* [PATCH v9 0/4] add NXP RTC driver support for S32G2/S32G3 SoCs
@ 2025-04-03 10:33 Ciprian Costea
2025-04-03 10:33 ` [PATCH v9 1/4] dt-bindings: rtc: add schema for NXP " Ciprian Costea
` (4 more replies)
0 siblings, 5 replies; 19+ messages in thread
From: Ciprian Costea @ 2025-04-03 10:33 UTC (permalink / raw)
To: Alexandre Belloni, Rob Herring, Krzysztof Kozlowski, Conor Dooley,
Catalin Marinas, Will Deacon
Cc: linux-rtc, devicetree, linux-kernel, linux-arm-kernel,
NXP S32 Linux, imx, Christophe Lizzi, Alberto Ruiz,
Enric Balletbo, Eric Chanudet, Ciprian Marian Costea
From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
This patch series add support for the NXP RTC hardware module present on
S32G2/S32G3 SoCs.
RTC tracks clock time during system suspend. It is used as a time-based
wakeup source for the S32G2/S32G3 SoCs.
RTC is not battery-powered and it is not kept alive during system reset.
Following is an example of Suspend to RAM trigger on S32G2/S32G3 SoCs,
using userspace tools such as rtcwake:
# rtcwake -s 120 -m mem
# rtcwake: assuming RTC uses UTC ...
# rtcwake: wakeup from "mem" using /dev/rtc0 at Wed Feb 6 06:28:36 2036
#
Changes in V9:
- Added missing 'bitfield.h' include file
- Moved reserved clk mask check earlier to stop execution in case a
reserved clk is being used.
- Used 'DIV_ROUND_UP' macro when determining the rtc clock frequency.
- Fixed some indentation inconsistencies.
- Added Frank Li's Reviewed-by tag.
Changes in V8:
- Added description of uncommon S32G RTC hardware behaviour on resume.
Changes in V7:
- Changed 'ioread32/iowrite32' calls to more common 'readl/writel'.
- Removed calculations performed in 'suspend' and 'resume' routines by
using RTC API ('APIVAL') functionality instead of relying on setting
'RTCVAL'.
- Simplified 'set_alarm' and 'read_time/read_alarm' by usage of
RTC APIVAL functionality and relying on kernel system time.
- Added comment about the available RTC hardware frequency divisors in the
RTC driver.
- Used both available frequency divisors to increase the RTC rollover
time.
- Removed a redundant error check on 'priv->irq'.
Changes in V6:
- Fixed a coding style issue regarding kernel doc reported by kernel test robot
- Refactored small sections from the S32G RTC driver without impacting
functionality
- Fixed an error probe path issue on S32G RTC driver.
- Added 'rtc' schema in S32G-RTC bindings
Changes in V5:
- Removed rollover support.
- Removed clock switching support between Runtime and Suspend. A clock source
which is always available has been used instead.
- Enabled 512 value RTC hardware frequency divisor to achieve higher rollover
time
- Removed unneeded 'remove' callback.
- Decreased driver print verbosity on error paths.
- Provided 'clock-names' actual names in bindings documentation
- Remove __maybe_unused notations. Used the DEFINE_SIMPLE_DEV_PM_OPS() and
pm_sleep_ptr() macros to handle the .suspend/.resume callbacks.
- Fixed some alignment issues.
Changes in V4:
- Dropped 'assigned-*' clock management approach. Simplified RTC Runtime
and Suspend/Standby clock configuration.
- Simplified error paths on probe function
- Removed node label from bindings example
- Several cosmetic coding style fixes
Changes in V3:
- Removed 'nxp,s32g3-rtc' compatible string
- Change to 'remove' callback from 'remove_new'
- Used 'dev.parent' from 'struct rtc_device' instead of defining a
specific 'struct device' in driver data
- Fixed several errors reported by kernel test robot
- Removed 'assigned-clocks', 'assigned-clock-parents' and
'assigned-clock-rates' from required properties in the binding
documentation.
- Refactored S32G RTC driver such that a default clock source and
divisors configuration will be applied in case 'assigned-clocks' and
'assigned-clock-parents' properties are missing.
Changes in V2:
- Removed 'clksel' support from dts bindings. Used clock parents support
from CCF to better illustrate the RTC hardware IP from S32G2/S32G3.
- Removed frequency dividers support from dts bindings. Used assigned
clock frequencies support from CCF instead.
- Reduced the interval while RTC is voluntarily disabled to a simple
register write in order to avoid any race condition between a possbile
rollover and 'suspend' callback execution flow.
- Updated bindings documentation with respect to clocking support.
- Fixed a potential unused variable warning reported by kernel test robot.
- Updated to usage of 'devm_rtc_allocate_device' and 'devm_rtc_register_device'
instead of deprecated 'devm_rtc_device_register'.
Ciprian Marian Costea (4):
dt-bindings: rtc: add schema for NXP S32G2/S32G3 SoCs
rtc: s32g: add NXP S32G2/S32G3 SoC support
arm64: defconfig: add S32G RTC module support
MAINTAINERS: add NXP S32G RTC driver
.../devicetree/bindings/rtc/nxp,s32g-rtc.yaml | 72 ++++
MAINTAINERS | 2 +
arch/arm64/configs/defconfig | 1 +
drivers/rtc/Kconfig | 11 +
drivers/rtc/Makefile | 1 +
drivers/rtc/rtc-s32g.c | 385 ++++++++++++++++++
6 files changed, 472 insertions(+)
create mode 100644 Documentation/devicetree/bindings/rtc/nxp,s32g-rtc.yaml
create mode 100644 drivers/rtc/rtc-s32g.c
--
2.45.2
^ permalink raw reply [flat|nested] 19+ messages in thread
* [PATCH v9 1/4] dt-bindings: rtc: add schema for NXP S32G2/S32G3 SoCs
2025-04-03 10:33 [PATCH v9 0/4] add NXP RTC driver support for S32G2/S32G3 SoCs Ciprian Costea
@ 2025-04-03 10:33 ` Ciprian Costea
2025-04-03 10:33 ` [PATCH v9 2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support Ciprian Costea
` (3 subsequent siblings)
4 siblings, 0 replies; 19+ messages in thread
From: Ciprian Costea @ 2025-04-03 10:33 UTC (permalink / raw)
To: Alexandre Belloni, Rob Herring, Krzysztof Kozlowski, Conor Dooley,
Catalin Marinas, Will Deacon
Cc: linux-rtc, devicetree, linux-kernel, linux-arm-kernel,
NXP S32 Linux, imx, Christophe Lizzi, Alberto Ruiz,
Enric Balletbo, Eric Chanudet, Ciprian Marian Costea,
Bogdan-Gabriel Roman, Ghennadi Procopciuc
From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
RTC tracks clock time during system suspend and it is used as a wakeup
source on S32G2/S32G3 architecture.
RTC from S32G2/S32G3 is not battery-powered and it is not kept alive
during system reset.
Co-developed-by: Bogdan-Gabriel Roman <bogdan-gabriel.roman@nxp.com>
Signed-off-by: Bogdan-Gabriel Roman <bogdan-gabriel.roman@nxp.com>
Co-developed-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
Signed-off-by: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
---
.../devicetree/bindings/rtc/nxp,s32g-rtc.yaml | 72 +++++++++++++++++++
1 file changed, 72 insertions(+)
create mode 100644 Documentation/devicetree/bindings/rtc/nxp,s32g-rtc.yaml
diff --git a/Documentation/devicetree/bindings/rtc/nxp,s32g-rtc.yaml b/Documentation/devicetree/bindings/rtc/nxp,s32g-rtc.yaml
new file mode 100644
index 000000000000..40fd2fa298fe
--- /dev/null
+++ b/Documentation/devicetree/bindings/rtc/nxp,s32g-rtc.yaml
@@ -0,0 +1,72 @@
+# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/rtc/nxp,s32g-rtc.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: NXP S32G2/S32G3 Real Time Clock (RTC)
+
+maintainers:
+ - Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
+ - Ciprian Marian Costea <ciprianmarian.costea@nxp.com>
+
+description:
+ RTC hardware module present on S32G2/S32G3 SoCs is used as a wakeup source.
+ It is not kept alive during system reset and it is not battery-powered.
+
+allOf:
+ - $ref: rtc.yaml#
+
+properties:
+ compatible:
+ oneOf:
+ - enum:
+ - nxp,s32g2-rtc
+ - items:
+ - const: nxp,s32g3-rtc
+ - const: nxp,s32g2-rtc
+
+ reg:
+ maxItems: 1
+
+ interrupts:
+ maxItems: 1
+
+ clocks:
+ items:
+ - description: ipg clock drives the access to the RTC iomapped registers
+ - description: Clock source for the RTC module. Can be selected between
+ 4 different clock sources using an integrated hardware mux.
+ On S32G2/S32G3 SoCs, 'source0' is the SIRC clock (~32KHz) and it is
+ available during standby and runtime. 'source1' is reserved and cannot
+ be used. 'source2' is the FIRC clock and it is only available during
+ runtime providing a better resolution (~48MHz). 'source3' is an external
+ RTC clock source which can be additionally added in hardware.
+
+ clock-names:
+ items:
+ - const: ipg
+ - enum: [ source0, source1, source2, source3 ]
+
+required:
+ - compatible
+ - reg
+ - interrupts
+ - clocks
+ - clock-names
+
+additionalProperties: false
+
+examples:
+ - |
+ #include <dt-bindings/interrupt-controller/arm-gic.h>
+ #include <dt-bindings/interrupt-controller/irq.h>
+
+ rtc@40060000 {
+ compatible = "nxp,s32g3-rtc",
+ "nxp,s32g2-rtc";
+ reg = <0x40060000 0x1000>;
+ interrupts = <GIC_SPI 121 IRQ_TYPE_LEVEL_HIGH>;
+ clocks = <&clks 54>, <&clks 55>;
+ clock-names = "ipg", "source0";
+ };
--
2.45.2
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH v9 2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support
2025-04-03 10:33 [PATCH v9 0/4] add NXP RTC driver support for S32G2/S32G3 SoCs Ciprian Costea
2025-04-03 10:33 ` [PATCH v9 1/4] dt-bindings: rtc: add schema for NXP " Ciprian Costea
@ 2025-04-03 10:33 ` Ciprian Costea
2025-04-09 14:14 ` Enric Balletbo i Serra
2025-04-03 10:33 ` [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support Ciprian Costea
` (2 subsequent siblings)
4 siblings, 1 reply; 19+ messages in thread
From: Ciprian Costea @ 2025-04-03 10:33 UTC (permalink / raw)
To: Alexandre Belloni, Rob Herring, Krzysztof Kozlowski, Conor Dooley,
Catalin Marinas, Will Deacon
Cc: linux-rtc, devicetree, linux-kernel, linux-arm-kernel,
NXP S32 Linux, imx, Christophe Lizzi, Alberto Ruiz,
Enric Balletbo, Eric Chanudet, Ciprian Marian Costea,
Bogdan Hamciuc, Ghennadi Procopciuc, Frank Li
From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
Add a RTC driver for NXP S32G2/S32G3 SoCs.
RTC tracks clock time during system suspend. It can be a wakeup source
for the S32G2/S32G3 SoC based boards.
The RTC module from S32G2/S32G3 is not battery-powered and it is not kept
alive during system reset.
Co-developed-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
Signed-off-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
Co-developed-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
Signed-off-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
Signed-off-by: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
---
drivers/rtc/Kconfig | 11 ++
drivers/rtc/Makefile | 1 +
drivers/rtc/rtc-s32g.c | 385 +++++++++++++++++++++++++++++++++++++++++
3 files changed, 397 insertions(+)
create mode 100644 drivers/rtc/rtc-s32g.c
diff --git a/drivers/rtc/Kconfig b/drivers/rtc/Kconfig
index 0bbbf778ecfa..510dc2db745d 100644
--- a/drivers/rtc/Kconfig
+++ b/drivers/rtc/Kconfig
@@ -2103,4 +2103,15 @@ config RTC_DRV_AMLOGIC_A4
This driver can also be built as a module. If so, the module
will be called "rtc-amlogic-a4".
+config RTC_DRV_S32G
+ tristate "RTC driver for S32G2/S32G3 SoCs"
+ depends on ARCH_S32 || COMPILE_TEST
+ depends on COMMON_CLK
+ help
+ Say yes to enable RTC driver for platforms based on the
+ S32G2/S32G3 SoC family.
+
+ This RTC module can be used as a wakeup source.
+ Please note that it is not battery-powered.
+
endif # RTC_CLASS
diff --git a/drivers/rtc/Makefile b/drivers/rtc/Makefile
index 489b4ab07068..e4b616ecd5ce 100644
--- a/drivers/rtc/Makefile
+++ b/drivers/rtc/Makefile
@@ -161,6 +161,7 @@ obj-$(CONFIG_RTC_DRV_RX8111) += rtc-rx8111.o
obj-$(CONFIG_RTC_DRV_RX8581) += rtc-rx8581.o
obj-$(CONFIG_RTC_DRV_RZN1) += rtc-rzn1.o
obj-$(CONFIG_RTC_DRV_RENESAS_RTCA3) += rtc-renesas-rtca3.o
+obj-$(CONFIG_RTC_DRV_S32G) += rtc-s32g.o
obj-$(CONFIG_RTC_DRV_S35390A) += rtc-s35390a.o
obj-$(CONFIG_RTC_DRV_S3C) += rtc-s3c.o
obj-$(CONFIG_RTC_DRV_S5M) += rtc-s5m.o
diff --git a/drivers/rtc/rtc-s32g.c b/drivers/rtc/rtc-s32g.c
new file mode 100644
index 000000000000..3a0818e972eb
--- /dev/null
+++ b/drivers/rtc/rtc-s32g.c
@@ -0,0 +1,385 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2025 NXP
+ */
+
+#include <linux/bitfield.h>
+#include <linux/clk.h>
+#include <linux/iopoll.h>
+#include <linux/of_irq.h>
+#include <linux/platform_device.h>
+#include <linux/rtc.h>
+
+#define RTCC_OFFSET 0x4ul
+#define RTCS_OFFSET 0x8ul
+#define APIVAL_OFFSET 0x10ul
+
+/* RTCC fields */
+#define RTCC_CNTEN BIT(31)
+#define RTCC_APIEN BIT(15)
+#define RTCC_APIIE BIT(14)
+#define RTCC_CLKSEL_MASK GENMASK(13, 12)
+#define RTCC_DIV512EN BIT(11)
+#define RTCC_DIV32EN BIT(10)
+
+/* RTCS fields */
+#define RTCS_INV_API BIT(17)
+#define RTCS_APIF BIT(13)
+
+#define APIVAL_MAX_VAL GENMASK(31, 0)
+#define RTC_SYNCH_TIMEOUT (100 * USEC_PER_MSEC)
+
+/*
+ * S32G2 and S32G3 SoCs have RTC clock source1 reserved and
+ * should not be used.
+ */
+#define RTC_CLK_SRC1_RESERVED BIT(1)
+
+/*
+ * S32G RTC module has a 512 value and a 32 value hardware frequency
+ * divisors (DIV512 and DIV32) which could be used to achieve higher
+ * counter ranges by lowering the RTC frequency.
+ */
+enum {
+ DIV1 = 1,
+ DIV32 = 32,
+ DIV512 = 512,
+ DIV512_32 = 16384
+};
+
+static const char *const rtc_clk_src[] = {
+ "source0",
+ "source1",
+ "source2",
+ "source3"
+};
+
+struct rtc_priv {
+ struct rtc_device *rdev;
+ void __iomem *rtc_base;
+ struct clk *ipg;
+ struct clk *clk_src;
+ const struct rtc_soc_data *rtc_data;
+ u64 rtc_hz;
+ time64_t sleep_sec;
+ int irq;
+ u32 clk_src_idx;
+};
+
+struct rtc_soc_data {
+ u32 clk_div;
+ u32 reserved_clk_mask;
+};
+
+static const struct rtc_soc_data rtc_s32g2_data = {
+ .clk_div = DIV512_32,
+ .reserved_clk_mask = RTC_CLK_SRC1_RESERVED,
+};
+
+static irqreturn_t s32g_rtc_handler(int irq, void *dev)
+{
+ struct rtc_priv *priv = platform_get_drvdata(dev);
+ u32 status;
+
+ status = readl(priv->rtc_base + RTCS_OFFSET);
+
+ if (status & RTCS_APIF) {
+ writel(0x0, priv->rtc_base + APIVAL_OFFSET);
+ writel(status | RTCS_APIF, priv->rtc_base + RTCS_OFFSET);
+ }
+
+ rtc_update_irq(priv->rdev, 1, RTC_IRQF | RTC_AF);
+
+ return IRQ_HANDLED;
+}
+
+/*
+ * The function is not really getting time from the RTC since the S32G RTC
+ * has several limitations. Thus, to setup alarm use system time.
+ */
+static int s32g_rtc_read_time(struct device *dev,
+ struct rtc_time *tm)
+{
+ struct rtc_priv *priv = dev_get_drvdata(dev);
+ time64_t sec;
+
+ if (check_add_overflow(ktime_get_real_seconds(),
+ priv->sleep_sec, &sec))
+ return -ERANGE;
+
+ rtc_time64_to_tm(sec, tm);
+
+ return 0;
+}
+
+static int s32g_rtc_read_alarm(struct device *dev, struct rtc_wkalrm *alrm)
+{
+ struct rtc_priv *priv = dev_get_drvdata(dev);
+ u32 rtcc, rtcs;
+
+ rtcc = readl(priv->rtc_base + RTCC_OFFSET);
+ rtcs = readl(priv->rtc_base + RTCS_OFFSET);
+
+ alrm->enabled = rtcc & RTCC_APIIE;
+ if (alrm->enabled)
+ alrm->pending = !(rtcs & RTCS_APIF);
+
+ return 0;
+}
+
+static int s32g_rtc_alarm_irq_enable(struct device *dev, unsigned int enabled)
+{
+ struct rtc_priv *priv = dev_get_drvdata(dev);
+ u32 rtcc;
+
+ /* RTC API functionality is used both for triggering interrupts
+ * and as a wakeup event. Hence it should always be enabled.
+ */
+ rtcc = readl(priv->rtc_base + RTCC_OFFSET);
+ rtcc |= RTCC_APIEN | RTCC_APIIE;
+ writel(rtcc, priv->rtc_base + RTCC_OFFSET);
+
+ return 0;
+}
+
+static int s32g_rtc_set_alarm(struct device *dev, struct rtc_wkalrm *alrm)
+{
+ struct rtc_priv *priv = dev_get_drvdata(dev);
+ unsigned long long cycles;
+ long long t_offset;
+ time64_t alrm_time;
+ u32 rtcs;
+ int ret;
+
+ alrm_time = rtc_tm_to_time64(&alrm->time);
+ t_offset = alrm_time - ktime_get_real_seconds() - priv->sleep_sec;
+ if (t_offset < 0)
+ return -ERANGE;
+
+ cycles = t_offset * priv->rtc_hz;
+ if (cycles > APIVAL_MAX_VAL)
+ return -ERANGE;
+
+ /* APIVAL could have been reset from the IRQ handler.
+ * Hence, we wait in case there is a synchronization process.
+ */
+ ret = read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
+ 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
+ if (ret)
+ return ret;
+
+ writel(cycles, priv->rtc_base + APIVAL_OFFSET);
+
+ return read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
+ 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
+}
+
+/*
+ * Disable the 32-bit free running counter.
+ * This allows Clock Source and Divisors selection
+ * to be performed without causing synchronization issues.
+ */
+static void s32g_rtc_disable(struct rtc_priv *priv)
+{
+ u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
+
+ rtcc &= ~RTCC_CNTEN;
+ writel(rtcc, priv->rtc_base + RTCC_OFFSET);
+}
+
+static void s32g_rtc_enable(struct rtc_priv *priv)
+{
+ u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
+
+ rtcc |= RTCC_CNTEN;
+ writel(rtcc, priv->rtc_base + RTCC_OFFSET);
+}
+
+static int rtc_clk_src_setup(struct rtc_priv *priv)
+{
+ u32 rtcc;
+
+ rtcc = FIELD_PREP(RTCC_CLKSEL_MASK, priv->clk_src_idx);
+
+ switch (priv->rtc_data->clk_div) {
+ case DIV512_32:
+ rtcc |= RTCC_DIV512EN;
+ rtcc |= RTCC_DIV32EN;
+ break;
+ case DIV512:
+ rtcc |= RTCC_DIV512EN;
+ break;
+ case DIV32:
+ rtcc |= RTCC_DIV32EN;
+ break;
+ case DIV1:
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ rtcc |= RTCC_APIEN | RTCC_APIIE;
+ /*
+ * Make sure the CNTEN is 0 before we configure
+ * the clock source and dividers.
+ */
+ s32g_rtc_disable(priv);
+ writel(rtcc, priv->rtc_base + RTCC_OFFSET);
+ s32g_rtc_enable(priv);
+
+ return 0;
+}
+
+static const struct rtc_class_ops rtc_ops = {
+ .read_time = s32g_rtc_read_time,
+ .read_alarm = s32g_rtc_read_alarm,
+ .set_alarm = s32g_rtc_set_alarm,
+ .alarm_irq_enable = s32g_rtc_alarm_irq_enable,
+};
+
+static int rtc_clk_dts_setup(struct rtc_priv *priv,
+ struct device *dev)
+{
+ u32 i;
+
+ priv->ipg = devm_clk_get_enabled(dev, "ipg");
+ if (IS_ERR(priv->ipg))
+ return dev_err_probe(dev, PTR_ERR(priv->ipg),
+ "Failed to get 'ipg' clock\n");
+
+ for (i = 0; i < ARRAY_SIZE(rtc_clk_src); i++) {
+ if (priv->rtc_data->reserved_clk_mask & BIT(i))
+ return -EOPNOTSUPP;
+
+ priv->clk_src = devm_clk_get_enabled(dev, rtc_clk_src[i]);
+ if (!IS_ERR(priv->clk_src)) {
+ priv->clk_src_idx = i;
+ break;
+ }
+ }
+
+ if (IS_ERR(priv->clk_src))
+ return dev_err_probe(dev, PTR_ERR(priv->clk_src),
+ "Failed to get rtc module clock source\n");
+
+ return 0;
+}
+
+static int s32g_rtc_probe(struct platform_device *pdev)
+{
+ struct device *dev = &pdev->dev;
+ struct rtc_priv *priv;
+ unsigned long rtc_hz;
+ int ret;
+
+ priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
+ if (!priv)
+ return -ENOMEM;
+
+ priv->rtc_data = of_device_get_match_data(dev);
+ if (!priv->rtc_data)
+ return -ENODEV;
+
+ priv->rtc_base = devm_platform_ioremap_resource(pdev, 0);
+ if (IS_ERR(priv->rtc_base))
+ return PTR_ERR(priv->rtc_base);
+
+ device_init_wakeup(dev, true);
+
+ ret = rtc_clk_dts_setup(priv, dev);
+ if (ret)
+ return ret;
+
+ priv->rdev = devm_rtc_allocate_device(dev);
+ if (IS_ERR(priv->rdev))
+ return PTR_ERR(priv->rdev);
+
+ ret = rtc_clk_src_setup(priv);
+ if (ret)
+ return ret;
+
+ priv->irq = platform_get_irq(pdev, 0);
+ if (priv->irq < 0) {
+ ret = priv->irq;
+ goto disable_rtc;
+ }
+
+ rtc_hz = clk_get_rate(priv->clk_src);
+ if (!rtc_hz) {
+ dev_err(dev, "Failed to get RTC frequency\n");
+ ret = -EINVAL;
+ goto disable_rtc;
+ }
+
+ priv->rtc_hz = DIV_ROUND_UP(rtc_hz, priv->rtc_data->clk_div);
+
+ platform_set_drvdata(pdev, priv);
+ priv->rdev->ops = &rtc_ops;
+
+ ret = devm_request_irq(dev, priv->irq,
+ s32g_rtc_handler, 0, dev_name(dev), pdev);
+ if (ret) {
+ dev_err(dev, "Request interrupt %d failed, error: %d\n",
+ priv->irq, ret);
+ goto disable_rtc;
+ }
+
+ ret = devm_rtc_register_device(priv->rdev);
+ if (ret)
+ goto disable_rtc;
+
+ return 0;
+
+disable_rtc:
+ s32g_rtc_disable(priv);
+ return ret;
+}
+
+static int s32g_rtc_suspend(struct device *dev)
+{
+ struct rtc_priv *priv = dev_get_drvdata(dev);
+ u32 apival = readl(priv->rtc_base + APIVAL_OFFSET);
+
+ if (check_add_overflow(priv->sleep_sec, div64_u64(apival, priv->rtc_hz),
+ &priv->sleep_sec)) {
+ dev_warn(dev, "Overflow on sleep cycles occurred. Resetting to 0.\n");
+ priv->sleep_sec = 0;
+ }
+
+ return 0;
+}
+
+static int s32g_rtc_resume(struct device *dev)
+{
+ struct rtc_priv *priv = dev_get_drvdata(dev);
+
+ /* The transition from resume to run is a reset event.
