smatch_data/db/smdb.py

   1 #!/usr/bin/python
   2
   3 # Copyright (C) 2013 Oracle.
   4 #
   5 # Licensed under the Open Software License version 1.1
   6
   7 import sqlite3
   8 import sys
   9 import re
  10
  11 try:
  12     con = sqlite3.connect('smatch_db.sqlite')
  13 except sqlite3.Error, e:
  14     print "Error %s:" % e.args[0]
  15     sys.exit(1)
  16
  17 def usage():
  18     print "%s" %(sys.argv[0])
  19     print "<function> - how a function is called"
  20     print "return_states <function> - what a function returns"
  21     print "call_tree <function> - show the call tree"
  22     print "where <struct_type> <member> - where a struct member is set"
  23     print "type_size <struct_type> <member> - how a struct member is allocated"
  24     print "data_info <struct_type> <member> - information about a given data type"
  25     print "function_ptr <function> - which function pointers point to this"
  26     print "trace_param <function> <param> - trace where a parameter came from"
  27     sys.exit(1)
  28
  29 function_ptrs = []
  30 searched_ptrs = []
  31 def get_function_pointers_helper(func):
  32     cur = con.cursor()
  33     cur.execute("select distinct ptr from function_ptr where function = '%s';" %(func))
  34     for row in cur:
  35         ptr = row[0]
  36         if ptr in function_ptrs:
  37             continue
  38         function_ptrs.append(ptr)
  39         if not ptr in searched_ptrs:
  40             searched_ptrs.append(ptr)
  41             get_function_pointers_helper(ptr)
  42
  43 def get_function_pointers(func):
  44     global function_ptrs
  45     global searched_ptrs
  46     function_ptrs = [func]
  47     searched_ptrs = [func]
  48     get_function_pointers_helper(func)
  49     return function_ptrs
  50
  51 db_types = {   0: "INTERNAL",
  52              101: "PARAM_CLEARED",
  53              103: "PARAM_LIMIT",
  54              104: "PARAM_FILTER",
  55             1001: "PARAM_VALUE",
  56             1002: "BUF_SIZE",
  57             1003: "USER_DATA",
  58             1004: "CAPPED_DATA",
  59             1005: "RETURN_VALUE",
  60             1006: "DEREFERENCE",
  61             1007: "RANGE_CAP",
  62             1008: "LOCK_HELD",
  63             1009: "LOCK_RELEASED",
  64             1010: "ABSOLUTE_LIMITS",
  65             1012: "PARAM_ADD",
  66             1013: "PARAM_FREED",
  67             1014: "DATA_SOURCE",
  68             1015: "FUZZY_MAX",
  69             1016: "STR_LEN",
  70             1017: "ARRAY_LEN",
  71             1018: "CAPABLE",
  72             1019: "NS_CAPABLE",
  73             1022: "TYPE_LINK",
  74             1023: "UNTRACKED_PARAM",
  75             1024: "CULL_PATH",
  76             1025: "PARAM_SET",
  77             1026: "PARAM_USED",
  78             1027: "BYTE_UNITS",
  79             1028: "PARAM_COMPARE",
  80             8017: "USER_DATA2",
  81             8018: "NO_OVERFLOW",
  82             8019: "NO_OVERFLOW_SIMPLE",
  83             8020: "LOCKED",
  84             8021: "UNLOCKED",
  85             8023: "ATOMIC_INC",
  86             8024: "ATOMIC_DEC",
  87 };
  88
  89 def add_range(rl, min_val, max_val):
  90     check_next = 0
  91     done = 0
  92     ret = []
  93     idx = 0
  94
  95     if not ret:
  96         return [[min_val, max_val]]
  97
  98     for idx in range(len(rl)):
  99         cur_min = rl[idx][0]
 100         cur_max = rl[idx][1]
 101
 102         # we already merged the new range but we might need to change later
 103         # ranges if they over lap with more than one
 104         if check_next:
 105             # join with added range
 106             if max_val + 1 == cur_min:
 107                 ret[len(ret) - 1][1] = cur_max
 108                 done = 1
