From: Dongha Lee <p@sswd.pw>

The code for invalidating VNCR entries in both kvm_invalidate_vncr_ipa()
and invalidate_vncr_va() incorrectly uses a bitwise AND with `(size - 1)`
instead of `~(size - 1)` to align the start address. This results
in masking the address bits instead of aligning them down to the start
of the block.

This bug may cause stale VNCR TLB entries to remain valid even after a
TLBI or MMU notifier, leading to incorrect memory translation and
unexpected guest behavior.

Credit
Team 0xB6 in bob14:
DongHa Lee (@GAP-dev)
Gyujeong Jin (@gyutrange)
Daehyeon Ko (@4ncienth)
Geonha Lee (@leegn4a)
Hyungyu Oh (@ohhyungyu)
Jaewon Yang (@R4mbb)

Link: https://lore.kernel.org/r/20250903123949.24858-1-p@sswd.pw
Reviewed-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Dongha Lee <p@sswd.pw>
---
 arch/arm64/kvm/nested.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/kvm/nested.c b/arch/arm64/kvm/nested.c
index 77db81bae86f..d0ddce877b5d 100644
--- a/arch/arm64/kvm/nested.c
+++ b/arch/arm64/kvm/nested.c
@@ -847,7 +847,7 @@ static void kvm_invalidate_vncr_ipa(struct kvm *kvm, u64 start, u64 end)
 
 		ipa_size = ttl_to_size(pgshift_level_to_ttl(vt->wi.pgshift,
 							    vt->wr.level));
-		ipa_start = vt->wr.pa & (ipa_size - 1);
+		ipa_start = vt->wr.pa & ~(ipa_size - 1);
 		ipa_end = ipa_start + ipa_size;
 
 		if (ipa_end <= start || ipa_start >= end)
@@ -887,7 +887,7 @@ static void invalidate_vncr_va(struct kvm *kvm,
 
 		va_size = ttl_to_size(pgshift_level_to_ttl(vt->wi.pgshift,
 							   vt->wr.level));
-		va_start = vt->gva & (va_size - 1);
+		va_start = vt->gva & ~(va_size - 1);
 		va_end = va_start + va_size;
 
 		switch (scope->type) {
-- 
2.43.0

On Sat, 06 Sep 2025 13:07:24 +0900, p@sswd.pw wrote:
> The code for invalidating VNCR entries in both kvm_invalidate_vncr_ipa()
> and invalidate_vncr_va() incorrectly uses a bitwise AND with `(size - 1)`
> instead of `~(size - 1)` to align the start address. This results
> in masking the address bits instead of aligning them down to the start
> of the block.
> 
> This bug may cause stale VNCR TLB entries to remain valid even after a
> TLBI or MMU notifier, leading to incorrect memory translation and
> unexpected guest behavior.
> 
> [...]

Applied to fixes, thanks!

[1/1] KVM: arm64: nv: Fix incorrect VNCR invalidation range calculation
      https://git.kernel.org/kvmarm/kvmarm/c/5b9c1beaa1fd

--
Best,
Oliver