[PATCH 6.18.y 1/2] KVM: Harden and prepare for modifying existing guest_memfd memslots

[Sasha Levin <sashal@kernel.org>]

From: Sean Christopherson <seanjc@google.com>

[ Upstream commit af62fe2494da84eb01752282c8228c9bb3fe9f67 ]

Unbind guest_memfd memslots if KVM commits a MOVE or FLAGS_ONLY memslot
change to harden against use-after-free, and to prepare for eventually
supporting dirty logging on guest_memfd memslots, at which point
FLAGS_ONLY changes will be expected/supported.

Add two separate WARNs, once to yell if a guest_memfd memslot is moved
(which KVM is never expected to allow/support), and again if the unbind()
is triggered, to help detect uAPI goofs prior to deliberately allowing
FLAGS_ONLY changes.

Link: https://patch.msgid.link/20251202020334.1171351-3-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
Stable-dep-of: 513333a55460 ("KVM: Only bind memslot to guest_memfd instance for CREATE operations")
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 virt/kvm/kvm_main.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 23e7d1bb80b04..297fd43b1da6c 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -1748,6 +1748,12 @@ static void kvm_commit_memory_region(struct kvm *kvm,
 		kvm_free_memslot(kvm, old);
 		break;
 	case KVM_MR_MOVE:
+		/*
+		 * Moving a guest_memfd memslot isn't supported, and will never
+		 * be supported.
+		 */
+		WARN_ON_ONCE(old->flags & KVM_MEM_GUEST_MEMFD);
+		fallthrough;
 	case KVM_MR_FLAGS_ONLY:
 		/*
 		 * Free the dirty bitmap as needed; the below check encompasses
@@ -1756,6 +1762,15 @@ static void kvm_commit_memory_region(struct kvm *kvm,
 		if (old->dirty_bitmap && !new->dirty_bitmap)
 			kvm_destroy_dirty_bitmap(old);
 
+		/*
+		 * Unbind the guest_memfd instance as needed; the @new slot has
+		 * already created its own binding.  TODO: Drop the WARN when
+		 * dirty logging guest_memfd memslots is supported.  Until then,
+		 * flags-only changes on guest_memfd slots should be impossible.
+		 */
+		if (WARN_ON_ONCE(old->flags & KVM_MEM_GUEST_MEMFD))
+			kvm_gmem_unbind(old);
+
 		/*
 		 * The final quirk.  Free the detached, old slot, but only its
 		 * memory, not any metadata.  Metadata, including arch specific
-- 
2.53.0