ARM: reset: Move SYSRESET condition from Makefile into source file
RFC PATCH 1/1 ARM: reset: Move SYSRESET condition from Makefile into source file
In case CONFIG_SYSRESET is set, do_reset from reset.c will not be available anywere, even if SYSRESET is disabled for SPL in the board specific header file like this:
#if defined(CONFIG_SPL_BUILD)
#undef CONFIG_WDT
#undef CONFIG_WATCHDOG
#undef CONFIG_SYSRESET
#define CONFIG_HW_WATCHDOG
#endif'do_reset' is called from SPL for instance from the panic handler in case SPL_USB_SDP is enabled and PANIC_HANG is not set.
Setting PANIC_HANG would solve this issue, but it also changes the behavior in case a panic occurs.
--- arch/arm/lib/Makefile | 2 -- arch/arm/lib/reset.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-)
arch/arm/lib/Makefile +0 -2
@@ -56,9 +56,7 @@ obj-y += interrupts_64.o else obj-y += interrupts.o endif -ifndef CONFIG_SYSRESET obj-y += reset.o -endif obj-y += cache.o obj-$(CONFIG_SYS_ARM_CACHE_CP15) += cache-cp15.o
arch/arm/lib/reset.c +2 -0
@@ -22,6 +22,7 @@#include <common.h> +#if !defined(CONFIG_SYSRESET )__weak void reset_misc (void ) { }
@@ -40,3 +41,4 @@ int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/*NOTREACHED*/ return 0; } +#endif
Signature · 1 line
2.21.0
RFC PATCH 1/1 ARM: reset: Move SYSRESET condition from Makefile into source file
Quoted 50 lines · Claudius Heine
In case
CONFIG_SYSRESETis set,do_resetfromreset.cwill not be available anywere, even if SYSRESET is disabled for SPL in the board specific header file like this:#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif'
do_reset' is called from SPL for instance from the panic handler in caseSPL_USB_SDPis enabled andPANIC_HANGis not set.Setting
PANIC_HANGwould solve this issue, but it also changes the behavior in case a panic occurs.Signed-off-by Claudius Heine <[email protected]>--- arch/arm/lib/Makefile | 2 -- arch/arm/lib/reset.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-)arch/arm/lib/Makefile +0
-2@@ -56,9 +56,7 @@ obj-y += interrupts_64.o else obj-y += interrupts.o endif -ifndef CONFIG_SYSRESET obj-y += reset.o -endif obj-y += cache.o obj-$(CONFIG_SYS_ARM_CACHE_CP15) += cache-cp15.oarch/arm/lib/reset.c +1
-0@@ -22,6 +22,7 @@#include <common.h> +#if !defined(CONFIG_SYSRESET )__weak void reset_misc (void ) { }
arch/arm/lib/reset.c int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 44
Quoted 4 lines
arch/arm/lib/reset.c +1
-0@@ -40,3 +41,4 @@ int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/*NOTREACHED*/ return 0; } +#endif
Does this mean there's now one huge ifdef around the entire source file? That's odd.
RFC PATCH 1/1 ARM: reset: Move SYSRESET condition from Makefile into source file
Hi Marek,
Quoted 46 lines · Marek Vasut
Quoted 45 lines · Claudius Heine
In case
CONFIG_SYSRESETis set,do_resetfromreset.cwill not be available anywere, even if SYSRESET is disabled for SPL in the board specific header file like this:#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif'
do_reset' is called from SPL for instance from the panic handler in caseSPL_USB_SDPis enabled andPANIC_HANGis not set.Setting
PANIC_HANGwould solve this issue, but it also changes the behavior in case a panic occurs.Signed-off-by Claudius Heine <[email protected]>--- arch/arm/lib/Makefile | 2 -- arch/arm/lib/reset.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-)arch/arm/lib/Makefile +0
-2@@ -56,9 +56,7 @@ obj-y += interrupts_64.o else obj-y += interrupts.o endif -ifndef CONFIG_SYSRESET obj-y += reset.o -endif obj-y += cache.o obj-$(CONFIG_SYS_ARM_CACHE_CP15) += cache-cp15.oarch/arm/lib/reset.c +0
-0@@ -22,6 +22,7 @@#include <common.h>
Quoted 12 lines
arch/arm/lib/reset.c int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 44
Quoted 9 lines
arch/arm/lib/reset.c +2
-0+#if !defined(CONFIG_SYSRESET )__weak void reset_misc (void ) { }@@ -40,3 +41,4 @@ int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/*NOTREACHED*/ return 0; } +#endif Does this mean there's now one huge ifdef around the entire source file? That's odd.
