fix

Signed-off-by: Geliang Tang <[email protected]>

net/mptcp/bpf.c

@@ -15,6 +15,7 @@
 #include <linux/btf_ids.h>
 #include <net/bpf_sk_storage.h>
 #include "protocol.h"
+#include "mib.h"
 
 #ifdef CONFIG_BPF_JIT
 static struct bpf_struct_ops bpf_mptcp_pm_ops,
@@ -686,6 +687,11 @@ __bpf_kfunc static void bpf_ipv6_addr_set_v4mapped(const __be32 addr,
 #endif
 }
 
+__bpf_kfunc static void mptcp_inc_stats_rmsubflow(struct sock *sk)
+{
+	MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_RMSUBFLOW);
+}
+
 __bpf_kfunc static bool bpf_mptcp_subflow_queues_empty(struct sock *sk)
 {
 	return tcp_rtx_queue_empty(sk);
@@ -737,6 +743,7 @@ BTF_ID_FLAGS(func, mptcp_pm_remove_addr_entry, KF_SLEEPABLE)
 BTF_ID_FLAGS(func, __mptcp_subflow_connect, KF_SLEEPABLE)
 BTF_ID_FLAGS(func, mptcp_subflow_shutdown, KF_SLEEPABLE)
 BTF_ID_FLAGS(func, mptcp_close_ssk, KF_SLEEPABLE)
+BTF_ID_FLAGS(func, mptcp_inc_stats_rmsubflow)
 BTF_ID_FLAGS(func, mptcp_pm_nl_mp_prio_send_ack, KF_SLEEPABLE)
 BTF_ID_FLAGS(func, mptcp_subflow_active)
 BTF_ID_FLAGS(func, mptcp_set_timeout)

net/mptcp/pm_netlink.c

@@ -508,7 +508,20 @@ __lookup_addr_by_id(struct pm_nl_pernet *pernet, unsigned int id)
 }
 
 static struct mptcp_pm_addr_entry *
-__lookup_addr(struct pm_nl_pernet *pernet, const struct mptcp_addr_info *info)
+__lookup_addr_by_id_rcu(struct pm_nl_pernet *pernet, unsigned int id)
+{
+	struct mptcp_pm_addr_entry *entry;
+
+	list_for_each_entry_rcu(entry, &pernet->local_addr_list, list) {
+		if (entry->addr.id == id)
+			return entry;
+	}
+	return NULL;
+}
+
+static struct mptcp_pm_addr_entry *
+__lookup_addr_rcu(struct pm_nl_pernet *pernet,
+		  const struct mptcp_addr_info *info)
 {
 	struct mptcp_pm_addr_entry *entry;
 
@@ -544,7 +557,7 @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk)
 
 		mptcp_local_address((struct sock_common *)msk->first, &mpc_addr);
 		rcu_read_lock();
-		entry = __lookup_addr(pernet, &mpc_addr);
+		entry = __lookup_addr_rcu(pernet, &mpc_addr);
 		if (entry) {
 			__clear_bit(entry->addr.id, msk->pm.id_avail_bitmap.map);
 			msk->mpc_endpoint_id = entry->addr.id;
@@ -1127,14 +1140,13 @@ int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct mptcp_pm_addr_entry
 {
 	struct mptcp_pm_addr_entry *entry;
 	struct pm_nl_pernet *pernet;
-	int ret = -1;
+	int ret;
 
 	pernet = pm_nl_get_pernet_from_msk(msk);
 
 	rcu_read_lock();
-	entry = __lookup_addr(pernet, &local->addr);
-	if (entry)
-		ret = entry->addr.id;
+	entry = __lookup_addr_rcu(pernet, &local->addr);
+	ret = entry ? entry->addr.id : -1;
 	rcu_read_unlock();
 	if (ret >= 0)
 		return ret;
@@ -1156,12 +1168,11 @@ u8 mptcp_pm_nl_get_flags(struct mptcp_sock *msk, struct mptcp_addr_info *skc)
 {
 	struct pm_nl_pernet *pernet = pm_nl_get_pernet_from_msk(msk);
 	struct mptcp_pm_addr_entry *entry;
-	u8 flags = 0;
+	u8 flags;
 
 	rcu_read_lock();
-	entry = __lookup_addr(pernet, skc);
-	if (entry)
-		flags = entry->flags;
+	entry = __lookup_addr_rcu(pernet, skc);
+	flags = entry ? entry->flags : 0;
 	rcu_read_unlock();
 
 	return flags;
@@ -1779,13 +1790,13 @@ static int mptcp_pm_nl_get_addr(u8 id, struct mptcp_pm_addr_entry *addr,
 
 	pernet = pm_nl_get_pernet(net);
 
