[PATCH v2 2/2] KVM: arm64: Check for kvm_vma_mte_allowed in the critical section

[Marc Zyngier]

On page fault, we find about the VMA that backs the page fault
early on, and quickly release the mmap_read_lock. However, using
the VMA pointer after the critical section is pretty dangerous,
as a teardown may happen in the meantime and the VMA be long gone.

Move the sampling of the MTE permission early, and NULL-ify the
VMA pointer after that, just to be on the safe side.

Signed-off-by: Marc Zyngier <maz at kernel.org>
Cc: stable at vger.kernel.org
---
 arch/arm64/kvm/mmu.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
index d3d4cdc0f617..e95593736ae3 100644
--- a/arch/arm64/kvm/mmu.c
+++ b/arch/arm64/kvm/mmu.c
@@ -1218,7 +1218,7 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 {
 	int ret = 0;
 	bool write_fault, writable, force_pte = false;
-	bool exec_fault;
+	bool exec_fault, mte_allowed;
 	bool device = false;
 	unsigned long mmu_seq;
 	struct kvm *kvm = vcpu->kvm;
@@ -1309,6 +1309,10 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 		fault_ipa &= ~(vma_pagesize - 1);
 
 	gfn = fault_ipa >> PAGE_SHIFT;
+	mte_allowed = kvm_vma_mte_allowed(vma);
+
+	/* Don't use the VMA after the unlock -- it may have vanished */
+	vma = NULL;
 
 	/*
 	 * Read mmu_invalidate_seq so that KVM can detect if the results of
@@ -1379,7 +1383,7 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 
 	if (fault_status != ESR_ELx_FSC_PERM && !device && kvm_has_mte(kvm)) {
 		/* Check the VMM hasn't introduced a new disallowed VMA */
-		if (kvm_vma_mte_allowed(vma)) {
+		if (mte_allowed) {
 			sanitise_mte_tags(kvm, pfn, vma_pagesize);
 		} else {
 			ret = -EFAULT;
-- 
2.34.1