 109                 break
 110             # don't overlap
 111             if max_val < cur_min:
 112                 ret.append([cur_min, cur_max])
 113                 done = 1
 114                 break
 115             # partially overlap
 116             if max_val < cur_max:
 117                 ret[len(ret) - 1][1] = cur_max
 118                 done = 1
 119                 break
 120             # completely overlap
 121             continue
 122
 123         # join 2 ranges into one
 124         if max_val + 1 == cur_min:
 125             ret.append([min_val, cur_max])
 126             done = 1
 127             break
 128         # range is entirely below
 129         if max_val < cur_min:
 130             ret.append([min_val, max_val])
 131             ret.append([cur_min, cur_max])
 132             done = 1
 133             break
 134         # range is partially below
 135         if min_val < cur_min:
 136             if max_val <= cur_max:
 137                 ret.append([min_val, cur_max])
 138                 done = 1
 139                 break
 140             else:
 141                 ret.append([min_val, max_val])
 142                 check_next = 1
 143                 continue
 144         # range already included
 145         if max_val <= cur_max:
 146             ret.append([cur_min, cur_max])
 147             done = 1
 148             break;
 149         # range partially above
 150         if min_val <= cur_max:
 151             ret.append([cur_min, max_val])
 152             check_next = 1
 153             continue
 154         # join 2 ranges on the other side
 155         if min_val - 1 == cur_max:
 156             ret.append([cur_min, max_val])
 157             check_next = 1
 158             continue
 159         # range is above
 160         ret.append([cur_min, cur_max])
 161
 162     if idx + 1 < len(rl):          # we hit a break statement
 163         ret.append(rl[idx + 1:])
 164     elif done:                     # we hit a break on the last iteration
 165         pass
 166     elif not check_next:           # it's past the end of the rl
 167         ret.append([min_val, max_val])
 168
 169     return ret;
 170
 171 def rl_union(rl1, rl2):
 172     ret = []
 173     for r in rl1:
 174         ret = add_range(ret, r[0], r[1])
 175     for r in rl2:
 176         ret = add_range(ret, r[0], r[1])
 177
 178     if (rl1 or rl2) and not ret:
 179         print "bug: merging %s + %s gives empty" %(rl1, rl2)
 180
 181     return ret
 182
 183 def txt_to_val(txt):
 184     if txt == "s64min":
 185         return -(2**63)
 186     elif txt == "s32min":
 187         return -(2**31)
 188     elif txt == "s16min":
 189         return -(2**15)
 190     elif txt == "s64max":
 191         return 2**63 - 1
 192     elif txt == "s32max":
 193         return 2**31 - 1
 194     elif txt == "s16max":
 195         return 2**15 - 1
 196     elif txt == "u64max":
 197         return 2**64 - 1
 198     elif txt == "u32max":
 199         return 2**32 - 1
 200     elif txt == "u16max":
 201         return 2**16 - 1
 202     else:
 203         try:
 204             return int(txt)
 205         except ValueError:
 206             return 0
 207
 208 def val_to_txt(val):
 209     if val == -(2**63):
 210         return "s64min"
 211     elif val == -(2**31):
 212         return "s32min"
 213     elif val == -(2**15):
 214         return "s16min"
 215     elif val == 2**63 - 1:
 216         return "s64max"
 217     elif val == 2**31 - 1:
 218         return "s32max"
 219     elif val == 2**15 - 1:
 220         return "s16max"
 221     elif val == 2**64 - 1:
 222         return "u64max"
 223     elif val == 2**32 - 1:
 224         return "u32max"
 225     elif val == 2**16 - 1:
 226         return "u16max"
 227     elif val < 0:
 228         return "(%d)" %(val)
 229     else:
 230         return "%d" %(val)
 231
 232 def get_next_str(txt):
 233     val = ""
 234     parsed = 0
 235
 236     if txt[0] == '(':
 237         parsed += 1
 238         for char in txt[1:]:
 239             if char == ')':
 240                 break
 241             parsed += 1
 242         val = txt[1:parsed]
 243         parsed += 1
 244     elif txt[0] == 's' or txt[0] == 'u':
 245         parsed += 6
 246         val = txt[:parsed]
 247     else:
 248         if txt[0] == '-':
 249             parsed += 1
 250         for char in txt[parsed:]:
 251             if char == '-':
 252                 break
 253             parsed += 1
 254         val = txt[:parsed]
 255     return [parsed, val]
 256
 257 def txt_to_rl(txt):
 258     ret = []
 259     pairs = txt.split(",")
 260     for pair in pairs:
 261         cnt, min_str = get_next_str(pair)
 262         if cnt == len(pair):
 263             max_str = min_str
 264         else:
 265             cnt, max_str = get_next_str(pair[cnt + 1:])
 266         min_val = txt_to_val(min_str)
 267         max_val = txt_to_val(max_str)
 268         ret.append([min_val, max_val])
 269
 270 #    Hm...  Smatch won't call INT_MAX s32max if the variable is unsigned.
 271 #    if txt != rl_to_txt(ret):
 272 #        print "bug: converting: text = %s rl = %s internal = %s" %(txt, rl_to_txt(ret), ret)
 273
 274     return ret
 275
 276 def rl_to_txt(rl):
 277     ret = ""
 278     for idx in range(len(rl)):
 279         cur_min = rl[idx][0]
 280         cur_max = rl[idx][1]
 281
 282         if idx != 0:
 283             ret += ","
 284
 285         if cur_min == cur_max:
 286             ret += val_to_txt(cur_min)
 287         else:
 288             ret += val_to_txt(cur_min)
 289             ret += "-"
 290             ret += val_to_txt(cur_max)
 291     return ret
 292
 293 def type_to_str(type_int):
 294
 295     t = int(type_int)
 296     if db_types.has_key(t):
 297         return db_types[t]
 298     return type_int
 299
 300 def type_to_int(type_string):
 301     for k in db_types.keys():
 302         if db_types[k] == type_string:
 303             return k
 304     return -1
 305
 306 def display_caller_info(printed, cur):
 307     for txt in cur:
 308         if not printed:
 309             print "file | caller | function | type | parameter | key | value |"
 310         printed = 1
 311         print "%20s | %20s | %20s |" %(txt[0], txt[1], txt[2]),
 312         print " %10s |" %(type_to_str(txt[5])),
 313         print " %d | %s | %s" %(txt[6], txt[7], txt[8])
 314     return printed
 315
 316 def get_caller_info(ptrs, my_type):
 317     cur = con.cursor()
 318     printed = 0
 319     type_filter = ""
 320     if my_type != "":
 321         type_filter = "and type = %d" %(type_to_int(my_type))
 322     for ptr in ptrs:
 323         cur.execute("select * from caller_info where function = '%s' %s;" %(ptr, type_filter))
 324         printed = display_caller_info(printed, cur)
 325
 326 def print_caller_info(func, my_type = ""):
 327     ptrs = get_function_pointers(func)
 328     get_caller_info(ptrs, my_type)
 329
 330 def merge_values(param_names, vals, cur):
 331     for txt in cur:
 332         parameter = int(txt[0])
 333         name = txt[1]
 334         rl = txt_to_rl(txt[2])
 335         if parameter in param_names:
 336             name = name.replace("$", param_names[parameter])
 337
 338         if not parameter in vals:
 339             vals[parameter] = {}
 340
 341         if name in vals[parameter]:
 342             vals[parameter][name] = [vals[parameter][name][0] + 1, rl_union(vals[parameter][name][1], rl)]
 343         else:
 344             vals[parameter][name] = [1, rl]