+ * This leads to the RTC registers being reset after resume from
+ * suspend. It is uncommon, but this behaviour has been observed
+ * on S32G RTC after issuing a Suspend to RAM operation.
+ * Thus, reconfigure RTC registers on the resume path.
+ */
+ return rtc_clk_src_setup(priv);
+}
+
+static const struct of_device_id rtc_dt_ids[] = {
+ { .compatible = "nxp,s32g2-rtc", .data = &rtc_s32g2_data },
+ { /* sentinel */ },
+};
+
+static DEFINE_SIMPLE_DEV_PM_OPS(s32g_rtc_pm_ops,
+ s32g_rtc_suspend, s32g_rtc_resume);
+
+static struct platform_driver s32g_rtc_driver = {
+ .driver = {
+ .name = "s32g-rtc",
+ .pm = pm_sleep_ptr(&s32g_rtc_pm_ops),
+ .of_match_table = rtc_dt_ids,
+ },
+ .probe = s32g_rtc_probe,
+};
+module_platform_driver(s32g_rtc_driver);
+
+MODULE_AUTHOR("NXP");
+MODULE_DESCRIPTION("NXP RTC driver for S32G2/S32G3");
+MODULE_LICENSE("GPL");
--
2.45.2
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support
2025-04-03 10:33 [PATCH v9 0/4] add NXP RTC driver support for S32G2/S32G3 SoCs Ciprian Costea
2025-04-03 10:33 ` [PATCH v9 1/4] dt-bindings: rtc: add schema for NXP " Ciprian Costea
2025-04-03 10:33 ` [PATCH v9 2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support Ciprian Costea
@ 2025-04-03 10:33 ` Ciprian Costea
2025-04-04 6:17 ` Krzysztof Kozlowski
2025-04-03 10:33 ` [PATCH v9 4/4] MAINTAINERS: add NXP S32G RTC driver Ciprian Costea
2025-05-25 21:33 ` (subset) [PATCH v9 0/4] add NXP RTC driver support for S32G2/S32G3 SoCs Alexandre Belloni
4 siblings, 1 reply; 19+ messages in thread
From: Ciprian Costea @ 2025-04-03 10:33 UTC (permalink / raw)
To: Alexandre Belloni, Rob Herring, Krzysztof Kozlowski, Conor Dooley,
Catalin Marinas, Will Deacon
Cc: linux-rtc, devicetree, linux-kernel, linux-arm-kernel,
NXP S32 Linux, imx, Christophe Lizzi, Alberto Ruiz,
Enric Balletbo, Eric Chanudet, Ciprian Marian Costea
From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
The RTC hardware module present on S32G based SoCs tracks clock time
during system suspend and it is used as a wakeup source on S32G2/S32G3
architecture.
Signed-off-by: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
---
arch/arm64/configs/defconfig | 1 +
1 file changed, 1 insertion(+)
diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig
index cb7da4415599..c24b88b0206e 100644
--- a/arch/arm64/configs/defconfig
+++ b/arch/arm64/configs/defconfig
@@ -1227,6 +1227,7 @@ CONFIG_RTC_DRV_DA9063=m
CONFIG_RTC_DRV_EFI=y
CONFIG_RTC_DRV_CROS_EC=y
CONFIG_RTC_DRV_FSL_FTM_ALARM=m
+CONFIG_RTC_DRV_S32G=m
CONFIG_RTC_DRV_S3C=y
CONFIG_RTC_DRV_PL031=y
CONFIG_RTC_DRV_SUN6I=y
--
2.45.2
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH v9 4/4] MAINTAINERS: add NXP S32G RTC driver
2025-04-03 10:33 [PATCH v9 0/4] add NXP RTC driver support for S32G2/S32G3 SoCs Ciprian Costea
` (2 preceding siblings ...)
2025-04-03 10:33 ` [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support Ciprian Costea
@ 2025-04-03 10:33 ` Ciprian Costea
2025-05-25 21:33 ` (subset) [PATCH v9 0/4] add NXP RTC driver support for S32G2/S32G3 SoCs Alexandre Belloni
4 siblings, 0 replies; 19+ messages in thread
From: Ciprian Costea @ 2025-04-03 10:33 UTC (permalink / raw)
To: Alexandre Belloni, Rob Herring, Krzysztof Kozlowski, Conor Dooley,
Catalin Marinas, Will Deacon
Cc: linux-rtc, devicetree, linux-kernel, linux-arm-kernel,
NXP S32 Linux, imx, Christophe Lizzi, Alberto Ruiz,
Enric Balletbo, Eric Chanudet, Ciprian Marian Costea
From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
Add the NXP S32G RTC driver as maintained so further patches on this
driver can be reviewed under this architecture.
Signed-off-by: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
---
MAINTAINERS | 2 ++
1 file changed, 2 insertions(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index 873aa2cce4d7..515ed8115922 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -2873,8 +2873,10 @@ R: Ghennadi Procopciuc <ghennadi.procopciuc@oss.nxp.com>
R: NXP S32 Linux Team <s32@nxp.com>
L: linux-arm-kernel@lists.infradead.org (moderated for non-subscribers)
S: Maintained
+F: Documentation/devicetree/bindings/rtc/nxp,s32g-rtc.yaml
F: arch/arm64/boot/dts/freescale/s32g*.dts*
F: drivers/pinctrl/nxp/
+F: drivers/rtc/rtc-s32g.c
ARM/NXP S32G/S32R DWMAC ETHERNET DRIVER
M: Jan Petrous <jan.petrous@oss.nxp.com>
--
2.45.2
^ permalink raw reply related [flat|nested] 19+ messages in thread
* Re: [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support
2025-04-03 10:33 ` [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support Ciprian Costea
@ 2025-04-04 6:17 ` Krzysztof Kozlowski
2025-04-04 6:24 ` Ciprian Marian Costea
0 siblings, 1 reply; 19+ messages in thread
From: Krzysztof Kozlowski @ 2025-04-04 6:17 UTC (permalink / raw)
To: Ciprian Costea, Alexandre Belloni, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Catalin Marinas, Will Deacon
Cc: linux-rtc, devicetree, linux-kernel, linux-arm-kernel,
NXP S32 Linux, imx, Christophe Lizzi, Alberto Ruiz,
Enric Balletbo, Eric Chanudet
On 03/04/2025 12:33, Ciprian Costea wrote:
> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>
> The RTC hardware module present on S32G based SoCs tracks clock time
> during system suspend and it is used as a wakeup source on S32G2/S32G3
> architecture.
Which boards are using it? I don't see any DTS (nowhere), so I do not
see single reason for this patch.
Best regards,
Krzysztof
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support
2025-04-04 6:17 ` Krzysztof Kozlowski
@ 2025-04-04 6:24 ` Ciprian Marian Costea
2025-04-04 7:15 ` Krzysztof Kozlowski
0 siblings, 1 reply; 19+ messages in thread
From: Ciprian Marian Costea @ 2025-04-04 6:24 UTC (permalink / raw)
To: Krzysztof Kozlowski, Alexandre Belloni, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Catalin Marinas, Will Deacon
Cc: linux-rtc, devicetree, linux-kernel, linux-arm-kernel,
NXP S32 Linux, imx, Christophe Lizzi, Alberto Ruiz,
Enric Balletbo, Eric Chanudet
On 4/4/2025 9:17 AM, Krzysztof Kozlowski wrote:
> On 03/04/2025 12:33, Ciprian Costea wrote:
>> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>>
>> The RTC hardware module present on S32G based SoCs tracks clock time
>> during system suspend and it is used as a wakeup source on S32G2/S32G3
>> architecture.
> Which boards are using it? I don't see any DTS (nowhere), so I do not
> see single reason for this patch.
>
> Best regards,
> Krzysztof
The RTC module is used by all the currently supported S32G2/S32G3
boards, so currently they are: S32G274A-EVB, S32G274A-RDB2, S32G399A-RDB3.
I do see your point in the fact that this driver should be enabled as
module only after platforms are actually using it.
So, would it be better for me to send a V10 in this series with the DTS
patch added ?
Best Regards,
Ciprian
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support
2025-04-04 6:24 ` Ciprian Marian Costea
@ 2025-04-04 7:15 ` Krzysztof Kozlowski
2025-04-04 7:19 ` Ciprian Marian Costea
0 siblings, 1 reply; 19+ messages in thread
From: Krzysztof Kozlowski @ 2025-04-04 7:15 UTC (permalink / raw)
To: Ciprian Marian Costea, Alexandre Belloni, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Catalin Marinas, Will Deacon
Cc: linux-rtc, devicetree, linux-kernel, linux-arm-kernel,
NXP S32 Linux, imx, Christophe Lizzi, Alberto Ruiz,
Enric Balletbo, Eric Chanudet
On 04/04/2025 08:24, Ciprian Marian Costea wrote:
> On 4/4/2025 9:17 AM, Krzysztof Kozlowski wrote:
>> On 03/04/2025 12:33, Ciprian Costea wrote:
>>> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>>>
>>> The RTC hardware module present on S32G based SoCs tracks clock time
>>> during system suspend and it is used as a wakeup source on S32G2/S32G3
>>> architecture.
>> Which boards are using it? I don't see any DTS (nowhere), so I do not
>> see single reason for this patch.
>>
>> Best regards,
>> Krzysztof
>
> The RTC module is used by all the currently supported S32G2/S32G3
> boards, so currently they are: S32G274A-EVB, S32G274A-RDB2, S32G399A-RDB3.
I don't think so. I looked at these DTS and there is no RTC.
> I do see your point in the fact that this driver should be enabled as
> module only after platforms are actually using it.
No, post the user. I don't see the point of sending defconfig patch with
RTC patchset anyway. That's different subsystem.
>
> So, would it be better for me to send a V10 in this series with the DTS
> patch added ?
No, separate patchsets.
Best regards,
Krzysztof
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support
2025-04-04 7:15 ` Krzysztof Kozlowski
@ 2025-04-04 7:19 ` Ciprian Marian Costea
2025-04-04 8:53 ` Alexandre Belloni
0 siblings, 1 reply; 19+ messages in thread
From: Ciprian Marian Costea @ 2025-04-04 7:19 UTC (permalink / raw)
To: Krzysztof Kozlowski, Alexandre Belloni, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Catalin Marinas, Will Deacon
Cc: linux-rtc, devicetree, linux-kernel, linux-arm-kernel,
NXP S32 Linux, imx, Christophe Lizzi, Alberto Ruiz,
Enric Balletbo, Eric Chanudet
On 4/4/2025 10:15 AM, Krzysztof Kozlowski wrote:
> On 04/04/2025 08:24, Ciprian Marian Costea wrote:
>> On 4/4/2025 9:17 AM, Krzysztof Kozlowski wrote:
>>> On 03/04/2025 12:33, Ciprian Costea wrote:
>>>> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>>>>
>>>> The RTC hardware module present on S32G based SoCs tracks clock time
>>>> during system suspend and it is used as a wakeup source on S32G2/S32G3
>>>> architecture.
>>> Which boards are using it? I don't see any DTS (nowhere), so I do not
>>> see single reason for this patch.
>>>
>>> Best regards,
>>> Krzysztof
>>
>> The RTC module is used by all the currently supported S32G2/S32G3
>> boards, so currently they are: S32G274A-EVB, S32G274A-RDB2, S32G399A-RDB3.
>
> I don't think so. I looked at these DTS and there is no RTC.
>
>> I do see your point in the fact that this driver should be enabled as
>> module only after platforms are actually using it.
>
> No, post the user. I don't see the point of sending defconfig patch with
> RTC patchset anyway. That's different subsystem.
>
>>
>> So, would it be better for me to send a V10 in this series with the DTS
>> patch added ?
>
> No, separate patchsets.
Ok. I will send out a V10 in which I will drop this current patch from
the patchset. Also, I will send the DTS patch which adds S32G274A-EVB,
S32G274-RDB2 and S32G399A-RDB3 usage of the RTC after this patchset gets
accepted.
Best Regards,
Ciprian
>
> Best regards,
> Krzysztof
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support
2025-04-04 7:19 ` Ciprian Marian Costea
@ 2025-04-04 8:53 ` Alexandre Belloni
2025-04-04 11:36 ` Ciprian Marian Costea
2025-05-20 9:01 ` Ciprian Marian Costea
0 siblings, 2 replies; 19+ messages in thread
From: Alexandre Belloni @ 2025-04-04 8:53 UTC (permalink / raw)
To: Ciprian Marian Costea
Cc: Krzysztof Kozlowski, Rob Herring, Krzysztof Kozlowski,
Conor Dooley, Catalin Marinas, Will Deacon, linux-rtc, devicetree,
linux-kernel, linux-arm-kernel, NXP S32 Linux, imx,
Christophe Lizzi, Alberto Ruiz, Enric Balletbo, Eric Chanudet
On 04/04/2025 10:19:56+0300, Ciprian Marian Costea wrote:
> On 4/4/2025 10:15 AM, Krzysztof Kozlowski wrote:
> > On 04/04/2025 08:24, Ciprian Marian Costea wrote:
> > > On 4/4/2025 9:17 AM, Krzysztof Kozlowski wrote:
> > > > On 03/04/2025 12:33, Ciprian Costea wrote:
> > > > > From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
> > > > >
> > > > > The RTC hardware module present on S32G based SoCs tracks clock time
> > > > > during system suspend and it is used as a wakeup source on S32G2/S32G3
> > > > > architecture.
> > > > Which boards are using it? I don't see any DTS (nowhere), so I do not
> > > > see single reason for this patch.
> > > >
> > > > Best regards,
> > > > Krzysztof
> > >
> > > The RTC module is used by all the currently supported S32G2/S32G3
> > > boards, so currently they are: S32G274A-EVB, S32G274A-RDB2, S32G399A-RDB3.
> >
> > I don't think so. I looked at these DTS and there is no RTC.
> >
> > > I do see your point in the fact that this driver should be enabled as
> > > module only after platforms are actually using it.
> >
> > No, post the user. I don't see the point of sending defconfig patch with
> > RTC patchset anyway. That's different subsystem.
> >
> > >
> > > So, would it be better for me to send a V10 in this series with the DTS
> > > patch added ?
> >
> > No, separate patchsets.
>
> Ok. I will send out a V10 in which I will drop this current patch from the
> patchset. Also, I will send the DTS patch which adds S32G274A-EVB,
> S32G274-RDB2 and S32G399A-RDB3 usage of the RTC after this patchset gets
> accepted.
>
I don't need V10, I can apply V9 without 3/4 and 4/4
> Best Regards,
> Ciprian
>
> >
> > Best regards,
> > Krzysztof
>
--
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support
2025-04-04 8:53 ` Alexandre Belloni
@ 2025-04-04 11:36 ` Ciprian Marian Costea
2025-05-20 9:01 ` Ciprian Marian Costea
1 sibling, 0 replies; 19+ messages in thread
From: Ciprian Marian Costea @ 2025-04-04 11:36 UTC (permalink / raw)
To: Alexandre Belloni
Cc: Krzysztof Kozlowski, Rob Herring, Krzysztof Kozlowski,
Conor Dooley, Catalin Marinas, Will Deacon, linux-rtc, devicetree,
linux-kernel, linux-arm-kernel, NXP S32 Linux, imx,
Christophe Lizzi, Alberto Ruiz, Enric Balletbo, Eric Chanudet
On 4/4/2025 11:53 AM, Alexandre Belloni wrote:
> On 04/04/2025 10:19:56+0300, Ciprian Marian Costea wrote:
>> On 4/4/2025 10:15 AM, Krzysztof Kozlowski wrote:
>>> On 04/04/2025 08:24, Ciprian Marian Costea wrote:
>>>> On 4/4/2025 9:17 AM, Krzysztof Kozlowski wrote:
>>>>> On 03/04/2025 12:33, Ciprian Costea wrote:
>>>>>> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>>>>>>
>>>>>> The RTC hardware module present on S32G based SoCs tracks clock time
>>>>>> during system suspend and it is used as a wakeup source on S32G2/S32G3
>>>>>> architecture.
>>>>> Which boards are using it? I don't see any DTS (nowhere), so I do not
>>>>> see single reason for this patch.
>>>>>
>>>>> Best regards,
>>>>> Krzysztof
>>>>
>>>> The RTC module is used by all the currently supported S32G2/S32G3
>>>> boards, so currently they are: S32G274A-EVB, S32G274A-RDB2, S32G399A-RDB3.
>>>
>>> I don't think so. I looked at these DTS and there is no RTC.
>>>
>>>> I do see your point in the fact that this driver should be enabled as
>>>> module only after platforms are actually using it.
>>>
>>> No, post the user. I don't see the point of sending defconfig patch with
>>> RTC patchset anyway. That's different subsystem.
>>>
>>>>
>>>> So, would it be better for me to send a V10 in this series with the DTS
>>>> patch added ?
>>>
>>> No, separate patchsets.
>>
>> Ok. I will send out a V10 in which I will drop this current patch from the
>> patchset. Also, I will send the DTS patch which adds S32G274A-EVB,
>> S32G274-RDB2 and S32G399A-RDB3 usage of the RTC after this patchset gets
>> accepted.
>>
>
> I don't need V10, I can apply V9 without 3/4 and 4/4
>
Ok. Thanks for letting me know.
Ciprian
>> Best Regards,
>> Ciprian
>>
>>>
>>> Best regards,
>>> Krzysztof
>>
>
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support
2025-04-03 10:33 ` [PATCH v9 2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support Ciprian Costea
@ 2025-04-09 14:14 ` Enric Balletbo i Serra
2025-04-09 14:26 ` Ciprian Marian Costea
0 siblings, 1 reply; 19+ messages in thread
From: Enric Balletbo i Serra @ 2025-04-09 14:14 UTC (permalink / raw)
To: Ciprian Costea
Cc: Alexandre Belloni, Rob Herring, Krzysztof Kozlowski, Conor Dooley,
Catalin Marinas, Will Deacon, linux-rtc, devicetree, linux-kernel,
linux-arm-kernel, NXP S32 Linux, imx, Christophe Lizzi,
Alberto Ruiz, Eric Chanudet, Bogdan Hamciuc, Ghennadi Procopciuc,
Frank Li
Hi Ciprian,
Many thanks for your patch.
On Thu, Apr 3, 2025 at 12:34 PM Ciprian Costea
<ciprianmarian.costea@oss.nxp.com> wrote:
>
> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>
> Add a RTC driver for NXP S32G2/S32G3 SoCs.
>
> RTC tracks clock time during system suspend. It can be a wakeup source
> for the S32G2/S32G3 SoC based boards.
>
> The RTC module from S32G2/S32G3 is not battery-powered and it is not kept
> alive during system reset.
>
> Co-developed-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
> Signed-off-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
> Co-developed-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
> Signed-off-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
> Signed-off-by: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
> Reviewed-by: Frank Li <Frank.Li@nxp.com>
I went through the code and it looks good to me, so:
However, I did a basic rtctest selftest in my S32G-VNP-RDB3 board,
and some tests failed (it shows a small deviation), any idea what
could be wrong?
This is the output of the test:
# ./rtctest
TAP version 13
1..8
# Starting 8 tests from 1 test cases.
# RUN rtc.date_read ...
# rtctest.c:59:date_read:Current RTC date/time is 28/01/2025 00:17:49.
# OK rtc.date_read
ok 1 rtc.date_read
# RUN rtc.date_read_loop ...
# rtctest.c:124:date_read_loop:Continuously reading RTC time for 30s
(with 11ms breaks after every read).
# rtctest.c:151:date_read_loop:Performed 2638 RTC time reads.
# OK rtc.date_read_loop
ok 2 rtc.date_read_loop
# RUN rtc.uie_read ...
# OK rtc.uie_read
ok 3 rtc.uie_read
# RUN rtc.uie_select ...
# OK rtc.uie_select
ok 4 rtc.uie_select
# RUN rtc.alarm_alm_set ...
# rtctest.c:262:alarm_alm_set:Alarm time now set to 00:18:31.
# rtctest.c:282:alarm_alm_set:data: 1a0
# OK rtc.alarm_alm_set
ok 5 rtc.alarm_alm_set
# RUN rtc.alarm_wkalm_set ...
# rtctest.c:334:alarm_wkalm_set:Alarm time now set to 28/01/2025 00:18:34.
# rtctest.c:353:alarm_wkalm_set:Expected new (1738023515) == secs (1738023514)
# alarm_wkalm_set: Test terminated by assertion
# FAIL rtc.alarm_wkalm_set
not ok 6 rtc.alarm_wkalm_set
# RUN rtc.alarm_alm_set_minute ...
# rtctest.c:394:alarm_alm_set_minute:Alarm time now set to 00:19:00.
# rtctest.c:414:alarm_alm_set_minute:data: 1a0
# rtctest.c:420:alarm_alm_set_minute:Expected new (1738023541) == secs
(1738023540)
# alarm_alm_set_minute: Test terminated by assertion
# FAIL rtc.alarm_alm_set_minute
not ok 7 rtc.alarm_alm_set_minute
# RUN rtc.alarm_wkalm_set_minute ...
# rtctest.c:464:alarm_wkalm_set_minute:Alarm time now set to
28/01/2025 00:20:00.
# rtctest.c:483:alarm_wkalm_set_minute:Expected new (1738023601) ==
secs (1738023600)
# alarm_wkalm_set_minute: Test terminated by assertion
# FAIL rtc.alarm_wkalm_set_minute
not ok 8 rtc.alarm_wkalm_set_minute
# FAILED: 5 / 8 tests passed.
# Totals: pass:5 fail:3 xfail:0 xpass:0 skip:0 error:0
Best regards,
Enric
> ---
> drivers/rtc/Kconfig | 11 ++
> drivers/rtc/Makefile | 1 +
> drivers/rtc/rtc-s32g.c | 385 +++++++++++++++++++++++++++++++++++++++++
> 3 files changed, 397 insertions(+)
> create mode 100644 drivers/rtc/rtc-s32g.c
>
> diff --git a/drivers/rtc/Kconfig b/drivers/rtc/Kconfig
> index 0bbbf778ecfa..510dc2db745d 100644
> --- a/drivers/rtc/Kconfig
> +++ b/drivers/rtc/Kconfig
> @@ -2103,4 +2103,15 @@ config RTC_DRV_AMLOGIC_A4
> This driver can also be built as a module. If so, the module
> will be called "rtc-amlogic-a4".
>
> +config RTC_DRV_S32G
> + tristate "RTC driver for S32G2/S32G3 SoCs"
> + depends on ARCH_S32 || COMPILE_TEST
> + depends on COMMON_CLK
> + help
> + Say yes to enable RTC driver for platforms based on the
> + S32G2/S32G3 SoC family.
> +
> + This RTC module can be used as a wakeup source.
> + Please note that it is not battery-powered.
