Re: [PATCH 2/2] KVM: arm64: Remove superfluous aligning of gfn for dirty logging

[Leonardo Bras <leo.bras@arm.com>]

On Fri, Jun 05, 2026 at 04:32:48PM +0100, Wei-Lin Chang wrote:
> Stage-2 mapping size is forced to PAGE_SIZE when dirty logging is
> enabled for a memslot, therefore we don't need to align it down to
> a possibly larger vma size or THP adjusted size, they won't happen.
> 

IIRC there was some effort being made in terms of tracking in LEVEL2 block 
granularity instead of LEVEL3 (PAGE_SIZE). It makes sense as some platforms 
can have very fast networking so setup time is more relevant than actual 
sending time.

Would not making this change affect such effort?

Thanks!
Leo


> Signed-off-by: Wei-Lin Chang <weilin.chang@arm.com>
> ---
>  arch/arm64/kvm/mmu.c | 9 ++++-----
>  1 file changed, 4 insertions(+), 5 deletions(-)
> 
> diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
> index 06c46124d3e7..d1f6ff7c2943 100644
> --- a/arch/arm64/kvm/mmu.c
> +++ b/arch/arm64/kvm/mmu.c
> @@ -2050,13 +2050,12 @@ static int kvm_s2_fault_map(const struct kvm_s2_fault_desc *s2fd,
>  
>  	/*
>  	 * Mark the page dirty only if the fault is handled successfully,
> -	 * making sure we adjust the canonical IPA if the mapping size has
> -	 * been updated (via a THP upgrade, for example).
> +	 * mapping size is forced to PAGE_SIZE if dirty logging is enabled,
> +	 * so we don't have to adjust the canonical IPA here.
>  	 */
>  	if (writable && !ret) {
> -		phys_addr_t ipa = gfn_to_gpa(get_canonical_gfn(s2fd, s2vi));
> -		ipa &= ~(mapping_size - 1);
> -		mark_page_dirty_in_slot(kvm, s2fd->memslot, gpa_to_gfn(ipa));
> +		gfn_t canonical_gfn = get_canonical_gfn(s2fd, s2vi);
> +		mark_page_dirty_in_slot(kvm, s2fd->memslot, canonical_gfn);
>  	}
>  
>  	if (ret != -EAGAIN)
> -- 
> 2.43.0
>