Right. Other suggestions?
Maybe having 'do_reset' here as a weak instead, so that sysreset can overwrite it? But then the other definitions in arch/*/lib/reset.c should probably be the same for consistency sake?
I tried with this patch not to change anything in case SYSRESET is enabled too much and since the file isn't too large I thought that would be ok for now.
regards
Signature · 6 lines
DENX Software Engineering GmbH, Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-54 Fax: (+49)-8142-66989-80 Email: ch at denx.de
PGP key: 6FF2 E59F 00C6 BC28 31D8 64C1 1173 CB19 9808 B153
Keyserver: hkp://pool.sks-keyservers.netRFC PATCH 1/1 ARM: reset: Move SYSRESET condition from Makefile into source file
Quoted 64 lines · Claudius Heine
Hi Marek,
Quoted 46 lines · Marek Vasut
Quoted 45 lines · Claudius Heine
In case
CONFIG_SYSRESETis set,do_resetfromreset.cwill not be available anywere, even if SYSRESET is disabled for SPL in the board specific header file like this:#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif'
do_reset' is called from SPL for instance from the panic handler in caseSPL_USB_SDPis enabled andPANIC_HANGis not set.Setting
PANIC_HANGwould solve this issue, but it also changes the behavior in case a panic occurs.Signed-off-by Claudius Heine <[email protected]>--- arch/arm/lib/Makefile | 2 -- arch/arm/lib/reset.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-)arch/arm/lib/Makefile +0
-2@@ -56,9 +56,7 @@ obj-y += interrupts_64.o else obj-y += interrupts.o endif -ifndef CONFIG_SYSRESET obj-y += reset.o -endif obj-y += cache.o obj-$(CONFIG_SYS_ARM_CACHE_CP15) += cache-cp15.oarch/arm/lib/reset.c +0
-0@@ -22,6 +22,7 @@#include <common.h> Quoted 12 lines
arch/arm/lib/reset.c int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 44
Quoted 9 lines
arch/arm/lib/reset.c +2
-0+#if !defined(CONFIG_SYSRESET )__weak void reset_misc (void ) { }@@ -40,3 +41,4 @@ int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/*NOTREACHED*/ return 0; } +#endif Does this mean there's now one huge ifdef around the entire source file? That's odd.
Right. Other suggestions?
Quoted 7 lines
Maybe having '
do_reset' here as a weak instead, so that sysreset can overwrite it? But then the other definitions inarch/*/lib/reset.cshould probably be the same for consistency sake?I tried with this patch not to change anything in case SYSRESET is enabled too much and since the file isn't too large I thought that would be ok for now.
What if sysreset implemented do_reset ? Wouldn't that solve the issue ?
RFC PATCH 1/1 ARM: reset: Move SYSRESET condition from Makefile into source file
Quoted 69 lines · Marek Vasut
Quoted 68 lines · Claudius Heine
Hi Marek,
Quoted 46 lines · Marek Vasut
Quoted 45 lines · Claudius Heine
In case
CONFIG_SYSRESETis set,do_resetfromreset.cwill not be available anywere, even if SYSRESET is disabled for SPL in the board specific header file like this:#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif'
do_reset' is called from SPL for instance from the panic handler in caseSPL_USB_SDPis enabled andPANIC_HANGis not set.Setting
PANIC_HANGwould solve this issue, but it also changes the behavior in case a panic occurs.Signed-off-by Claudius Heine <[email protected]>--- arch/arm/lib/Makefile | 2 -- arch/arm/lib/reset.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-)arch/arm/lib/Makefile +0
-2@@ -56,9 +56,7 @@ obj-y += interrupts_64.o else obj-y += interrupts.o endif -ifndef CONFIG_SYSRESET obj-y += reset.o -endif obj-y += cache.o obj-$(CONFIG_SYS_ARM_CACHE_CP15) += cache-cp15.oarch/arm/lib/reset.c +0
-0@@ -22,6 +22,7 @@#include <common.h> Quoted 12 lines
arch/arm/lib/reset.c int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 44
Quoted 9 lines
arch/arm/lib/reset.c +2
-0+#if !defined(CONFIG_SYSRESET )__weak void reset_misc (void ) { }@@ -40,3 +41,4 @@ int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/*NOTREACHED*/ return 0; } +#endif Does this mean there's now one huge ifdef around the entire source file? That's odd.
