[PATCH] arm64: clear_page() shouldn't use DC ZVA when DCZID_EL0.DZP == 1

[Robin Murphy]

On 2021-10-26 04:48, Reiji Watanabe wrote:
> Currently, clear_page() uses DC ZVA instruction unconditionally.  But it
> should make sure that DCZID_EL0.DZP, which indicates whether or not use
> of DC ZVA instruction is prohibited, is zero when using the instruction.
> Use stp as memset does instead when DCZID_EL0.DZP == 1.
> 
> Signed-off-by: Reiji Watanabe <reijiw at google.com>
> ---
>   arch/arm64/lib/clear_page.S | 11 +++++++++++
>   1 file changed, 11 insertions(+)
> 
> diff --git a/arch/arm64/lib/clear_page.S b/arch/arm64/lib/clear_page.S
> index b84b179edba3..7ce1bfa4081c 100644
> --- a/arch/arm64/lib/clear_page.S
> +++ b/arch/arm64/lib/clear_page.S
> @@ -16,6 +16,7 @@
>    */
>   SYM_FUNC_START_PI(clear_page)
>   	mrs	x1, dczid_el0
> +	tbnz	x1, #4, 2f	/* Branch if DC GVA is prohibited */
>   	and	w1, w1, #0xf
>   	mov	x2, #4
>   	lsl	x1, x2, x1
> @@ -25,5 +26,15 @@ SYM_FUNC_START_PI(clear_page)
>   	tst	x0, #(PAGE_SIZE - 1)
>   	b.ne	1b
>   	ret
> +
> +2:	mov	x1, #(PAGE_SIZE)
> +	sub	x0, x0, #16	/* Pre-bias. */

Out of curiosity, what's this for? It's not like we need to worry about 
PAGE_SIZE or page addresses being misaligned. I don't really see why 
we'd need a different condition from the DC ZVA loop.

Robin.

> +3:	stp	xzr, xzr, [x0, #16]
> +	stp	xzr, xzr, [x0, #32]
> +	stp	xzr, xzr, [x0, #48]
> +	stp	xzr, xzr, [x0, #64]!
> +	subs	x1, x1, #64
> +	b.gt	3b
> +	ret
>   SYM_FUNC_END_PI(clear_page)
>   EXPORT_SYMBOL(clear_page)
>