> +
> endif # RTC_CLASS
> diff --git a/drivers/rtc/Makefile b/drivers/rtc/Makefile
> index 489b4ab07068..e4b616ecd5ce 100644
> --- a/drivers/rtc/Makefile
> +++ b/drivers/rtc/Makefile
> @@ -161,6 +161,7 @@ obj-$(CONFIG_RTC_DRV_RX8111) += rtc-rx8111.o
> obj-$(CONFIG_RTC_DRV_RX8581) += rtc-rx8581.o
> obj-$(CONFIG_RTC_DRV_RZN1) += rtc-rzn1.o
> obj-$(CONFIG_RTC_DRV_RENESAS_RTCA3) += rtc-renesas-rtca3.o
> +obj-$(CONFIG_RTC_DRV_S32G) += rtc-s32g.o
> obj-$(CONFIG_RTC_DRV_S35390A) += rtc-s35390a.o
> obj-$(CONFIG_RTC_DRV_S3C) += rtc-s3c.o
> obj-$(CONFIG_RTC_DRV_S5M) += rtc-s5m.o
> diff --git a/drivers/rtc/rtc-s32g.c b/drivers/rtc/rtc-s32g.c
> new file mode 100644
> index 000000000000..3a0818e972eb
> --- /dev/null
> +++ b/drivers/rtc/rtc-s32g.c
> @@ -0,0 +1,385 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * Copyright 2025 NXP
> + */
> +
> +#include <linux/bitfield.h>
> +#include <linux/clk.h>
> +#include <linux/iopoll.h>
> +#include <linux/of_irq.h>
> +#include <linux/platform_device.h>
> +#include <linux/rtc.h>
> +
> +#define RTCC_OFFSET 0x4ul
> +#define RTCS_OFFSET 0x8ul
> +#define APIVAL_OFFSET 0x10ul
> +
> +/* RTCC fields */
> +#define RTCC_CNTEN BIT(31)
> +#define RTCC_APIEN BIT(15)
> +#define RTCC_APIIE BIT(14)
> +#define RTCC_CLKSEL_MASK GENMASK(13, 12)
> +#define RTCC_DIV512EN BIT(11)
> +#define RTCC_DIV32EN BIT(10)
> +
> +/* RTCS fields */
> +#define RTCS_INV_API BIT(17)
> +#define RTCS_APIF BIT(13)
> +
> +#define APIVAL_MAX_VAL GENMASK(31, 0)
> +#define RTC_SYNCH_TIMEOUT (100 * USEC_PER_MSEC)
> +
> +/*
> + * S32G2 and S32G3 SoCs have RTC clock source1 reserved and
> + * should not be used.
> + */
> +#define RTC_CLK_SRC1_RESERVED BIT(1)
> +
> +/*
> + * S32G RTC module has a 512 value and a 32 value hardware frequency
> + * divisors (DIV512 and DIV32) which could be used to achieve higher
> + * counter ranges by lowering the RTC frequency.
> + */
> +enum {
> + DIV1 = 1,
> + DIV32 = 32,
> + DIV512 = 512,
> + DIV512_32 = 16384
> +};
> +
> +static const char *const rtc_clk_src[] = {
> + "source0",
> + "source1",
> + "source2",
> + "source3"
> +};
> +
> +struct rtc_priv {
> + struct rtc_device *rdev;
> + void __iomem *rtc_base;
> + struct clk *ipg;
> + struct clk *clk_src;
> + const struct rtc_soc_data *rtc_data;
> + u64 rtc_hz;
> + time64_t sleep_sec;
> + int irq;
> + u32 clk_src_idx;
> +};
> +
> +struct rtc_soc_data {
> + u32 clk_div;
> + u32 reserved_clk_mask;
> +};
> +
> +static const struct rtc_soc_data rtc_s32g2_data = {
> + .clk_div = DIV512_32,
> + .reserved_clk_mask = RTC_CLK_SRC1_RESERVED,
> +};
> +
> +static irqreturn_t s32g_rtc_handler(int irq, void *dev)
> +{
> + struct rtc_priv *priv = platform_get_drvdata(dev);
> + u32 status;
> +
> + status = readl(priv->rtc_base + RTCS_OFFSET);
> +
> + if (status & RTCS_APIF) {
> + writel(0x0, priv->rtc_base + APIVAL_OFFSET);
> + writel(status | RTCS_APIF, priv->rtc_base + RTCS_OFFSET);
> + }
> +
> + rtc_update_irq(priv->rdev, 1, RTC_IRQF | RTC_AF);
> +
> + return IRQ_HANDLED;
> +}
> +
> +/*
> + * The function is not really getting time from the RTC since the S32G RTC
> + * has several limitations. Thus, to setup alarm use system time.
> + */
> +static int s32g_rtc_read_time(struct device *dev,
> + struct rtc_time *tm)
> +{
> + struct rtc_priv *priv = dev_get_drvdata(dev);
> + time64_t sec;
> +
> + if (check_add_overflow(ktime_get_real_seconds(),
> + priv->sleep_sec, &sec))
> + return -ERANGE;
> +
> + rtc_time64_to_tm(sec, tm);
> +
> + return 0;
> +}
> +
> +static int s32g_rtc_read_alarm(struct device *dev, struct rtc_wkalrm *alrm)
> +{
> + struct rtc_priv *priv = dev_get_drvdata(dev);
> + u32 rtcc, rtcs;
> +
> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> + rtcs = readl(priv->rtc_base + RTCS_OFFSET);
> +
> + alrm->enabled = rtcc & RTCC_APIIE;
> + if (alrm->enabled)
> + alrm->pending = !(rtcs & RTCS_APIF);
> +
> + return 0;
> +}
> +
> +static int s32g_rtc_alarm_irq_enable(struct device *dev, unsigned int enabled)
> +{
> + struct rtc_priv *priv = dev_get_drvdata(dev);
> + u32 rtcc;
> +
> + /* RTC API functionality is used both for triggering interrupts
> + * and as a wakeup event. Hence it should always be enabled.
> + */
> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> + rtcc |= RTCC_APIEN | RTCC_APIIE;
> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> +
> + return 0;
> +}
> +
> +static int s32g_rtc_set_alarm(struct device *dev, struct rtc_wkalrm *alrm)
> +{
> + struct rtc_priv *priv = dev_get_drvdata(dev);
> + unsigned long long cycles;
> + long long t_offset;
> + time64_t alrm_time;
> + u32 rtcs;
> + int ret;
> +
> + alrm_time = rtc_tm_to_time64(&alrm->time);
> + t_offset = alrm_time - ktime_get_real_seconds() - priv->sleep_sec;
> + if (t_offset < 0)
> + return -ERANGE;
> +
> + cycles = t_offset * priv->rtc_hz;
> + if (cycles > APIVAL_MAX_VAL)
> + return -ERANGE;
> +
> + /* APIVAL could have been reset from the IRQ handler.
> + * Hence, we wait in case there is a synchronization process.
> + */
> + ret = read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
> + 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
> + if (ret)
> + return ret;
> +
> + writel(cycles, priv->rtc_base + APIVAL_OFFSET);
> +
> + return read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
> + 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
> +}
> +
> +/*
> + * Disable the 32-bit free running counter.
> + * This allows Clock Source and Divisors selection
> + * to be performed without causing synchronization issues.
> + */
> +static void s32g_rtc_disable(struct rtc_priv *priv)
> +{
> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> +
> + rtcc &= ~RTCC_CNTEN;
> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> +}
> +
> +static void s32g_rtc_enable(struct rtc_priv *priv)
> +{
> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> +
> + rtcc |= RTCC_CNTEN;
> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> +}
> +
> +static int rtc_clk_src_setup(struct rtc_priv *priv)
> +{
> + u32 rtcc;
> +
> + rtcc = FIELD_PREP(RTCC_CLKSEL_MASK, priv->clk_src_idx);
> +
> + switch (priv->rtc_data->clk_div) {
> + case DIV512_32:
> + rtcc |= RTCC_DIV512EN;
> + rtcc |= RTCC_DIV32EN;
> + break;
> + case DIV512:
> + rtcc |= RTCC_DIV512EN;
> + break;
> + case DIV32:
> + rtcc |= RTCC_DIV32EN;
> + break;
> + case DIV1:
> + break;
> + default:
> + return -EINVAL;
> + }
> +
> + rtcc |= RTCC_APIEN | RTCC_APIIE;
> + /*
> + * Make sure the CNTEN is 0 before we configure
> + * the clock source and dividers.
> + */
> + s32g_rtc_disable(priv);
> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> + s32g_rtc_enable(priv);
> +
> + return 0;
> +}
> +
> +static const struct rtc_class_ops rtc_ops = {
> + .read_time = s32g_rtc_read_time,
> + .read_alarm = s32g_rtc_read_alarm,
> + .set_alarm = s32g_rtc_set_alarm,
> + .alarm_irq_enable = s32g_rtc_alarm_irq_enable,
> +};
> +
> +static int rtc_clk_dts_setup(struct rtc_priv *priv,
> + struct device *dev)
> +{
> + u32 i;
> +
> + priv->ipg = devm_clk_get_enabled(dev, "ipg");
> + if (IS_ERR(priv->ipg))
> + return dev_err_probe(dev, PTR_ERR(priv->ipg),
> + "Failed to get 'ipg' clock\n");
> +
> + for (i = 0; i < ARRAY_SIZE(rtc_clk_src); i++) {
> + if (priv->rtc_data->reserved_clk_mask & BIT(i))
> + return -EOPNOTSUPP;
> +
> + priv->clk_src = devm_clk_get_enabled(dev, rtc_clk_src[i]);
> + if (!IS_ERR(priv->clk_src)) {
> + priv->clk_src_idx = i;
> + break;
> + }
> + }
> +
> + if (IS_ERR(priv->clk_src))
> + return dev_err_probe(dev, PTR_ERR(priv->clk_src),
> + "Failed to get rtc module clock source\n");
> +
> + return 0;
> +}
> +
> +static int s32g_rtc_probe(struct platform_device *pdev)
> +{
> + struct device *dev = &pdev->dev;
> + struct rtc_priv *priv;
> + unsigned long rtc_hz;
> + int ret;
> +
> + priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
> + if (!priv)
> + return -ENOMEM;
> +
> + priv->rtc_data = of_device_get_match_data(dev);
> + if (!priv->rtc_data)
> + return -ENODEV;
> +
> + priv->rtc_base = devm_platform_ioremap_resource(pdev, 0);
> + if (IS_ERR(priv->rtc_base))
> + return PTR_ERR(priv->rtc_base);
> +
> + device_init_wakeup(dev, true);
> +
> + ret = rtc_clk_dts_setup(priv, dev);
> + if (ret)
> + return ret;
> +
> + priv->rdev = devm_rtc_allocate_device(dev);
> + if (IS_ERR(priv->rdev))
> + return PTR_ERR(priv->rdev);
> +
> + ret = rtc_clk_src_setup(priv);
> + if (ret)
> + return ret;
> +
> + priv->irq = platform_get_irq(pdev, 0);
> + if (priv->irq < 0) {
> + ret = priv->irq;
> + goto disable_rtc;
> + }
> +
> + rtc_hz = clk_get_rate(priv->clk_src);
> + if (!rtc_hz) {
> + dev_err(dev, "Failed to get RTC frequency\n");
> + ret = -EINVAL;
> + goto disable_rtc;
> + }
> +
> + priv->rtc_hz = DIV_ROUND_UP(rtc_hz, priv->rtc_data->clk_div);
> +
> + platform_set_drvdata(pdev, priv);
> + priv->rdev->ops = &rtc_ops;
> +
> + ret = devm_request_irq(dev, priv->irq,
> + s32g_rtc_handler, 0, dev_name(dev), pdev);
> + if (ret) {
> + dev_err(dev, "Request interrupt %d failed, error: %d\n",
> + priv->irq, ret);
> + goto disable_rtc;
> + }
> +
> + ret = devm_rtc_register_device(priv->rdev);
> + if (ret)
> + goto disable_rtc;
> +
> + return 0;
> +
> +disable_rtc:
> + s32g_rtc_disable(priv);
> + return ret;
> +}
> +
> +static int s32g_rtc_suspend(struct device *dev)
> +{
> + struct rtc_priv *priv = dev_get_drvdata(dev);
> + u32 apival = readl(priv->rtc_base + APIVAL_OFFSET);
> +
> + if (check_add_overflow(priv->sleep_sec, div64_u64(apival, priv->rtc_hz),
> + &priv->sleep_sec)) {
> + dev_warn(dev, "Overflow on sleep cycles occurred. Resetting to 0.\n");
> + priv->sleep_sec = 0;
> + }
> +
> + return 0;
> +}
> +
> +static int s32g_rtc_resume(struct device *dev)
> +{
> + struct rtc_priv *priv = dev_get_drvdata(dev);
> +
> + /* The transition from resume to run is a reset event.
> + * This leads to the RTC registers being reset after resume from
> + * suspend. It is uncommon, but this behaviour has been observed
> + * on S32G RTC after issuing a Suspend to RAM operation.
> + * Thus, reconfigure RTC registers on the resume path.
> + */
> + return rtc_clk_src_setup(priv);
> +}
> +
> +static const struct of_device_id rtc_dt_ids[] = {
> + { .compatible = "nxp,s32g2-rtc", .data = &rtc_s32g2_data },
> + { /* sentinel */ },
> +};
> +
> +static DEFINE_SIMPLE_DEV_PM_OPS(s32g_rtc_pm_ops,
> + s32g_rtc_suspend, s32g_rtc_resume);
> +
> +static struct platform_driver s32g_rtc_driver = {
> + .driver = {
> + .name = "s32g-rtc",
> + .pm = pm_sleep_ptr(&s32g_rtc_pm_ops),
> + .of_match_table = rtc_dt_ids,
> + },
> + .probe = s32g_rtc_probe,
> +};
> +module_platform_driver(s32g_rtc_driver);
> +
> +MODULE_AUTHOR("NXP");
> +MODULE_DESCRIPTION("NXP RTC driver for S32G2/S32G3");
> +MODULE_LICENSE("GPL");
> --
> 2.45.2
>
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support
2025-04-09 14:14 ` Enric Balletbo i Serra
@ 2025-04-09 14:26 ` Ciprian Marian Costea
2025-04-09 14:42 ` Enric Balletbo i Serra
0 siblings, 1 reply; 19+ messages in thread
From: Ciprian Marian Costea @ 2025-04-09 14:26 UTC (permalink / raw)
To: Enric Balletbo i Serra
Cc: Alexandre Belloni, Rob Herring, Krzysztof Kozlowski, Conor Dooley,
Catalin Marinas, Will Deacon, linux-rtc, devicetree, linux-kernel,
linux-arm-kernel, NXP S32 Linux, imx, Christophe Lizzi,
Alberto Ruiz, Eric Chanudet, Bogdan Hamciuc, Ghennadi Procopciuc,
Frank Li
On 4/9/2025 5:14 PM, Enric Balletbo i Serra wrote:
> Hi Ciprian,
>
> Many thanks for your patch.
>
> On Thu, Apr 3, 2025 at 12:34 PM Ciprian Costea
> <ciprianmarian.costea@oss.nxp.com> wrote:
>>
>> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>>
>> Add a RTC driver for NXP S32G2/S32G3 SoCs.
>>
>> RTC tracks clock time during system suspend. It can be a wakeup source
>> for the S32G2/S32G3 SoC based boards.
>>
>> The RTC module from S32G2/S32G3 is not battery-powered and it is not kept
>> alive during system reset.
>>
>> Co-developed-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
>> Signed-off-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
>> Co-developed-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
>> Signed-off-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
>> Signed-off-by: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>> Reviewed-by: Frank Li <Frank.Li@nxp.com>
>
> I went through the code and it looks good to me, so:
>
> However, I did a basic rtctest selftest in my S32G-VNP-RDB3 board,
> and some tests failed (it shows a small deviation), any idea what
> could be wrong?
>
> This is the output of the test:
>
> # ./rtctest
> TAP version 13
> 1..8
> # Starting 8 tests from 1 test cases.
> # RUN rtc.date_read ...
> # rtctest.c:59:date_read:Current RTC date/time is 28/01/2025 00:17:49.
> # OK rtc.date_read
> ok 1 rtc.date_read
> # RUN rtc.date_read_loop ...
> # rtctest.c:124:date_read_loop:Continuously reading RTC time for 30s
> (with 11ms breaks after every read).
> # rtctest.c:151:date_read_loop:Performed 2638 RTC time reads.
> # OK rtc.date_read_loop
> ok 2 rtc.date_read_loop
> # RUN rtc.uie_read ...
> # OK rtc.uie_read
> ok 3 rtc.uie_read
> # RUN rtc.uie_select ...
> # OK rtc.uie_select
> ok 4 rtc.uie_select
> # RUN rtc.alarm_alm_set ...
> # rtctest.c:262:alarm_alm_set:Alarm time now set to 00:18:31.
> # rtctest.c:282:alarm_alm_set:data: 1a0
> # OK rtc.alarm_alm_set
> ok 5 rtc.alarm_alm_set
> # RUN rtc.alarm_wkalm_set ...
> # rtctest.c:334:alarm_wkalm_set:Alarm time now set to 28/01/2025 00:18:34.
> # rtctest.c:353:alarm_wkalm_set:Expected new (1738023515) == secs (1738023514)
> # alarm_wkalm_set: Test terminated by assertion
> # FAIL rtc.alarm_wkalm_set
> not ok 6 rtc.alarm_wkalm_set
> # RUN rtc.alarm_alm_set_minute ...
> # rtctest.c:394:alarm_alm_set_minute:Alarm time now set to 00:19:00.
> # rtctest.c:414:alarm_alm_set_minute:data: 1a0
> # rtctest.c:420:alarm_alm_set_minute:Expected new (1738023541) == secs
> (1738023540)
> # alarm_alm_set_minute: Test terminated by assertion
> # FAIL rtc.alarm_alm_set_minute
> not ok 7 rtc.alarm_alm_set_minute
> # RUN rtc.alarm_wkalm_set_minute ...
> # rtctest.c:464:alarm_wkalm_set_minute:Alarm time now set to
> 28/01/2025 00:20:00.
> # rtctest.c:483:alarm_wkalm_set_minute:Expected new (1738023601) ==
> secs (1738023600)
> # alarm_wkalm_set_minute: Test terminated by assertion
> # FAIL rtc.alarm_wkalm_set_minute
> not ok 8 rtc.alarm_wkalm_set_minute
> # FAILED: 5 / 8 tests passed.
> # Totals: pass:5 fail:3 xfail:0 xpass:0 skip:0 error:0
>
> Best regards,
> Enric
>
Hello Enric,
In order to obtain a better resolution/precision you could try to only
use DIV512 hardware divisor instead of both DIV32 and DIV512.
Regards,
Ciprian
>> ---
>> drivers/rtc/Kconfig | 11 ++
>> drivers/rtc/Makefile | 1 +
>> drivers/rtc/rtc-s32g.c | 385 +++++++++++++++++++++++++++++++++++++++++
>> 3 files changed, 397 insertions(+)
>> create mode 100644 drivers/rtc/rtc-s32g.c
>>
>> diff --git a/drivers/rtc/Kconfig b/drivers/rtc/Kconfig
>> index 0bbbf778ecfa..510dc2db745d 100644
>> --- a/drivers/rtc/Kconfig
>> +++ b/drivers/rtc/Kconfig
>> @@ -2103,4 +2103,15 @@ config RTC_DRV_AMLOGIC_A4
>> This driver can also be built as a module. If so, the module
>> will be called "rtc-amlogic-a4".
>>
>> +config RTC_DRV_S32G
>> + tristate "RTC driver for S32G2/S32G3 SoCs"
>> + depends on ARCH_S32 || COMPILE_TEST
>> + depends on COMMON_CLK
>> + help
>> + Say yes to enable RTC driver for platforms based on the
>> + S32G2/S32G3 SoC family.
>> +
>> + This RTC module can be used as a wakeup source.
>> + Please note that it is not battery-powered.
>> +
>> endif # RTC_CLASS
>> diff --git a/drivers/rtc/Makefile b/drivers/rtc/Makefile
>> index 489b4ab07068..e4b616ecd5ce 100644
>> --- a/drivers/rtc/Makefile
>> +++ b/drivers/rtc/Makefile
>> @@ -161,6 +161,7 @@ obj-$(CONFIG_RTC_DRV_RX8111) += rtc-rx8111.o
>> obj-$(CONFIG_RTC_DRV_RX8581) += rtc-rx8581.o
>> obj-$(CONFIG_RTC_DRV_RZN1) += rtc-rzn1.o
>> obj-$(CONFIG_RTC_DRV_RENESAS_RTCA3) += rtc-renesas-rtca3.o
>> +obj-$(CONFIG_RTC_DRV_S32G) += rtc-s32g.o
>> obj-$(CONFIG_RTC_DRV_S35390A) += rtc-s35390a.o
>> obj-$(CONFIG_RTC_DRV_S3C) += rtc-s3c.o
>> obj-$(CONFIG_RTC_DRV_S5M) += rtc-s5m.o
>> diff --git a/drivers/rtc/rtc-s32g.c b/drivers/rtc/rtc-s32g.c
>> new file mode 100644
>> index 000000000000..3a0818e972eb
>> --- /dev/null
>> +++ b/drivers/rtc/rtc-s32g.c
>> @@ -0,0 +1,385 @@
>> +// SPDX-License-Identifier: GPL-2.0-or-later
>> +/*
>> + * Copyright 2025 NXP
>> + */
>> +
>> +#include <linux/bitfield.h>
>> +#include <linux/clk.h>
>> +#include <linux/iopoll.h>
>> +#include <linux/of_irq.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/rtc.h>
>> +
>> +#define RTCC_OFFSET 0x4ul
>> +#define RTCS_OFFSET 0x8ul
>> +#define APIVAL_OFFSET 0x10ul
>> +
>> +/* RTCC fields */
>> +#define RTCC_CNTEN BIT(31)
>> +#define RTCC_APIEN BIT(15)
>> +#define RTCC_APIIE BIT(14)
>> +#define RTCC_CLKSEL_MASK GENMASK(13, 12)
>> +#define RTCC_DIV512EN BIT(11)
>> +#define RTCC_DIV32EN BIT(10)
>> +
>> +/* RTCS fields */
>> +#define RTCS_INV_API BIT(17)
>> +#define RTCS_APIF BIT(13)
>> +
>> +#define APIVAL_MAX_VAL GENMASK(31, 0)
>> +#define RTC_SYNCH_TIMEOUT (100 * USEC_PER_MSEC)
>> +
>> +/*
>> + * S32G2 and S32G3 SoCs have RTC clock source1 reserved and
>> + * should not be used.
>> + */
>> +#define RTC_CLK_SRC1_RESERVED BIT(1)
>> +
>> +/*
>> + * S32G RTC module has a 512 value and a 32 value hardware frequency
>> + * divisors (DIV512 and DIV32) which could be used to achieve higher
>> + * counter ranges by lowering the RTC frequency.
>> + */
>> +enum {
>> + DIV1 = 1,
>> + DIV32 = 32,
>> + DIV512 = 512,
>> + DIV512_32 = 16384
>> +};
>> +
>> +static const char *const rtc_clk_src[] = {
>> + "source0",
>> + "source1",
>> + "source2",
>> + "source3"
>> +};
>> +
>> +struct rtc_priv {
>> + struct rtc_device *rdev;
>> + void __iomem *rtc_base;
>> + struct clk *ipg;
>> + struct clk *clk_src;
>> + const struct rtc_soc_data *rtc_data;
>> + u64 rtc_hz;
>> + time64_t sleep_sec;
>> + int irq;
>> + u32 clk_src_idx;
>> +};
>> +
>> +struct rtc_soc_data {
>> + u32 clk_div;
>> + u32 reserved_clk_mask;
>> +};
>> +
>> +static const struct rtc_soc_data rtc_s32g2_data = {
>> + .clk_div = DIV512_32,
>> + .reserved_clk_mask = RTC_CLK_SRC1_RESERVED,
>> +};
>> +
>> +static irqreturn_t s32g_rtc_handler(int irq, void *dev)
>> +{
>> + struct rtc_priv *priv = platform_get_drvdata(dev);
>> + u32 status;
>> +
>> + status = readl(priv->rtc_base + RTCS_OFFSET);
>> +
>> + if (status & RTCS_APIF) {
>> + writel(0x0, priv->rtc_base + APIVAL_OFFSET);
>> + writel(status | RTCS_APIF, priv->rtc_base + RTCS_OFFSET);
>> + }
>> +
>> + rtc_update_irq(priv->rdev, 1, RTC_IRQF | RTC_AF);
>> +
>> + return IRQ_HANDLED;
>> +}
>> +
>> +/*
>> + * The function is not really getting time from the RTC since the S32G RTC
>> + * has several limitations. Thus, to setup alarm use system time.
