fix: avoid panic when peer is blacklisted after connection

pubsub.go

@@ -504,6 +504,7 @@ func (p *PubSub) processLoop(ctx context.Context) {
 			if p.blacklist.Contains(pid) {
 				log.Warn("closing stream for blacklisted peer: ", pid)
 				close(ch)
+				delete(p.peers, pid)
 				s.Reset()
 				continue
 			}

pubsub_test.go

@@ -0,0 +1,49 @@
+package pubsub
+
+import (
+	"context"
+	"testing"
+	"time"
+)
+
+// See https://github.com/libp2p/go-libp2p-pubsub/issues/426
+func TestPubSubRemovesBlacklistedPeer(t *testing.T) {
+	ctx, cancel := context.WithCancel(context.Background())
+
+	hosts := getNetHosts(t, ctx, 2)
+
+	bl := NewMapBlacklist()
+
+	psubs0 := getPubsub(ctx, hosts[0])
+	psubs1 := getPubsub(ctx, hosts[1], WithBlacklist(bl))
+	connect(t, hosts[0], hosts[1])
+
+	// Bad peer is blacklisted after it has connected.
+	// Calling p.BlacklistPeer directly does the right thing but we should also clean
+	// up the peer if it has been added the the blacklist by another means.
+	bl.Add(hosts[0].ID())
+
+	_, err := psubs0.Subscribe("test")
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	sub1, err := psubs1.Subscribe("test")
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	time.Sleep(time.Millisecond * 100)
+
+	psubs0.Publish("test", []byte("message"))
+
+	wctx, cancel2 := context.WithTimeout(ctx, 1*time.Second)
+	defer cancel2()
+
+	_, _ = sub1.Next(wctx)
+
+	// Explicitly cancel context so PubSub cleans up peer channels.
+	// Issue 426 reports a panic due to a peer channel being closed twice.
+	cancel()
+	time.Sleep(time.Millisecond * 100)
+}