[PATCH 1/3] ARM: Define KERNEL_START and KERNEL_END

Chris Brandt Chris.Brandt at renesas.com
Tue Dec 6 14:43:40 PST 2016


On 12/6/2016, Florian Fainelli wrote:
> diff --git a/arch/arm/mm/mmu.c b/arch/arm/mm/mmu.c
> index 4001dd15818d..18ef688a796e 100644
> --- a/arch/arm/mm/mmu.c
> +++ b/arch/arm/mm/mmu.c
> @@ -1437,12 +1437,8 @@ static void __init kmap_init(void)
>  static void __init map_lowmem(void)
>  {
>  	struct memblock_region *reg;
> -#ifdef CONFIG_XIP_KERNEL
> -	phys_addr_t kernel_x_start = round_down(__pa(_sdata), SECTION_SIZE);
> -#else
> -	phys_addr_t kernel_x_start = round_down(__pa(_stext), SECTION_SIZE);
> -#endif
> -	phys_addr_t kernel_x_end = round_up(__pa(__init_end), SECTION_SIZE);
> +	phys_addr_t kernel_x_start = round_down(__pa(KERNEL_START),
> SECTION_SIZE);
> +	phys_addr_t kernel_x_end = round_down(__pa(_end), SECTION_SIZE);

Why are you changing the end of executable kernel (hence the 'x' in
kernel_x_end) from __init_end to _end which basically maps the entire
kernel image including text and data?

Doing so would then change data from MT_MEMORY_RW into MT_MEMORY_RWX.

I would think it would create some type of security risk to allow
data to be executable.



More information about the linux-arm-kernel mailing list