[PATCH v2 0/2] KVM: arm64: Dirty logging fixes/cleanups

[PATCH v2 0/2] KVM: arm64: Dirty logging fixes/cleanups

[Wei-Lin Chang]

Hi,

These are two fixes/cleanups I found when reading the code. It fixes
dirty logging detection in ring-only mode and updates a stale comment.
Please view the patches for more detail.

* Changes from v1 ([1]):

  - Keep memslot_is_logging() and use kvm_slot_dirty_track_enabled() in
    it. This preserves the RO memslot check.

  - Change patch 2 to updating the stale comment only, the original
    change is obsolete now.

  - Collect Reviewed-by and Tested-by tags from Fuad (thanks!).

Thanks!

[1]: https://lore.kernel.org/kvmarm/20260605153248.2412064-1-weilin.chang@arm.com/

Wei-Lin Chang (2):
  KVM: arm64: Fix dirty logging detection in ring-only mode
  KVM: arm64: Remove stale remark on canonical IPA alignment

 arch/arm64/kvm/mmu.c | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)

-- 
2.43.0

[PATCH v2 1/2] KVM: arm64: Fix dirty logging detection in ring-only mode

[Wei-Lin Chang]

When checking whether a memslot has dirty logging enabled, the
KVM_MEM_LOG_DIRTY_PAGES flag is the source of truth. Previously in
memslot_is_logging() we were testing the dirty bitmap only and did not
consider dirty ring.

With the ring and no bitmap, a logged memslot can still get block
mappings, so guest writes to most of a block are never logged and a
migration can copy stale memory.

Fix this by using kvm_slot_dirty_track_enabled() instead which checks
KVM_MEM_LOG_DIRTY_PAGES, within memslot_is_logging().

Fixes: 9cb1096f8590 ("KVM: arm64: Enable ring-based dirty memory tracking")
Reviewed-by: Fuad Tabba <fuad.tabba@linux.dev>
Tested-by: Fuad Tabba <fuad.tabba@linux.dev>
Signed-off-by: Wei-Lin Chang <weilin.chang@arm.com>
Cc: stable@vger.kernel.org
---
 arch/arm64/kvm/mmu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
index 705f6d84770e..c7df646af79f 100644
--- a/arch/arm64/kvm/mmu.c
+++ b/arch/arm64/kvm/mmu.c
@@ -176,7 +176,7 @@ static int kvm_mmu_split_huge_pages(struct kvm *kvm, phys_addr_t addr,
 
 static bool memslot_is_logging(struct kvm_memory_slot *memslot)
 {
-	return memslot->dirty_bitmap && !(memslot->flags & KVM_MEM_READONLY);
+	return kvm_slot_dirty_track_enabled(memslot) && !(memslot->flags & KVM_MEM_READONLY);
 }
 
 static int pkvm_flush_remote_tlbs(struct kvm *kvm)
-- 
2.43.0

Re: [PATCH v2 1/2] KVM: arm64: Fix dirty logging detection in ring-only mode

[Gavin Shan]

On 10/30/26 3:52 AM, Wei-Lin Chang wrote:
> When checking whether a memslot has dirty logging enabled, the
> KVM_MEM_LOG_DIRTY_PAGES flag is the source of truth. Previously in
> memslot_is_logging() we were testing the dirty bitmap only and did not
> consider dirty ring.
> 
> With the ring and no bitmap, a logged memslot can still get block
> mappings, so guest writes to most of a block are never logged and a
> migration can copy stale memory.
> 
> Fix this by using kvm_slot_dirty_track_enabled() instead which checks
> KVM_MEM_LOG_DIRTY_PAGES, within memslot_is_logging().
> 
> Fixes: 9cb1096f8590 ("KVM: arm64: Enable ring-based dirty memory tracking")
> Reviewed-by: Fuad Tabba <fuad.tabba@linux.dev>
> Tested-by: Fuad Tabba <fuad.tabba@linux.dev>
> Signed-off-by: Wei-Lin Chang <weilin.chang@arm.com>
> Cc: stable@vger.kernel.org
> ---
>   arch/arm64/kvm/mmu.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
Reviewed-by: Gavin Shan <gshan@redhat.com>

Re: [PATCH v2 1/2] KVM: arm64: Fix dirty logging detection in ring-only mode

[Leonardo Bras]

On Thu, Oct 29, 2026 at 05:52:37PM +0000, Wei-Lin Chang wrote:
> When checking whether a memslot has dirty logging enabled, the
> KVM_MEM_LOG_DIRTY_PAGES flag is the source of truth. Previously in
> memslot_is_logging() we were testing the dirty bitmap only and did not
> consider dirty ring.
> 
> With the ring and no bitmap, a logged memslot can still get block
> mappings, so guest writes to most of a block are never logged and a
> migration can copy stale memory.
> 
> Fix this by using kvm_slot_dirty_track_enabled() instead which checks
> KVM_MEM_LOG_DIRTY_PAGES, within memslot_is_logging().
> 
> Fixes: 9cb1096f8590 ("KVM: arm64: Enable ring-based dirty memory tracking")
> Reviewed-by: Fuad Tabba <fuad.tabba@linux.dev>
> Tested-by: Fuad Tabba <fuad.tabba@linux.dev>
> Signed-off-by: Wei-Lin Chang <weilin.chang@arm.com>
> Cc: stable@vger.kernel.org
> ---
>  arch/arm64/kvm/mmu.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
> index 705f6d84770e..c7df646af79f 100644
> --- a/arch/arm64/kvm/mmu.c
> +++ b/arch/arm64/kvm/mmu.c
> @@ -176,7 +176,7 @@ static int kvm_mmu_split_huge_pages(struct kvm *kvm, phys_addr_t addr,
>  
>  static bool memslot_is_logging(struct kvm_memory_slot *memslot)
>  {
> -	return memslot->dirty_bitmap && !(memslot->flags & KVM_MEM_READONLY);
> +	return kvm_slot_dirty_track_enabled(memslot) && !(memslot->flags & KVM_MEM_READONLY);
>  }
>  
>  static int pkvm_flush_remote_tlbs(struct kvm *kvm)
> -- 
> 2.43.0
> 