>> + */
>> +static int s32g_rtc_read_time(struct device *dev,
>> + struct rtc_time *tm)
>> +{
>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>> + time64_t sec;
>> +
>> + if (check_add_overflow(ktime_get_real_seconds(),
>> + priv->sleep_sec, &sec))
>> + return -ERANGE;
>> +
>> + rtc_time64_to_tm(sec, tm);
>> +
>> + return 0;
>> +}
>> +
>> +static int s32g_rtc_read_alarm(struct device *dev, struct rtc_wkalrm *alrm)
>> +{
>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>> + u32 rtcc, rtcs;
>> +
>> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>> + rtcs = readl(priv->rtc_base + RTCS_OFFSET);
>> +
>> + alrm->enabled = rtcc & RTCC_APIIE;
>> + if (alrm->enabled)
>> + alrm->pending = !(rtcs & RTCS_APIF);
>> +
>> + return 0;
>> +}
>> +
>> +static int s32g_rtc_alarm_irq_enable(struct device *dev, unsigned int enabled)
>> +{
>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>> + u32 rtcc;
>> +
>> + /* RTC API functionality is used both for triggering interrupts
>> + * and as a wakeup event. Hence it should always be enabled.
>> + */
>> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>> + rtcc |= RTCC_APIEN | RTCC_APIIE;
>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>> +
>> + return 0;
>> +}
>> +
>> +static int s32g_rtc_set_alarm(struct device *dev, struct rtc_wkalrm *alrm)
>> +{
>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>> + unsigned long long cycles;
>> + long long t_offset;
>> + time64_t alrm_time;
>> + u32 rtcs;
>> + int ret;
>> +
>> + alrm_time = rtc_tm_to_time64(&alrm->time);
>> + t_offset = alrm_time - ktime_get_real_seconds() - priv->sleep_sec;
>> + if (t_offset < 0)
>> + return -ERANGE;
>> +
>> + cycles = t_offset * priv->rtc_hz;
>> + if (cycles > APIVAL_MAX_VAL)
>> + return -ERANGE;
>> +
>> + /* APIVAL could have been reset from the IRQ handler.
>> + * Hence, we wait in case there is a synchronization process.
>> + */
>> + ret = read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
>> + 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
>> + if (ret)
>> + return ret;
>> +
>> + writel(cycles, priv->rtc_base + APIVAL_OFFSET);
>> +
>> + return read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
>> + 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
>> +}
>> +
>> +/*
>> + * Disable the 32-bit free running counter.
>> + * This allows Clock Source and Divisors selection
>> + * to be performed without causing synchronization issues.
>> + */
>> +static void s32g_rtc_disable(struct rtc_priv *priv)
>> +{
>> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>> +
>> + rtcc &= ~RTCC_CNTEN;
>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>> +}
>> +
>> +static void s32g_rtc_enable(struct rtc_priv *priv)
>> +{
>> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>> +
>> + rtcc |= RTCC_CNTEN;
>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>> +}
>> +
>> +static int rtc_clk_src_setup(struct rtc_priv *priv)
>> +{
>> + u32 rtcc;
>> +
>> + rtcc = FIELD_PREP(RTCC_CLKSEL_MASK, priv->clk_src_idx);
>> +
>> + switch (priv->rtc_data->clk_div) {
>> + case DIV512_32:
>> + rtcc |= RTCC_DIV512EN;
>> + rtcc |= RTCC_DIV32EN;
>> + break;
>> + case DIV512:
>> + rtcc |= RTCC_DIV512EN;
>> + break;
>> + case DIV32:
>> + rtcc |= RTCC_DIV32EN;
>> + break;
>> + case DIV1:
>> + break;
>> + default:
>> + return -EINVAL;
>> + }
>> +
>> + rtcc |= RTCC_APIEN | RTCC_APIIE;
>> + /*
>> + * Make sure the CNTEN is 0 before we configure
>> + * the clock source and dividers.
>> + */
>> + s32g_rtc_disable(priv);
>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>> + s32g_rtc_enable(priv);
>> +
>> + return 0;
>> +}
>> +
>> +static const struct rtc_class_ops rtc_ops = {
>> + .read_time = s32g_rtc_read_time,
>> + .read_alarm = s32g_rtc_read_alarm,
>> + .set_alarm = s32g_rtc_set_alarm,
>> + .alarm_irq_enable = s32g_rtc_alarm_irq_enable,
>> +};
>> +
>> +static int rtc_clk_dts_setup(struct rtc_priv *priv,
>> + struct device *dev)
>> +{
>> + u32 i;
>> +
>> + priv->ipg = devm_clk_get_enabled(dev, "ipg");
>> + if (IS_ERR(priv->ipg))
>> + return dev_err_probe(dev, PTR_ERR(priv->ipg),
>> + "Failed to get 'ipg' clock\n");
>> +
>> + for (i = 0; i < ARRAY_SIZE(rtc_clk_src); i++) {
>> + if (priv->rtc_data->reserved_clk_mask & BIT(i))
>> + return -EOPNOTSUPP;
>> +
>> + priv->clk_src = devm_clk_get_enabled(dev, rtc_clk_src[i]);
>> + if (!IS_ERR(priv->clk_src)) {
>> + priv->clk_src_idx = i;
>> + break;
>> + }
>> + }
>> +
>> + if (IS_ERR(priv->clk_src))
>> + return dev_err_probe(dev, PTR_ERR(priv->clk_src),
>> + "Failed to get rtc module clock source\n");
>> +
>> + return 0;
>> +}
>> +
>> +static int s32g_rtc_probe(struct platform_device *pdev)
>> +{
>> + struct device *dev = &pdev->dev;
>> + struct rtc_priv *priv;
>> + unsigned long rtc_hz;
>> + int ret;
>> +
>> + priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
>> + if (!priv)
>> + return -ENOMEM;
>> +
>> + priv->rtc_data = of_device_get_match_data(dev);
>> + if (!priv->rtc_data)
>> + return -ENODEV;
>> +
>> + priv->rtc_base = devm_platform_ioremap_resource(pdev, 0);
>> + if (IS_ERR(priv->rtc_base))
>> + return PTR_ERR(priv->rtc_base);
>> +
>> + device_init_wakeup(dev, true);
>> +
>> + ret = rtc_clk_dts_setup(priv, dev);
>> + if (ret)
>> + return ret;
>> +
>> + priv->rdev = devm_rtc_allocate_device(dev);
>> + if (IS_ERR(priv->rdev))
>> + return PTR_ERR(priv->rdev);
>> +
>> + ret = rtc_clk_src_setup(priv);
>> + if (ret)
>> + return ret;
>> +
>> + priv->irq = platform_get_irq(pdev, 0);
>> + if (priv->irq < 0) {
>> + ret = priv->irq;
>> + goto disable_rtc;
>> + }
>> +
>> + rtc_hz = clk_get_rate(priv->clk_src);
>> + if (!rtc_hz) {
>> + dev_err(dev, "Failed to get RTC frequency\n");
>> + ret = -EINVAL;
>> + goto disable_rtc;
>> + }
>> +
>> + priv->rtc_hz = DIV_ROUND_UP(rtc_hz, priv->rtc_data->clk_div);
>> +
>> + platform_set_drvdata(pdev, priv);
>> + priv->rdev->ops = &rtc_ops;
>> +
>> + ret = devm_request_irq(dev, priv->irq,
>> + s32g_rtc_handler, 0, dev_name(dev), pdev);
>> + if (ret) {
>> + dev_err(dev, "Request interrupt %d failed, error: %d\n",
>> + priv->irq, ret);
>> + goto disable_rtc;
>> + }
>> +
>> + ret = devm_rtc_register_device(priv->rdev);
>> + if (ret)
>> + goto disable_rtc;
>> +
>> + return 0;
>> +
>> +disable_rtc:
>> + s32g_rtc_disable(priv);
>> + return ret;
>> +}
>> +
>> +static int s32g_rtc_suspend(struct device *dev)
>> +{
>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>> + u32 apival = readl(priv->rtc_base + APIVAL_OFFSET);
>> +
>> + if (check_add_overflow(priv->sleep_sec, div64_u64(apival, priv->rtc_hz),
>> + &priv->sleep_sec)) {
>> + dev_warn(dev, "Overflow on sleep cycles occurred. Resetting to 0.\n");
>> + priv->sleep_sec = 0;
>> + }
>> +
>> + return 0;
>> +}
>> +
>> +static int s32g_rtc_resume(struct device *dev)
>> +{
>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>> +
>> + /* The transition from resume to run is a reset event.
>> + * This leads to the RTC registers being reset after resume from
>> + * suspend. It is uncommon, but this behaviour has been observed
>> + * on S32G RTC after issuing a Suspend to RAM operation.
>> + * Thus, reconfigure RTC registers on the resume path.
>> + */
>> + return rtc_clk_src_setup(priv);
>> +}
>> +
>> +static const struct of_device_id rtc_dt_ids[] = {
>> + { .compatible = "nxp,s32g2-rtc", .data = &rtc_s32g2_data },
>> + { /* sentinel */ },
>> +};
>> +
>> +static DEFINE_SIMPLE_DEV_PM_OPS(s32g_rtc_pm_ops,
>> + s32g_rtc_suspend, s32g_rtc_resume);
>> +
>> +static struct platform_driver s32g_rtc_driver = {
>> + .driver = {
>> + .name = "s32g-rtc",
>> + .pm = pm_sleep_ptr(&s32g_rtc_pm_ops),
>> + .of_match_table = rtc_dt_ids,
>> + },
>> + .probe = s32g_rtc_probe,
>> +};
>> +module_platform_driver(s32g_rtc_driver);
>> +
>> +MODULE_AUTHOR("NXP");
>> +MODULE_DESCRIPTION("NXP RTC driver for S32G2/S32G3");
>> +MODULE_LICENSE("GPL");
>> --
>> 2.45.2
>>
>
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support
2025-04-09 14:26 ` Ciprian Marian Costea
@ 2025-04-09 14:42 ` Enric Balletbo i Serra
2025-04-09 14:58 ` Ciprian Marian Costea
0 siblings, 1 reply; 19+ messages in thread
From: Enric Balletbo i Serra @ 2025-04-09 14:42 UTC (permalink / raw)
To: Ciprian Marian Costea
Cc: Alexandre Belloni, Rob Herring, Krzysztof Kozlowski, Conor Dooley,
Catalin Marinas, Will Deacon, linux-rtc, devicetree, linux-kernel,
linux-arm-kernel, NXP S32 Linux, imx, Christophe Lizzi,
Alberto Ruiz, Eric Chanudet, Bogdan Hamciuc, Ghennadi Procopciuc,
Frank Li
Hi Ciprian,
On Wed, Apr 9, 2025 at 4:26 PM Ciprian Marian Costea
<ciprianmarian.costea@oss.nxp.com> wrote:
>
> On 4/9/2025 5:14 PM, Enric Balletbo i Serra wrote:
> > Hi Ciprian,
> >
> > Many thanks for your patch.
> >
> > On Thu, Apr 3, 2025 at 12:34 PM Ciprian Costea
> > <ciprianmarian.costea@oss.nxp.com> wrote:
> >>
> >> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
> >>
> >> Add a RTC driver for NXP S32G2/S32G3 SoCs.
> >>
> >> RTC tracks clock time during system suspend. It can be a wakeup source
> >> for the S32G2/S32G3 SoC based boards.
> >>
> >> The RTC module from S32G2/S32G3 is not battery-powered and it is not kept
> >> alive during system reset.
> >>
> >> Co-developed-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
> >> Signed-off-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
> >> Co-developed-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
> >> Signed-off-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
> >> Signed-off-by: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
> >> Reviewed-by: Frank Li <Frank.Li@nxp.com>
> >
> > I went through the code and it looks good to me, so:
> >
> > However, I did a basic rtctest selftest in my S32G-VNP-RDB3 board,
> > and some tests failed (it shows a small deviation), any idea what
> > could be wrong?
> >
> > This is the output of the test:
> >
> > # ./rtctest
> > TAP version 13
> > 1..8
> > # Starting 8 tests from 1 test cases.
> > # RUN rtc.date_read ...
> > # rtctest.c:59:date_read:Current RTC date/time is 28/01/2025 00:17:49.
> > # OK rtc.date_read
> > ok 1 rtc.date_read
> > # RUN rtc.date_read_loop ...
> > # rtctest.c:124:date_read_loop:Continuously reading RTC time for 30s
> > (with 11ms breaks after every read).
> > # rtctest.c:151:date_read_loop:Performed 2638 RTC time reads.
> > # OK rtc.date_read_loop
> > ok 2 rtc.date_read_loop
> > # RUN rtc.uie_read ...
> > # OK rtc.uie_read
> > ok 3 rtc.uie_read
> > # RUN rtc.uie_select ...
> > # OK rtc.uie_select
> > ok 4 rtc.uie_select
> > # RUN rtc.alarm_alm_set ...
> > # rtctest.c:262:alarm_alm_set:Alarm time now set to 00:18:31.
> > # rtctest.c:282:alarm_alm_set:data: 1a0
> > # OK rtc.alarm_alm_set
> > ok 5 rtc.alarm_alm_set
> > # RUN rtc.alarm_wkalm_set ...
> > # rtctest.c:334:alarm_wkalm_set:Alarm time now set to 28/01/2025 00:18:34.
> > # rtctest.c:353:alarm_wkalm_set:Expected new (1738023515) == secs (1738023514)
> > # alarm_wkalm_set: Test terminated by assertion
> > # FAIL rtc.alarm_wkalm_set
> > not ok 6 rtc.alarm_wkalm_set
> > # RUN rtc.alarm_alm_set_minute ...
> > # rtctest.c:394:alarm_alm_set_minute:Alarm time now set to 00:19:00.
> > # rtctest.c:414:alarm_alm_set_minute:data: 1a0
> > # rtctest.c:420:alarm_alm_set_minute:Expected new (1738023541) == secs
> > (1738023540)
> > # alarm_alm_set_minute: Test terminated by assertion
> > # FAIL rtc.alarm_alm_set_minute
> > not ok 7 rtc.alarm_alm_set_minute
> > # RUN rtc.alarm_wkalm_set_minute ...
> > # rtctest.c:464:alarm_wkalm_set_minute:Alarm time now set to
> > 28/01/2025 00:20:00.
> > # rtctest.c:483:alarm_wkalm_set_minute:Expected new (1738023601) ==
> > secs (1738023600)
> > # alarm_wkalm_set_minute: Test terminated by assertion
> > # FAIL rtc.alarm_wkalm_set_minute
> > not ok 8 rtc.alarm_wkalm_set_minute
> > # FAILED: 5 / 8 tests passed.
> > # Totals: pass:5 fail:3 xfail:0 xpass:0 skip:0 error:0
> >
> > Best regards,
> > Enric
> >
>
> Hello Enric,
>
> In order to obtain a better resolution/precision you could try to only
> use DIV512 hardware divisor instead of both DIV32 and DIV512.
>
That makes the tests pass, so I'm wondering if by default you should
set DIV512 in platform data instead. With that change you can add my
Tested-by: Enric Balletbo i Serra <eballetbo@kernel.org>
# ./rtctest
TAP version 13
1..8
# Starting 8 tests from 1 test cases.
# RUN rtc.date_read ...
# rtctest.c:59:date_read:Current RTC date/time is 28/01/2025 00:00:34.
# OK rtc.date_read
ok 1 rtc.date_read
# RUN rtc.date_read_loop ...
# rtctest.c:124:date_read_loop:Continuously reading RTC time for 30s
(with 11ms breaks after every read).
# rtctest.c:151:date_read_loop:Performed 2656 RTC time reads.
# OK rtc.date_read_loop
ok 2 rtc.date_read_loop
# RUN rtc.uie_read ...
# OK rtc.uie_read
ok 3 rtc.uie_read
# RUN rtc.uie_select ...
# OK rtc.uie_select
ok 4 rtc.uie_select
# RUN rtc.alarm_alm_set ...
# rtctest.c:262:alarm_alm_set:Alarm time now set to 00:01:14.
# rtctest.c:282:alarm_alm_set:data: 1a0
# OK rtc.alarm_alm_set
ok 5 rtc.alarm_alm_set
# RUN rtc.alarm_wkalm_set ...
# rtctest.c:334:alarm_wkalm_set:Alarm time now set to 28/01/2025 00:01:17.
# OK rtc.alarm_wkalm_set
ok 6 rtc.alarm_wkalm_set
# RUN rtc.alarm_alm_set_minute ...
# rtctest.c:394:alarm_alm_set_minute:Alarm time now set to 00:02:00.
# rtctest.c:414:alarm_alm_set_minute:data: 1a0
# OK rtc.alarm_alm_set_minute
ok 7 rtc.alarm_alm_set_minute
# RUN rtc.alarm_wkalm_set_minute ...
# rtctest.c:464:alarm_wkalm_set_minute:Alarm time now set to
28/01/2025 00:03:00.
# OK rtc.alarm_wkalm_set_minute
ok 8 rtc.alarm_wkalm_set_minute
# PASSED: 8 / 8 tests passed.
# Totals: pass:8 fail:0 xfail:0 xpass:0 skip:0 error:0
Thanks,
Enric
> Regards,
> Ciprian
>
> >> ---
> >> drivers/rtc/Kconfig | 11 ++
> >> drivers/rtc/Makefile | 1 +
> >> drivers/rtc/rtc-s32g.c | 385 +++++++++++++++++++++++++++++++++++++++++
> >> 3 files changed, 397 insertions(+)
> >> create mode 100644 drivers/rtc/rtc-s32g.c
> >>
> >> diff --git a/drivers/rtc/Kconfig b/drivers/rtc/Kconfig
> >> index 0bbbf778ecfa..510dc2db745d 100644
> >> --- a/drivers/rtc/Kconfig
> >> +++ b/drivers/rtc/Kconfig
> >> @@ -2103,4 +2103,15 @@ config RTC_DRV_AMLOGIC_A4
> >> This driver can also be built as a module. If so, the module
> >> will be called "rtc-amlogic-a4".
> >>
> >> +config RTC_DRV_S32G
> >> + tristate "RTC driver for S32G2/S32G3 SoCs"
> >> + depends on ARCH_S32 || COMPILE_TEST
> >> + depends on COMMON_CLK
> >> + help
> >> + Say yes to enable RTC driver for platforms based on the
> >> + S32G2/S32G3 SoC family.
> >> +
> >> + This RTC module can be used as a wakeup source.
> >> + Please note that it is not battery-powered.
> >> +
> >> endif # RTC_CLASS
> >> diff --git a/drivers/rtc/Makefile b/drivers/rtc/Makefile
> >> index 489b4ab07068..e4b616ecd5ce 100644
> >> --- a/drivers/rtc/Makefile
> >> +++ b/drivers/rtc/Makefile
> >> @@ -161,6 +161,7 @@ obj-$(CONFIG_RTC_DRV_RX8111) += rtc-rx8111.o
> >> obj-$(CONFIG_RTC_DRV_RX8581) += rtc-rx8581.o
> >> obj-$(CONFIG_RTC_DRV_RZN1) += rtc-rzn1.o
> >> obj-$(CONFIG_RTC_DRV_RENESAS_RTCA3) += rtc-renesas-rtca3.o
> >> +obj-$(CONFIG_RTC_DRV_S32G) += rtc-s32g.o
> >> obj-$(CONFIG_RTC_DRV_S35390A) += rtc-s35390a.o
> >> obj-$(CONFIG_RTC_DRV_S3C) += rtc-s3c.o
> >> obj-$(CONFIG_RTC_DRV_S5M) += rtc-s5m.o
> >> diff --git a/drivers/rtc/rtc-s32g.c b/drivers/rtc/rtc-s32g.c
> >> new file mode 100644
> >> index 000000000000..3a0818e972eb
> >> --- /dev/null
> >> +++ b/drivers/rtc/rtc-s32g.c
> >> @@ -0,0 +1,385 @@
> >> +// SPDX-License-Identifier: GPL-2.0-or-later
> >> +/*
> >> + * Copyright 2025 NXP
> >> + */
> >> +
> >> +#include <linux/bitfield.h>
> >> +#include <linux/clk.h>
> >> +#include <linux/iopoll.h>
> >> +#include <linux/of_irq.h>
> >> +#include <linux/platform_device.h>
> >> +#include <linux/rtc.h>
> >> +
> >> +#define RTCC_OFFSET 0x4ul
> >> +#define RTCS_OFFSET 0x8ul
> >> +#define APIVAL_OFFSET 0x10ul
> >> +
> >> +/* RTCC fields */
> >> +#define RTCC_CNTEN BIT(31)
> >> +#define RTCC_APIEN BIT(15)
> >> +#define RTCC_APIIE BIT(14)
> >> +#define RTCC_CLKSEL_MASK GENMASK(13, 12)
> >> +#define RTCC_DIV512EN BIT(11)
> >> +#define RTCC_DIV32EN BIT(10)
> >> +
> >> +/* RTCS fields */
> >> +#define RTCS_INV_API BIT(17)
> >> +#define RTCS_APIF BIT(13)
> >> +
> >> +#define APIVAL_MAX_VAL GENMASK(31, 0)
> >> +#define RTC_SYNCH_TIMEOUT (100 * USEC_PER_MSEC)
> >> +
> >> +/*
> >> + * S32G2 and S32G3 SoCs have RTC clock source1 reserved and
> >> + * should not be used.
> >> + */
> >> +#define RTC_CLK_SRC1_RESERVED BIT(1)
> >> +
> >> +/*
> >> + * S32G RTC module has a 512 value and a 32 value hardware frequency
> >> + * divisors (DIV512 and DIV32) which could be used to achieve higher
> >> + * counter ranges by lowering the RTC frequency.
> >> + */
> >> +enum {
> >> + DIV1 = 1,
> >> + DIV32 = 32,
> >> + DIV512 = 512,
> >> + DIV512_32 = 16384
> >> +};
> >> +
> >> +static const char *const rtc_clk_src[] = {
> >> + "source0",
> >> + "source1",
> >> + "source2",
> >> + "source3"
> >> +};
> >> +
> >> +struct rtc_priv {
> >> + struct rtc_device *rdev;
> >> + void __iomem *rtc_base;
> >> + struct clk *ipg;
> >> + struct clk *clk_src;
> >> + const struct rtc_soc_data *rtc_data;
> >> + u64 rtc_hz;
> >> + time64_t sleep_sec;
> >> + int irq;
> >> + u32 clk_src_idx;
> >> +};
> >> +
> >> +struct rtc_soc_data {
> >> + u32 clk_div;
> >> + u32 reserved_clk_mask;
> >> +};
> >> +
> >> +static const struct rtc_soc_data rtc_s32g2_data = {
> >> + .clk_div = DIV512_32,
> >> + .reserved_clk_mask = RTC_CLK_SRC1_RESERVED,
> >> +};
> >> +
> >> +static irqreturn_t s32g_rtc_handler(int irq, void *dev)
> >> +{
> >> + struct rtc_priv *priv = platform_get_drvdata(dev);
> >> + u32 status;
> >> +
> >> + status = readl(priv->rtc_base + RTCS_OFFSET);
> >> +
> >> + if (status & RTCS_APIF) {
> >> + writel(0x0, priv->rtc_base + APIVAL_OFFSET);
> >> + writel(status | RTCS_APIF, priv->rtc_base + RTCS_OFFSET);
> >> + }
> >> +
> >> + rtc_update_irq(priv->rdev, 1, RTC_IRQF | RTC_AF);
> >> +
> >> + return IRQ_HANDLED;
> >> +}
> >> +
> >> +/*
> >> + * The function is not really getting time from the RTC since the S32G RTC
> >> + * has several limitations. Thus, to setup alarm use system time.
> >> + */
> >> +static int s32g_rtc_read_time(struct device *dev,
> >> + struct rtc_time *tm)
> >> +{
> >> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >> + time64_t sec;
> >> +
> >> + if (check_add_overflow(ktime_get_real_seconds(),
> >> + priv->sleep_sec, &sec))
> >> + return -ERANGE;
> >> +
> >> + rtc_time64_to_tm(sec, tm);
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static int s32g_rtc_read_alarm(struct device *dev, struct rtc_wkalrm *alrm)
> >> +{
> >> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >> + u32 rtcc, rtcs;
> >> +
> >> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> >> + rtcs = readl(priv->rtc_base + RTCS_OFFSET);
> >> +
> >> + alrm->enabled = rtcc & RTCC_APIIE;
> >> + if (alrm->enabled)
> >> + alrm->pending = !(rtcs & RTCS_APIF);
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static int s32g_rtc_alarm_irq_enable(struct device *dev, unsigned int enabled)
> >> +{
> >> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >> + u32 rtcc;
> >> +
> >> + /* RTC API functionality is used both for triggering interrupts
> >> + * and as a wakeup event. Hence it should always be enabled.
