Skip to contentSkip to conversation

rockchip: odroid-m1s/rk3566 watchdog support

PATCH Superseded Andreas Zdziarstek 2 participants 6 messages Jun 29

Subsystems (3)

v2 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

v2 cover rockchip: odroid-m1s/rk3566 watchdog support

This series makes the Synopsys DesignWare watchdog usable on rk3566/rk3568 and enables it on the Hardkernel ODROID-M1S.

On current next the dw-wdt driver already probes (d62801d09441), but a timeout or "wdt expire" still hangs the SoC: CRU_GLB_RST_CON routes the watchdog to the second global reset by default, which causes a hung SoC.

Patch 1 routes the watchdog to the first global reset instead, same as the PX30 implementation. Patch 2 does the same for the TSADC, so a thermal shutdown also resets the whole SoC. Patch 3 enables the watchdog (wdt command only, no autostart to be non-breaking) on the ODROID-M1S.

Verified on the ODROID-M1S (RK3566): a watchdog timeout cleanly reboots the board.

Changes in v2:
- patch 1: access CRU_GLB_RST_CON through the rk3568_cru struct, matching PX30's arch_cpu_init() (review comment by Quentin)
- patch 2 (new): also route the TSADC to a first global reset (review comment by Quentin)
- patch 3 (was 2/2): unchanged

On the TSADC (patch 2): while checking how the ODROID-M1S comes up, I noticed the rkbin DDR init (probably, at least I don't know what else could set it at that stage) already sets the TSADC bit in CRU_GLB_RST_CON before U-Boot runs. So, on this board the patch is effectively a no-op. It still seems right to configure it explicitly in U-Boot rather than depend on the firmware blob, and it matches what PX30 does.

Jonas: on v1 you raised two points on patch 1 - guarding the write on XPL_BUILD, and clearing SOC_CON1 so the watchdog reset behaves like the TF-A/PSCI reset. I replied on the thread [2]: I kept the write unconditional to match PX30 (and because U-Boot proper can be entered without our SPL having run, e.g. using an rkbin miniloader), and I posed the question if the watchdog actually *should* drop BOOT_MODE_REG for a clean recovery rather than preserve it.

I've left both as they were in v2, for now. Happy to do a v3, however, if you disagree on these points.

[2] https://lists.denx.de/pipermail/u-boot/2026-June/622663.html

Andreas Zdziarstek (3):
rockchip: rk3568: make the WDT trigger a first global reset
rockchip: rk3568: make the TSADC trigger a first global reset
rockchip: odroid-m1s: enable watchdog support

 arch/arm/mach-rockchip/rk3568/rk3568.c | 14 ++++++++++++++
 configs/odroid-m1s-rk3566_defconfig    |  5 +++++
 2 files changed, 19 insertions(+)

base-commit: 835a18f80f25731dc818bf9b771bfa111ea3dbeb

Signature · 1 line
2.53.0
#1 lore raw

PATCH v2 1/3 rockchip: rk3568: make the WDT trigger a first global reset

Default is a second global reset which causes a system hang on a timeout in U-Boot.

Access the register through the rk3568_cru struct to match how PX30 does it in arch_cpu_init().

Verified on the Hardkernel ODROID-M1S (RK3566): a watchdog timeout now cleanly reboots the board.

---
 arch/arm/mach-rockchip/rk3568/rk3568.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
arch/arm/mach-rockchip/rk3568/rk3568.c +10 -0
index 2b1eafee37c..87bd443e9c4 100644
@@ -9,6 +9,7 @@
 #include <misc.h>
 #include <asm/armv8/mmu.h>
 #include <asm/arch-rockchip/bootrom.h>
+#include <asm/arch-rockchip/cru_rk3568.h>
 #include <asm/arch-rockchip/grf_rk3568.h>
 #include <asm/arch-rockchip/hardware.h>
 #include <dt-bindings/clock/rk3568-cru.h>
@@ -37,6 +38,9 @@
 #define CPU_GRF_BASE		0xfdc30000
 #define GRF_CORE_PVTPLL_CON0	(0x10)
 
