xfrm: destroy xfrm_state synchronously on net exit path

xfrm_state_put() moves struct xfrm_state to the GC list
and schedules the GC work to clean it up. On net exit call
path, xfrm_state_flush() is called to clean up and
xfrm_flush_gc() is called to wait for the GC work to complete
before exit.

However, this doesn't work because one of the ->destructor(),
ipcomp_destroy(), schedules the same GC work again inside
the GC work. It is hard to wait for such a nested async
callback. This is also why syzbot still reports the following
warning:

 WARNING: CPU: 1 PID: 33 at net/ipv6/xfrm6_tunnel.c:351 xfrm6_tunnel_net_exit+0x2cb/0x500 net/ipv6/xfrm6_tunnel.c:351
 ...
  ops_exit_list.isra.0+0xb0/0x160 net/core/net_namespace.c:153
  cleanup_net+0x51d/0xb10 net/core/net_namespace.c:551
  process_one_work+0xd0c/0x1ce0 kernel/workqueue.c:2153
  worker_thread+0x143/0x14a0 kernel/workqueue.c:2296
  kthread+0x357/0x430 kernel/kthread.c:246
  ret_from_fork+0x3a/0x50 arch/x86/entry/entry_64.S:352

In fact, it is perfectly fine to bypass GC and destroy xfrm_state
synchronously on net exit call path, because it is in process context
and doesn't need a work struct to do any blocking work.

This patch introduces xfrm_state_put_sync() which simply bypasses
GC, and lets its callers to decide whether to use this synchronous
version. On net exit path, xfrm_state_fini() and
xfrm6_tunnel_net_exit() use it. And, as ipcomp_destroy() itself is
blocking, it can use xfrm_state_put_sync() directly too.

Also rename xfrm_state_gc_destroy() to ___xfrm_state_destroy() to
reflect this change.

Fixes: b48c05ab5d32 ("xfrm: Fix warning in xfrm6_tunnel_net_exit.")
Reported-and-tested-by: syzbot+e9aebef558e3ed673934@syzkaller.appspotmail.com
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>

2 files changed, 20 insertions, 12 deletions

diff --git a/net/xfrm/xfrm_state.c b/net/xfrm/xfrm_state.c
index 23c92891758a..1bb971f46fc6 100644
--- a/net/xfrm/xfrm_state.c
+++ b/net/xfrm/xfrm_state.c
@@ -432,7 +432,7 @@ void xfrm_state_free(struct xfrm_state *x)
 }
 EXPORT_SYMBOL(xfrm_state_free);
 
-static void xfrm_state_gc_destroy(struct xfrm_state *x)
+static void ___xfrm_state_destroy(struct xfrm_state *x)
 {
 	tasklet_hrtimer_cancel(&x->mtimer);
 	del_timer_sync(&x->rtimer);
@@ -474,7 +474,7 @@ static void xfrm_state_gc_task(struct work_struct *work)
 	synchronize_rcu();
 
 	hlist_for_each_entry_safe(x, tmp, &gc_list, gclist)
-		xfrm_state_gc_destroy(x);
+		___xfrm_state_destroy(x);
 }
 
 static enum hrtimer_restart xfrm_timer_handler(struct hrtimer *me)
@@ -598,14 +598,19 @@ struct xfrm_state *xfrm_state_alloc(struct net *net)
 }
 EXPORT_SYMBOL(xfrm_state_alloc);
 
-void __xfrm_state_destroy(struct xfrm_state *x)
+void __xfrm_state_destroy(struct xfrm_state *x, bool sync)
 {
 	WARN_ON(x->km.state != XFRM_STATE_DEAD);
 
-	spin_lock_bh(&xfrm_state_gc_lock);
-	hlist_add_head(&x->gclist, &xfrm_state_gc_list);
-	spin_unlock_bh(&xfrm_state_gc_lock);
-	schedule_work(&xfrm_state_gc_work);
+	if (sync) {
+		synchronize_rcu();
+		___xfrm_state_destroy(x);
+	} else {
+		spin_lock_bh(&xfrm_state_gc_lock);
+		hlist_add_head(&x->gclist, &xfrm_state_gc_list);
+		spin_unlock_bh(&xfrm_state_gc_lock);
+		schedule_work(&xfrm_state_gc_work);
+	}
 }
 EXPORT_SYMBOL(__xfrm_state_destroy);
 
@@ -708,7 +713,7 @@ xfrm_dev_state_flush_secctx_check(struct net *net, struct net_device *dev, bool
 }
 #endif
 
-int xfrm_state_flush(struct net *net, u8 proto, bool task_valid)
+int xfrm_state_flush(struct net *net, u8 proto, bool task_valid, bool sync)
 {
 	int i, err = 0, cnt = 0;
 
@@ -730,7 +735,10 @@ restart:
 				err = xfrm_state_delete(x);
 				xfrm_audit_state_delete(x, err ? 0 : 1,
 							task_valid);
-				xfrm_state_put(x);
+				if (sync)
+					xfrm_state_put_sync(x);
+				else
+					xfrm_state_put(x);
 				if (!err)
 					cnt++;
 
@@ -2215,7 +2223,7 @@ void xfrm_state_delete_tunnel(struct xfrm_state *x)
 		if (atomic_read(&t->tunnel_users) == 2)
 			xfrm_state_delete(t);
 		atomic_dec(&t->tunnel_users);
-		xfrm_state_put(t);
+		xfrm_state_put_sync(t);
 		x->tunnel = NULL;
 	}
 }
@@ -2375,8 +2383,8 @@ void xfrm_state_fini(struct net *net)
 	unsigned int sz;
 
 	flush_work(&net->xfrm.state_hash_work);
-	xfrm_state_flush(net, IPSEC_PROTO_ANY, false);
 	flush_work(&xfrm_state_gc_work);
+	xfrm_state_flush(net, IPSEC_PROTO_ANY, false, true);
 
 	WARN_ON(!list_empty(&net->xfrm.state_all));
 
diff --git a/net/xfrm/xfrm_user.c b/net/xfrm/xfrm_user.c
index c6d26afcf89d..a131f9ff979e 100644
--- a/net/xfrm/xfrm_user.c
+++ b/net/xfrm/xfrm_user.c
@@ -1932,7 +1932,7 @@ static int xfrm_flush_sa(struct sk_buff *skb, struct nlmsghdr *nlh,
 	struct xfrm_usersa_flush *p = nlmsg_data(nlh);
 	int err;
 
-	err = xfrm_state_flush(net, p->proto, true);
+	err = xfrm_state_flush(net, p->proto, true, false);
 	if (err) {
 		if (err == -ESRCH) /* empty table */
 			return 0;