aspeed: watchdog: fix system reset and read alt-boot option
Subsystems (2)
cover aspeed: watchdog: fix system reset and read alt-boot option
This series provides a fix to correctly set the reset mode of the control register. Previously, configuring anything other than a full chip reset would not reset anything when the watchdog timer expires. The series also provides a patch to read in the existing aspeed,alt-boot boolean option from the device-tree and set the appropriate bit in the control register.
Milton Miller (2):
watchdog: aspeed: Fix translation of reset mode to ctrl register
watchdog: aspeed: Allow configuring for alternate boot
drivers/watchdog/aspeed_wdt.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-)
Signature · 1 line
1.8.3.1
PATCH 1/2 watchdog: aspeed: Fix translation of reset mode to ctrl register
From: Milton Miller <[email protected]>
Assert RESET_SYSTEM bit for any reset and set MODE field from reset type.
The watchdog control register has a RESET_SYSTEM bit that is really closer to activate a reset, and RESET_SYSTEM_MODE field that chooses how much to reset.
Before this patch, a node without these optional property would do a SOC reset, but a node with properties requesting a cpu or SOC reset would do nothing and a node requesting a system reset would do a SOC reset.
--- drivers/watchdog/aspeed_wdt.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)
drivers/watchdog/aspeed_wdt.c +6 -3
@@ -232,11 +232,14 @@ static int aspeed_wdt_probe(struct platform_device *pdev) wdt->ctrl |=WDT_CTRL_RESET_MODE_SOC |WDT_CTRL_RESET_SYSTEM ; }else {if (!strcmp (reset_type,"cpu" )) - wdt->ctrl |=WDT_CTRL_RESET_MODE_ARM_CPU ; + wdt->ctrl |=WDT_CTRL_RESET_MODE_ARM_CPU | +WDT_CTRL_RESET_SYSTEM ;else if (!strcmp (reset_type,"soc" )) - wdt->ctrl |=WDT_CTRL_RESET_MODE_SOC ; + wdt->ctrl |=WDT_CTRL_RESET_MODE_SOC | +WDT_CTRL_RESET_SYSTEM ;else if (!strcmp (reset_type,"system" )) - wdt->ctrl |=WDT_CTRL_RESET_SYSTEM ; + wdt->ctrl |=WDT_CTRL_RESET_MODE_FULL_CHIP | +WDT_CTRL_RESET_SYSTEM ;else if (strcmp (reset_type,"none" ))return -EINVAL ; }
Signature · 1 line
1.8.3.1
Re: [PATCH 1/2] watchdog: aspeed: Fix translation of reset mode to ctrl register
Quoted 9 lines · Eddie James
From: Milton Miller <[email protected]>
Assert
RESET_SYSTEMbit for any reset and set MODE field from reset type.The watchdog control register has a
RESET_SYSTEMbit that is really closer to activate a reset, andRESET_SYSTEM_MODEfield that chooses how much to reset.
Quoted 8 lines
Before this patch, a node without these optional property would do a SOC reset, but a node with properties requesting a cpu or SOC reset would do nothing and a node requesting a system reset would do a SOC reset.
