version 0.5

Docker/README.md

@@ -1,18 +1,18 @@
-# Docker for Uroboros version 0.4
+# Docker for Uroboros version 0.5
 
 ## Docker build
 
 
 Then use `docker build` to build the docker image from Dockerfile.
 
 ```shell
-docker build -f Dockerfile -t uroboros:0.4 .
+docker build -f Dockerfile -t uroboros:0.5 .
 ```
 
 ## Docker usage
 
 run the Docker container with `docker run`.
 
 ```shell
-docker run -it uroboros:0.4
+docker run -it uroboros:0.5
 ```

README.md

@@ -1,4 +1,4 @@
-# Uroboros: Infrastructure for Reassembleable Disassembling and Transformation (Version 0.4)
+# Uroboros: Infrastructure for Reassembleable Disassembling and Transformation (Version 0.5)
 
 # Installation
 

instruction_to_reproduce_experimenttal_results_reported_in_Wang_et_al._2015.md

@@ -1,4 +1,4 @@
-Please use Uroboros (version 0.4) for the tests below. 
+Please use Uroboros (version 0.5) for the tests below. 
 
 # Coreutils:
 

src/cfg.ml

@@ -114,7 +114,7 @@ class cfg =
            begin
              match c with
              | Jump _ -> true (* jump ops *)
-             | CALL -> true (* call *)
+             (*| CALL -> true (* call *)*)
              | RET ->  true
              | RETN -> true
              | _ -> false

src/extern_symbol_process64.py

@@ -41,8 +41,8 @@ def check_32():
 
     for i in range(len(lines)):
         l = lines[i]
-        if "#" in l and not "+" in l:
-            m_s = re.search(pat_s, l)
+        m_s = re.search(pat_s, l)
+        if "#" in l and not "+" in l and m_s:
             m_d = re.search(pat_d, l)
             try:
                 src = m_s.group(1) # let it crash it not

src/init.ml

@@ -22,8 +22,7 @@ object (self)
 
 	print_string "1: linearly disassemble\n";
 
-    ret := Sys.command("objdump -Dr -j \
-                        .text "^f^" > "^f^".temp");
+    ret := Sys.command("objdump -Dr -j .text "^f^" > "^f^".temp");
     self#checkret ret (f^".temp");
 
     Sys.command("python pic_process.py "^f^" "^(string_of_bool !is_32));

src/pic_process.py

@@ -67,7 +67,7 @@ def text_process_unstrip(f):
     info_dump(f)
     pic_map = info_collect()
 
-    for i in xrange(1,len(ls)):
+    for i in range(1,len(ls)):
         l = ls[i]
         if "<__i686.get_pc_thunk.bx>" in l and "call" in l:
             t = ls[i+1]
@@ -96,7 +96,7 @@ def text_process_unstrip(f):
 def thunk_identify(ls):
     global step
 
-    for i in xrange(step, len(ls)):
+    for i in range(step, len(ls)):
         l = ls[i]
         m = re.search(r'([0-9A-Fa-f]*)(.+)(mov\s+\(%esp\),)(%e\w{2})', l.strip())
         if m != None:
@@ -157,7 +157,7 @@ def text_process_strip(f):
     while True:
         pc_thunk_addr, register = thunk_identify(ls)
         if pc_thunk_addr == None: break
-        for i in xrange(1,len(ls)):
+        for i in range(1,len(ls)):
             l = ls[i]
             if re.search(r'call\s+'+pc_thunk_addr, l) != None:
                 t = ls[i+1]

src/pic_process64.py

@@ -35,7 +35,7 @@
         l = lines[i]
         if "#" in l:
             des = l.split('#')[1].split()[0]
-            #symbols.append(des+"\n")
+            if des.startswith('0x'): des = des[2:]
             m = re.search(pat, l)
             try:
                 sub = m.group(0) # let it crash it not
@@ -44,8 +44,8 @@
                 l = l.replace(sub, sub1)
                 lines[i] = l+"\n"
             except Exception:
-                print "exception in pic processing of 64-bit ELF"
-                print l
+                print("exception in pic processing of 64-bit ELF")
+                print(l)
     with open(fn + '.temp', 'w') as f:
         f.writelines(lines)
 

src/post_process.py

@@ -36,6 +36,20 @@ def check_32():
         if m and re.search(r'add\s+%r\w{2},'+m.group(1),lines[i-1]):
             lines[i-1] = "nop\n"
 