> >> + */
> >> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> >> + rtcc |= RTCC_APIEN | RTCC_APIIE;
> >> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static int s32g_rtc_set_alarm(struct device *dev, struct rtc_wkalrm *alrm)
> >> +{
> >> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >> + unsigned long long cycles;
> >> + long long t_offset;
> >> + time64_t alrm_time;
> >> + u32 rtcs;
> >> + int ret;
> >> +
> >> + alrm_time = rtc_tm_to_time64(&alrm->time);
> >> + t_offset = alrm_time - ktime_get_real_seconds() - priv->sleep_sec;
> >> + if (t_offset < 0)
> >> + return -ERANGE;
> >> +
> >> + cycles = t_offset * priv->rtc_hz;
> >> + if (cycles > APIVAL_MAX_VAL)
> >> + return -ERANGE;
> >> +
> >> + /* APIVAL could have been reset from the IRQ handler.
> >> + * Hence, we wait in case there is a synchronization process.
> >> + */
> >> + ret = read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
> >> + 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
> >> + if (ret)
> >> + return ret;
> >> +
> >> + writel(cycles, priv->rtc_base + APIVAL_OFFSET);
> >> +
> >> + return read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
> >> + 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
> >> +}
> >> +
> >> +/*
> >> + * Disable the 32-bit free running counter.
> >> + * This allows Clock Source and Divisors selection
> >> + * to be performed without causing synchronization issues.
> >> + */
> >> +static void s32g_rtc_disable(struct rtc_priv *priv)
> >> +{
> >> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> >> +
> >> + rtcc &= ~RTCC_CNTEN;
> >> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> >> +}
> >> +
> >> +static void s32g_rtc_enable(struct rtc_priv *priv)
> >> +{
> >> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> >> +
> >> + rtcc |= RTCC_CNTEN;
> >> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> >> +}
> >> +
> >> +static int rtc_clk_src_setup(struct rtc_priv *priv)
> >> +{
> >> + u32 rtcc;
> >> +
> >> + rtcc = FIELD_PREP(RTCC_CLKSEL_MASK, priv->clk_src_idx);
> >> +
> >> + switch (priv->rtc_data->clk_div) {
> >> + case DIV512_32:
> >> + rtcc |= RTCC_DIV512EN;
> >> + rtcc |= RTCC_DIV32EN;
> >> + break;
> >> + case DIV512:
> >> + rtcc |= RTCC_DIV512EN;
> >> + break;
> >> + case DIV32:
> >> + rtcc |= RTCC_DIV32EN;
> >> + break;
> >> + case DIV1:
> >> + break;
> >> + default:
> >> + return -EINVAL;
> >> + }
> >> +
> >> + rtcc |= RTCC_APIEN | RTCC_APIIE;
> >> + /*
> >> + * Make sure the CNTEN is 0 before we configure
> >> + * the clock source and dividers.
> >> + */
> >> + s32g_rtc_disable(priv);
> >> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> >> + s32g_rtc_enable(priv);
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static const struct rtc_class_ops rtc_ops = {
> >> + .read_time = s32g_rtc_read_time,
> >> + .read_alarm = s32g_rtc_read_alarm,
> >> + .set_alarm = s32g_rtc_set_alarm,
> >> + .alarm_irq_enable = s32g_rtc_alarm_irq_enable,
> >> +};
> >> +
> >> +static int rtc_clk_dts_setup(struct rtc_priv *priv,
> >> + struct device *dev)
> >> +{
> >> + u32 i;
> >> +
> >> + priv->ipg = devm_clk_get_enabled(dev, "ipg");
> >> + if (IS_ERR(priv->ipg))
> >> + return dev_err_probe(dev, PTR_ERR(priv->ipg),
> >> + "Failed to get 'ipg' clock\n");
> >> +
> >> + for (i = 0; i < ARRAY_SIZE(rtc_clk_src); i++) {
> >> + if (priv->rtc_data->reserved_clk_mask & BIT(i))
> >> + return -EOPNOTSUPP;
> >> +
> >> + priv->clk_src = devm_clk_get_enabled(dev, rtc_clk_src[i]);
> >> + if (!IS_ERR(priv->clk_src)) {
> >> + priv->clk_src_idx = i;
> >> + break;
> >> + }
> >> + }
> >> +
> >> + if (IS_ERR(priv->clk_src))
> >> + return dev_err_probe(dev, PTR_ERR(priv->clk_src),
> >> + "Failed to get rtc module clock source\n");
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static int s32g_rtc_probe(struct platform_device *pdev)
> >> +{
> >> + struct device *dev = &pdev->dev;
> >> + struct rtc_priv *priv;
> >> + unsigned long rtc_hz;
> >> + int ret;
> >> +
> >> + priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
> >> + if (!priv)
> >> + return -ENOMEM;
> >> +
> >> + priv->rtc_data = of_device_get_match_data(dev);
> >> + if (!priv->rtc_data)
> >> + return -ENODEV;
> >> +
> >> + priv->rtc_base = devm_platform_ioremap_resource(pdev, 0);
> >> + if (IS_ERR(priv->rtc_base))
> >> + return PTR_ERR(priv->rtc_base);
> >> +
> >> + device_init_wakeup(dev, true);
> >> +
> >> + ret = rtc_clk_dts_setup(priv, dev);
> >> + if (ret)
> >> + return ret;
> >> +
> >> + priv->rdev = devm_rtc_allocate_device(dev);
> >> + if (IS_ERR(priv->rdev))
> >> + return PTR_ERR(priv->rdev);
> >> +
> >> + ret = rtc_clk_src_setup(priv);
> >> + if (ret)
> >> + return ret;
> >> +
> >> + priv->irq = platform_get_irq(pdev, 0);
> >> + if (priv->irq < 0) {
> >> + ret = priv->irq;
> >> + goto disable_rtc;
> >> + }
> >> +
> >> + rtc_hz = clk_get_rate(priv->clk_src);
> >> + if (!rtc_hz) {
> >> + dev_err(dev, "Failed to get RTC frequency\n");
> >> + ret = -EINVAL;
> >> + goto disable_rtc;
> >> + }
> >> +
> >> + priv->rtc_hz = DIV_ROUND_UP(rtc_hz, priv->rtc_data->clk_div);
> >> +
> >> + platform_set_drvdata(pdev, priv);
> >> + priv->rdev->ops = &rtc_ops;
> >> +
> >> + ret = devm_request_irq(dev, priv->irq,
> >> + s32g_rtc_handler, 0, dev_name(dev), pdev);
> >> + if (ret) {
> >> + dev_err(dev, "Request interrupt %d failed, error: %d\n",
> >> + priv->irq, ret);
> >> + goto disable_rtc;
> >> + }
> >> +
> >> + ret = devm_rtc_register_device(priv->rdev);
> >> + if (ret)
> >> + goto disable_rtc;
> >> +
> >> + return 0;
> >> +
> >> +disable_rtc:
> >> + s32g_rtc_disable(priv);
> >> + return ret;
> >> +}
> >> +
> >> +static int s32g_rtc_suspend(struct device *dev)
> >> +{
> >> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >> + u32 apival = readl(priv->rtc_base + APIVAL_OFFSET);
> >> +
> >> + if (check_add_overflow(priv->sleep_sec, div64_u64(apival, priv->rtc_hz),
> >> + &priv->sleep_sec)) {
> >> + dev_warn(dev, "Overflow on sleep cycles occurred. Resetting to 0.\n");
> >> + priv->sleep_sec = 0;
> >> + }
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static int s32g_rtc_resume(struct device *dev)
> >> +{
> >> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >> +
> >> + /* The transition from resume to run is a reset event.
> >> + * This leads to the RTC registers being reset after resume from
> >> + * suspend. It is uncommon, but this behaviour has been observed
> >> + * on S32G RTC after issuing a Suspend to RAM operation.
> >> + * Thus, reconfigure RTC registers on the resume path.
> >> + */
> >> + return rtc_clk_src_setup(priv);
> >> +}
> >> +
> >> +static const struct of_device_id rtc_dt_ids[] = {
> >> + { .compatible = "nxp,s32g2-rtc", .data = &rtc_s32g2_data },
> >> + { /* sentinel */ },
> >> +};
> >> +
> >> +static DEFINE_SIMPLE_DEV_PM_OPS(s32g_rtc_pm_ops,
> >> + s32g_rtc_suspend, s32g_rtc_resume);
> >> +
> >> +static struct platform_driver s32g_rtc_driver = {
> >> + .driver = {
> >> + .name = "s32g-rtc",
> >> + .pm = pm_sleep_ptr(&s32g_rtc_pm_ops),
> >> + .of_match_table = rtc_dt_ids,
> >> + },
> >> + .probe = s32g_rtc_probe,
> >> +};
> >> +module_platform_driver(s32g_rtc_driver);
> >> +
> >> +MODULE_AUTHOR("NXP");
> >> +MODULE_DESCRIPTION("NXP RTC driver for S32G2/S32G3");
> >> +MODULE_LICENSE("GPL");
> >> --
> >> 2.45.2
> >>
> >
>
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support
2025-04-09 14:42 ` Enric Balletbo i Serra
@ 2025-04-09 14:58 ` Ciprian Marian Costea
2025-04-10 7:08 ` Enric Balletbo i Serra
2025-04-11 8:09 ` Ciprian Marian Costea
0 siblings, 2 replies; 19+ messages in thread
From: Ciprian Marian Costea @ 2025-04-09 14:58 UTC (permalink / raw)
To: Enric Balletbo i Serra
Cc: Alexandre Belloni, Rob Herring, Krzysztof Kozlowski, Conor Dooley,
Catalin Marinas, Will Deacon, linux-rtc, devicetree, linux-kernel,
linux-arm-kernel, NXP S32 Linux, imx, Christophe Lizzi,
Alberto Ruiz, Eric Chanudet, Bogdan Hamciuc, Ghennadi Procopciuc,
Frank Li
On 4/9/2025 5:42 PM, Enric Balletbo i Serra wrote:
> Hi Ciprian,
>
>
> On Wed, Apr 9, 2025 at 4:26 PM Ciprian Marian Costea
> <ciprianmarian.costea@oss.nxp.com> wrote:
>>
>> On 4/9/2025 5:14 PM, Enric Balletbo i Serra wrote:
>>> Hi Ciprian,
>>>
>>> Many thanks for your patch.
>>>
>>> On Thu, Apr 3, 2025 at 12:34 PM Ciprian Costea
>>> <ciprianmarian.costea@oss.nxp.com> wrote:
>>>>
>>>> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>>>>
>>>> Add a RTC driver for NXP S32G2/S32G3 SoCs.
>>>>
>>>> RTC tracks clock time during system suspend. It can be a wakeup source
>>>> for the S32G2/S32G3 SoC based boards.
>>>>
>>>> The RTC module from S32G2/S32G3 is not battery-powered and it is not kept
>>>> alive during system reset.
>>>>
>>>> Co-developed-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
>>>> Signed-off-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
>>>> Co-developed-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
>>>> Signed-off-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
>>>> Signed-off-by: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>>>> Reviewed-by: Frank Li <Frank.Li@nxp.com>
>>>
>>> I went through the code and it looks good to me, so:
>>>
>>> However, I did a basic rtctest selftest in my S32G-VNP-RDB3 board,
>>> and some tests failed (it shows a small deviation), any idea what
>>> could be wrong?
>>>
>>> This is the output of the test:
>>>
>>> # ./rtctest
>>> TAP version 13
>>> 1..8
>>> # Starting 8 tests from 1 test cases.
>>> # RUN rtc.date_read ...
>>> # rtctest.c:59:date_read:Current RTC date/time is 28/01/2025 00:17:49.
>>> # OK rtc.date_read
>>> ok 1 rtc.date_read
>>> # RUN rtc.date_read_loop ...
>>> # rtctest.c:124:date_read_loop:Continuously reading RTC time for 30s
>>> (with 11ms breaks after every read).
>>> # rtctest.c:151:date_read_loop:Performed 2638 RTC time reads.
>>> # OK rtc.date_read_loop
>>> ok 2 rtc.date_read_loop
>>> # RUN rtc.uie_read ...
>>> # OK rtc.uie_read
>>> ok 3 rtc.uie_read
>>> # RUN rtc.uie_select ...
>>> # OK rtc.uie_select
>>> ok 4 rtc.uie_select
>>> # RUN rtc.alarm_alm_set ...
>>> # rtctest.c:262:alarm_alm_set:Alarm time now set to 00:18:31.
>>> # rtctest.c:282:alarm_alm_set:data: 1a0
>>> # OK rtc.alarm_alm_set
>>> ok 5 rtc.alarm_alm_set
>>> # RUN rtc.alarm_wkalm_set ...
>>> # rtctest.c:334:alarm_wkalm_set:Alarm time now set to 28/01/2025 00:18:34.
>>> # rtctest.c:353:alarm_wkalm_set:Expected new (1738023515) == secs (1738023514)
>>> # alarm_wkalm_set: Test terminated by assertion
>>> # FAIL rtc.alarm_wkalm_set
>>> not ok 6 rtc.alarm_wkalm_set
>>> # RUN rtc.alarm_alm_set_minute ...
>>> # rtctest.c:394:alarm_alm_set_minute:Alarm time now set to 00:19:00.
>>> # rtctest.c:414:alarm_alm_set_minute:data: 1a0
>>> # rtctest.c:420:alarm_alm_set_minute:Expected new (1738023541) == secs
>>> (1738023540)
>>> # alarm_alm_set_minute: Test terminated by assertion
>>> # FAIL rtc.alarm_alm_set_minute
>>> not ok 7 rtc.alarm_alm_set_minute
>>> # RUN rtc.alarm_wkalm_set_minute ...
>>> # rtctest.c:464:alarm_wkalm_set_minute:Alarm time now set to
>>> 28/01/2025 00:20:00.
>>> # rtctest.c:483:alarm_wkalm_set_minute:Expected new (1738023601) ==
>>> secs (1738023600)
>>> # alarm_wkalm_set_minute: Test terminated by assertion
>>> # FAIL rtc.alarm_wkalm_set_minute
>>> not ok 8 rtc.alarm_wkalm_set_minute
>>> # FAILED: 5 / 8 tests passed.
>>> # Totals: pass:5 fail:3 xfail:0 xpass:0 skip:0 error:0
>>>
>>> Best regards,
>>> Enric
>>>
>>
>> Hello Enric,
>>
>> In order to obtain a better resolution/precision you could try to only
>> use DIV512 hardware divisor instead of both DIV32 and DIV512.
>>
>
> That makes the tests pass, so I'm wondering if by default you should
> set DIV512 in platform data instead. With that change you can add my
>
> Tested-by: Enric Balletbo i Serra <eballetbo@kernel.org>
>
This change would decrease the RTC rollover time. Initially indeed only
DIV512 was enabled, but during review [1] I've agreed enabling both
hardware divisors to increase the RTC rollover time. Personally, I am ok
with the current 1 second resolution capability.
Alexandre Belloni, are you ok with the current version of this driver ?
Or should I send a V10 patch in which I enable back only the DIV512
divisor in order to increase the RTC precision at the downside of losing
rollover time.
[1]
https://lore.kernel.org/all/9f4fc9d8-ec7f-43b4-8bdd-01a4ba4855c5@oss.nxp.com/
Regards,
Ciprian
> # ./rtctest
> TAP version 13
> 1..8
> # Starting 8 tests from 1 test cases.
> # RUN rtc.date_read ...
> # rtctest.c:59:date_read:Current RTC date/time is 28/01/2025 00:00:34.
> # OK rtc.date_read
> ok 1 rtc.date_read
> # RUN rtc.date_read_loop ...
> # rtctest.c:124:date_read_loop:Continuously reading RTC time for 30s
> (with 11ms breaks after every read).
> # rtctest.c:151:date_read_loop:Performed 2656 RTC time reads.
> # OK rtc.date_read_loop
> ok 2 rtc.date_read_loop
> # RUN rtc.uie_read ...
> # OK rtc.uie_read
> ok 3 rtc.uie_read
> # RUN rtc.uie_select ...
> # OK rtc.uie_select
> ok 4 rtc.uie_select
> # RUN rtc.alarm_alm_set ...
> # rtctest.c:262:alarm_alm_set:Alarm time now set to 00:01:14.
> # rtctest.c:282:alarm_alm_set:data: 1a0
> # OK rtc.alarm_alm_set
> ok 5 rtc.alarm_alm_set
> # RUN rtc.alarm_wkalm_set ...
> # rtctest.c:334:alarm_wkalm_set:Alarm time now set to 28/01/2025 00:01:17.
> # OK rtc.alarm_wkalm_set
> ok 6 rtc.alarm_wkalm_set
> # RUN rtc.alarm_alm_set_minute ...
> # rtctest.c:394:alarm_alm_set_minute:Alarm time now set to 00:02:00.
> # rtctest.c:414:alarm_alm_set_minute:data: 1a0
> # OK rtc.alarm_alm_set_minute
> ok 7 rtc.alarm_alm_set_minute
> # RUN rtc.alarm_wkalm_set_minute ...
> # rtctest.c:464:alarm_wkalm_set_minute:Alarm time now set to
> 28/01/2025 00:03:00.
> # OK rtc.alarm_wkalm_set_minute
> ok 8 rtc.alarm_wkalm_set_minute
> # PASSED: 8 / 8 tests passed.
> # Totals: pass:8 fail:0 xfail:0 xpass:0 skip:0 error:0
>
>
> Thanks,
> Enric
>
>> Regards,
>> Ciprian
>>
>>>> ---
>>>> drivers/rtc/Kconfig | 11 ++
>>>> drivers/rtc/Makefile | 1 +
>>>> drivers/rtc/rtc-s32g.c | 385 +++++++++++++++++++++++++++++++++++++++++
>>>> 3 files changed, 397 insertions(+)
>>>> create mode 100644 drivers/rtc/rtc-s32g.c
>>>>
>>>> diff --git a/drivers/rtc/Kconfig b/drivers/rtc/Kconfig
>>>> index 0bbbf778ecfa..510dc2db745d 100644
>>>> --- a/drivers/rtc/Kconfig
>>>> +++ b/drivers/rtc/Kconfig
>>>> @@ -2103,4 +2103,15 @@ config RTC_DRV_AMLOGIC_A4
>>>> This driver can also be built as a module. If so, the module
>>>> will be called "rtc-amlogic-a4".
>>>>
>>>> +config RTC_DRV_S32G
>>>> + tristate "RTC driver for S32G2/S32G3 SoCs"
>>>> + depends on ARCH_S32 || COMPILE_TEST
>>>> + depends on COMMON_CLK
>>>> + help
>>>> + Say yes to enable RTC driver for platforms based on the
>>>> + S32G2/S32G3 SoC family.
>>>> +
>>>> + This RTC module can be used as a wakeup source.
>>>> + Please note that it is not battery-powered.
>>>> +
>>>> endif # RTC_CLASS
>>>> diff --git a/drivers/rtc/Makefile b/drivers/rtc/Makefile
>>>> index 489b4ab07068..e4b616ecd5ce 100644
>>>> --- a/drivers/rtc/Makefile
>>>> +++ b/drivers/rtc/Makefile
>>>> @@ -161,6 +161,7 @@ obj-$(CONFIG_RTC_DRV_RX8111) += rtc-rx8111.o
>>>> obj-$(CONFIG_RTC_DRV_RX8581) += rtc-rx8581.o
>>>> obj-$(CONFIG_RTC_DRV_RZN1) += rtc-rzn1.o
>>>> obj-$(CONFIG_RTC_DRV_RENESAS_RTCA3) += rtc-renesas-rtca3.o
>>>> +obj-$(CONFIG_RTC_DRV_S32G) += rtc-s32g.o
>>>> obj-$(CONFIG_RTC_DRV_S35390A) += rtc-s35390a.o
>>>> obj-$(CONFIG_RTC_DRV_S3C) += rtc-s3c.o
>>>> obj-$(CONFIG_RTC_DRV_S5M) += rtc-s5m.o
>>>> diff --git a/drivers/rtc/rtc-s32g.c b/drivers/rtc/rtc-s32g.c
>>>> new file mode 100644
>>>> index 000000000000..3a0818e972eb
>>>> --- /dev/null
>>>> +++ b/drivers/rtc/rtc-s32g.c
>>>> @@ -0,0 +1,385 @@
>>>> +// SPDX-License-Identifier: GPL-2.0-or-later
>>>> +/*
>>>> + * Copyright 2025 NXP
>>>> + */
>>>> +
>>>> +#include <linux/bitfield.h>
>>>> +#include <linux/clk.h>
>>>> +#include <linux/iopoll.h>
>>>> +#include <linux/of_irq.h>
>>>> +#include <linux/platform_device.h>
>>>> +#include <linux/rtc.h>
>>>> +
>>>> +#define RTCC_OFFSET 0x4ul
>>>> +#define RTCS_OFFSET 0x8ul
>>>> +#define APIVAL_OFFSET 0x10ul
>>>> +
>>>> +/* RTCC fields */
>>>> +#define RTCC_CNTEN BIT(31)
>>>> +#define RTCC_APIEN BIT(15)
>>>> +#define RTCC_APIIE BIT(14)
>>>> +#define RTCC_CLKSEL_MASK GENMASK(13, 12)
>>>> +#define RTCC_DIV512EN BIT(11)
>>>> +#define RTCC_DIV32EN BIT(10)
>>>> +
>>>> +/* RTCS fields */
>>>> +#define RTCS_INV_API BIT(17)
>>>> +#define RTCS_APIF BIT(13)
>>>> +
>>>> +#define APIVAL_MAX_VAL GENMASK(31, 0)
>>>> +#define RTC_SYNCH_TIMEOUT (100 * USEC_PER_MSEC)
>>>> +
>>>> +/*
>>>> + * S32G2 and S32G3 SoCs have RTC clock source1 reserved and
>>>> + * should not be used.
>>>> + */
>>>> +#define RTC_CLK_SRC1_RESERVED BIT(1)
>>>> +
>>>> +/*
>>>> + * S32G RTC module has a 512 value and a 32 value hardware frequency
>>>> + * divisors (DIV512 and DIV32) which could be used to achieve higher
>>>> + * counter ranges by lowering the RTC frequency.
>>>> + */
>>>> +enum {
>>>> + DIV1 = 1,
>>>> + DIV32 = 32,
>>>> + DIV512 = 512,
>>>> + DIV512_32 = 16384
>>>> +};
>>>> +
>>>> +static const char *const rtc_clk_src[] = {
>>>> + "source0",
>>>> + "source1",
>>>> + "source2",
>>>> + "source3"
>>>> +};
>>>> +
>>>> +struct rtc_priv {
>>>> + struct rtc_device *rdev;
>>>> + void __iomem *rtc_base;
>>>> + struct clk *ipg;
>>>> + struct clk *clk_src;
>>>> + const struct rtc_soc_data *rtc_data;
>>>> + u64 rtc_hz;
>>>> + time64_t sleep_sec;
>>>> + int irq;
>>>> + u32 clk_src_idx;
>>>> +};
>>>> +
>>>> +struct rtc_soc_data {
>>>> + u32 clk_div;
>>>> + u32 reserved_clk_mask;
>>>> +};
>>>> +
>>>> +static const struct rtc_soc_data rtc_s32g2_data = {
>>>> + .clk_div = DIV512_32,
>>>> + .reserved_clk_mask = RTC_CLK_SRC1_RESERVED,
>>>> +};
>>>> +
>>>> +static irqreturn_t s32g_rtc_handler(int irq, void *dev)
>>>> +{
>>>> + struct rtc_priv *priv = platform_get_drvdata(dev);
>>>> + u32 status;
>>>> +
>>>> + status = readl(priv->rtc_base + RTCS_OFFSET);
>>>> +
>>>> + if (status & RTCS_APIF) {
>>>> + writel(0x0, priv->rtc_base + APIVAL_OFFSET);
>>>> + writel(status | RTCS_APIF, priv->rtc_base + RTCS_OFFSET);
>>>> + }
>>>> +
>>>> + rtc_update_irq(priv->rdev, 1, RTC_IRQF | RTC_AF);
>>>> +
>>>> + return IRQ_HANDLED;
>>>> +}
>>>> +
>>>> +/*
>>>> + * The function is not really getting time from the RTC since the S32G RTC
>>>> + * has several limitations. Thus, to setup alarm use system time.