Fixes b7f0b8ad25f3 ("drivers/watchdog: ASPEED reference dev tree properties for config")Signed-off-by Milton Miller <[email protected]>Signed-off-by Eddie James <[email protected]>
Quoted 29 lines
--- drivers/watchdog/aspeed_wdt.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)drivers/watchdog/aspeed_wdt.c +6
-3@@ -232,11 +232,14 @@ static int aspeed_wdt_probe(struct platform_device *pdev) wdt->ctrl |=WDT_CTRL_RESET_MODE_SOC |WDT_CTRL_RESET_SYSTEM ; }else {if (!strcmp (reset_type,"cpu" )) - wdt->ctrl |=WDT_CTRL_RESET_MODE_ARM_CPU ; + wdt->ctrl |=WDT_CTRL_RESET_MODE_ARM_CPU | +WDT_CTRL_RESET_SYSTEM ;else if (!strcmp (reset_type,"soc" )) - wdt->ctrl |=WDT_CTRL_RESET_MODE_SOC ; + wdt->ctrl |=WDT_CTRL_RESET_MODE_SOC | +WDT_CTRL_RESET_SYSTEM ;else if (!strcmp (reset_type,"system" )) - wdt->ctrl |=WDT_CTRL_RESET_SYSTEM ; + wdt->ctrl |=WDT_CTRL_RESET_MODE_FULL_CHIP | +WDT_CTRL_RESET_SYSTEM ;else if (strcmp (reset_type,"none" ))return -EINVAL ; }Signature · 1 line
1.8.3.1
PATCH 2/2 watchdog: aspeed: Allow configuring for alternate boot
In reply to Eddie JamesFrom: Milton Miller <[email protected]>
Allow the device tree to specify a watchdog to fallover to the alternate boot source.
The aspeeed watchdog can set a latch directing flash chip select 0 to chip select 1, allowing boot from an alternate media if the watchdog is not reset in time. On the ast2400 bank 1 also goes to flash bank 1, while on the ast2500 the chip selects are swapped.
--- drivers/watchdog/aspeed_wdt.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/watchdog/aspeed_wdt.c +3 -0
@@ -46,6 +46,7 @@ struct aspeed_wdt_config {#define WDT_RELOAD_VALUE 0x04#define WDT_RESTART 0x08#define WDT_CTRL 0x0C +#define WDT_CTRL_BOOT_SECONDARY BIT(7)#define WDT_CTRL_RESET_MODE_SOC (0x00 << 5)#define WDT_CTRL_RESET_MODE_FULL_CHIP (0x01 << 5)#define WDT_CTRL_RESET_MODE_ARM_CPU (0x10 << 5)
@@ -245,6 +246,8 @@ static int aspeed_wdt_probe(struct platform_device *pdev) }if (of_property_read_bool (np,"aspeed,external-signal" )) wdt->ctrl |=WDT_CTRL_WDT_EXT ; +if (of_property_read_bool (np,"aspeed,alt-boot" )) + wdt->ctrl |=WDT_CTRL_BOOT_SECONDARY ;if (readl (wdt->base +WDT_CTRL ) &WDT_CTRL_ENABLE ) {/*
Signature · 1 line
1.8.3.1
Re: [PATCH 2/2] watchdog: aspeed: Allow configuring for alternate boot
Quoted 5 lines · Eddie James
From: Milton Miller <[email protected]>
Allow the device tree to specify a watchdog to fallover to the alternate boot source.
Quoted 7 lines
The aspeeed watchdog can set a latch directing flash chip select 0 to chip select 1, allowing boot from an alternate media if the watchdog is not reset in time. On the ast2400 bank 1 also goes to flash bank 1, while on the ast2500 the chip selects are swapped.
Signed-off-by Milton Miller <[email protected]>Signed-off-by Eddie James <[email protected]>
This is already documented in the bindings document, so the property should be ok.
Quoted 28 lines
--- drivers/watchdog/aspeed_wdt.c | 3 +++ 1 file changed, 3 insertions(+)drivers/watchdog/aspeed_wdt.c +3
-0@@ -46,6 +46,7 @@ struct aspeed_wdt_config {#define WDT_RELOAD_VALUE 0x04#define WDT_RESTART 0x08#define WDT_CTRL 0x0C +#define WDT_CTRL_BOOT_SECONDARY BIT(7)#define WDT_CTRL_RESET_MODE_SOC (0x00 << 5)#define WDT_CTRL_RESET_MODE_FULL_CHIP (0x01 << 5)#define WDT_CTRL_RESET_MODE_ARM_CPU (0x10 << 5)@@ -245,6 +246,8 @@ static int aspeed_wdt_probe(struct platform_device *pdev) }if (of_property_read_bool (np,"aspeed,external-signal" )) wdt->ctrl |=WDT_CTRL_WDT_EXT ; +if (of_property_read_bool (np,"aspeed,alt-boot" )) + wdt->ctrl |=WDT_CTRL_BOOT_SECONDARY ;if (readl (wdt->base +WDT_CTRL ) &WDT_CTRL_ENABLE ) {/* Signature · 1 line
1.8.3.1
Re: [PATCH 2/2] watchdog: aspeed: Allow configuring for alternate boot
Quoted 11 lines · Guenter Roeck
Quoted 10 lines · Eddie James
From: Milton Miller <[email protected]>
Allow the device tree to specify a watchdog to fallover to the alternate boot source.