+pre_symbol = ''
+pre_idx = -1
+for i in range(ll):
+    l = lines[i]
+    if l.startswith('BB_'):
+        if len(lines[i+1].strip()) == 0 or lines[i+1].startswith('S_0x'):
+            lines[i] = ""
+    if l.startswith('S_0x'):
+        if l == pre_symbol:
+            lines[pre_idx] = ''
+        pre_symbol = l
+        pre_idx = i
+
+
 for i in range(ll):
     l = lines[i]
     if ".text" in l:
@@ -129,13 +143,7 @@ def check_exe():
 
     def help(l):
         if main_symbol1 != "" and main_symbol1 in l and l.startswith(main_symbol1):
-    	#main_s1 = "S"+main_symbol
-    	#if main_s1 in l:
-    	    #l = l.replace(main_s1, ".globl main\nmain")
     	    l = ".globl main\nmain:\n"+l
-    	#else:
-    	    #l = l.replace(main_symbol, ".globl main\nmain")
-    	#    l = ".globl main\nmain:\n"+l
         return l
     #print lines
     lines = map(lambda l : help(l), lines)

src/reassemble_symbol_get.ml

@@ -969,7 +969,7 @@ end
       let p = object(sp)
         method process lbs =
           let dec_hex (s:int) : string =
-            (Printf.sprintf "S_0x%X : " s) 
+            (Printf.sprintf "S_0x%X:\n" s) 
           in
           let rec help loc_list =
             match loc_list with
@@ -1195,7 +1195,7 @@ class instrhandler instr_list des'  =
         "S_0x"^(Printf.sprintf "%X" s) in
       let help l =
         let d = dec_hex l.loc_addr in
-        print_string (l.loc_label^" : "^d^"\n") in
+        print_string (l.loc_label^":\n"^d^"\n") in
       List.iter help locs
 
     method get_instr_list =
@@ -1299,7 +1299,7 @@ class instrhandler instr_list des'  =
           if dh = lhd then
             (
               let lhs = dec_hex lh.loc_addr in
-              let label' = do_update lh.loc_label (lhs^" : ") in
+              let label' = do_update lh.loc_label (lhs^":\n") in
               let lh' = {lh with loc_label=label'} in
               help (lh'::acc) lt dt
             )
@@ -1328,7 +1328,7 @@ class instrhandler instr_list des'  =
 
     method update_loc locs d =
       let dec_hex (s:int) : string =
-        "S_0x"^(Printf.sprintf "%X : " s) in
+        "S_0x"^(Printf.sprintf "%X:\n" s) in
       let identify_des addr1 addr2 =
         addr1.loc_addr = (int_of_string addr2)
       and lift_addr addr =
@@ -1365,7 +1365,7 @@ class funchandler instr_list u_funcs'  =
         "S_0x"^(Printf.sprintf "%X" s) in
       let help l =
         let d = dec_hex l.loc_addr in
-        print_string (l.loc_label^" : "^d^"\n") in
+        print_string (l.loc_label^":\n"^d^"\n") in
       List.iter help locs
 
 

src/uroboros.py

@@ -121,7 +121,7 @@ def process(f, i):
 def iterate (f, iterations):
     print("start to process binary: " + f)
 
-    for i in xrange(0, iterations):
+    for i in range(0, iterations):
         print ("########## iteration round "+str(i+1) + " begin ! ###########")
         if process(f, i):
             pass
@@ -220,7 +220,7 @@ def set_assumption (l):
 (n-byte alignment) are set by default,
 while assumption two and three need to be configured. For example, setting
 assumption two and three: -a 2 -a 3''')
-    p.add_argument('--version', action='version', version='Uroboros v0.4')
+    p.add_argument('--version', action='version', version='Uroboros 0.5')
 
     args = p.parse_args()
     b = args.binary