>>>> + */
>>>> +static int s32g_rtc_read_time(struct device *dev,
>>>> + struct rtc_time *tm)
>>>> +{
>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>> + time64_t sec;
>>>> +
>>>> + if (check_add_overflow(ktime_get_real_seconds(),
>>>> + priv->sleep_sec, &sec))
>>>> + return -ERANGE;
>>>> +
>>>> + rtc_time64_to_tm(sec, tm);
>>>> +
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static int s32g_rtc_read_alarm(struct device *dev, struct rtc_wkalrm *alrm)
>>>> +{
>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>> + u32 rtcc, rtcs;
>>>> +
>>>> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>>>> + rtcs = readl(priv->rtc_base + RTCS_OFFSET);
>>>> +
>>>> + alrm->enabled = rtcc & RTCC_APIIE;
>>>> + if (alrm->enabled)
>>>> + alrm->pending = !(rtcs & RTCS_APIF);
>>>> +
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static int s32g_rtc_alarm_irq_enable(struct device *dev, unsigned int enabled)
>>>> +{
>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>> + u32 rtcc;
>>>> +
>>>> + /* RTC API functionality is used both for triggering interrupts
>>>> + * and as a wakeup event. Hence it should always be enabled.
>>>> + */
>>>> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>>>> + rtcc |= RTCC_APIEN | RTCC_APIIE;
>>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>>>> +
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static int s32g_rtc_set_alarm(struct device *dev, struct rtc_wkalrm *alrm)
>>>> +{
>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>> + unsigned long long cycles;
>>>> + long long t_offset;
>>>> + time64_t alrm_time;
>>>> + u32 rtcs;
>>>> + int ret;
>>>> +
>>>> + alrm_time = rtc_tm_to_time64(&alrm->time);
>>>> + t_offset = alrm_time - ktime_get_real_seconds() - priv->sleep_sec;
>>>> + if (t_offset < 0)
>>>> + return -ERANGE;
>>>> +
>>>> + cycles = t_offset * priv->rtc_hz;
>>>> + if (cycles > APIVAL_MAX_VAL)
>>>> + return -ERANGE;
>>>> +
>>>> + /* APIVAL could have been reset from the IRQ handler.
>>>> + * Hence, we wait in case there is a synchronization process.
>>>> + */
>>>> + ret = read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
>>>> + 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
>>>> + if (ret)
>>>> + return ret;
>>>> +
>>>> + writel(cycles, priv->rtc_base + APIVAL_OFFSET);
>>>> +
>>>> + return read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
>>>> + 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
>>>> +}
>>>> +
>>>> +/*
>>>> + * Disable the 32-bit free running counter.
>>>> + * This allows Clock Source and Divisors selection
>>>> + * to be performed without causing synchronization issues.
>>>> + */
>>>> +static void s32g_rtc_disable(struct rtc_priv *priv)
>>>> +{
>>>> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>>>> +
>>>> + rtcc &= ~RTCC_CNTEN;
>>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>>>> +}
>>>> +
>>>> +static void s32g_rtc_enable(struct rtc_priv *priv)
>>>> +{
>>>> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>>>> +
>>>> + rtcc |= RTCC_CNTEN;
>>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>>>> +}
>>>> +
>>>> +static int rtc_clk_src_setup(struct rtc_priv *priv)
>>>> +{
>>>> + u32 rtcc;
>>>> +
>>>> + rtcc = FIELD_PREP(RTCC_CLKSEL_MASK, priv->clk_src_idx);
>>>> +
>>>> + switch (priv->rtc_data->clk_div) {
>>>> + case DIV512_32:
>>>> + rtcc |= RTCC_DIV512EN;
>>>> + rtcc |= RTCC_DIV32EN;
>>>> + break;
>>>> + case DIV512:
>>>> + rtcc |= RTCC_DIV512EN;
>>>> + break;
>>>> + case DIV32:
>>>> + rtcc |= RTCC_DIV32EN;
>>>> + break;
>>>> + case DIV1:
>>>> + break;
>>>> + default:
>>>> + return -EINVAL;
>>>> + }
>>>> +
>>>> + rtcc |= RTCC_APIEN | RTCC_APIIE;
>>>> + /*
>>>> + * Make sure the CNTEN is 0 before we configure
>>>> + * the clock source and dividers.
>>>> + */
>>>> + s32g_rtc_disable(priv);
>>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>>>> + s32g_rtc_enable(priv);
>>>> +
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static const struct rtc_class_ops rtc_ops = {
>>>> + .read_time = s32g_rtc_read_time,
>>>> + .read_alarm = s32g_rtc_read_alarm,
>>>> + .set_alarm = s32g_rtc_set_alarm,
>>>> + .alarm_irq_enable = s32g_rtc_alarm_irq_enable,
>>>> +};
>>>> +
>>>> +static int rtc_clk_dts_setup(struct rtc_priv *priv,
>>>> + struct device *dev)
>>>> +{
>>>> + u32 i;
>>>> +
>>>> + priv->ipg = devm_clk_get_enabled(dev, "ipg");
>>>> + if (IS_ERR(priv->ipg))
>>>> + return dev_err_probe(dev, PTR_ERR(priv->ipg),
>>>> + "Failed to get 'ipg' clock\n");
>>>> +
>>>> + for (i = 0; i < ARRAY_SIZE(rtc_clk_src); i++) {
>>>> + if (priv->rtc_data->reserved_clk_mask & BIT(i))
>>>> + return -EOPNOTSUPP;
>>>> +
>>>> + priv->clk_src = devm_clk_get_enabled(dev, rtc_clk_src[i]);
>>>> + if (!IS_ERR(priv->clk_src)) {
>>>> + priv->clk_src_idx = i;
>>>> + break;
>>>> + }
>>>> + }
>>>> +
>>>> + if (IS_ERR(priv->clk_src))
>>>> + return dev_err_probe(dev, PTR_ERR(priv->clk_src),
>>>> + "Failed to get rtc module clock source\n");
>>>> +
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static int s32g_rtc_probe(struct platform_device *pdev)
>>>> +{
>>>> + struct device *dev = &pdev->dev;
>>>> + struct rtc_priv *priv;
>>>> + unsigned long rtc_hz;
>>>> + int ret;
>>>> +
>>>> + priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
>>>> + if (!priv)
>>>> + return -ENOMEM;
>>>> +
>>>> + priv->rtc_data = of_device_get_match_data(dev);
>>>> + if (!priv->rtc_data)
>>>> + return -ENODEV;
>>>> +
>>>> + priv->rtc_base = devm_platform_ioremap_resource(pdev, 0);
>>>> + if (IS_ERR(priv->rtc_base))
>>>> + return PTR_ERR(priv->rtc_base);
>>>> +
>>>> + device_init_wakeup(dev, true);
>>>> +
>>>> + ret = rtc_clk_dts_setup(priv, dev);
>>>> + if (ret)
>>>> + return ret;
>>>> +
>>>> + priv->rdev = devm_rtc_allocate_device(dev);
>>>> + if (IS_ERR(priv->rdev))
>>>> + return PTR_ERR(priv->rdev);
>>>> +
>>>> + ret = rtc_clk_src_setup(priv);
>>>> + if (ret)
>>>> + return ret;
>>>> +
>>>> + priv->irq = platform_get_irq(pdev, 0);
>>>> + if (priv->irq < 0) {
>>>> + ret = priv->irq;
>>>> + goto disable_rtc;
>>>> + }
>>>> +
>>>> + rtc_hz = clk_get_rate(priv->clk_src);
>>>> + if (!rtc_hz) {
>>>> + dev_err(dev, "Failed to get RTC frequency\n");
>>>> + ret = -EINVAL;
>>>> + goto disable_rtc;
>>>> + }
>>>> +
>>>> + priv->rtc_hz = DIV_ROUND_UP(rtc_hz, priv->rtc_data->clk_div);
>>>> +
>>>> + platform_set_drvdata(pdev, priv);
>>>> + priv->rdev->ops = &rtc_ops;
>>>> +
>>>> + ret = devm_request_irq(dev, priv->irq,
>>>> + s32g_rtc_handler, 0, dev_name(dev), pdev);
>>>> + if (ret) {
>>>> + dev_err(dev, "Request interrupt %d failed, error: %d\n",
>>>> + priv->irq, ret);
>>>> + goto disable_rtc;
>>>> + }
>>>> +
>>>> + ret = devm_rtc_register_device(priv->rdev);
>>>> + if (ret)
>>>> + goto disable_rtc;
>>>> +
>>>> + return 0;
>>>> +
>>>> +disable_rtc:
>>>> + s32g_rtc_disable(priv);
>>>> + return ret;
>>>> +}
>>>> +
>>>> +static int s32g_rtc_suspend(struct device *dev)
>>>> +{
>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>> + u32 apival = readl(priv->rtc_base + APIVAL_OFFSET);
>>>> +
>>>> + if (check_add_overflow(priv->sleep_sec, div64_u64(apival, priv->rtc_hz),
>>>> + &priv->sleep_sec)) {
>>>> + dev_warn(dev, "Overflow on sleep cycles occurred. Resetting to 0.\n");
>>>> + priv->sleep_sec = 0;
>>>> + }
>>>> +
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static int s32g_rtc_resume(struct device *dev)
>>>> +{
>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>> +
>>>> + /* The transition from resume to run is a reset event.
>>>> + * This leads to the RTC registers being reset after resume from
>>>> + * suspend. It is uncommon, but this behaviour has been observed
>>>> + * on S32G RTC after issuing a Suspend to RAM operation.
>>>> + * Thus, reconfigure RTC registers on the resume path.
>>>> + */
>>>> + return rtc_clk_src_setup(priv);
>>>> +}
>>>> +
>>>> +static const struct of_device_id rtc_dt_ids[] = {
>>>> + { .compatible = "nxp,s32g2-rtc", .data = &rtc_s32g2_data },
>>>> + { /* sentinel */ },
>>>> +};
>>>> +
>>>> +static DEFINE_SIMPLE_DEV_PM_OPS(s32g_rtc_pm_ops,
>>>> + s32g_rtc_suspend, s32g_rtc_resume);
>>>> +
>>>> +static struct platform_driver s32g_rtc_driver = {
>>>> + .driver = {
>>>> + .name = "s32g-rtc",
>>>> + .pm = pm_sleep_ptr(&s32g_rtc_pm_ops),
>>>> + .of_match_table = rtc_dt_ids,
>>>> + },
>>>> + .probe = s32g_rtc_probe,
>>>> +};
>>>> +module_platform_driver(s32g_rtc_driver);
>>>> +
>>>> +MODULE_AUTHOR("NXP");
>>>> +MODULE_DESCRIPTION("NXP RTC driver for S32G2/S32G3");
>>>> +MODULE_LICENSE("GPL");
>>>> --
>>>> 2.45.2
>>>>
>>>
>>
>
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support
2025-04-09 14:58 ` Ciprian Marian Costea
@ 2025-04-10 7:08 ` Enric Balletbo i Serra
2025-04-11 8:09 ` Ciprian Marian Costea
1 sibling, 0 replies; 19+ messages in thread
From: Enric Balletbo i Serra @ 2025-04-10 7:08 UTC (permalink / raw)
To: Ciprian Marian Costea
Cc: Alexandre Belloni, Rob Herring, Krzysztof Kozlowski, Conor Dooley,
Catalin Marinas, Will Deacon, linux-rtc, devicetree, linux-kernel,
linux-arm-kernel, NXP S32 Linux, imx, Christophe Lizzi,
Alberto Ruiz, Eric Chanudet, Bogdan Hamciuc, Ghennadi Procopciuc,
Frank Li
Hi Ciprian,
On Wed, Apr 9, 2025 at 4:58 PM Ciprian Marian Costea
<ciprianmarian.costea@oss.nxp.com> wrote:
>
> On 4/9/2025 5:42 PM, Enric Balletbo i Serra wrote:
> > Hi Ciprian,
> >
> >
> > On Wed, Apr 9, 2025 at 4:26 PM Ciprian Marian Costea
> > <ciprianmarian.costea@oss.nxp.com> wrote:
> >>
> >> On 4/9/2025 5:14 PM, Enric Balletbo i Serra wrote:
> >>> Hi Ciprian,
> >>>
> >>> Many thanks for your patch.
> >>>
> >>> On Thu, Apr 3, 2025 at 12:34 PM Ciprian Costea
> >>> <ciprianmarian.costea@oss.nxp.com> wrote:
> >>>>
> >>>> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
> >>>>
> >>>> Add a RTC driver for NXP S32G2/S32G3 SoCs.
> >>>>
> >>>> RTC tracks clock time during system suspend. It can be a wakeup source
> >>>> for the S32G2/S32G3 SoC based boards.
> >>>>
> >>>> The RTC module from S32G2/S32G3 is not battery-powered and it is not kept
> >>>> alive during system reset.
> >>>>
> >>>> Co-developed-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
> >>>> Signed-off-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
> >>>> Co-developed-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
> >>>> Signed-off-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
> >>>> Signed-off-by: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
> >>>> Reviewed-by: Frank Li <Frank.Li@nxp.com>
> >>>
> >>> I went through the code and it looks good to me, so:
> >>>
> >>> However, I did a basic rtctest selftest in my S32G-VNP-RDB3 board,
> >>> and some tests failed (it shows a small deviation), any idea what
> >>> could be wrong?
> >>>
> >>> This is the output of the test:
> >>>
> >>> # ./rtctest
> >>> TAP version 13
> >>> 1..8
> >>> # Starting 8 tests from 1 test cases.
> >>> # RUN rtc.date_read ...
> >>> # rtctest.c:59:date_read:Current RTC date/time is 28/01/2025 00:17:49.
> >>> # OK rtc.date_read
> >>> ok 1 rtc.date_read
> >>> # RUN rtc.date_read_loop ...
> >>> # rtctest.c:124:date_read_loop:Continuously reading RTC time for 30s
> >>> (with 11ms breaks after every read).
> >>> # rtctest.c:151:date_read_loop:Performed 2638 RTC time reads.
> >>> # OK rtc.date_read_loop
> >>> ok 2 rtc.date_read_loop
> >>> # RUN rtc.uie_read ...
> >>> # OK rtc.uie_read
> >>> ok 3 rtc.uie_read
> >>> # RUN rtc.uie_select ...
> >>> # OK rtc.uie_select
> >>> ok 4 rtc.uie_select
> >>> # RUN rtc.alarm_alm_set ...
> >>> # rtctest.c:262:alarm_alm_set:Alarm time now set to 00:18:31.
> >>> # rtctest.c:282:alarm_alm_set:data: 1a0
> >>> # OK rtc.alarm_alm_set
> >>> ok 5 rtc.alarm_alm_set
> >>> # RUN rtc.alarm_wkalm_set ...
> >>> # rtctest.c:334:alarm_wkalm_set:Alarm time now set to 28/01/2025 00:18:34.
> >>> # rtctest.c:353:alarm_wkalm_set:Expected new (1738023515) == secs (1738023514)
> >>> # alarm_wkalm_set: Test terminated by assertion
> >>> # FAIL rtc.alarm_wkalm_set
> >>> not ok 6 rtc.alarm_wkalm_set
> >>> # RUN rtc.alarm_alm_set_minute ...
> >>> # rtctest.c:394:alarm_alm_set_minute:Alarm time now set to 00:19:00.
> >>> # rtctest.c:414:alarm_alm_set_minute:data: 1a0
> >>> # rtctest.c:420:alarm_alm_set_minute:Expected new (1738023541) == secs
> >>> (1738023540)
> >>> # alarm_alm_set_minute: Test terminated by assertion
> >>> # FAIL rtc.alarm_alm_set_minute
> >>> not ok 7 rtc.alarm_alm_set_minute
> >>> # RUN rtc.alarm_wkalm_set_minute ...
> >>> # rtctest.c:464:alarm_wkalm_set_minute:Alarm time now set to
> >>> 28/01/2025 00:20:00.
> >>> # rtctest.c:483:alarm_wkalm_set_minute:Expected new (1738023601) ==
> >>> secs (1738023600)
> >>> # alarm_wkalm_set_minute: Test terminated by assertion
> >>> # FAIL rtc.alarm_wkalm_set_minute
> >>> not ok 8 rtc.alarm_wkalm_set_minute
> >>> # FAILED: 5 / 8 tests passed.
> >>> # Totals: pass:5 fail:3 xfail:0 xpass:0 skip:0 error:0
> >>>
> >>> Best regards,
> >>> Enric
> >>>
> >>
> >> Hello Enric,
> >>
> >> In order to obtain a better resolution/precision you could try to only
> >> use DIV512 hardware divisor instead of both DIV32 and DIV512.
> >>
> >
> > That makes the tests pass, so I'm wondering if by default you should
> > set DIV512 in platform data instead. With that change you can add my
> >
> > Tested-by: Enric Balletbo i Serra <eballetbo@kernel.org>
> >
>
> This change would decrease the RTC rollover time. Initially indeed only
> DIV512 was enabled, but during review [1] I've agreed enabling both
> hardware divisors to increase the RTC rollover time. Personally, I am ok
> with the current 1 second resolution capability.
>
I'd say that having rtc selftests failing will add a bit of headache
to those that run integration tests, as the test will need to be
marked as an exception. Another option would be to make the tests a
bit more flexible, but I'm not sure this is the right way. I guess we
can make this user configurable via a module parameter so the user has
the final decision. Just my two cents.
> Alexandre Belloni, are you ok with the current version of this driver ?
> Or should I send a V10 patch in which I enable back only the DIV512
> divisor in order to increase the RTC precision at the downside of losing
> rollover time.
>
Agree, up to the maintainer to decide.
Regards,
Enric
> [1]
> https://lore.kernel.org/all/9f4fc9d8-ec7f-43b4-8bdd-01a4ba4855c5@oss.nxp.com/
>
> Regards,
> Ciprian
>
> > # ./rtctest
> > TAP version 13
> > 1..8
> > # Starting 8 tests from 1 test cases.
> > # RUN rtc.date_read ...
> > # rtctest.c:59:date_read:Current RTC date/time is 28/01/2025 00:00:34.
> > # OK rtc.date_read
> > ok 1 rtc.date_read
> > # RUN rtc.date_read_loop ...
> > # rtctest.c:124:date_read_loop:Continuously reading RTC time for 30s
> > (with 11ms breaks after every read).
> > # rtctest.c:151:date_read_loop:Performed 2656 RTC time reads.
> > # OK rtc.date_read_loop
> > ok 2 rtc.date_read_loop
> > # RUN rtc.uie_read ...
> > # OK rtc.uie_read
> > ok 3 rtc.uie_read
> > # RUN rtc.uie_select ...
> > # OK rtc.uie_select
> > ok 4 rtc.uie_select
> > # RUN rtc.alarm_alm_set ...
> > # rtctest.c:262:alarm_alm_set:Alarm time now set to 00:01:14.
> > # rtctest.c:282:alarm_alm_set:data: 1a0
> > # OK rtc.alarm_alm_set
> > ok 5 rtc.alarm_alm_set
> > # RUN rtc.alarm_wkalm_set ...
> > # rtctest.c:334:alarm_wkalm_set:Alarm time now set to 28/01/2025 00:01:17.
> > # OK rtc.alarm_wkalm_set
> > ok 6 rtc.alarm_wkalm_set
> > # RUN rtc.alarm_alm_set_minute ...
> > # rtctest.c:394:alarm_alm_set_minute:Alarm time now set to 00:02:00.
> > # rtctest.c:414:alarm_alm_set_minute:data: 1a0
> > # OK rtc.alarm_alm_set_minute
> > ok 7 rtc.alarm_alm_set_minute
> > # RUN rtc.alarm_wkalm_set_minute ...
> > # rtctest.c:464:alarm_wkalm_set_minute:Alarm time now set to
> > 28/01/2025 00:03:00.
> > # OK rtc.alarm_wkalm_set_minute
> > ok 8 rtc.alarm_wkalm_set_minute
> > # PASSED: 8 / 8 tests passed.
> > # Totals: pass:8 fail:0 xfail:0 xpass:0 skip:0 error:0
> >
> >
> > Thanks,
> > Enric
> >
> >> Regards,
> >> Ciprian
> >>
> >>>> ---
> >>>> drivers/rtc/Kconfig | 11 ++
> >>>> drivers/rtc/Makefile | 1 +
> >>>> drivers/rtc/rtc-s32g.c | 385 +++++++++++++++++++++++++++++++++++++++++
> >>>> 3 files changed, 397 insertions(+)
> >>>> create mode 100644 drivers/rtc/rtc-s32g.c
> >>>>
> >>>> diff --git a/drivers/rtc/Kconfig b/drivers/rtc/Kconfig
> >>>> index 0bbbf778ecfa..510dc2db745d 100644
> >>>> --- a/drivers/rtc/Kconfig
> >>>> +++ b/drivers/rtc/Kconfig
> >>>> @@ -2103,4 +2103,15 @@ config RTC_DRV_AMLOGIC_A4
> >>>> This driver can also be built as a module. If so, the module
> >>>> will be called "rtc-amlogic-a4".
> >>>>
> >>>> +config RTC_DRV_S32G
> >>>> + tristate "RTC driver for S32G2/S32G3 SoCs"
> >>>> + depends on ARCH_S32 || COMPILE_TEST
> >>>> + depends on COMMON_CLK
> >>>> + help
> >>>> + Say yes to enable RTC driver for platforms based on the
> >>>> + S32G2/S32G3 SoC family.
> >>>> +
> >>>> + This RTC module can be used as a wakeup source.
> >>>> + Please note that it is not battery-powered.
> >>>> +
> >>>> endif # RTC_CLASS
> >>>> diff --git a/drivers/rtc/Makefile b/drivers/rtc/Makefile
> >>>> index 489b4ab07068..e4b616ecd5ce 100644
> >>>> --- a/drivers/rtc/Makefile
> >>>> +++ b/drivers/rtc/Makefile
> >>>> @@ -161,6 +161,7 @@ obj-$(CONFIG_RTC_DRV_RX8111) += rtc-rx8111.o
> >>>> obj-$(CONFIG_RTC_DRV_RX8581) += rtc-rx8581.o
> >>>> obj-$(CONFIG_RTC_DRV_RZN1) += rtc-rzn1.o
> >>>> obj-$(CONFIG_RTC_DRV_RENESAS_RTCA3) += rtc-renesas-rtca3.o
> >>>> +obj-$(CONFIG_RTC_DRV_S32G) += rtc-s32g.o
> >>>> obj-$(CONFIG_RTC_DRV_S35390A) += rtc-s35390a.o
> >>>> obj-$(CONFIG_RTC_DRV_S3C) += rtc-s3c.o
> >>>> obj-$(CONFIG_RTC_DRV_S5M) += rtc-s5m.o
> >>>> diff --git a/drivers/rtc/rtc-s32g.c b/drivers/rtc/rtc-s32g.c
> >>>> new file mode 100644
> >>>> index 000000000000..3a0818e972eb
> >>>> --- /dev/null
> >>>> +++ b/drivers/rtc/rtc-s32g.c
> >>>> @@ -0,0 +1,385 @@
> >>>> +// SPDX-License-Identifier: GPL-2.0-or-later
> >>>> +/*
> >>>> + * Copyright 2025 NXP
> >>>> + */
> >>>> +
> >>>> +#include <linux/bitfield.h>
> >>>> +#include <linux/clk.h>
> >>>> +#include <linux/iopoll.h>
> >>>> +#include <linux/of_irq.h>
> >>>> +#include <linux/platform_device.h>
> >>>> +#include <linux/rtc.h>
> >>>> +
> >>>> +#define RTCC_OFFSET 0x4ul
> >>>> +#define RTCS_OFFSET 0x8ul
> >>>> +#define APIVAL_OFFSET 0x10ul
> >>>> +
> >>>> +/* RTCC fields */
> >>>> +#define RTCC_CNTEN BIT(31)
> >>>> +#define RTCC_APIEN BIT(15)
> >>>> +#define RTCC_APIIE BIT(14)
> >>>> +#define RTCC_CLKSEL_MASK GENMASK(13, 12)
> >>>> +#define RTCC_DIV512EN BIT(11)
> >>>> +#define RTCC_DIV32EN BIT(10)
> >>>> +
> >>>> +/* RTCS fields */
> >>>> +#define RTCS_INV_API BIT(17)
> >>>> +#define RTCS_APIF BIT(13)
> >>>> +
> >>>> +#define APIVAL_MAX_VAL GENMASK(31, 0)
> >>>> +#define RTC_SYNCH_TIMEOUT (100 * USEC_PER_MSEC)
> >>>> +
> >>>> +/*
> >>>> + * S32G2 and S32G3 SoCs have RTC clock source1 reserved and
> >>>> + * should not be used.