The aspeeed watchdog can set a latch directing flash chip select 0 to chip select 1, allowing boot from an alternate media if the watchdog is not reset in time. On the ast2400 bank 1 also goes to flash bank 1, while on the ast2500 the chip selects are swapped.
Quoted 6 lines
Quoted 2 lines
Signed-off-by Milton Miller <[email protected]>Signed-off-by Eddie James <[email protected]>This is already documented in the bindings document, so the property should be ok.
Reviewed-by Guenter Roeck <[email protected]>
Thanks. On a related note, our system has need of a way to determine if it's currently booted from the alternate chip or not. It seems the only way to tell is to look at the ast2400/2500 watchdog timeout status register. We can add some debugfs to the aspeed driver I guess, but I wonder if this is a common enough situation that the watchdog core could use some debugfs or sysfs for providing that info?
Thanks,
Eddie
Quoted 29 lines
Quoted 29 lines
--- drivers/watchdog/aspeed_wdt.c | 3 +++ 1 file changed, 3 insertions(+)drivers/watchdog/aspeed_wdt.c +3
-0@@ -46,6 +46,7 @@ struct aspeed_wdt_config {#define WDT_RELOAD_VALUE 0x04#define WDT_RESTART 0x08#define WDT_CTRL 0x0C +#define WDT_CTRL_BOOT_SECONDARY BIT(7)#define WDT_CTRL_RESET_MODE_SOC (0x00 << 5)#define WDT_CTRL_RESET_MODE_FULL_CHIP (0x01 << 5)#define WDT_CTRL_RESET_MODE_ARM_CPU (0x10 << 5)@@ -245,6 +246,8 @@ static int aspeed_wdt_probe(struct platform_device *pdev) }if (of_property_read_bool (np,"aspeed,external-signal" )) wdt->ctrl |=WDT_CTRL_WDT_EXT ; +if (of_property_read_bool (np,"aspeed,alt-boot" )) + wdt->ctrl |=WDT_CTRL_BOOT_SECONDARY ;if (readl (wdt->base +WDT_CTRL ) &WDT_CTRL_ENABLE ) {/* Signature · 1 line
1.8.3.1
Guenter Roeck
maintainerRe: [PATCH 2/2] watchdog: aspeed: Allow configuring for alternate boot
Quoted 19 lines · Eddie James
Quoted 16 lines · Guenter Roeck
Quoted 5 lines · Eddie James
From: Milton Miller <[email protected]>
Allow the device tree to specify a watchdog to fallover to the alternate boot source.
Quoted 7 lines
The aspeeed watchdog can set a latch directing flash chip select 0 to chip select 1, allowing boot from an alternate media if the watchdog is not reset in time. On the ast2400 bank 1 also goes to flash bank 1, while on the ast2500 the chip selects are swapped.
Signed-off-by Milton Miller <[email protected]>Signed-off-by Eddie James <[email protected]>This is already documented in the bindings document, so the property should be ok.
Quoted 9 lines
Quoted 1 line
Reviewed-by Guenter Roeck <[email protected]>Thanks. On a related note, our system has need of a way to determine if it's currently booted from the alternate chip or not. It seems the only way to tell is to look at the ast2400/2500 watchdog timeout status register. We can add some debugfs to the aspeed driver I guess, but I wonder if this is a common enough situation that the watchdog core could use some debugfs or sysfs for providing that info?