Right. Other suggestions?
Maybe having '
do_reset' here as a weak instead, so that sysreset can overwrite it? But then the other definitions inarch/*/lib/reset.cshould probably be the same for consistency sake?
Quoted 5 lines
Quoted 3 lines
I tried with this patch not to change anything in case SYSRESET is enabled too much and since the file isn't too large I thought that would be ok for now.
What if sysreset implemented
do_reset? Wouldn't that solve the issue ?
Not sure what you mean... sysreset implements do_reset:
https://gitlab.denx.de/u-boot/u-boot/blob/master/drivers/sysreset/sysreset-uclass.c#L112
But the SPL does not have sysreset in this case, so it needs something different.
Signature · 6 lines
DENX Software Engineering GmbH, Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-54 Fax: (+49)-8142-66989-80 Email: ch at denx.de
PGP key: 6FF2 E59F 00C6 BC28 31D8 64C1 1173 CB19 9808 B153
Keyserver: hkp://pool.sks-keyservers.netRFC PATCH 1/1 ARM: reset: Move SYSRESET condition from Makefile into source file
Quoted 74 lines · Claudius Heine
Quoted 73 lines · Marek Vasut
Quoted 71 lines · Claudius Heine
Hi Marek,
Quoted 46 lines · Marek Vasut
Quoted 45 lines · Claudius Heine
In case
CONFIG_SYSRESETis set,do_resetfromreset.cwill not be available anywere, even if SYSRESET is disabled for SPL in the board specific header file like this:#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif'
do_reset' is called from SPL for instance from the panic handler in caseSPL_USB_SDPis enabled andPANIC_HANGis not set.Setting
PANIC_HANGwould solve this issue, but it also changes the behavior in case a panic occurs.Signed-off-by Claudius Heine <[email protected]>--- arch/arm/lib/Makefile | 2 -- arch/arm/lib/reset.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-)arch/arm/lib/Makefile +0
-2@@ -56,9 +56,7 @@ obj-y += interrupts_64.o else obj-y += interrupts.o endif -ifndef CONFIG_SYSRESET obj-y += reset.o -endif obj-y += cache.o obj-$(CONFIG_SYS_ARM_CACHE_CP15) += cache-cp15.oarch/arm/lib/reset.c +0
-0@@ -22,6 +22,7 @@#include <common.h> Quoted 12 lines
arch/arm/lib/reset.c int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 44
Quoted 9 lines
arch/arm/lib/reset.c +2
-0+#if !defined(CONFIG_SYSRESET )__weak void reset_misc (void ) { }@@ -40,3 +41,4 @@ int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/*NOTREACHED*/ return 0; } +#endif Does this mean there's now one huge ifdef around the entire source file? That's odd.
Right. Other suggestions?
Maybe having '
do_reset' here as a weak instead, so that sysreset can overwrite it? But then the other definitions inarch/*/lib/reset.cshould probably be the same for consistency sake?I tried with this patch not to change anything in case SYSRESET is enabled too much and since the file isn't too large I thought that would be ok for now.
Quoted 8 lines
Quoted 1 line
What if sysreset implemented
do_reset? Wouldn't that solve the issue ?Not sure what you mean... sysreset implements
do_reset:https://gitlab.denx.de/u-boot/u-boot/blob/master/drivers/sysreset/sysreset-uclass.c#L112
But the SPL does not have sysreset in this case, so it needs something different.
Oh, so you need CONFIG_$(SPL_TPL_)SYSRESET then ?