Reviewed-by: Leonardo Bras <leo.bras@arm.com>

Awesome, thanks for taking the suggestion!
Leo

[PATCH v2 2/2] KVM: arm64: Remove stale remark on canonical IPA alignment

[Wei-Lin Chang]

Commit 3b8ce2b5132ae ("KVM: arm64: Use a variable for the canonical IPA
in kvm_s2_fault_map()") created a variable to store the canonical IPA,
instead of calculating it when needed. It moved the aligning just before
mark_page_dirty_in_slot() to an earlier place but did not update the
comment.

Remove the stale part of the comment.

Signed-off-by: Wei-Lin Chang <weilin.chang@arm.com>
---
 arch/arm64/kvm/mmu.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
index c7df646af79f..695c6229a216 100644
--- a/arch/arm64/kvm/mmu.c
+++ b/arch/arm64/kvm/mmu.c
@@ -2232,11 +2232,7 @@ static int kvm_s2_fault_map(const struct kvm_s2_fault_desc *s2fd,
 	kvm_fault_unlock(kvm);
 	kfree(mapping);
 
-	/*
-	 * 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).
-	 */
+	/* Mark the page dirty only if the fault is handled successfully. */
 	if (writable && !ret)
 		mark_page_dirty_in_slot(kvm, s2fd->memslot,
 					gpa_to_gfn(canonical_ipa));
-- 
2.43.0

Re: [PATCH v2 2/2] KVM: arm64: Remove stale remark on canonical IPA alignment

[Fuad Tabba]

On Thu, 29 Oct 2026 at 17:53, Wei-Lin Chang <weilin.chang@arm.com> wrote:
>
> Commit 3b8ce2b5132ae ("KVM: arm64: Use a variable for the canonical IPA
> in kvm_s2_fault_map()") created a variable to store the canonical IPA,
> instead of calculating it when needed. It moved the aligning just before
> mark_page_dirty_in_slot() to an earlier place but did not update the
> comment.
>
> Remove the stale part of the comment.
>
> Signed-off-by: Wei-Lin Chang <weilin.chang@arm.com>

Reviewed-by: Fuad Tabba <fuad.tabba@linux.dev>

Cheers,
/fuad

Re: [PATCH v2 2/2] KVM: arm64: Remove stale remark on canonical IPA alignment

[Gavin Shan]

On 10/30/26 3:52 AM, Wei-Lin Chang wrote:
> Commit 3b8ce2b5132ae ("KVM: arm64: Use a variable for the canonical IPA
> in kvm_s2_fault_map()") created a variable to store the canonical IPA,
> instead of calculating it when needed. It moved the aligning just before
> mark_page_dirty_in_slot() to an earlier place but did not update the
> comment.
> 
> Remove the stale part of the comment.
> 
> Signed-off-by: Wei-Lin Chang <weilin.chang@arm.com>
> ---
>   arch/arm64/kvm/mmu.c | 6 +-----
>   1 file changed, 1 insertion(+), 5 deletions(-)
> 
Reviewed-by: Gavin Shan <gshan@redhat.com>

Re: [PATCH v2 2/2] KVM: arm64: Remove stale remark on canonical IPA alignment

[Leonardo Bras]

On Thu, Oct 29, 2026 at 05:52:38PM +0000, Wei-Lin Chang wrote:
> Commit 3b8ce2b5132ae ("KVM: arm64: Use a variable for the canonical IPA
> in kvm_s2_fault_map()") created a variable to store the canonical IPA,
> instead of calculating it when needed. It moved the aligning just before
> mark_page_dirty_in_slot() to an earlier place but did not update the
> comment.
> 
> Remove the stale part of the comment.
> 
> Signed-off-by: Wei-Lin Chang <weilin.chang@arm.com>
> ---
>  arch/arm64/kvm/mmu.c | 6 +-----
>  1 file changed, 1 insertion(+), 5 deletions(-)
> 
> diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
> index c7df646af79f..695c6229a216 100644
> --- a/arch/arm64/kvm/mmu.c
> +++ b/arch/arm64/kvm/mmu.c
> @@ -2232,11 +2232,7 @@ static int kvm_s2_fault_map(const struct kvm_s2_fault_desc *s2fd,
>  	kvm_fault_unlock(kvm);
>  	kfree(mapping);
>  
> -	/*
> -	 * 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).
> -	 */
> +	/* Mark the page dirty only if the fault is handled successfully. */
>  	if (writable && !ret)
>  		mark_page_dirty_in_slot(kvm, s2fd->memslot,
>  					gpa_to_gfn(canonical_ipa));
> -- 
> 2.43.0
> 

Reviewed-by: Leonardo Bras <leo.bras@arm.com>