On Wed, Jun 07, 2023 at 09:58:41AM +0530, Vignesh Raghavendra wrote:
> If early stack usage is exactly same as SIZE_LIMIT_PROVIDE_STACK, then
> its most likely to have overflowed beyond this limit and corrupted any
> regions beyond stack. Warn about this to the user.
> 
> Signed-off-by: Vignesh Raghavendra <vigne...@ti.com>
> ---
> 
> I found this useful when debugging slient corruption of code/data leading
> to random failures post relocation.
> 
>  common/spl/spl.c | 7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/common/spl/spl.c b/common/spl/spl.c
> index 72078a8ebc8e..9a9c8fad7673 100644
> --- a/common/spl/spl.c
> +++ b/common/spl/spl.c
> @@ -950,6 +950,13 @@ __weak void spl_relocate_stack_check(void)
>       }
>       printf("SPL initial stack usage: %lu bytes\n",
>              CONFIG_VAL(SIZE_LIMIT_PROVIDE_STACK) - i);
> +
> +     /*
> +      * If we used up all of the SIZE_LIMIT_PROVIDE_STACK, then here is high
> +      * possibility of stack overflow, warn the user accordingly
> +      */
> +     if (!i)
> +             printf("SPL possible initial stack overflow detected!!\n");
>  #endif

Since we're already inside of CONFIG_SPL_SYS_REPORT_STACK_F_USAGE being
enabled, we should rework the previous print to include something like
"%lu bytes free".

-- 
Tom

Attachment: signature.asc
Description: PGP signature

Reply via email to