RFC PATCH 1/1 ARM: reset: Move SYSRESET condition from Makefile into source file
Quoted 77 lines · Marek Vasut
Quoted 76 lines · Claudius Heine
Quoted 74 lines · Marek Vasut
Quoted 64 lines · Claudius Heine
Hi Marek,
Quoted 46 lines · Marek Vasut
Quoted 45 lines · Claudius Heine
In case
CONFIG_SYSRESETis set,do_resetfromreset.cwill not be available anywere, even if SYSRESET is disabled for SPL in the board specific header file like this:#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif'
do_reset' is called from SPL for instance from the panic handler in caseSPL_USB_SDPis enabled andPANIC_HANGis not set.Setting
PANIC_HANGwould solve this issue, but it also changes the behavior in case a panic occurs.Signed-off-by Claudius Heine <[email protected]>--- arch/arm/lib/Makefile | 2 -- arch/arm/lib/reset.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-)arch/arm/lib/Makefile +0
-2@@ -56,9 +56,7 @@ obj-y += interrupts_64.o else obj-y += interrupts.o endif -ifndef CONFIG_SYSRESET obj-y += reset.o -endif obj-y += cache.o obj-$(CONFIG_SYS_ARM_CACHE_CP15) += cache-cp15.oarch/arm/lib/reset.c +0
-0@@ -22,6 +22,7 @@#include <common.h> Quoted 12 lines
arch/arm/lib/reset.c int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 44
Quoted 9 lines
arch/arm/lib/reset.c +2
-0+#if !defined(CONFIG_SYSRESET )__weak void reset_misc (void ) { }@@ -40,3 +41,4 @@ int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/*NOTREACHED*/ return 0; } +#endif Does this mean there's now one huge ifdef around the entire source file? That's odd.
Right. Other suggestions?
Quoted 7 lines
Maybe having '
do_reset' here as a weak instead, so that sysreset can overwrite it? But then the other definitions inarch/*/lib/reset.cshould probably be the same for consistency sake?I tried with this patch not to change anything in case SYSRESET is enabled too much and since the file isn't too large I thought that would be ok for now.
What if sysreset implemented
do_reset? Wouldn't that solve the issue ?
Quoted 8 lines
Quoted 6 lines
Not sure what you mean... sysreset implements
do_reset:https://gitlab.denx.de/u-boot/u-boot/blob/master/drivers/sysreset/sysreset-uclass.c#L112
But the SPL does not have sysreset in this case, so it needs something different.
Oh, so you need CONFIG_$(SPL_TPL_)SYSRESET then ?
Well that would probably not enough. I would also need settings for the watchdog, because the SPL does not have DM support, so while u-boot uses CONFIG_WATCHDOG the SPL uses CONFIG_HW_WATCHDOG.
Easier that changing all this is something like this in the board header file (as I described in the commit description):
#if defined(CONFIG_SPL_BUILD)
#undef CONFIG_WDT
#undef CONFIG_WATCHDOG
#undef CONFIG_SYSRESET
#define CONFIG_HW_WATCHDOG
#endifIn case of imx6, that way the SPL uses the hw_watchdog_reset from the imx watchdog driver instead of the 'watchdog_reset'.
'watchdog_reset' is not available since that is implemented in wdt-uclass.c and CONFIG_SPL_WDT depends on SPL_DM.
RFC PATCH 1/1 ARM: reset: Move SYSRESET condition from Makefile into source file
Quoted 94 lines · Claudius Heine
Quoted 77 lines · Marek Vasut
Quoted 76 lines · Claudius Heine
Quoted 74 lines · Marek Vasut
Quoted 64 lines · Claudius Heine
Hi Marek,
Quoted 46 lines · Marek Vasut
Quoted 45 lines · Claudius Heine
In case
CONFIG_SYSRESETis set,do_resetfromreset.cwill not be available anywere, even if SYSRESET is disabled for SPL in the board specific header file like this:#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif'
do_reset' is called from SPL for instance from the panic handler in caseSPL_USB_SDPis enabled andPANIC_HANGis not set.Setting
PANIC_HANGwould solve this issue, but it also changes the behavior in case a panic occurs.Signed-off-by Claudius Heine <[email protected]>--- arch/arm/lib/Makefile | 2 -- arch/arm/lib/reset.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-)arch/arm/lib/Makefile +0
-2@@ -56,9 +56,7 @@ obj-y += interrupts_64.o else obj-y += interrupts.o endif -ifndef CONFIG_SYSRESET obj-y += reset.o -endif obj-y += cache.o obj-$(CONFIG_SYS_ARM_CACHE_CP15) += cache-cp15.oarch/arm/lib/reset.c +0
-0@@ -22,6 +22,7 @@#include <common.h> Quoted 12 lines
arch/arm/lib/reset.c int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 44
Quoted 9 lines
arch/arm/lib/reset.c +2
-0+#if !defined(CONFIG_SYSRESET )__weak void reset_misc (void ) { }@@ -40,3 +41,4 @@ int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/*NOTREACHED*/ return 0; } +#endif Does this mean there's now one huge ifdef around the entire source file? That's odd.
