Fix data race in regular expression processing (#4065)

Author: mangalaman93

query/common_test.go

@@ -264,6 +264,8 @@ previous_model                 : uid @reverse .
 created_at                     : datetime @index(hour) .
 updated_at                     : datetime @index(year) .
 number                         : int @index(int) .
+firstName                      : string .
+lastName                       : string .
 `
 
 func populateCluster() {
@@ -548,6 +550,18 @@ func populateCluster() {
 		<202> <model> "Prius" .
 		<202> <model> "プリウス"@jp .
 		<202> <dgraph.type> "CarModel" .
+
+		# data for regexp testing
+		_:luke <firstName> "Luke" .
+		_:luke <lastName> "Skywalker" .
+		_:leia <firstName> "Princess" .
+		_:leia <lastName> "Leia" .
+		_:han <firstName> "Han" .
+		_:han <lastName> "Solo" .
+		_:har <firstName> "Harrison" .
+		_:har <lastName> "Ford" .
+		_:ss <firstName> "Steven" .
+		_:ss <lastName> "Spielberg" .
 	`)
 
 	addGeoPointToCluster(1, "loc", []float64{1.1, 2.0})

query/query3_test.go

@@ -2137,3 +2137,61 @@ func TestQueryMultipleTypes(t *testing.T) {
 	{"name":"Person", "fields":[{"name":"name", "type": "default"},
 		{"name":"pet", "type":"default"}]}]}}`, js)
 }
+
+func TestRegexInFilterNoDataOnRoot(t *testing.T) {
+	query := `
+		{
+			q(func: has(nonExistent)) @filter(regexp(make, /.*han/i)) {
+				uid
+				firstName
+				lastName
+			}
+		}
+	`
+	res := processQueryNoErr(t, query)
+	require.JSONEq(t, `{"data":{"q":[]}}`, res)
+}
+
+func TestRegexInFilterIndexedPredOnRoot(t *testing.T) {
+	query := `
+		{
+			q(func: regexp(name, /.*nonExistent/i)) {
+				uid
+				firstName
+				lastName
+			}
+		}
+	`
+	res := processQueryNoErr(t, query)
+	require.JSONEq(t, `{"data":{"q":[]}}`, res)
+}
+
+func TestMultiRegexInFilter(t *testing.T) {
+	query := `
+		{
+			q(func: has(full_name)) @filter(regexp(full_name, /.*michonne/i) OR regexp(name, /.*michonne/i)) {
+				expand(_all_)
+			}
+		}
+	`
+	res := processQueryNoErr(t, query)
+	require.JSONEq(t, `{"data": {"q": [{"name": "Michonne"}]}}`, res)
+}
+
+func TestMultiRegexInFilter2(t *testing.T) {
+	query := `
+		{
+			q(func: has(firstName)) @filter(regexp(firstName, /.*han/i) OR regexp(lastName, /.*han/i)) {
+				firstName
+				lastName
+			}
+		}
+	`
+
+	// run 20 times ensure that there is no data race
+	// https://github.com/dgraph-io/dgraph/issues/4030
+	for i := 0; i < 20; i++ {
+		res := processQueryNoErr(t, query)
+		require.JSONEq(t, `{"data": {"q": [{"firstName": "Han", "lastName":"Solo"}]}}`, res)
+	}
+}

worker/task.go

@@ -772,8 +772,9 @@ type queryState struct {
 	cache *posting.LocalCache
 }
 
-func (qs *queryState) helpProcessTask(
-	ctx context.Context, q *pb.Query, gid uint32) (*pb.Result, error) {
+func (qs *queryState) helpProcessTask(ctx context.Context, q *pb.Query, gid uint32) (
+	*pb.Result, error) {
+
 	span := otrace.FromContext(ctx)
 	out := new(pb.Result)
 	attr := q.Attr
@@ -854,8 +855,6 @@ func (qs *queryState) helpProcessTask(
 	}
 
 	if srcFn.fnType == regexFn {
-		// Go through the indexkeys for the predicate and match them with
-		// the regex matcher.
 		span.Annotate(nil, "handleRegexFunction")
 		if err := qs.handleRegexFunction(ctx, funcArgs{q, gid, srcFn, out}); err != nil {
 			return nil, err
@@ -961,8 +960,18 @@ func (qs *queryState) handleRegexFunction(ctx context.Context, arg funcArgs) err
 	// Here we determine the list of uids to match.
 	switch {
 	// If this is a filter eval, use the given uid list (good)
-	case arg.q.UidList != nil && len(arg.q.UidList.Uids) != 0:
-		uids = arg.q.UidList
+	case arg.q.UidList != nil:
+		// These UIDs are copied into arg.out.UidMatrix which is later updated while
+		// processing the query. The below trick makes a copy of the list to avoid the
+		// race conditions later. I (Aman) did a race condition tests to ensure that we
+		// do not have more race condition in similar code in the rest of the file.
+		// The race condition was found only here because in filter condition, even when
+		// predicates do not have indexes, we allow regexp queries (for example, we do
+		// not support eq/gt/lt/le in @filter, see #4077), and this was new code that
+		// was added just to support the aforementioned case, the race condition is only
+		// in this part of the code.
+		uids = &pb.List{}
+		uids.Uids = append(arg.q.UidList.Uids[:0:0], arg.q.UidList.Uids...)
 
 	// Prefer to use an index (fast)
 	case useIndex: