Re: [PATCH] armv8: mmu: don't switch to emergency tlb when adding a dynamic mapping

[Marc Zyngier <maz@kernel.org>]

On Wed, 08 Jan 2025 14:22:24 +0000,
Caleb Connolly <caleb.connolly@linaro.org> wrote:
> 
> This seems to cause crashes on a bunch of Qualcomm platforms. It's safer
> to just update the live table and flush it.

You may want to provide a bit more information, because that's not
much to go on, really.

> 
> Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
> ---
>  arch/arm/cpu/armv8/cache_v8.c | 11 ++---------
>  arch/arm/include/asm/system.h |  3 +--
>  drivers/soc/qcom/cmd-db.c     |  2 +-
>  3 files changed, 4 insertions(+), 12 deletions(-)
> 
> diff --git a/arch/arm/cpu/armv8/cache_v8.c b/arch/arm/cpu/armv8/cache_v8.c
> index e6be6359c5d9..43051d156122 100644
> --- a/arch/arm/cpu/armv8/cache_v8.c
> +++ b/arch/arm/cpu/armv8/cache_v8.c
> @@ -338,9 +338,9 @@ static void map_range(u64 virt, u64 phys, u64 size, int level,
>  		size -= next_size;
>  	}
>  }
>  
> -void mmu_map_region(phys_addr_t addr, u64 size, bool emergency)
> +void mmu_map_region(phys_addr_t addr, u64 size)
>  {
>  	u64 va_bits;
>  	int level = 0;
>  	u64 attrs = PTE_BLOCK_MEMTYPE(MT_NORMAL) | PTE_BLOCK_INNER_SHARE;
> @@ -350,19 +350,12 @@ void mmu_map_region(phys_addr_t addr, u64 size, bool emergency)
>  	get_tcr(NULL, &va_bits);
>  	if (va_bits < 39)
>  		level = 1;
>  
> -	if (emergency)
> -		map_range(addr, addr, size, level,
> -			  (u64 *)gd->arch.tlb_emerg, attrs);
> -
> -	/* Switch pagetables while we update the primary one */
> -	__asm_switch_ttbr(gd->arch.tlb_emerg);
> -
>  	map_range(addr, addr, size, level,
>  		  (u64 *)gd->arch.tlb_addr, attrs);
>  
> -	__asm_switch_ttbr(gd->arch.tlb_addr);
> +	flush_dcache_range(gd->arch.tlb_addr, gd->arch.tlb_size);

Why would you invalidate anything when *mapping* something? By
definition, if there was nothing mapped before, there is nothing to
invalidate (hint: the architecture forbids negative caching).

	M.

-- 
Without deviation from the norm, progress is not possible.