Right. Other suggestions?
Quoted 7 lines
Maybe having '
do_reset' here as a weak instead, so that sysreset can overwrite it? But then the other definitions inarch/*/lib/reset.cshould probably be the same for consistency sake?I tried with this patch not to change anything in case SYSRESET is enabled too much and since the file isn't too large I thought that would be ok for now.
What if sysreset implemented
do_reset? Wouldn't that solve the issue ?Quoted 8 lines
Quoted 6 lines
Not sure what you mean... sysreset implements
do_reset:https://gitlab.denx.de/u-boot/u-boot/blob/master/drivers/sysreset/sysreset-uclass.c#L112
But the SPL does not have sysreset in this case, so it needs something different.
Oh, so you need CONFIG_$(SPL_TPL_)SYSRESET then ?
Well that would probably not enough. I would also need settings for the watchdog, because the SPL does not have DM support, so while u-boot uses
CONFIG_WATCHDOGthe SPL usesCONFIG_HW_WATCHDOG.Easier that changing all this is something like this in the board header file (as I described in the commit description):
Quoted 6 lines
#if defined(CONFIG_SPL_BUILD)
#undef CONFIG_WDT
#undef CONFIG_WATCHDOG
#undef CONFIG_SYSRESET
#define CONFIG_HW_WATCHDOG
#endifCan't we add DM watchdog to SPL instead ?
Quoted 5 lines
In case of imx6, that way the SPL uses the
hw_watchdog_resetfrom the imx watchdog driver instead of the 'watchdog_reset'.'
watchdog_reset' is not available since that is implemented inwdt-uclass.candCONFIG_SPL_WDTdepends onSPL_DM.
RFC PATCH 1/1 ARM: reset: Move SYSRESET condition from Makefile into source file
Quoted 95 lines · Marek Vasut
Quoted 94 lines · Claudius Heine
Quoted 77 lines · Marek Vasut
Quoted 76 lines · Claudius Heine
Quoted 74 lines · Marek Vasut
Quoted 64 lines · Claudius Heine
Hi Marek,
Quoted 46 lines · Marek Vasut
Quoted 45 lines · Claudius Heine
In case
CONFIG_SYSRESETis set,do_resetfromreset.cwill not be available anywere, even if SYSRESET is disabled for SPL in the board specific header file like this:#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif'
do_reset' is called from SPL for instance from the panic handler in caseSPL_USB_SDPis enabled andPANIC_HANGis not set.Setting
PANIC_HANGwould solve this issue, but it also changes the behavior in case a panic occurs.Signed-off-by Claudius Heine <[email protected]>--- arch/arm/lib/Makefile | 2 -- arch/arm/lib/reset.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-)arch/arm/lib/Makefile +0
-2@@ -56,9 +56,7 @@ obj-y += interrupts_64.o else obj-y += interrupts.o endif -ifndef CONFIG_SYSRESET obj-y += reset.o -endif obj-y += cache.o obj-$(CONFIG_SYS_ARM_CACHE_CP15) += cache-cp15.oarch/arm/lib/reset.c +0
-0@@ -22,6 +22,7 @@#include <common.h> Quoted 12 lines
arch/arm/lib/reset.c int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 44
Quoted 9 lines
arch/arm/lib/reset.c +2
-0+#if !defined(CONFIG_SYSRESET )__weak void reset_misc (void ) { }@@ -40,3 +41,4 @@ int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/*NOTREACHED*/ return 0; } +#endif Does this mean there's now one huge ifdef around the entire source file? That's odd.
Right. Other suggestions?