+#define CRU_BASE		0xfdd20000
+#define WDT_GLB_SRST_CTRL	BIT(1)
+
 /* PMU_GRF_GPIO0D_IOMUX_L */
 enum {
 	GPIO0D1_SHIFT		= 4,
@@ -116,6 +120,8 @@ void board_debug_uart_init(void)
 
 int arch_cpu_init(void)
 {
+	static struct rk3568_cru * const cru = (void *)CRU_BASE;
+
 #ifdef CONFIG_XPL_BUILD
 	/*
 	 * When perform idle operation, corresponding clock can
@@ -146,6 +152,10 @@ int arch_cpu_init(void)
 	writel((PMU_PD_VO_DWN_ENA << 16),
 	       PMU_BASE_ADDR + PMU_PWR_GATE_SFTCON);
 #endif
+
+	/* Make WDT trigger a first global reset */
+	setbits_le32(&cru->glb_rst_con, WDT_GLB_SRST_CTRL);
+
 	return 0;
 }
 
Signature · 1 line
2.53.0
#2 lore raw

Re: [PATCH v2 1/3] rockchip: rk3568: make the WDT trigger a first global reset

Hi Andreas,

Quoted 46 lines · Andreas Zdziarstek

Default is a second global reset which causes a system hang on a timeout in U-Boot.

Access the register through the rk3568_cru struct to match how PX30 does it in arch_cpu_init().

Verified on the Hardkernel ODROID-M1S (RK3566): a watchdog timeout now cleanly reboots the board.

---
 arch/arm/mach-rockchip/rk3568/rk3568.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
arch/arm/mach-rockchip/rk3568/rk3568.c +6 -0
index 2b1eafee37c..87bd443e9c4 100644
@@ -9,6 +9,7 @@
 #include <misc.h>
 #include <asm/armv8/mmu.h>
 #include <asm/arch-rockchip/bootrom.h>
+#include <asm/arch-rockchip/cru_rk3568.h>
 #include <asm/arch-rockchip/grf_rk3568.h>
 #include <asm/arch-rockchip/hardware.h>
 #include <dt-bindings/clock/rk3568-cru.h>
@@ -37,6 +38,9 @@
 #define CPU_GRF_BASE		0xfdc30000
 #define GRF_CORE_PVTPLL_CON0	(0x10)
 
+#define CRU_BASE		0xfdd20000
+#define WDT_GLB_SRST_CTRL	BIT(1)
+
 /* PMU_GRF_GPIO0D_IOMUX_L */
 enum {
 	GPIO0D1_SHIFT		= 4,
@@ -116,6 +120,8 @@ void board_debug_uart_init(void)
 
 int arch_cpu_init(void)
 {
+	static struct rk3568_cru * const cru = (void *)CRU_BASE;
+
 #ifdef CONFIG_XPL_BUILD
 	/*
 	 * When perform idle operation, corresponding clock can

arch/arm/mach-rockchip/rk3568/rk3568.c int arch_cpu_init(void) · line 157

Quoted 6 lines
arch/arm/mach-rockchip/rk3568/rk3568.c +3 -0
@@ -146,6 +152,10 @@ int arch_cpu_init(void)
 	writel((PMU_PD_VO_DWN_ENA << 16),
 	       PMU_BASE_ADDR + PMU_PWR_GATE_SFTCON);
 #endif
+
+	/* Make WDT trigger a first global reset */
+	setbits_le32(&cru->glb_rst_con, WDT_GLB_SRST_CTRL);

As mentioned in v1, please keep this inside XPL_BUILD.

SPL+U-Boot proper combined should be seen as a single firmware, we do not consider vendor SPL + mainline U-Boot proper a supported combo.

If you also want to configure this in U-Boot proper, maybe probe of WDT or CRU driver is a better fit?

arch_cpu_init() is mainly for one-off configuration that is needed very early or otherwise does not really fit anywhere in the lazy loading of driver model.

Regards,
Jonas

Quoted 4 lines
arch/arm/mach-rockchip/rk3568/rk3568.c +1 -0
+
 	return 0;
 }
 
#3 lore raw

PATCH v2 2/3 rockchip: rk3568: make the TSADC trigger a first global reset

In reply to Andreas Zdziarstek

Same rationale as the watchdog: the default second global reset leaves the SoC hung. Set the TSADC to the first global reset as well, so a thermal shutdown reliably resets the whole SoC.

