[PATCH v2 ipsec-next] xfrm: check state validity under lock in xfrm_state_find()

[Tristan Madani]

From: Tristan Madani <tristan@xxxxxxxxxxxxxxxxxxx>

The output-cache insertion path in xfrm_state_find() checks
x->km.state == XFRM_STATE_VALID before acquiring xfrm_state_lock,
then inserts the state into pol->state_cache_list under the lock.

While this insertion block is currently unreachable -- cached is set
unconditionally at the cached: label, making the !cached guard
always false -- the correct pattern is to perform the validity
check under the lock, matching the fix in commit ddd3d01329203
("xfrm: Fix xfrm state cache insertion race") which addressed the
same TOCTOU for the input cache path.

Move the XFRM_STATE_VALID check inside the critical section so the
pattern is correct when the cache insertion becomes reachable.

Signed-off-by: Tristan Madani <tristan@xxxxxxxxxxxxxxxxxxx>
---
Changes in v2:
- Reclassify as hardening: the insertion block is currently unreachable
  (cached is unconditionally true at the cached: label), as identified
  by sashiko AI review. Drop Fixes: and Cc: stable tags accordingly.
  Move from ipsec to ipsec-next tree since this is a cleanup.
- Reword commit message to describe the current state honestly and
  frame as a preventive correctness improvement.
 net/xfrm/xfrm_state.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/net/xfrm/xfrm_state.c b/net/xfrm/xfrm_state.c
index e45aa1ed5b965..03baa50f0df7c 100644
--- a/net/xfrm/xfrm_state.c
+++ b/net/xfrm/xfrm_state.c
@@ -1632,10 +1632,11 @@ xfrm_state_find(const xfrm_address_t *daddr, const xfrm_address_t *saddr,
 		*err = acquire_in_progress ? -EAGAIN : error;
 	}
 
-	if (x && x->km.state == XFRM_STATE_VALID && !cached &&
+	if (x && !cached &&
 	    (!(pol->flags & XFRM_POLICY_CPU_ACQUIRE) || x->pcpu_num == pcpu_id)) {
 		spin_lock_bh(&net->xfrm.xfrm_state_lock);
-		if (hlist_unhashed(&x->state_cache))
+		if (x->km.state == XFRM_STATE_VALID &&
+		    hlist_unhashed(&x->state_cache))
 			hlist_add_head_rcu(&x->state_cache, &pol->state_cache_list);
 		spin_unlock_bh(&net->xfrm.xfrm_state_lock);
 	}
-- 
2.47.3