kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] KVM: arm64: pkvm: Use the mm_ops indirection for cache maintenance
@ 2022-01-14 12:50 Marc Zyngier
  2022-01-14 13:49 ` Quentin Perret
  0 siblings, 1 reply; 2+ messages in thread
From: Marc Zyngier @ 2022-01-14 12:50 UTC (permalink / raw)
  To: kvmarm, kvm, linux-arm-kernel
  Cc: James Morse, Suzuki K Poulose, Alexandru Elisei, Will Deacon,
	Fuad Tabba, Quentin Perret

CMOs issued from EL2 cannot directly use the kernel helpers,
as EL2 doesn't have a mapping of the guest pages. Oops.

Instead, use the mm_ops indirection to use helpers that will
perform a mapping at EL2 and allow the CMO to be effective.

Fixes: 25aa28691bb9 ("KVM: arm64: Move guest CMOs to the fault handlers")
Signed-off-by: Marc Zyngier <maz@kernel.org>
---
 arch/arm64/kvm/hyp/pgtable.c | 18 ++++++------------
 1 file changed, 6 insertions(+), 12 deletions(-)

diff --git a/arch/arm64/kvm/hyp/pgtable.c b/arch/arm64/kvm/hyp/pgtable.c
index 844a6f003fd5..2cb3867eb7c2 100644
--- a/arch/arm64/kvm/hyp/pgtable.c
+++ b/arch/arm64/kvm/hyp/pgtable.c
@@ -983,13 +983,9 @@ static int stage2_unmap_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
 	 */
 	stage2_put_pte(ptep, mmu, addr, level, mm_ops);
 
-	if (need_flush) {
-		kvm_pte_t *pte_follow = kvm_pte_follow(pte, mm_ops);
-
-		dcache_clean_inval_poc((unsigned long)pte_follow,
-				    (unsigned long)pte_follow +
-					    kvm_granule_size(level));
-	}
+	if (need_flush && mm_ops->dcache_clean_inval_poc)
+		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(pte, mm_ops),
+					       kvm_granule_size(level));
 
 	if (childp)
 		mm_ops->put_page(childp);
@@ -1151,15 +1147,13 @@ static int stage2_flush_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
 	struct kvm_pgtable *pgt = arg;
 	struct kvm_pgtable_mm_ops *mm_ops = pgt->mm_ops;
 	kvm_pte_t pte = *ptep;
-	kvm_pte_t *pte_follow;
 
 	if (!kvm_pte_valid(pte) || !stage2_pte_cacheable(pgt, pte))
 		return 0;
 
-	pte_follow = kvm_pte_follow(pte, mm_ops);
-	dcache_clean_inval_poc((unsigned long)pte_follow,
-			    (unsigned long)pte_follow +
-				    kvm_granule_size(level));
+	if (mm_ops->dcache_clean_inval_poc)
+		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(pte, mm_ops),
+					       kvm_granule_size(level));
 	return 0;
 }
 
-- 
2.30.2


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH] KVM: arm64: pkvm: Use the mm_ops indirection for cache maintenance
  2022-01-14 12:50 [PATCH] KVM: arm64: pkvm: Use the mm_ops indirection for cache maintenance Marc Zyngier
@ 2022-01-14 13:49 ` Quentin Perret
  0 siblings, 0 replies; 2+ messages in thread
From: Quentin Perret @ 2022-01-14 13:49 UTC (permalink / raw)
  To: Marc Zyngier
  Cc: kvmarm, kvm, linux-arm-kernel, James Morse, Suzuki K Poulose,
	Alexandru Elisei, Will Deacon, Fuad Tabba

On Friday 14 Jan 2022 at 12:50:38 (+0000), Marc Zyngier wrote:
> CMOs issued from EL2 cannot directly use the kernel helpers,
> as EL2 doesn't have a mapping of the guest pages. Oops.
> 
> Instead, use the mm_ops indirection to use helpers that will
> perform a mapping at EL2 and allow the CMO to be effective.

Right, we were clearly lucky not to use those paths at EL2 _yet_, but
that's going to change soon and this is better for consistency, so:

Reviewed-by: Quentin Perret <qperret@google.com>

> Fixes: 25aa28691bb9 ("KVM: arm64: Move guest CMOs to the fault handlers")
> Signed-off-by: Marc Zyngier <maz@kernel.org>
> ---
>  arch/arm64/kvm/hyp/pgtable.c | 18 ++++++------------
>  1 file changed, 6 insertions(+), 12 deletions(-)
> 
> diff --git a/arch/arm64/kvm/hyp/pgtable.c b/arch/arm64/kvm/hyp/pgtable.c
> index 844a6f003fd5..2cb3867eb7c2 100644
> --- a/arch/arm64/kvm/hyp/pgtable.c
> +++ b/arch/arm64/kvm/hyp/pgtable.c
> @@ -983,13 +983,9 @@ static int stage2_unmap_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
>  	 */
>  	stage2_put_pte(ptep, mmu, addr, level, mm_ops);
>  
> -	if (need_flush) {
> -		kvm_pte_t *pte_follow = kvm_pte_follow(pte, mm_ops);
> -
> -		dcache_clean_inval_poc((unsigned long)pte_follow,
> -				    (unsigned long)pte_follow +
> -					    kvm_granule_size(level));
> -	}
> +	if (need_flush && mm_ops->dcache_clean_inval_poc)
> +		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(pte, mm_ops),
> +					       kvm_granule_size(level));
>  
>  	if (childp)
>  		mm_ops->put_page(childp);
> @@ -1151,15 +1147,13 @@ static int stage2_flush_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
>  	struct kvm_pgtable *pgt = arg;
>  	struct kvm_pgtable_mm_ops *mm_ops = pgt->mm_ops;
>  	kvm_pte_t pte = *ptep;
> -	kvm_pte_t *pte_follow;
>  
>  	if (!kvm_pte_valid(pte) || !stage2_pte_cacheable(pgt, pte))
>  		return 0;
>  
> -	pte_follow = kvm_pte_follow(pte, mm_ops);
> -	dcache_clean_inval_poc((unsigned long)pte_follow,
> -			    (unsigned long)pte_follow +
> -				    kvm_granule_size(level));
> +	if (mm_ops->dcache_clean_inval_poc)
> +		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(pte, mm_ops),
> +					       kvm_granule_size(level));
>  	return 0;
>  }
>  
> -- 
> 2.30.2
> 

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2022-01-14 13:49 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-01-14 12:50 [PATCH] KVM: arm64: pkvm: Use the mm_ops indirection for cache maintenance Marc Zyngier
2022-01-14 13:49 ` Quentin Perret

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).