[PATCH] board_init: Do not reserve MALLOC_F area on stack if non-zero MALLOC_F_ADDR
In case the MALLOC_F_ADDR is set to non-zero value, the early malloc area is not going to be placed just below stack top, but elsewhere. Do not reserve MALLOC_F bytes in this case, as that wastes stack space and may even cause insufficient stack space in SPL.
This functionality is particularly useful on i.MX8M, where the insufficient stack space can be triggered.
Signed-off-by: Marek Vasut marex@denx.de Cc: Albert ARIBAUD albert.u.boot@aribaud.net Cc: Fabio Estevam festevam@denx.de Cc: Peng Fan peng.fan@nxp.com Cc: Simon Glass sjg@chromium.org Cc: Stefano Babic sbabic@denx.de Cc: Thomas Chou thomas@wytron.com.tw Cc: Tom Rini trini@konsulko.com --- common/init/board_init.c | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/common/init/board_init.c b/common/init/board_init.c index eab5ee13953..6a550261778 100644 --- a/common/init/board_init.c +++ b/common/init/board_init.c @@ -78,8 +78,10 @@ __weak void board_init_f_init_stack_protection(void) ulong board_init_f_alloc_reserve(ulong top) { /* Reserve early malloc arena */ +#ifndef CONFIG_MALLOC_F_ADDR #if CONFIG_VAL(SYS_MALLOC_F_LEN) top -= CONFIG_VAL(SYS_MALLOC_F_LEN); +#endif #endif /* LAST : reserve GD (rounded up to a multiple of 16 bytes) */ top = rounddown(top-sizeof(struct global_data), 16);
Hi Marek,
On Sat, 25 Jun 2022 at 11:58, Marek Vasut marex@denx.de wrote:
In case the MALLOC_F_ADDR is set to non-zero value, the early malloc area is not going to be placed just below stack top, but elsewhere. Do not reserve MALLOC_F bytes in this case, as that wastes stack space and may even cause insufficient stack space in SPL.
This functionality is particularly useful on i.MX8M, where the insufficient stack space can be triggered.
Signed-off-by: Marek Vasut marex@denx.de Cc: Albert ARIBAUD albert.u.boot@aribaud.net Cc: Fabio Estevam festevam@denx.de Cc: Peng Fan peng.fan@nxp.com Cc: Simon Glass sjg@chromium.org Cc: Stefano Babic sbabic@denx.de Cc: Thomas Chou thomas@wytron.com.tw Cc: Tom Rini trini@konsulko.com
common/init/board_init.c | 2 ++ 1 file changed, 2 insertions(+)
Please can you migrate the option to Kconfig first? I suspect that will simplify the logic and avoid the #ifndef
diff --git a/common/init/board_init.c b/common/init/board_init.c index eab5ee13953..6a550261778 100644 --- a/common/init/board_init.c +++ b/common/init/board_init.c @@ -78,8 +78,10 @@ __weak void board_init_f_init_stack_protection(void) ulong board_init_f_alloc_reserve(ulong top) { /* Reserve early malloc arena */ +#ifndef CONFIG_MALLOC_F_ADDR #if CONFIG_VAL(SYS_MALLOC_F_LEN) top -= CONFIG_VAL(SYS_MALLOC_F_LEN); +#endif #endif /* LAST : reserve GD (rounded up to a multiple of 16 bytes) */ top = rounddown(top-sizeof(struct global_data), 16); -- 2.35.1
Regards, Simon
On Thu, Jun 30, 2022 at 04:06:22AM -0600, Simon Glass wrote:
Hi Marek,
On Sat, 25 Jun 2022 at 11:58, Marek Vasut marex@denx.de wrote:
In case the MALLOC_F_ADDR is set to non-zero value, the early malloc area is not going to be placed just below stack top, but elsewhere. Do not reserve MALLOC_F bytes in this case, as that wastes stack space and may even cause insufficient stack space in SPL.
This functionality is particularly useful on i.MX8M, where the insufficient stack space can be triggered.
