Best regards,
-- 
Alexandre Belloni <alexandre.belloni@bootlin.com>

Re: [PATCH v3 0/5] rtc: nintendo: Add a RTC driver for the GameCube, Wii and Wii U

[Emmanuel Gil Peyrot]

[-- Attachment #1: Type: text/plain, Size: 1793 bytes --]

On Thu, Dec 16, 2021 at 10:49:44AM +0100, Alexandre Belloni wrote:
> On Wed, 15 Dec 2021 18:54:56 +0100, Emmanuel Gil Peyrot wrote:
> > These three consoles share a device, the MX23L4005, which contains a
> > clock and 64 bytes of SRAM storage, and is exposed on the EXI bus
> > (similar to SPI) on channel 0, device 1.  This driver allows it to be
> > used as a Linux RTC device, where time can be read and set.
> > 
> > The hardware also exposes two timers, one which shuts down the console
> > and one which powers it on, but these aren’t supported currently.
> > 
> > [...]
> 
> Applied, thanks!
> 
> [1/5] rtc: gamecube: Add a RTC driver for the GameCube, Wii and Wii U
>       commit: 86559400b3ef9de93ba50523cffe767c35cd531a
> [2/5] rtc: gamecube: Report low battery as invalid data
>       commit: 322539a014bcd24cbb9281832c09b24e07912237
> [3/5] powerpc: wii.dts: Expose HW_SRNPROT on this platform
>       commit: 5479618e1e2641dd57352a73b7b7b2f6908fbeee
> [4/5] powerpc: gamecube_defconfig: Enable the RTC driver
>       commit: 57bd7d356506b713d0df8d8e42da7810a18864df
> [5/5] powerpc: wii_defconfig: Enable the RTC driver
>       commit: 69e8ba80ddda4db31e59facbf2db19773ad3785b
> 
> This one didn't apply ceanly but I believe I did the right thing. Can you check?

I believe you didn’t, at least that commit[1] seems to have one “+” too
many in the modified line, whereas the previous one[2] doesn’t.

But thanks for applying them!

> 
> 
> Best regards,
> -- 
> Alexandre Belloni <alexandre.belloni@bootlin.com>

[1] https://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux.git/commit/?id=69e8ba80dd
[2] https://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux.git/commit/?id=57bd7d3565

-- 
Emmanuel Gil Peyrot

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH v3 0/5] rtc: nintendo: Add a RTC driver for the GameCube, Wii and Wii U

[Alexandre Belloni]

On 16/12/2021 21:22:20+0100, Emmanuel Gil Peyrot wrote:
> On Thu, Dec 16, 2021 at 10:49:44AM +0100, Alexandre Belloni wrote:
> > On Wed, 15 Dec 2021 18:54:56 +0100, Emmanuel Gil Peyrot wrote:
> > > These three consoles share a device, the MX23L4005, which contains a
> > > clock and 64 bytes of SRAM storage, and is exposed on the EXI bus
> > > (similar to SPI) on channel 0, device 1.  This driver allows it to be
> > > used as a Linux RTC device, where time can be read and set.
> > > 
> > > The hardware also exposes two timers, one which shuts down the console
> > > and one which powers it on, but these aren’t supported currently.
> > > 
> > > [...]
> > 
> > Applied, thanks!
> > 
> > [1/5] rtc: gamecube: Add a RTC driver for the GameCube, Wii and Wii U
> >       commit: 86559400b3ef9de93ba50523cffe767c35cd531a
> > [2/5] rtc: gamecube: Report low battery as invalid data
> >       commit: 322539a014bcd24cbb9281832c09b24e07912237
> > [3/5] powerpc: wii.dts: Expose HW_SRNPROT on this platform
> >       commit: 5479618e1e2641dd57352a73b7b7b2f6908fbeee
> > [4/5] powerpc: gamecube_defconfig: Enable the RTC driver
> >       commit: 57bd7d356506b713d0df8d8e42da7810a18864df
> > [5/5] powerpc: wii_defconfig: Enable the RTC driver
> >       commit: 69e8ba80ddda4db31e59facbf2db19773ad3785b
> > 
> > This one didn't apply ceanly but I believe I did the right thing. Can you check?
> 
> I believe you didn’t, at least that commit[1] seems to have one “+” too
> many in the modified line, whereas the previous one[2] doesn’t.
> 

I knew I needed you to check, this is fixed now.

https://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux.git/commit/?id=c636783d594f6cfc95db51c796761719317ce5eb


-- 
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

Re: [PATCH v3 0/5] rtc: nintendo: Add a RTC driver for the GameCube, Wii and Wii U

[Emmanuel Gil Peyrot]

[-- Attachment #1: Type: text/plain, Size: 2069 bytes --]

On Thu, Dec 16, 2021 at 09:56:06PM +0100, Alexandre Belloni wrote:
> On 16/12/2021 21:22:20+0100, Emmanuel Gil Peyrot wrote:
> > On Thu, Dec 16, 2021 at 10:49:44AM +0100, Alexandre Belloni wrote:
> > > On Wed, 15 Dec 2021 18:54:56 +0100, Emmanuel Gil Peyrot wrote:
> > > > These three consoles share a device, the MX23L4005, which contains a
> > > > clock and 64 bytes of SRAM storage, and is exposed on the EXI bus
> > > > (similar to SPI) on channel 0, device 1.  This driver allows it to be
> > > > used as a Linux RTC device, where time can be read and set.
> > > > 
> > > > The hardware also exposes two timers, one which shuts down the console
> > > > and one which powers it on, but these aren’t supported currently.
> > > > 
> > > > [...]
> > > 
> > > Applied, thanks!
> > > 
> > > [1/5] rtc: gamecube: Add a RTC driver for the GameCube, Wii and Wii U
> > >       commit: 86559400b3ef9de93ba50523cffe767c35cd531a
> > > [2/5] rtc: gamecube: Report low battery as invalid data
> > >       commit: 322539a014bcd24cbb9281832c09b24e07912237
> > > [3/5] powerpc: wii.dts: Expose HW_SRNPROT on this platform
> > >       commit: 5479618e1e2641dd57352a73b7b7b2f6908fbeee
> > > [4/5] powerpc: gamecube_defconfig: Enable the RTC driver
> > >       commit: 57bd7d356506b713d0df8d8e42da7810a18864df
> > > [5/5] powerpc: wii_defconfig: Enable the RTC driver
> > >       commit: 69e8ba80ddda4db31e59facbf2db19773ad3785b
> > > 
> > > This one didn't apply ceanly but I believe I did the right thing. Can you check?
> > 
> > I believe you didn’t, at least that commit[1] seems to have one “+” too
> > many in the modified line, whereas the previous one[2] doesn’t.
> > 
> 
> I knew I needed you to check, this is fixed now.
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux.git/commit/?id=c636783d594f6cfc95db51c796761719317ce5eb

Perfect, thanks a lot!

> 
> 
> -- 
> Alexandre Belloni, co-owner and COO, Bootlin
> Embedded Linux and Kernel engineering
> https://bootlin.com

-- 
Emmanuel Gil Peyrot

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]