If you have the means to determine that the watchdog timed out in the previous boot, can you just set the WDIOF_CARDRESET status bit ?
Guenter
Quoted 32 lines
Thanks,
EddieQuoted 28 lines
Quoted 28 lines
--- drivers/watchdog/aspeed_wdt.c | 3 +++ 1 file changed, 3 insertions(+)drivers/watchdog/aspeed_wdt.c +3
-0@@ -46,6 +46,7 @@ struct aspeed_wdt_config {#define WDT_RELOAD_VALUE 0x04#define WDT_RESTART 0x08#define WDT_CTRL 0x0C +#define WDT_CTRL_BOOT_SECONDARY BIT(7)#define WDT_CTRL_RESET_MODE_SOC (0x00 << 5)#define WDT_CTRL_RESET_MODE_FULL_CHIP (0x01 << 5)#define WDT_CTRL_RESET_MODE_ARM_CPU (0x10 << 5)@@ -245,6 +246,8 @@ static int aspeed_wdt_probe(struct platform_device *pdev) }if (of_property_read_bool (np,"aspeed,external-signal" )) wdt->ctrl |=WDT_CTRL_WDT_EXT ; +if (of_property_read_bool (np,"aspeed,alt-boot" )) + wdt->ctrl |=WDT_CTRL_BOOT_SECONDARY ;if (readl (wdt->base +WDT_CTRL ) &WDT_CTRL_ENABLE ) {/* Signature · 1 line
1.8.3.1
Joel Stanley
maintainerRe: [PATCH 2/2] watchdog: aspeed: Allow configuring for alternate boot
In reply to Eddie JamesQuoted 29 lines · Eddie James
From: Milton Miller <[email protected]>
Allow the device tree to specify a watchdog to fallover to the alternate boot source.
The aspeeed watchdog can set a latch directing flash chip select 0 to chip select 1, allowing boot from an alternate media if the watchdog is not reset in time. On the ast2400 bank 1 also goes to flash bank 1, while on the ast2500 the chip selects are swapped.
Signed-off-by Milton Miller <[email protected]>Signed-off-by Eddie James <[email protected]>--- drivers/watchdog/aspeed_wdt.c | 3 +++ 1 file changed, 3 insertions(+)drivers/watchdog/aspeed_wdt.c +1
-0@@ -46,6 +46,7 @@ struct aspeed_wdt_config {#define WDT_RELOAD_VALUE 0x04#define WDT_RESTART 0x08#define WDT_CTRL 0x0C +#define WDT_CTRL_BOOT_SECONDARY BIT(7)#define WDT_CTRL_RESET_MODE_SOC (0x00 << 5)#define WDT_CTRL_RESET_MODE_FULL_CHIP (0x01 << 5)#define WDT_CTRL_RESET_MODE_ARM_CPU (0x10 << 5)
drivers/watchdog/aspeed_wdt.c static int aspeed_wdt_probe(struct platform_device *pdev) · line 250
Quoted 5 lines
drivers/watchdog/aspeed_wdt.c +2
-0@@ -245,6 +246,8 @@ static int aspeed_wdt_probe(struct platform_device *pdev) }if (of_property_read_bool (np,"aspeed,external-signal" )) wdt->ctrl |=WDT_CTRL_WDT_EXT ; +if (of_property_read_bool (np,"aspeed,alt-boot" )) + wdt->ctrl |=WDT_CTRL_BOOT_SECONDARY ;
If a user sets this property on the only watchdog in the system, then they will trigger this behaviour when doing a normal 'reboot'. That would not be desirable.
We could mitigate this by:
- not registering the watchdog as a reboot device if this property is enabled
- clearing the WDT_CTRL_BOOT_SECONDARY bit in the aspeed_wdt_restart path
The second option is neater as it allows the watchdog to behave normally when only one is enabled. The first would be confusing if a system didn't have more than one watchdog enabled.
Cheers,
Joel