Re: [PATCH for-next] xen/arm: mm: flush_page_to_ram() only need to clean to PoC

[Bertrand Marquis <Bertrand.Marquis@arm.com>]

Hi Julien,

> On 20 Feb 2021, at 17:54, Julien Grall <julien@xen.org> wrote:
> 
> From: Julien Grall <jgrall@amazon.com>
> 
> At the moment, flush_page_to_ram() is both cleaning and invalidate to
> PoC the page. However, the cache line can be speculated and pull in the
> cache right after as it is part of the direct map.

If we go further through this logic maybe all calls to
clean_and_invalidate_dcache_va_range could be transformed in a
clean_dcache_va_range.

> 
> So it is pointless to try to invalidate the line in the data cache.
> 

But what about processors which would not speculate ?

Do you expect any performance optimization here ?

If so it might be good to explain it as I am not quite sure I get it.

Cheers
Bertrand

> Signed-off-by: Julien Grall <jgrall@amazon.com>
> ---
> xen/arch/arm/mm.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/xen/arch/arm/mm.c b/xen/arch/arm/mm.c
> index 59f8a3f15fd1..2f11d214e184 100644
> --- a/xen/arch/arm/mm.c
> +++ b/xen/arch/arm/mm.c
> @@ -529,7 +529,7 @@ void flush_page_to_ram(unsigned long mfn, bool sync_icache)
> {
>     void *v = map_domain_page(_mfn(mfn));
> 
> -    clean_and_invalidate_dcache_va_range(v, PAGE_SIZE);
> +    clean_dcache_va_range(v, PAGE_SIZE);
>     unmap_domain_page(v);
> 
>     /*
> -- 
> 2.17.1
> 
>