Signed-off-by: Marek Vasut marex@denx.de Cc: Albert ARIBAUD albert.u.boot@aribaud.net Cc: Fabio Estevam festevam@denx.de Cc: Peng Fan peng.fan@nxp.com Cc: Simon Glass sjg@chromium.org Cc: Stefano Babic sbabic@denx.de Cc: Thomas Chou thomas@wytron.com.tw Cc: Tom Rini trini@konsulko.com
common/init/board_init.c | 2 ++ 1 file changed, 2 insertions(+)
Please can you migrate the option to Kconfig first? I suspect that will simplify the logic and avoid the #ifndef
That was my first reaction as well. But, I'm not so sure after looking at things more. An issue is that CONFIG_MALLOC_F_ADDR should probably be CONFIG_SPL_MALLOC_F_ADDR as it's only used in SPL. But it's also not an either/or around CONFIG_SPL_SYS_MALLOC_F_LEN.
It would be good to move to Kconfig (but I also don't see a common default for the handful of SoCs using it), and maybe a !CONFIG_VAL() test instead of ifndef.
diff --git a/common/init/board_init.c b/common/init/board_init.c index eab5ee13953..6a550261778 100644 --- a/common/init/board_init.c +++ b/common/init/board_init.c @@ -78,8 +78,10 @@ __weak void board_init_f_init_stack_protection(void) ulong board_init_f_alloc_reserve(ulong top) { /* Reserve early malloc arena */ +#ifndef CONFIG_MALLOC_F_ADDR #if CONFIG_VAL(SYS_MALLOC_F_LEN) top -= CONFIG_VAL(SYS_MALLOC_F_LEN); +#endif #endif /* LAST : reserve GD (rounded up to a multiple of 16 bytes) */ top = rounddown(top-sizeof(struct global_data), 16); -- 2.35.1
Regards, Simon
On 6/30/22 15:36, Tom Rini wrote:
Hi,
[...]
In case the MALLOC_F_ADDR is set to non-zero value, the early malloc area is not going to be placed just below stack top, but elsewhere. Do not reserve MALLOC_F bytes in this case, as that wastes stack space and may even cause insufficient stack space in SPL.
This functionality is particularly useful on i.MX8M, where the insufficient stack space can be triggered.
[...]
common/init/board_init.c | 2 ++ 1 file changed, 2 insertions(+)
Please can you migrate the option to Kconfig first? I suspect that will simplify the logic and avoid the #ifndef
That was my first reaction as well. But, I'm not so sure after looking at things more. An issue is that CONFIG_MALLOC_F_ADDR should probably be CONFIG_SPL_MALLOC_F_ADDR as it's only used in SPL. But it's also not an either/or around CONFIG_SPL_SYS_MALLOC_F_LEN.
It would be good to move to Kconfig (but I also don't see a common default for the handful of SoCs using it), and maybe a !CONFIG_VAL() test instead of ifndef.
See [PATCH] board_init: Convert CONFIG_MALLOC_F_ADDR to Kconfig
On Sat, Jun 25, 2022 at 07:58:24PM +0200, Marek Vasut wrote:
In case the MALLOC_F_ADDR is set to non-zero value, the early malloc area is not going to be placed just below stack top, but elsewhere. Do not reserve MALLOC_F bytes in this case, as that wastes stack space and may even cause insufficient stack space in SPL.
This functionality is particularly useful on i.MX8M, where the insufficient stack space can be triggered.
Signed-off-by: Marek Vasut marex@denx.de Cc: Albert ARIBAUD albert.u.boot@aribaud.net Cc: Fabio Estevam festevam@denx.de Cc: Peng Fan peng.fan@nxp.com Cc: Simon Glass sjg@chromium.org Cc: Stefano Babic sbabic@denx.de Cc: Thomas Chou thomas@wytron.com.tw Cc: Tom Rini trini@konsulko.com
Applied to u-boot/next, thanks!
participants (3)
-
Marek Vasut -
Simon Glass -
Tom Rini