> >>>> + */
> >>>> +#define RTC_CLK_SRC1_RESERVED BIT(1)
> >>>> +
> >>>> +/*
> >>>> + * S32G RTC module has a 512 value and a 32 value hardware frequency
> >>>> + * divisors (DIV512 and DIV32) which could be used to achieve higher
> >>>> + * counter ranges by lowering the RTC frequency.
> >>>> + */
> >>>> +enum {
> >>>> + DIV1 = 1,
> >>>> + DIV32 = 32,
> >>>> + DIV512 = 512,
> >>>> + DIV512_32 = 16384
> >>>> +};
> >>>> +
> >>>> +static const char *const rtc_clk_src[] = {
> >>>> + "source0",
> >>>> + "source1",
> >>>> + "source2",
> >>>> + "source3"
> >>>> +};
> >>>> +
> >>>> +struct rtc_priv {
> >>>> + struct rtc_device *rdev;
> >>>> + void __iomem *rtc_base;
> >>>> + struct clk *ipg;
> >>>> + struct clk *clk_src;
> >>>> + const struct rtc_soc_data *rtc_data;
> >>>> + u64 rtc_hz;
> >>>> + time64_t sleep_sec;
> >>>> + int irq;
> >>>> + u32 clk_src_idx;
> >>>> +};
> >>>> +
> >>>> +struct rtc_soc_data {
> >>>> + u32 clk_div;
> >>>> + u32 reserved_clk_mask;
> >>>> +};
> >>>> +
> >>>> +static const struct rtc_soc_data rtc_s32g2_data = {
> >>>> + .clk_div = DIV512_32,
> >>>> + .reserved_clk_mask = RTC_CLK_SRC1_RESERVED,
> >>>> +};
> >>>> +
> >>>> +static irqreturn_t s32g_rtc_handler(int irq, void *dev)
> >>>> +{
> >>>> + struct rtc_priv *priv = platform_get_drvdata(dev);
> >>>> + u32 status;
> >>>> +
> >>>> + status = readl(priv->rtc_base + RTCS_OFFSET);
> >>>> +
> >>>> + if (status & RTCS_APIF) {
> >>>> + writel(0x0, priv->rtc_base + APIVAL_OFFSET);
> >>>> + writel(status | RTCS_APIF, priv->rtc_base + RTCS_OFFSET);
> >>>> + }
> >>>> +
> >>>> + rtc_update_irq(priv->rdev, 1, RTC_IRQF | RTC_AF);
> >>>> +
> >>>> + return IRQ_HANDLED;
> >>>> +}
> >>>> +
> >>>> +/*
> >>>> + * The function is not really getting time from the RTC since the S32G RTC
> >>>> + * has several limitations. Thus, to setup alarm use system time.
> >>>> + */
> >>>> +static int s32g_rtc_read_time(struct device *dev,
> >>>> + struct rtc_time *tm)
> >>>> +{
> >>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >>>> + time64_t sec;
> >>>> +
> >>>> + if (check_add_overflow(ktime_get_real_seconds(),
> >>>> + priv->sleep_sec, &sec))
> >>>> + return -ERANGE;
> >>>> +
> >>>> + rtc_time64_to_tm(sec, tm);
> >>>> +
> >>>> + return 0;
> >>>> +}
> >>>> +
> >>>> +static int s32g_rtc_read_alarm(struct device *dev, struct rtc_wkalrm *alrm)
> >>>> +{
> >>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >>>> + u32 rtcc, rtcs;
> >>>> +
> >>>> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> >>>> + rtcs = readl(priv->rtc_base + RTCS_OFFSET);
> >>>> +
> >>>> + alrm->enabled = rtcc & RTCC_APIIE;
> >>>> + if (alrm->enabled)
> >>>> + alrm->pending = !(rtcs & RTCS_APIF);
> >>>> +
> >>>> + return 0;
> >>>> +}
> >>>> +
> >>>> +static int s32g_rtc_alarm_irq_enable(struct device *dev, unsigned int enabled)
> >>>> +{
> >>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >>>> + u32 rtcc;
> >>>> +
> >>>> + /* RTC API functionality is used both for triggering interrupts
> >>>> + * and as a wakeup event. Hence it should always be enabled.
> >>>> + */
> >>>> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> >>>> + rtcc |= RTCC_APIEN | RTCC_APIIE;
> >>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> >>>> +
> >>>> + return 0;
> >>>> +}
> >>>> +
> >>>> +static int s32g_rtc_set_alarm(struct device *dev, struct rtc_wkalrm *alrm)
> >>>> +{
> >>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >>>> + unsigned long long cycles;
> >>>> + long long t_offset;
> >>>> + time64_t alrm_time;
> >>>> + u32 rtcs;
> >>>> + int ret;
> >>>> +
> >>>> + alrm_time = rtc_tm_to_time64(&alrm->time);
> >>>> + t_offset = alrm_time - ktime_get_real_seconds() - priv->sleep_sec;
> >>>> + if (t_offset < 0)
> >>>> + return -ERANGE;
> >>>> +
> >>>> + cycles = t_offset * priv->rtc_hz;
> >>>> + if (cycles > APIVAL_MAX_VAL)
> >>>> + return -ERANGE;
> >>>> +
> >>>> + /* APIVAL could have been reset from the IRQ handler.
> >>>> + * Hence, we wait in case there is a synchronization process.
> >>>> + */
> >>>> + ret = read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
> >>>> + 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
> >>>> + if (ret)
> >>>> + return ret;
> >>>> +
> >>>> + writel(cycles, priv->rtc_base + APIVAL_OFFSET);
> >>>> +
> >>>> + return read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
> >>>> + 0, RTC_SYNCH_TIMEOUT, false, priv->rtc_base + RTCS_OFFSET);
> >>>> +}
> >>>> +
> >>>> +/*
> >>>> + * Disable the 32-bit free running counter.
> >>>> + * This allows Clock Source and Divisors selection
> >>>> + * to be performed without causing synchronization issues.
> >>>> + */
> >>>> +static void s32g_rtc_disable(struct rtc_priv *priv)
> >>>> +{
> >>>> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> >>>> +
> >>>> + rtcc &= ~RTCC_CNTEN;
> >>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> >>>> +}
> >>>> +
> >>>> +static void s32g_rtc_enable(struct rtc_priv *priv)
> >>>> +{
> >>>> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
> >>>> +
> >>>> + rtcc |= RTCC_CNTEN;
> >>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> >>>> +}
> >>>> +
> >>>> +static int rtc_clk_src_setup(struct rtc_priv *priv)
> >>>> +{
> >>>> + u32 rtcc;
> >>>> +
> >>>> + rtcc = FIELD_PREP(RTCC_CLKSEL_MASK, priv->clk_src_idx);
> >>>> +
> >>>> + switch (priv->rtc_data->clk_div) {
> >>>> + case DIV512_32:
> >>>> + rtcc |= RTCC_DIV512EN;
> >>>> + rtcc |= RTCC_DIV32EN;
> >>>> + break;
> >>>> + case DIV512:
> >>>> + rtcc |= RTCC_DIV512EN;
> >>>> + break;
> >>>> + case DIV32:
> >>>> + rtcc |= RTCC_DIV32EN;
> >>>> + break;
> >>>> + case DIV1:
> >>>> + break;
> >>>> + default:
> >>>> + return -EINVAL;
> >>>> + }
> >>>> +
> >>>> + rtcc |= RTCC_APIEN | RTCC_APIIE;
> >>>> + /*
> >>>> + * Make sure the CNTEN is 0 before we configure
> >>>> + * the clock source and dividers.
> >>>> + */
> >>>> + s32g_rtc_disable(priv);
> >>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
> >>>> + s32g_rtc_enable(priv);
> >>>> +
> >>>> + return 0;
> >>>> +}
> >>>> +
> >>>> +static const struct rtc_class_ops rtc_ops = {
> >>>> + .read_time = s32g_rtc_read_time,
> >>>> + .read_alarm = s32g_rtc_read_alarm,
> >>>> + .set_alarm = s32g_rtc_set_alarm,
> >>>> + .alarm_irq_enable = s32g_rtc_alarm_irq_enable,
> >>>> +};
> >>>> +
> >>>> +static int rtc_clk_dts_setup(struct rtc_priv *priv,
> >>>> + struct device *dev)
> >>>> +{
> >>>> + u32 i;
> >>>> +
> >>>> + priv->ipg = devm_clk_get_enabled(dev, "ipg");
> >>>> + if (IS_ERR(priv->ipg))
> >>>> + return dev_err_probe(dev, PTR_ERR(priv->ipg),
> >>>> + "Failed to get 'ipg' clock\n");
> >>>> +
> >>>> + for (i = 0; i < ARRAY_SIZE(rtc_clk_src); i++) {
> >>>> + if (priv->rtc_data->reserved_clk_mask & BIT(i))
> >>>> + return -EOPNOTSUPP;
> >>>> +
> >>>> + priv->clk_src = devm_clk_get_enabled(dev, rtc_clk_src[i]);
> >>>> + if (!IS_ERR(priv->clk_src)) {
> >>>> + priv->clk_src_idx = i;
> >>>> + break;
> >>>> + }
> >>>> + }
> >>>> +
> >>>> + if (IS_ERR(priv->clk_src))
> >>>> + return dev_err_probe(dev, PTR_ERR(priv->clk_src),
> >>>> + "Failed to get rtc module clock source\n");
> >>>> +
> >>>> + return 0;
> >>>> +}
> >>>> +
> >>>> +static int s32g_rtc_probe(struct platform_device *pdev)
> >>>> +{
> >>>> + struct device *dev = &pdev->dev;
> >>>> + struct rtc_priv *priv;
> >>>> + unsigned long rtc_hz;
> >>>> + int ret;
> >>>> +
> >>>> + priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
> >>>> + if (!priv)
> >>>> + return -ENOMEM;
> >>>> +
> >>>> + priv->rtc_data = of_device_get_match_data(dev);
> >>>> + if (!priv->rtc_data)
> >>>> + return -ENODEV;
> >>>> +
> >>>> + priv->rtc_base = devm_platform_ioremap_resource(pdev, 0);
> >>>> + if (IS_ERR(priv->rtc_base))
> >>>> + return PTR_ERR(priv->rtc_base);
> >>>> +
> >>>> + device_init_wakeup(dev, true);
> >>>> +
> >>>> + ret = rtc_clk_dts_setup(priv, dev);
> >>>> + if (ret)
> >>>> + return ret;
> >>>> +
> >>>> + priv->rdev = devm_rtc_allocate_device(dev);
> >>>> + if (IS_ERR(priv->rdev))
> >>>> + return PTR_ERR(priv->rdev);
> >>>> +
> >>>> + ret = rtc_clk_src_setup(priv);
> >>>> + if (ret)
> >>>> + return ret;
> >>>> +
> >>>> + priv->irq = platform_get_irq(pdev, 0);
> >>>> + if (priv->irq < 0) {
> >>>> + ret = priv->irq;
> >>>> + goto disable_rtc;
> >>>> + }
> >>>> +
> >>>> + rtc_hz = clk_get_rate(priv->clk_src);
> >>>> + if (!rtc_hz) {
> >>>> + dev_err(dev, "Failed to get RTC frequency\n");
> >>>> + ret = -EINVAL;
> >>>> + goto disable_rtc;
> >>>> + }
> >>>> +
> >>>> + priv->rtc_hz = DIV_ROUND_UP(rtc_hz, priv->rtc_data->clk_div);
> >>>> +
> >>>> + platform_set_drvdata(pdev, priv);
> >>>> + priv->rdev->ops = &rtc_ops;
> >>>> +
> >>>> + ret = devm_request_irq(dev, priv->irq,
> >>>> + s32g_rtc_handler, 0, dev_name(dev), pdev);
> >>>> + if (ret) {
> >>>> + dev_err(dev, "Request interrupt %d failed, error: %d\n",
> >>>> + priv->irq, ret);
> >>>> + goto disable_rtc;
> >>>> + }
> >>>> +
> >>>> + ret = devm_rtc_register_device(priv->rdev);
> >>>> + if (ret)
> >>>> + goto disable_rtc;
> >>>> +
> >>>> + return 0;
> >>>> +
> >>>> +disable_rtc:
> >>>> + s32g_rtc_disable(priv);
> >>>> + return ret;
> >>>> +}
> >>>> +
> >>>> +static int s32g_rtc_suspend(struct device *dev)
> >>>> +{
> >>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >>>> + u32 apival = readl(priv->rtc_base + APIVAL_OFFSET);
> >>>> +
> >>>> + if (check_add_overflow(priv->sleep_sec, div64_u64(apival, priv->rtc_hz),
> >>>> + &priv->sleep_sec)) {
> >>>> + dev_warn(dev, "Overflow on sleep cycles occurred. Resetting to 0.\n");
> >>>> + priv->sleep_sec = 0;
> >>>> + }
> >>>> +
> >>>> + return 0;
> >>>> +}
> >>>> +
> >>>> +static int s32g_rtc_resume(struct device *dev)
> >>>> +{
> >>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
> >>>> +
> >>>> + /* The transition from resume to run is a reset event.
> >>>> + * This leads to the RTC registers being reset after resume from
> >>>> + * suspend. It is uncommon, but this behaviour has been observed
> >>>> + * on S32G RTC after issuing a Suspend to RAM operation.
> >>>> + * Thus, reconfigure RTC registers on the resume path.
> >>>> + */
> >>>> + return rtc_clk_src_setup(priv);
> >>>> +}
> >>>> +
> >>>> +static const struct of_device_id rtc_dt_ids[] = {
> >>>> + { .compatible = "nxp,s32g2-rtc", .data = &rtc_s32g2_data },
> >>>> + { /* sentinel */ },
> >>>> +};
> >>>> +
> >>>> +static DEFINE_SIMPLE_DEV_PM_OPS(s32g_rtc_pm_ops,
> >>>> + s32g_rtc_suspend, s32g_rtc_resume);
> >>>> +
> >>>> +static struct platform_driver s32g_rtc_driver = {
> >>>> + .driver = {
> >>>> + .name = "s32g-rtc",
> >>>> + .pm = pm_sleep_ptr(&s32g_rtc_pm_ops),
> >>>> + .of_match_table = rtc_dt_ids,
> >>>> + },
> >>>> + .probe = s32g_rtc_probe,
> >>>> +};
> >>>> +module_platform_driver(s32g_rtc_driver);
> >>>> +
> >>>> +MODULE_AUTHOR("NXP");
> >>>> +MODULE_DESCRIPTION("NXP RTC driver for S32G2/S32G3");
> >>>> +MODULE_LICENSE("GPL");
> >>>> --
> >>>> 2.45.2
> >>>>
> >>>
> >>
> >
>
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support
2025-04-09 14:58 ` Ciprian Marian Costea
2025-04-10 7:08 ` Enric Balletbo i Serra
@ 2025-04-11 8:09 ` Ciprian Marian Costea
1 sibling, 0 replies; 19+ messages in thread
From: Ciprian Marian Costea @ 2025-04-11 8:09 UTC (permalink / raw)
To: Alexandre Belloni
Cc: Rob Herring, Krzysztof Kozlowski, Conor Dooley, Catalin Marinas,
Will Deacon, linux-rtc, devicetree, linux-kernel,
linux-arm-kernel, NXP S32 Linux, imx, Christophe Lizzi,
Alberto Ruiz, Eric Chanudet, Bogdan Hamciuc, Ghennadi Procopciuc,
Frank Li, Enric Balletbo
On 4/9/2025 5:58 PM, Ciprian Marian Costea wrote:
> On 4/9/2025 5:42 PM, Enric Balletbo i Serra wrote:
>> Hi Ciprian,
>>
>>
>> On Wed, Apr 9, 2025 at 4:26 PM Ciprian Marian Costea
>> <ciprianmarian.costea@oss.nxp.com> wrote:
>>>
>>> On 4/9/2025 5:14 PM, Enric Balletbo i Serra wrote:
>>>> Hi Ciprian,
>>>>
>>>> Many thanks for your patch.
>>>>
>>>> On Thu, Apr 3, 2025 at 12:34 PM Ciprian Costea
>>>> <ciprianmarian.costea@oss.nxp.com> wrote:
>>>>>
>>>>> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>>>>>
>>>>> Add a RTC driver for NXP S32G2/S32G3 SoCs.
>>>>>
>>>>> RTC tracks clock time during system suspend. It can be a wakeup source
>>>>> for the S32G2/S32G3 SoC based boards.
>>>>>
>>>>> The RTC module from S32G2/S32G3 is not battery-powered and it is
>>>>> not kept
>>>>> alive during system reset.
>>>>>
>>>>> Co-developed-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
>>>>> Signed-off-by: Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
>>>>> Co-developed-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
>>>>> Signed-off-by: Ghennadi Procopciuc <Ghennadi.Procopciuc@nxp.com>
>>>>> Signed-off-by: Ciprian Marian Costea
>>>>> <ciprianmarian.costea@oss.nxp.com>
>>>>> Reviewed-by: Frank Li <Frank.Li@nxp.com>
>>>>
>>>> I went through the code and it looks good to me, so:
>>>>
>>>> However, I did a basic rtctest selftest in my S32G-VNP-RDB3 board,
>>>> and some tests failed (it shows a small deviation), any idea what
>>>> could be wrong?
>>>>
>>>> This is the output of the test:
>>>>
>>>> # ./rtctest
>>>> TAP version 13
>>>> 1..8
>>>> # Starting 8 tests from 1 test cases.
>>>> # RUN rtc.date_read ...
>>>> # rtctest.c:59:date_read:Current RTC date/time is 28/01/2025 00:17:49.
>>>> # OK rtc.date_read
>>>> ok 1 rtc.date_read
>>>> # RUN rtc.date_read_loop ...
>>>> # rtctest.c:124:date_read_loop:Continuously reading RTC time for 30s
>>>> (with 11ms breaks after every read).
>>>> # rtctest.c:151:date_read_loop:Performed 2638 RTC time reads.
>>>> # OK rtc.date_read_loop
>>>> ok 2 rtc.date_read_loop
>>>> # RUN rtc.uie_read ...
>>>> # OK rtc.uie_read
>>>> ok 3 rtc.uie_read
>>>> # RUN rtc.uie_select ...
>>>> # OK rtc.uie_select
>>>> ok 4 rtc.uie_select
>>>> # RUN rtc.alarm_alm_set ...
>>>> # rtctest.c:262:alarm_alm_set:Alarm time now set to 00:18:31.
>>>> # rtctest.c:282:alarm_alm_set:data: 1a0
>>>> # OK rtc.alarm_alm_set
>>>> ok 5 rtc.alarm_alm_set
>>>> # RUN rtc.alarm_wkalm_set ...
>>>> # rtctest.c:334:alarm_wkalm_set:Alarm time now set to 28/01/2025
>>>> 00:18:34.
>>>> # rtctest.c:353:alarm_wkalm_set:Expected new (1738023515) == secs
>>>> (1738023514)
>>>> # alarm_wkalm_set: Test terminated by assertion
>>>> # FAIL rtc.alarm_wkalm_set
>>>> not ok 6 rtc.alarm_wkalm_set
>>>> # RUN rtc.alarm_alm_set_minute ...
>>>> # rtctest.c:394:alarm_alm_set_minute:Alarm time now set to 00:19:00.
>>>> # rtctest.c:414:alarm_alm_set_minute:data: 1a0
>>>> # rtctest.c:420:alarm_alm_set_minute:Expected new (1738023541) == secs
>>>> (1738023540)
>>>> # alarm_alm_set_minute: Test terminated by assertion
>>>> # FAIL rtc.alarm_alm_set_minute
>>>> not ok 7 rtc.alarm_alm_set_minute
>>>> # RUN rtc.alarm_wkalm_set_minute ...
>>>> # rtctest.c:464:alarm_wkalm_set_minute:Alarm time now set to
>>>> 28/01/2025 00:20:00.
>>>> # rtctest.c:483:alarm_wkalm_set_minute:Expected new (1738023601) ==
>>>> secs (1738023600)
>>>> # alarm_wkalm_set_minute: Test terminated by assertion
>>>> # FAIL rtc.alarm_wkalm_set_minute
>>>> not ok 8 rtc.alarm_wkalm_set_minute
>>>> # FAILED: 5 / 8 tests passed.
>>>> # Totals: pass:5 fail:3 xfail:0 xpass:0 skip:0 error:0
>>>>
>>>> Best regards,
>>>> Enric
>>>>
>>>
>>> Hello Enric,
>>>
>>> In order to obtain a better resolution/precision you could try to only
>>> use DIV512 hardware divisor instead of both DIV32 and DIV512.
>>>
>>
>> That makes the tests pass, so I'm wondering if by default you should
>> set DIV512 in platform data instead. With that change you can add my
>>
>> Tested-by: Enric Balletbo i Serra <eballetbo@kernel.org>
>>
>
> This change would decrease the RTC rollover time. Initially indeed only
> DIV512 was enabled, but during review [1] I've agreed enabling both
> hardware divisors to increase the RTC rollover time. Personally, I am ok
> with the current 1 second resolution capability.
>
> Alexandre Belloni, are you ok with the current version of this driver ?
> Or should I send a V10 patch in which I enable back only the DIV512
> divisor in order to increase the RTC precision at the downside of losing
> rollover time.
>
> [1]
> https://lore.kernel.org/all/9f4fc9d8-ec7f-43b4-8bdd-01a4ba4855c5@oss.nxp.com/
>
> Regards,
> Ciprian
>
Hello Alexandre,
I just want to make sure this email reaches you with my above question,
since you were just in CC previously.
Sorry for the inconvenience, I just do not want all the effort invested
in this driver (development + review) to get lost.
Regards,
Ciprian
>> # ./rtctest
>> TAP version 13
>> 1..8
>> # Starting 8 tests from 1 test cases.
>> # RUN rtc.date_read ...
>> # rtctest.c:59:date_read:Current RTC date/time is 28/01/2025 00:00:34.
>> # OK rtc.date_read
>> ok 1 rtc.date_read
>> # RUN rtc.date_read_loop ...
>> # rtctest.c:124:date_read_loop:Continuously reading RTC time for 30s
>> (with 11ms breaks after every read).
>> # rtctest.c:151:date_read_loop:Performed 2656 RTC time reads.
>> # OK rtc.date_read_loop
>> ok 2 rtc.date_read_loop
>> # RUN rtc.uie_read ...
>> # OK rtc.uie_read
>> ok 3 rtc.uie_read
>> # RUN rtc.uie_select ...
>> # OK rtc.uie_select
>> ok 4 rtc.uie_select
>> # RUN rtc.alarm_alm_set ...
>> # rtctest.c:262:alarm_alm_set:Alarm time now set to 00:01:14.
>> # rtctest.c:282:alarm_alm_set:data: 1a0
>> # OK rtc.alarm_alm_set
>> ok 5 rtc.alarm_alm_set
>> # RUN rtc.alarm_wkalm_set ...
>> # rtctest.c:334:alarm_wkalm_set:Alarm time now set to 28/01/2025
>> 00:01:17.
>> # OK rtc.alarm_wkalm_set
>> ok 6 rtc.alarm_wkalm_set
>> # RUN rtc.alarm_alm_set_minute ...
>> # rtctest.c:394:alarm_alm_set_minute:Alarm time now set to 00:02:00.
>> # rtctest.c:414:alarm_alm_set_minute:data: 1a0
>> # OK rtc.alarm_alm_set_minute
>> ok 7 rtc.alarm_alm_set_minute
>> # RUN rtc.alarm_wkalm_set_minute ...
>> # rtctest.c:464:alarm_wkalm_set_minute:Alarm time now set to
>> 28/01/2025 00:03:00.