---
 arch/arm/mach-rockchip/rk3568/rk3568.c | 4 ++++
 1 file changed, 4 insertions(+)
arch/arm/mach-rockchip/rk3568/rk3568.c +4 -0
index 87bd443e9c4..50b61b7cd26 100644
@@ -40,6 +40,7 @@
 
 #define CRU_BASE		0xfdd20000
 #define WDT_GLB_SRST_CTRL	BIT(1)
+#define TSADC_GLB_SRST_CTRL	BIT(0)
 
 /* PMU_GRF_GPIO0D_IOMUX_L */
 enum {
@@ -156,6 +157,9 @@ int arch_cpu_init(void)
 	/* Make WDT trigger a first global reset */
 	setbits_le32(&cru->glb_rst_con, WDT_GLB_SRST_CTRL);
 
+	/* Make TSADC trigger a first global reset */
+	setbits_le32(&cru->glb_rst_con, TSADC_GLB_SRST_CTRL);
+
 	return 0;
 }
 
Signature · 1 line
2.53.0
#4 lore raw

Re: [PATCH v2 2/3] rockchip: rk3568: make the TSADC trigger a first global reset

Hi Andreas,

Quoted 22 lines · Andreas Zdziarstek

Same rationale as the watchdog: the default second global reset leaves the SoC hung. Set the TSADC to the first global reset as well, so a thermal shutdown reliably resets the whole SoC.

---
 arch/arm/mach-rockchip/rk3568/rk3568.c | 4 ++++
 1 file changed, 4 insertions(+)
arch/arm/mach-rockchip/rk3568/rk3568.c +1 -0
index 87bd443e9c4..50b61b7cd26 100644
@@ -40,6 +40,7 @@
 
 #define CRU_BASE		0xfdd20000
 #define WDT_GLB_SRST_CTRL	BIT(1)
+#define TSADC_GLB_SRST_CTRL	BIT(0)
 
 /* PMU_GRF_GPIO0D_IOMUX_L */
 enum {

arch/arm/mach-rockchip/rk3568/rk3568.c int arch_cpu_init(void) · line 161

Quoted 5 lines
arch/arm/mach-rockchip/rk3568/rk3568.c +2 -0
@@ -156,6 +157,9 @@ int arch_cpu_init(void)
 	/* Make WDT trigger a first global reset */
 	setbits_le32(&cru->glb_rst_con, WDT_GLB_SRST_CTRL);
 
+	/* Make TSADC trigger a first global reset */
+	setbits_le32(&cru->glb_rst_con, TSADC_GLB_SRST_CTRL);

There is likely no need to read and write this reg twice, please combine.

Regards,
Jonas

Quoted 4 lines
arch/arm/mach-rockchip/rk3568/rk3568.c +1 -0
+
 	return 0;
 }
 
#5 lore raw

PATCH v2 3/3 rockchip: odroid-m1s: enable watchdog support

In reply to Andreas Zdziarstek

The rk3566 ODROID-M1S has a DesignWare watchdog. Enable CONFIG_WDT, CONFIG_DESIGNWARE_WATCHDOG and CONFIG_CMD_WDT so it can be driven via the wdt command.

CONFIG_WATCHDOG and CONFIG_WATCHDOG_AUTOSTART are left explicitly unset to not cause any breaking changes on existing systems.

---
 configs/odroid-m1s-rk3566_defconfig | 5 +++++
 1 file changed, 5 insertions(+)
configs/odroid-m1s-rk3566_defconfig +5 -0
index d59882a2a59..943c0509e9e 100644
@@ -30,6 +30,7 @@ CONFIG_CMD_PCI=y
 CONFIG_CMD_USB=y
 CONFIG_CMD_ROCKUSB=y
 CONFIG_CMD_USB_MASS_STORAGE=y
+CONFIG_CMD_WDT=y
 # CONFIG_CMD_SETEXPR is not set
 CONFIG_CMD_INI=y
 CONFIG_CMD_PMIC=y
@@ -87,5 +88,9 @@ CONFIG_USB_DWC3_GENERIC=y
 CONFIG_USB_GADGET=y
 CONFIG_USB_GADGET_DOWNLOAD=y
 CONFIG_USB_FUNCTION_ROCKUSB=y
+# CONFIG_WATCHDOG is not set
+# CONFIG_WATCHDOG_AUTOSTART is not set
+CONFIG_DESIGNWARE_WATCHDOG=y
+CONFIG_WDT=y
 CONFIG_FS_CRAMFS=y
 CONFIG_ERRNO_STR=y
Signature · 1 line
2.53.0