KVM: x86/xen: Fix lockdep warning on "recursive" gpc locking

Commit Message

David Woodhouse Jan. 10, 2023, 3:25 p.m. UTC

From: David Woodhouse <[e-mail]>

In commit 5ec3289b31 ("KVM: x86/xen: Compatibility fixes for shared runstate
area") we declared it safe to obtain two gfn_to_pfn_cache locks at the same
time:
	/*
	 * The guest's runstate_info is split across two pages and we
	 * need to hold and validate both GPCs simultaneously. We can
	 * declare a lock ordering GPC1 > GPC2 because nothing else
	 * takes them more than one at a time.
	 */

However, we forgot to tell lockdep. Do so, by setting a subclass on the
first lock before taking the second.

Fixes: 5ec3289b31 ("KVM: x86/xen: Compatibility fixes for shared runstate area")
Suggested-by: Peter Zijlstra <[e-mail]>
Signed-off-by: David Woodhouse <[e-mail]>
---
 arch/x86/kvm/xen.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Patch

diff --git a/arch/x86/kvm/xen.c b/arch/x86/kvm/xen.c
index 402d9a34552c..07e61cc9881e 100644
--- a/arch/x86/kvm/xen.c
+++ b/arch/x86/kvm/xen.c
@@ -305,8 +305,10 @@  static void kvm_xen_update_runstate_guest(struct kvm_vcpu *v, bool atomic)
 		 * The guest's runstate_info is split across two pages and we
 		 * need to hold and validate both GPCs simultaneously. We can
 		 * declare a lock ordering GPC1 > GPC2 because nothing else
-		 * takes them more than one at a time.
+		 * takes them more than one at a time. Set a subclass on the
+		 * gpc1 lock to make lockdep shut up about it.
 		 */
+		lock_set_subclass(&gpc1->lock.dep_map, 1, _THIS_IP_);
 		read_lock(&gpc2->lock);
 
 		if (!kvm_gpc_check(gpc2, user_len2)) {