Quoted 7 lines
Maybe having '
do_reset' here as a weak instead, so that sysreset can overwrite it? But then the other definitions inarch/*/lib/reset.cshould probably be the same for consistency sake?I tried with this patch not to change anything in case SYSRESET is enabled too much and since the file isn't too large I thought that would be ok for now.
What if sysreset implemented
do_reset? Wouldn't that solve the issue ?Quoted 8 lines
Quoted 6 lines
Not sure what you mean... sysreset implements
do_reset:https://gitlab.denx.de/u-boot/u-boot/blob/master/drivers/sysreset/sysreset-uclass.c#L112
But the SPL does not have sysreset in this case, so it needs something different.
Oh, so you need CONFIG_$(SPL_TPL_)SYSRESET then ?
Well that would probably not enough. I would also need settings for the watchdog, because the SPL does not have DM support, so while u-boot uses
CONFIG_WATCHDOGthe SPL usesCONFIG_HW_WATCHDOG.Easier that changing all this is something like this in the board header file (as I described in the commit description):
Quoted 8 lines
Quoted 6 lines
#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endifCan't we add DM watchdog to SPL instead ?
Do you mean implementing SPL_DM support for this board so that we can use the DM watchdog and sysreset?
Well you know more about this than me. But it probably comes down to technical limitations like size of the SPL.
Lukasz has done something similar with the display5 board [1], but that uses PANIC_HANG to avoid this issue.
Quoted 6 lines
Quoted 6 lines
In case of imx6, that way the SPL uses the
hw_watchdog_resetfrom the imx watchdog driver instead of the 'watchdog_reset'.'
watchdog_reset' is not available since that is implemented inwdt-uclass.candCONFIG_SPL_WDTdepends onSPL_DM
RFC PATCH 1/1 ARM: reset: Move SYSRESET condition from Makefile into source file
Quoted 103 lines · Claudius Heine
Quoted 102 lines · Marek Vasut
Quoted 100 lines · Claudius Heine
Quoted 77 lines · Marek Vasut
Quoted 76 lines · Claudius Heine
Quoted 74 lines · Marek Vasut
Quoted 64 lines · Claudius Heine
Hi Marek,
Quoted 46 lines · Marek Vasut
Quoted 45 lines · Claudius Heine
In case
CONFIG_SYSRESETis set,do_resetfromreset.cwill not be available anywere, even if SYSRESET is disabled for SPL in the board specific header file like this:#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif'
do_reset' is called from SPL for instance from the panic handler in caseSPL_USB_SDPis enabled andPANIC_HANGis not set.Setting
PANIC_HANGwould solve this issue, but it also changes the behavior in case a panic occurs.Signed-off-by Claudius Heine <[email protected]>--- arch/arm/lib/Makefile | 2 -- arch/arm/lib/reset.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-)arch/arm/lib/Makefile +0
-2@@ -56,9 +56,7 @@ obj-y += interrupts_64.o else obj-y += interrupts.o endif -ifndef CONFIG_SYSRESET obj-y += reset.o -endif obj-y += cache.o obj-$(CONFIG_SYS_ARM_CACHE_CP15) += cache-cp15.oarch/arm/lib/reset.c +0
-0@@ -22,6 +22,7 @@#include <common.h> Quoted 12 lines
arch/arm/lib/reset.c int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 44
Quoted 9 lines
arch/arm/lib/reset.c +2
-0+#if !defined(CONFIG_SYSRESET )__weak void reset_misc (void ) { }@@ -40,3 +41,4 @@ int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/*NOTREACHED*/ return 0; } +#endif Does this mean there's now one huge ifdef around the entire source file? That's odd.
Right. Other suggestions?
Quoted 7 lines
Maybe having '
do_reset' here as a weak instead, so that sysreset can overwrite it? But then the other definitions inarch/*/lib/reset.cshould probably be the same for consistency sake?I tried with this patch not to change anything in case SYSRESET is enabled too much and since the file isn't too large I thought that would be ok for now.
What if sysreset implemented
do_reset? Wouldn't that solve the issue ?Quoted 8 lines
Quoted 6 lines
Not sure what you mean... sysreset implements
do_reset:https://gitlab.denx.de/u-boot/u-boot/blob/master/drivers/sysreset/sysreset-uclass.c#L112
But the SPL does not have sysreset in this case, so it needs something different.
