rlp: finalize listIterator on parse error to prevent non-advancing loops (#33245)

The list iterator previously returned true on parse errors without
advancing the input, which could lead to non-advancing infinite loops
for callers that do not check Err() inside the loop; to make iteration
safe while preserving error visibility, Next() now marks the iterator as
finished when readKind fails, returning true for the error step so
existing users that check Err() can handle it, and then false on
subsequent calls, and the function comment was updated to document this
behavior and the need to check Err().

Author: radik878

rlp/iterator.go

@@ -37,15 +37,24 @@ func NewListIterator(data RawValue) (*listIterator, error) {
 	return it, nil
 }
 
-// Next forwards the iterator one step, returns true if it was not at end yet
+// Next forwards the iterator one step.
+// Returns true if there is a next item or an error occurred on this step (check Err()).
+// On parse error, the iterator is marked finished and subsequent calls return false.
 func (it *listIterator) Next() bool {
 	if len(it.data) == 0 {
 		return false
 	}
 	_, t, c, err := readKind(it.data)
+	if err != nil {
+		it.next = nil
+		it.err = err
+		// Mark iteration as finished to avoid potential infinite loops on subsequent Next calls.
+		it.data = nil
+		return true
+	}
 	it.next = it.data[:t+c]
 	it.data = it.data[t+c:]
-	it.err = err
+	it.err = nil
 	return true
 }