>> # OK rtc.alarm_wkalm_set_minute
>> ok 8 rtc.alarm_wkalm_set_minute
>> # PASSED: 8 / 8 tests passed.
>> # Totals: pass:8 fail:0 xfail:0 xpass:0 skip:0 error:0
>>
>>
>> Thanks,
>> Enric
>>
>>> Regards,
>>> Ciprian
>>>
>>>>> ---
>>>>> drivers/rtc/Kconfig | 11 ++
>>>>> drivers/rtc/Makefile | 1 +
>>>>> drivers/rtc/rtc-s32g.c | 385
>>>>> +++++++++++++++++++++++++++++++++++++++++
>>>>> 3 files changed, 397 insertions(+)
>>>>> create mode 100644 drivers/rtc/rtc-s32g.c
>>>>>
>>>>> diff --git a/drivers/rtc/Kconfig b/drivers/rtc/Kconfig
>>>>> index 0bbbf778ecfa..510dc2db745d 100644
>>>>> --- a/drivers/rtc/Kconfig
>>>>> +++ b/drivers/rtc/Kconfig
>>>>> @@ -2103,4 +2103,15 @@ config RTC_DRV_AMLOGIC_A4
>>>>> This driver can also be built as a module. If so, the
>>>>> module
>>>>> will be called "rtc-amlogic-a4".
>>>>>
>>>>> +config RTC_DRV_S32G
>>>>> + tristate "RTC driver for S32G2/S32G3 SoCs"
>>>>> + depends on ARCH_S32 || COMPILE_TEST
>>>>> + depends on COMMON_CLK
>>>>> + help
>>>>> + Say yes to enable RTC driver for platforms based on the
>>>>> + S32G2/S32G3 SoC family.
>>>>> +
>>>>> + This RTC module can be used as a wakeup source.
>>>>> + Please note that it is not battery-powered.
>>>>> +
>>>>> endif # RTC_CLASS
>>>>> diff --git a/drivers/rtc/Makefile b/drivers/rtc/Makefile
>>>>> index 489b4ab07068..e4b616ecd5ce 100644
>>>>> --- a/drivers/rtc/Makefile
>>>>> +++ b/drivers/rtc/Makefile
>>>>> @@ -161,6 +161,7 @@ obj-$(CONFIG_RTC_DRV_RX8111) +=
>>>>> rtc-rx8111.o
>>>>> obj-$(CONFIG_RTC_DRV_RX8581) += rtc-rx8581.o
>>>>> obj-$(CONFIG_RTC_DRV_RZN1) += rtc-rzn1.o
>>>>> obj-$(CONFIG_RTC_DRV_RENESAS_RTCA3) += rtc-renesas-rtca3.o
>>>>> +obj-$(CONFIG_RTC_DRV_S32G) += rtc-s32g.o
>>>>> obj-$(CONFIG_RTC_DRV_S35390A) += rtc-s35390a.o
>>>>> obj-$(CONFIG_RTC_DRV_S3C) += rtc-s3c.o
>>>>> obj-$(CONFIG_RTC_DRV_S5M) += rtc-s5m.o
>>>>> diff --git a/drivers/rtc/rtc-s32g.c b/drivers/rtc/rtc-s32g.c
>>>>> new file mode 100644
>>>>> index 000000000000..3a0818e972eb
>>>>> --- /dev/null
>>>>> +++ b/drivers/rtc/rtc-s32g.c
>>>>> @@ -0,0 +1,385 @@
>>>>> +// SPDX-License-Identifier: GPL-2.0-or-later
>>>>> +/*
>>>>> + * Copyright 2025 NXP
>>>>> + */
>>>>> +
>>>>> +#include <linux/bitfield.h>
>>>>> +#include <linux/clk.h>
>>>>> +#include <linux/iopoll.h>
>>>>> +#include <linux/of_irq.h>
>>>>> +#include <linux/platform_device.h>
>>>>> +#include <linux/rtc.h>
>>>>> +
>>>>> +#define RTCC_OFFSET 0x4ul
>>>>> +#define RTCS_OFFSET 0x8ul
>>>>> +#define APIVAL_OFFSET 0x10ul
>>>>> +
>>>>> +/* RTCC fields */
>>>>> +#define RTCC_CNTEN BIT(31)
>>>>> +#define RTCC_APIEN BIT(15)
>>>>> +#define RTCC_APIIE BIT(14)
>>>>> +#define RTCC_CLKSEL_MASK GENMASK(13, 12)
>>>>> +#define RTCC_DIV512EN BIT(11)
>>>>> +#define RTCC_DIV32EN BIT(10)
>>>>> +
>>>>> +/* RTCS fields */
>>>>> +#define RTCS_INV_API BIT(17)
>>>>> +#define RTCS_APIF BIT(13)
>>>>> +
>>>>> +#define APIVAL_MAX_VAL GENMASK(31, 0)
>>>>> +#define RTC_SYNCH_TIMEOUT (100 * USEC_PER_MSEC)
>>>>> +
>>>>> +/*
>>>>> + * S32G2 and S32G3 SoCs have RTC clock source1 reserved and
>>>>> + * should not be used.
>>>>> + */
>>>>> +#define RTC_CLK_SRC1_RESERVED BIT(1)
>>>>> +
>>>>> +/*
>>>>> + * S32G RTC module has a 512 value and a 32 value hardware frequency
>>>>> + * divisors (DIV512 and DIV32) which could be used to achieve higher
>>>>> + * counter ranges by lowering the RTC frequency.
>>>>> + */
>>>>> +enum {
>>>>> + DIV1 = 1,
>>>>> + DIV32 = 32,
>>>>> + DIV512 = 512,
>>>>> + DIV512_32 = 16384
>>>>> +};
>>>>> +
>>>>> +static const char *const rtc_clk_src[] = {
>>>>> + "source0",
>>>>> + "source1",
>>>>> + "source2",
>>>>> + "source3"
>>>>> +};
>>>>> +
>>>>> +struct rtc_priv {
>>>>> + struct rtc_device *rdev;
>>>>> + void __iomem *rtc_base;
>>>>> + struct clk *ipg;
>>>>> + struct clk *clk_src;
>>>>> + const struct rtc_soc_data *rtc_data;
>>>>> + u64 rtc_hz;
>>>>> + time64_t sleep_sec;
>>>>> + int irq;
>>>>> + u32 clk_src_idx;
>>>>> +};
>>>>> +
>>>>> +struct rtc_soc_data {
>>>>> + u32 clk_div;
>>>>> + u32 reserved_clk_mask;
>>>>> +};
>>>>> +
>>>>> +static const struct rtc_soc_data rtc_s32g2_data = {
>>>>> + .clk_div = DIV512_32,
>>>>> + .reserved_clk_mask = RTC_CLK_SRC1_RESERVED,
>>>>> +};
>>>>> +
>>>>> +static irqreturn_t s32g_rtc_handler(int irq, void *dev)
>>>>> +{
>>>>> + struct rtc_priv *priv = platform_get_drvdata(dev);
>>>>> + u32 status;
>>>>> +
>>>>> + status = readl(priv->rtc_base + RTCS_OFFSET);
>>>>> +
>>>>> + if (status & RTCS_APIF) {
>>>>> + writel(0x0, priv->rtc_base + APIVAL_OFFSET);
>>>>> + writel(status | RTCS_APIF, priv->rtc_base +
>>>>> RTCS_OFFSET);
>>>>> + }
>>>>> +
>>>>> + rtc_update_irq(priv->rdev, 1, RTC_IRQF | RTC_AF);
>>>>> +
>>>>> + return IRQ_HANDLED;
>>>>> +}
>>>>> +
>>>>> +/*
>>>>> + * The function is not really getting time from the RTC since the
>>>>> S32G RTC
>>>>> + * has several limitations. Thus, to setup alarm use system time.
>>>>> + */
>>>>> +static int s32g_rtc_read_time(struct device *dev,
>>>>> + struct rtc_time *tm)
>>>>> +{
>>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>>> + time64_t sec;
>>>>> +
>>>>> + if (check_add_overflow(ktime_get_real_seconds(),
>>>>> + priv->sleep_sec, &sec))
>>>>> + return -ERANGE;
>>>>> +
>>>>> + rtc_time64_to_tm(sec, tm);
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +static int s32g_rtc_read_alarm(struct device *dev, struct
>>>>> rtc_wkalrm *alrm)
>>>>> +{
>>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>>> + u32 rtcc, rtcs;
>>>>> +
>>>>> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>>>>> + rtcs = readl(priv->rtc_base + RTCS_OFFSET);
>>>>> +
>>>>> + alrm->enabled = rtcc & RTCC_APIIE;
>>>>> + if (alrm->enabled)
>>>>> + alrm->pending = !(rtcs & RTCS_APIF);
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +static int s32g_rtc_alarm_irq_enable(struct device *dev, unsigned
>>>>> int enabled)
>>>>> +{
>>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>>> + u32 rtcc;
>>>>> +
>>>>> + /* RTC API functionality is used both for triggering
>>>>> interrupts
>>>>> + * and as a wakeup event. Hence it should always be enabled.
>>>>> + */
>>>>> + rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>>>>> + rtcc |= RTCC_APIEN | RTCC_APIIE;
>>>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +static int s32g_rtc_set_alarm(struct device *dev, struct
>>>>> rtc_wkalrm *alrm)
>>>>> +{
>>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>>> + unsigned long long cycles;
>>>>> + long long t_offset;
>>>>> + time64_t alrm_time;
>>>>> + u32 rtcs;
>>>>> + int ret;
>>>>> +
>>>>> + alrm_time = rtc_tm_to_time64(&alrm->time);
>>>>> + t_offset = alrm_time - ktime_get_real_seconds() -
>>>>> priv->sleep_sec;
>>>>> + if (t_offset < 0)
>>>>> + return -ERANGE;
>>>>> +
>>>>> + cycles = t_offset * priv->rtc_hz;
>>>>> + if (cycles > APIVAL_MAX_VAL)
>>>>> + return -ERANGE;
>>>>> +
>>>>> + /* APIVAL could have been reset from the IRQ handler.
>>>>> + * Hence, we wait in case there is a synchronization process.
>>>>> + */
>>>>> + ret = read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
>>>>> + 0, RTC_SYNCH_TIMEOUT, false,
>>>>> priv->rtc_base + RTCS_OFFSET);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> +
>>>>> + writel(cycles, priv->rtc_base + APIVAL_OFFSET);
>>>>> +
>>>>> + return read_poll_timeout(readl, rtcs, !(rtcs & RTCS_INV_API),
>>>>> + 0, RTC_SYNCH_TIMEOUT, false,
>>>>> priv->rtc_base + RTCS_OFFSET);
>>>>> +}
>>>>> +
>>>>> +/*
>>>>> + * Disable the 32-bit free running counter.
>>>>> + * This allows Clock Source and Divisors selection
>>>>> + * to be performed without causing synchronization issues.
>>>>> + */
>>>>> +static void s32g_rtc_disable(struct rtc_priv *priv)
>>>>> +{
>>>>> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>>>>> +
>>>>> + rtcc &= ~RTCC_CNTEN;
>>>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>>>>> +}
>>>>> +
>>>>> +static void s32g_rtc_enable(struct rtc_priv *priv)
>>>>> +{
>>>>> + u32 rtcc = readl(priv->rtc_base + RTCC_OFFSET);
>>>>> +
>>>>> + rtcc |= RTCC_CNTEN;
>>>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>>>>> +}
>>>>> +
>>>>> +static int rtc_clk_src_setup(struct rtc_priv *priv)
>>>>> +{
>>>>> + u32 rtcc;
>>>>> +
>>>>> + rtcc = FIELD_PREP(RTCC_CLKSEL_MASK, priv->clk_src_idx);
>>>>> +
>>>>> + switch (priv->rtc_data->clk_div) {
>>>>> + case DIV512_32:
>>>>> + rtcc |= RTCC_DIV512EN;
>>>>> + rtcc |= RTCC_DIV32EN;
>>>>> + break;
>>>>> + case DIV512:
>>>>> + rtcc |= RTCC_DIV512EN;
>>>>> + break;
>>>>> + case DIV32:
>>>>> + rtcc |= RTCC_DIV32EN;
>>>>> + break;
>>>>> + case DIV1:
>>>>> + break;
>>>>> + default:
>>>>> + return -EINVAL;
>>>>> + }
>>>>> +
>>>>> + rtcc |= RTCC_APIEN | RTCC_APIIE;
>>>>> + /*
>>>>> + * Make sure the CNTEN is 0 before we configure
>>>>> + * the clock source and dividers.
>>>>> + */
>>>>> + s32g_rtc_disable(priv);
>>>>> + writel(rtcc, priv->rtc_base + RTCC_OFFSET);
>>>>> + s32g_rtc_enable(priv);
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +static const struct rtc_class_ops rtc_ops = {
>>>>> + .read_time = s32g_rtc_read_time,
>>>>> + .read_alarm = s32g_rtc_read_alarm,
>>>>> + .set_alarm = s32g_rtc_set_alarm,
>>>>> + .alarm_irq_enable = s32g_rtc_alarm_irq_enable,
>>>>> +};
>>>>> +
>>>>> +static int rtc_clk_dts_setup(struct rtc_priv *priv,
>>>>> + struct device *dev)
>>>>> +{
>>>>> + u32 i;
>>>>> +
>>>>> + priv->ipg = devm_clk_get_enabled(dev, "ipg");
>>>>> + if (IS_ERR(priv->ipg))
>>>>> + return dev_err_probe(dev, PTR_ERR(priv->ipg),
>>>>> + "Failed to get 'ipg' clock\n");
>>>>> +
>>>>> + for (i = 0; i < ARRAY_SIZE(rtc_clk_src); i++) {
>>>>> + if (priv->rtc_data->reserved_clk_mask & BIT(i))
>>>>> + return -EOPNOTSUPP;
>>>>> +
>>>>> + priv->clk_src = devm_clk_get_enabled(dev,
>>>>> rtc_clk_src[i]);
>>>>> + if (!IS_ERR(priv->clk_src)) {
>>>>> + priv->clk_src_idx = i;
>>>>> + break;
>>>>> + }
>>>>> + }
>>>>> +
>>>>> + if (IS_ERR(priv->clk_src))
>>>>> + return dev_err_probe(dev, PTR_ERR(priv->clk_src),
>>>>> + "Failed to get rtc module clock
>>>>> source\n");
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +static int s32g_rtc_probe(struct platform_device *pdev)
>>>>> +{
>>>>> + struct device *dev = &pdev->dev;
>>>>> + struct rtc_priv *priv;
>>>>> + unsigned long rtc_hz;
>>>>> + int ret;
>>>>> +
>>>>> + priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
>>>>> + if (!priv)
>>>>> + return -ENOMEM;
>>>>> +
>>>>> + priv->rtc_data = of_device_get_match_data(dev);
>>>>> + if (!priv->rtc_data)
>>>>> + return -ENODEV;
>>>>> +
>>>>> + priv->rtc_base = devm_platform_ioremap_resource(pdev, 0);
>>>>> + if (IS_ERR(priv->rtc_base))
>>>>> + return PTR_ERR(priv->rtc_base);
>>>>> +
>>>>> + device_init_wakeup(dev, true);
>>>>> +
>>>>> + ret = rtc_clk_dts_setup(priv, dev);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> +
>>>>> + priv->rdev = devm_rtc_allocate_device(dev);
>>>>> + if (IS_ERR(priv->rdev))
>>>>> + return PTR_ERR(priv->rdev);
>>>>> +
>>>>> + ret = rtc_clk_src_setup(priv);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> +
>>>>> + priv->irq = platform_get_irq(pdev, 0);
>>>>> + if (priv->irq < 0) {
>>>>> + ret = priv->irq;
>>>>> + goto disable_rtc;
>>>>> + }
>>>>> +
>>>>> + rtc_hz = clk_get_rate(priv->clk_src);
>>>>> + if (!rtc_hz) {
>>>>> + dev_err(dev, "Failed to get RTC frequency\n");
>>>>> + ret = -EINVAL;
>>>>> + goto disable_rtc;
>>>>> + }
>>>>> +
>>>>> + priv->rtc_hz = DIV_ROUND_UP(rtc_hz, priv->rtc_data->clk_div);
>>>>> +
>>>>> + platform_set_drvdata(pdev, priv);
>>>>> + priv->rdev->ops = &rtc_ops;
>>>>> +
>>>>> + ret = devm_request_irq(dev, priv->irq,
>>>>> + s32g_rtc_handler, 0, dev_name(dev),
>>>>> pdev);
>>>>> + if (ret) {
>>>>> + dev_err(dev, "Request interrupt %d failed, error:
>>>>> %d\n",
>>>>> + priv->irq, ret);
>>>>> + goto disable_rtc;
>>>>> + }
>>>>> +
>>>>> + ret = devm_rtc_register_device(priv->rdev);
>>>>> + if (ret)
>>>>> + goto disable_rtc;
>>>>> +
>>>>> + return 0;
>>>>> +
>>>>> +disable_rtc:
>>>>> + s32g_rtc_disable(priv);
>>>>> + return ret;
>>>>> +}
>>>>> +
>>>>> +static int s32g_rtc_suspend(struct device *dev)
>>>>> +{
>>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>>> + u32 apival = readl(priv->rtc_base + APIVAL_OFFSET);
>>>>> +
>>>>> + if (check_add_overflow(priv->sleep_sec, div64_u64(apival,
>>>>> priv->rtc_hz),
>>>>> + &priv->sleep_sec)) {
>>>>> + dev_warn(dev, "Overflow on sleep cycles occurred.
>>>>> Resetting to 0.\n");
>>>>> + priv->sleep_sec = 0;
>>>>> + }
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +static int s32g_rtc_resume(struct device *dev)
>>>>> +{
>>>>> + struct rtc_priv *priv = dev_get_drvdata(dev);
>>>>> +
>>>>> + /* The transition from resume to run is a reset event.
>>>>> + * This leads to the RTC registers being reset after resume
>>>>> from
>>>>> + * suspend. It is uncommon, but this behaviour has been
>>>>> observed
>>>>> + * on S32G RTC after issuing a Suspend to RAM operation.
>>>>> + * Thus, reconfigure RTC registers on the resume path.
>>>>> + */
>>>>> + return rtc_clk_src_setup(priv);
>>>>> +}
>>>>> +
>>>>> +static const struct of_device_id rtc_dt_ids[] = {
>>>>> + { .compatible = "nxp,s32g2-rtc", .data = &rtc_s32g2_data },
>>>>> + { /* sentinel */ },
>>>>> +};
>>>>> +
>>>>> +static DEFINE_SIMPLE_DEV_PM_OPS(s32g_rtc_pm_ops,
>>>>> + s32g_rtc_suspend, s32g_rtc_resume);
>>>>> +
>>>>> +static struct platform_driver s32g_rtc_driver = {
>>>>> + .driver = {
>>>>> + .name = "s32g-rtc",
>>>>> + .pm = pm_sleep_ptr(&s32g_rtc_pm_ops),
>>>>> + .of_match_table = rtc_dt_ids,
>>>>> + },
>>>>> + .probe = s32g_rtc_probe,
>>>>> +};
>>>>> +module_platform_driver(s32g_rtc_driver);
>>>>> +
>>>>> +MODULE_AUTHOR("NXP");
>>>>> +MODULE_DESCRIPTION("NXP RTC driver for S32G2/S32G3");
>>>>> +MODULE_LICENSE("GPL");
>>>>> --
>>>>> 2.45.2
>>>>>
>>>>
>>>
>>
>
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support
2025-04-04 8:53 ` Alexandre Belloni
2025-04-04 11:36 ` Ciprian Marian Costea
@ 2025-05-20 9:01 ` Ciprian Marian Costea
1 sibling, 0 replies; 19+ messages in thread
From: Ciprian Marian Costea @ 2025-05-20 9:01 UTC (permalink / raw)
To: Alexandre Belloni
Cc: Krzysztof Kozlowski, Rob Herring, Krzysztof Kozlowski,
Conor Dooley, Catalin Marinas, Will Deacon, linux-rtc, devicetree,
linux-kernel, linux-arm-kernel, NXP S32 Linux, imx,
Christophe Lizzi, Alberto Ruiz, Enric Balletbo, Eric Chanudet
On 4/4/2025 11:53 AM, Alexandre Belloni wrote:
[ ... ]
>>
>> Ok. I will send out a V10 in which I will drop this current patch from the
>> patchset. Also, I will send the DTS patch which adds S32G274A-EVB,
>> S32G274-RDB2 and S32G399A-RDB3 usage of the RTC after this patchset gets
>> accepted.
>>
>
> I don't need V10, I can apply V9 without 3/4 and 4/4
>
>> Best Regards,
>> Ciprian
>>
Hello Alexandre,
Sorry for bothering. I am writing to ask about the status of this
patchset since it has been hanging for a while.
Are there any more changes required from your point of view ?
Regards,
Ciprian
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: (subset) [PATCH v9 0/4] add NXP RTC driver support for S32G2/S32G3 SoCs
2025-04-03 10:33 [PATCH v9 0/4] add NXP RTC driver support for S32G2/S32G3 SoCs Ciprian Costea
` (3 preceding siblings ...)
2025-04-03 10:33 ` [PATCH v9 4/4] MAINTAINERS: add NXP S32G RTC driver Ciprian Costea
@ 2025-05-25 21:33 ` Alexandre Belloni
4 siblings, 0 replies; 19+ messages in thread
From: Alexandre Belloni @ 2025-05-25 21:33 UTC (permalink / raw)
To: Rob Herring, Krzysztof Kozlowski, Conor Dooley, Catalin Marinas,
Will Deacon, Ciprian Costea
Cc: linux-rtc, devicetree, linux-kernel, linux-arm-kernel,
NXP S32 Linux, imx, Christophe Lizzi, Alberto Ruiz,
Enric Balletbo, Eric Chanudet
On Thu, 03 Apr 2025 13:33:42 +0300, Ciprian Costea wrote:
> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>
> This patch series add support for the NXP RTC hardware module present on
> S32G2/S32G3 SoCs.
>
> RTC tracks clock time during system suspend. It is used as a time-based
> wakeup source for the S32G2/S32G3 SoCs.
>
> [...]
Applied, thanks!
[1/4] dt-bindings: rtc: add schema for NXP S32G2/S32G3 SoCs
https://git.kernel.org/abelloni/c/8568d8b4daec
[2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support
https://git.kernel.org/abelloni/c/818b6709e1cb
Best regards,
--
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
^ permalink raw reply [flat|nested] 19+ messages in thread
end of thread, other threads:[~2025-05-25 21:36 UTC | newest]
Thread overview: 19+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-04-03 10:33 [PATCH v9 0/4] add NXP RTC driver support for S32G2/S32G3 SoCs Ciprian Costea
2025-04-03 10:33 ` [PATCH v9 1/4] dt-bindings: rtc: add schema for NXP " Ciprian Costea
2025-04-03 10:33 ` [PATCH v9 2/4] rtc: s32g: add NXP S32G2/S32G3 SoC support Ciprian Costea
2025-04-09 14:14 ` Enric Balletbo i Serra
2025-04-09 14:26 ` Ciprian Marian Costea
2025-04-09 14:42 ` Enric Balletbo i Serra
2025-04-09 14:58 ` Ciprian Marian Costea
2025-04-10 7:08 ` Enric Balletbo i Serra
2025-04-11 8:09 ` Ciprian Marian Costea
2025-04-03 10:33 ` [PATCH v9 3/4] arm64: defconfig: add S32G RTC module support Ciprian Costea
2025-04-04 6:17 ` Krzysztof Kozlowski
2025-04-04 6:24 ` Ciprian Marian Costea
2025-04-04 7:15 ` Krzysztof Kozlowski
2025-04-04 7:19 ` Ciprian Marian Costea
2025-04-04 8:53 ` Alexandre Belloni
2025-04-04 11:36 ` Ciprian Marian Costea
2025-05-20 9:01 ` Ciprian Marian Costea
2025-04-03 10:33 ` [PATCH v9 4/4] MAINTAINERS: add NXP S32G RTC driver Ciprian Costea
2025-05-25 21:33 ` (subset) [PATCH v9 0/4] add NXP RTC driver support for S32G2/S32G3 SoCs Alexandre Belloni
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).