-	spin_lock_bh(&pernet->lock);
-	entry = __lookup_addr_by_id(pernet, id);
+	rcu_read_lock();
+	entry = __lookup_addr_by_id_rcu(pernet, id);
 	if (entry) {
 		*addr = *entry;
 		ret = 0;
 	}
-	spin_unlock_bh(&pernet->lock);
+	rcu_read_unlock();
 
 	return ret;
 }
@@ -1849,9 +1860,7 @@ static int mptcp_pm_nl_dump_addr(struct mptcp_id_bitmap *bitmap,
 
 	pernet = pm_nl_get_pernet(net);
 
-	spin_lock_bh(&pernet->lock);
 	bitmap_copy(bitmap->map, pernet->id_bitmap.map, MPTCP_PM_MAX_ADDR_ID + 1);
-	spin_unlock_bh(&pernet->lock);
 
 	return 0;
 }
@@ -2047,25 +2056,25 @@ static int mptcp_pm_nl_set_flags(struct mptcp_pm_addr_entry *loc,
 	if (loc->flags & MPTCP_PM_ADDR_FLAG_BACKUP)
 		bkup = 1;
 
-	spin_lock_bh(&pernet->lock);
-	entry = lookup_by_id ? __lookup_addr_by_id(pernet, loc->addr.id) :
-			       __lookup_addr(pernet, &loc->addr);
+	rcu_read_lock();
+	entry = lookup_by_id ? __lookup_addr_by_id_rcu(pernet, loc->addr.id) :
+			       __lookup_addr_rcu(pernet, &loc->addr);
 	if (!entry) {
-		spin_unlock_bh(&pernet->lock);
+		rcu_read_unlock();
 		GENL_SET_ERR_MSG(info, "address not found");
 		return -EINVAL;
 	}
 	if ((loc->flags & MPTCP_PM_ADDR_FLAG_FULLMESH) &&
 	    (entry->flags & MPTCP_PM_ADDR_FLAG_SIGNAL)) {
-		spin_unlock_bh(&pernet->lock);
+		rcu_read_unlock();
 		GENL_SET_ERR_MSG(info, "invalid addr flags");
 		return -EINVAL;
 	}
 
 	changed = (loc->flags ^ entry->flags) & mask;
 	entry->flags = (entry->flags & ~mask) | (loc->flags & mask);
 	*loc = *entry;
-	spin_unlock_bh(&pernet->lock);
+	rcu_read_unlock();
 
 	mptcp_nl_set_flags(net, &loc->addr, bkup, changed);
 	return 0;

tools/testing/selftests/bpf/progs/mptcp_bpf.h

@@ -23,6 +23,7 @@ extern bool CONFIG_MPTCP_IPV6 __kconfig __weak;
 #define RCV_SHUTDOWN	1
 #define SEND_SHUTDOWN	2
 
+#define	ESRCH		3	/* No such process */
 #define	ENOMEM		12	/* Out of Memory */
 #define	EINVAL		22	/* Invalid argument */
 
@@ -132,6 +133,7 @@ extern void bpf_ipv6_addr_set_v4mapped(const __be32 addr,
 extern void mptcp_subflow_shutdown(struct sock *sk, struct sock *ssk, int how) __ksym;
 extern void mptcp_close_ssk(struct sock *sk, struct sock *ssk,
 			    struct mptcp_subflow_context *subflow) __ksym;
+extern void mptcp_inc_stats_rmsubflow(struct sock *sk) __ksym;
 
 extern int mptcp_pm_nl_mp_prio_send_ack(struct mptcp_sock *msk,
 					struct mptcp_addr_info *addr,

tools/testing/selftests/bpf/progs/mptcp_bpf_bytes.c

@@ -30,10 +30,10 @@ int BPF_PROG(trace_mptcp_sched_get_send, struct mptcp_sock *msk)
 		tp = bpf_core_cast(ssk, struct tcp_sock);
 
 		if (subflow->subflow_id == 1) {
-			bpf_printk("bytes 1: sent %lu received %lu subflows %u", tp->bytes_sent, tp->bytes_received, msk->pm.subflows);
+			//bpf_printk("bytes 1: sent %lu received %lu subflows %u", tp->bytes_sent, tp->bytes_received, msk->pm.subflows);
 			bytes_sent_1 += tp->bytes_sent;
 		} else if (subflow->subflow_id == 2) {
-			bpf_printk("bytes 2: sent %lu received %lu subflows %u", tp->bytes_sent, tp->bytes_received, msk->pm.subflows);
+			//bpf_printk("bytes 2: sent %lu received %lu subflows %u", tp->bytes_sent, tp->bytes_received, msk->pm.subflows);
 			bytes_sent_2 += tp->bytes_sent;
 		}
 	}

tools/testing/selftests/bpf/progs/mptcp_bpf_rr.c

@@ -60,6 +60,7 @@ int BPF_PROG(bpf_rr_get_subflow, struct mptcp_sock *msk,
 out:
 	next = bpf_core_cast(next, struct mptcp_subflow_context);
 	mptcp_subflow_set_scheduled(next, true);
+	//bpf_printk("rr subflow=%u/%u", next->subflow_id, msk->pm.subflows + 1);
 	ptr->last_snd = mptcp_subflow_tcp_sock(next);
 	return 0;
 }