KVM: arm64: Fix page leak in user_mem_abort()

The user_mem_abort() function acquires a page reference via
__kvm_faultin_pfn() early in its execution. However, the subsequent
checks for mismatched attributes between stage 1 and stage 2 mappings
would return an error code directly, bypassing the corresponding page
release.

Fix this by storing the error and releasing the unused page before
returning the error.

Fixes: 6d674e2 ("KVM: arm/arm64: Properly handle faulting of device mappings")
Fixes: 2a8dfab ("KVM: arm64: Block cacheable PFNMAP mapping")
Signed-off-by: Fuad Tabba <tabba@google.com>
Reviewed-by: Oliver Upton <oliver.upton@linux.dev>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Cc: stable@vger.kernel.org

Author: Fuad Tabba

arch/arm64/kvm/mmu.c

@@ -1670,7 +1670,7 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 			 * cache maintenance.
 			 */
 			if (!kvm_supports_cacheable_pfnmap())
-				return -EFAULT;
+				ret = -EFAULT;
 		} else {
 			/*
 			 * If the page was identified as device early by looking at
@@ -1693,7 +1693,12 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 	}
 
 	if (exec_fault && s2_force_noncacheable)
-		return -ENOEXEC;
+		ret = -ENOEXEC;
+
+	if (ret) {
+		kvm_release_page_unused(page);
+		return ret;
+	}
 
 	/*
 	 * Potentially reduce shadow S2 permissions to match the guest's own