Merge pull request #73 from Wenzel/fix/havoc_find_diffs

fuzzer: fix find_diffs comparison

kafl_fuzzer/common/util.py

@@ -10,6 +10,7 @@
 import string
 import logging
 from shutil import copyfile
+from typing import Optional, Tuple
 
 import psutil
 
@@ -78,16 +79,16 @@ def atomic_write(filename, data):
     os.chmod(f.name, 0o644)
     os.rename(f.name, filename)
 
-def read_binary_file(filename):
+def read_binary_file(filename) -> bytes:
     with open(filename, 'rb') as f:
         return f.read()
 
-def find_diffs(data_a, data_b):
-    first_diff = 0
-    last_diff = 0
+def find_diffs(data_a: bytes, data_b: bytes) -> Tuple[Optional[int], Optional[int]]:
+    first_diff = None
+    last_diff = None
     for i in range(min(len(data_a), len(data_b))):
-        if data_a[i] != data_b:
-            if first_diff == 0:
+        if data_a[i] != data_b[i]:
+            if first_diff is None:
                 first_diff = i
             last_diff = i
     return first_diff, last_diff

kafl_fuzzer/manager/node.py

@@ -34,11 +34,11 @@ def __init__(self, config, payload, bitmap, node_struct, write=True):
         self.set_state("initial", write=False)
 
     @staticmethod
-    def get_metadata(workdir, node_id):
+    def get_metadata(workdir, node_id) -> bytes:
         return msgpack.unpackb(read_binary_file(QueueNode.__get_metadata_filename(workdir, node_id)), strict_map_key=False)
 
     @staticmethod
-    def get_payload(workdir, node_struct):
+    def get_payload(workdir, node_struct) -> bytes:
         return read_binary_file(QueueNode.__get_payload_filename(workdir, node_struct['info']['exit_reason'], node_struct['id']))
 
     @staticmethod

kafl_fuzzer/technique/havoc.py

@@ -61,7 +61,7 @@ def mutate_seq_havoc_array(data, func, max_iterations, resize=False):
             data = handler(data)[:KAFL_MAX_FILE]
             func(data)
 
-def mutate_seq_splice_array(data, func, max_iterations, resize=False):
+def mutate_seq_splice_array(data: bytes, func, max_iterations, resize=False):
     global location_corpus
     havoc_rounds = 4
     splice_rounds = max_iterations//havoc_rounds

kafl_fuzzer/technique/havoc_handler.py

@@ -220,15 +220,15 @@ def havoc_perform_byte_seq_extra2(data):
     pass
 
 
-def havoc_splicing(data, files):
+def havoc_splicing(data: bytes, files):
     if len(data) < 2 or files is None:
         return data
 
     rand.shuffle(files)
     retry_limit = 64
 
     for file in files[:retry_limit]:
-        file_data = read_binary_file(file)
+        file_data: bytes = read_binary_file(file)
         if len(file_data) < 2:
             continue
 

kafl_fuzzer/worker/state_logic.py

@@ -92,7 +92,7 @@ def process_kickstart(self, kick_len):
         self.init_stage_info(metadata)
         self.handle_kickstart(kick_len, metadata)
 
-    def process_node(self, payload, metadata):
+    def process_node(self, payload: bytes, metadata):
         self.init_stage_info(metadata)
 
         if metadata["state"]["name"] == "initial":
@@ -263,7 +263,7 @@ def handle_redqueen(self, payload, metadata):
             self.__perform_redqueen(payload, metadata)
         self.redqueen_time += time.time() - redqueen_start_time
 
-    def handle_havoc(self, payload, metadata):
+    def handle_havoc(self, payload: bytes, metadata):
         havoc_afl = True
         havoc_splice = True
         havoc_radamsa = self.config.radamsa
@@ -499,7 +499,7 @@ def __perform_radamsa(self, payload_array, metadata):
         self.stage_update_label("radamsa")
         radamsa.mutate_seq_radamsa_array(payload_array, self.execute, radamsa_amount)
 
-    def __perform_havoc(self, payload_array, metadata, use_splicing):
+    def __perform_havoc(self, payload_array: bytes, metadata, use_splicing):
         perf = metadata["performance"]
         havoc_amount = havoc.havoc_range(self.HAVOC_MULTIPLIER / perf)
 

kafl_fuzzer/worker/worker.py

@@ -81,8 +81,8 @@ def handle_busy(self):
         self.conn.send_ready()
 
     def handle_node(self, msg):
-        meta_data = QueueNode.get_metadata(self.config.workdir, msg["task"]["nid"])
-        payload = QueueNode.get_payload(self.config.workdir, meta_data)
+        meta_data: bytes = QueueNode.get_metadata(self.config.workdir, msg["task"]["nid"])
+        payload: bytes = QueueNode.get_payload(self.config.workdir, meta_data)
 
         # fixme: determine globally based on all seen regulars
         t_dyn = self.t_soft + 1.2 * meta_data["info"]["performance"]