Oh, so you need CONFIG_$(SPL_TPL_)SYSRESET then ?
Well that would probably not enough. I would also need settings for the watchdog, because the SPL does not have DM support, so while u-boot uses
CONFIG_WATCHDOGthe SPL usesCONFIG_HW_WATCHDOG.Easier that changing all this is something like this in the board header file (as I described in the commit description):
#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif
Quoted 4 lines
Quoted 1 line
Can't we add DM watchdog to SPL instead ?
Do you mean implementing
SPL_DMsupport for this board so that we can use the DM watchdog and sysreset?
Isn't SPL DM already implemented ?
[...]
RFC PATCH 1/1 ARM: reset: Move SYSRESET condition from Makefile into source file
In reply to Claudius HeineQuoted 102 lines · Claudius Heine
Quoted 77 lines · Marek Vasut
Quoted 76 lines · Claudius Heine
Quoted 74 lines · Marek Vasut
Quoted 64 lines · Claudius Heine
Hi Marek,
Quoted 46 lines · Marek Vasut
Quoted 45 lines · Claudius Heine
In case
CONFIG_SYSRESETis set,do_resetfromreset.cwill not be available anywere, even if SYSRESET is disabled for SPL in the board specific header file like this:#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif'
do_reset' is called from SPL for instance from the panic handler in caseSPL_USB_SDPis enabled andPANIC_HANGis not set.Setting
PANIC_HANGwould solve this issue, but it also changes the behavior in case a panic occurs.Signed-off-by Claudius Heine <[email protected]>--- arch/arm/lib/Makefile | 2 -- arch/arm/lib/reset.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-)arch/arm/lib/Makefile +0
-2@@ -56,9 +56,7 @@ obj-y += interrupts_64.o else obj-y += interrupts.o endif -ifndef CONFIG_SYSRESET obj-y += reset.o -endif obj-y += cache.o obj-$(CONFIG_SYS_ARM_CACHE_CP15) += cache-cp15.oarch/arm/lib/reset.c +0
-0@@ -22,6 +22,7 @@#include <common.h> Quoted 12 lines
arch/arm/lib/reset.c int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 44
Quoted 9 lines
arch/arm/lib/reset.c +2
-0+#if !defined(CONFIG_SYSRESET )__weak void reset_misc (void ) { }@@ -40,3 +41,4 @@ int do_reset(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])/*NOTREACHED*/ return 0; } +#endif Does this mean there's now one huge ifdef around the entire source file? That's odd.
Right. Other suggestions?
Quoted 7 lines
Maybe having '
do_reset' here as a weak instead, so that sysreset can overwrite it? But then the other definitions inarch/*/lib/reset.cshould probably be the same for consistency sake?I tried with this patch not to change anything in case SYSRESET is enabled too much and since the file isn't too large I thought that would be ok for now.
What if sysreset implemented
do_reset? Wouldn't that solve the issue ?Quoted 8 lines
Quoted 6 lines
Not sure what you mean... sysreset implements
do_reset:https://gitlab.denx.de/u-boot/u-boot/blob/master/drivers/sysreset/sysreset-uclass.c#L112
But the SPL does not have sysreset in this case, so it needs something different.
Oh, so you need CONFIG_$(SPL_TPL_)SYSRESET then ?
Well that would probably not enough. I would also need settings for the watchdog, because the SPL does not have DM support, so while u-boot uses
CONFIG_WATCHDOGthe SPL usesCONFIG_HW_WATCHDOG.Easier that changing all this is something like this in the board header file (as I described in the commit description):
#if defined(CONFIG_SPL_BUILD) #undef CONFIG_WDT #undef CONFIG_WATCHDOG #undef CONFIG_SYSRESET #define CONFIG_HW_WATCHDOG #endif
Quoted 5 lines
In case of imx6, that way the SPL uses the
hw_watchdog_resetfrom the imx watchdog driver instead of the 'watchdog_reset'.'
watchdog_reset' is not available since that is implemented inwdt-uclass.candCONFIG_SPL_WDTdepends onSPL_DM.
That seems totally unrelated to this patch.
I think this patch should change the Makefile to use CONFIG_$(SPL_TPL_)SYSRESET and you need to solve your watchdog issue in a separate patch.
Regards,
Simon