 345
 346 def get_param_names(filename, func):
 347     cur = con.cursor()
 348     param_names = {}
 349     cur.execute("select parameter, value from parameter_name where file = '%s' and function = '%s';" %(filename, func))
 350     for txt in cur:
 351         parameter = int(txt[0])
 352         name = txt[1]
 353         param_names[parameter] = name
 354     return param_names
 355
 356 def get_caller_count(ptrs):
 357     cur = con.cursor()
 358     count = 0
 359     for ptr in ptrs:
 360         cur.execute("select count(distinct(call_id)) from caller_info where function = '%s';" %(ptr))
 361         for txt in cur:
 362             count += int(txt[0])
 363     return count
 364
 365 def print_merged_caller_values(filename, func, ptrs, param_names, call_cnt):
 366     cur = con.cursor()
 367     vals = {}
 368     for ptr in ptrs:
 369         cur.execute("select parameter, key, value from caller_info where function = '%s' and type = %d;" %(ptr, type_to_int("PARAM_VALUE")))
 370         merge_values(param_names, vals, cur);
 371
 372     for param in sorted(vals):
 373         for name in sorted(vals[param]):
 374             if vals[param][name][0] != call_cnt:
 375                 continue
 376             print "%d %s -> %s" %(param, name, rl_to_txt(vals[param][name][1]))
 377
 378
 379 def print_unmerged_caller_values(filename, func, ptrs, param_names):
 380     cur = con.cursor()
 381     for ptr in ptrs:
 382         prev = -1
 383         cur.execute("select file, caller, call_id, parameter, key, value from caller_info where function = '%s' and type = %d;" %(ptr, type_to_int("PARAM_VALUE")))
 384         for filename, caller, call_id, parameter, name, value in cur:
 385             if prev != int(call_id):
 386                 prev = int(call_id)
 387
 388             parameter = int(parameter)
 389             if param_names[parameter]:
 390                 name = name.replace("$", param_names[parameter])
 391             else:
 392                 name = name.replace("$", "$%d" %(parameter))
 393
 394             print "%s | %s | %s | %s" %(filename, caller, name, value)
 395         print "=========================="
 396
 397 def print_caller_values(filename, func, ptrs):
 398     param_names = get_param_names(filename, func)
 399     call_cnt = get_caller_count(ptrs)
 400
 401     print_merged_caller_values(filename, func, ptrs, param_names, call_cnt)
 402     print "=========================="
 403     print_unmerged_caller_values(filename, func, ptrs, param_names)
 404
 405 def caller_info_values(filename, func):
 406     ptrs = get_function_pointers(func)
 407     print_caller_values(filename, func, ptrs)
 408
 409 def print_return_states(func):
 410     cur = con.cursor()
 411     cur.execute("select * from return_states where function = '%s';" %(func))
 412     count = 0
 413     for txt in cur:
 414         printed = 1
 415         if count == 0:
 416             print "file | function | return_id | return_value | type | param | key | value |"
 417         count += 1
 418         print "%s | %s | %2s | %13s" %(txt[0], txt[1], txt[3], txt[4]),
 419         print "| %13s |" %(type_to_str(txt[6])),
 420         print " %2d | %20s | %20s |" %(txt[7], txt[8], txt[9])
 421
 422 def print_call_implies(func):
 423     cur = con.cursor()
 424     cur.execute("select * from call_implies where function = '%s';" %(func))
 425     count = 0
 426     for txt in cur:
 427         if not count:
 428             print "file | function | type | param | key | value |"
 429         count += 1
 430         print "%15s | %15s" %(txt[0], txt[1]),
 431         print "| %15s" %(type_to_str(txt[4])),
 432         print "| %3d | %15s | %15s |" %(txt[5], txt[6], txt[7])
 433
 434 def print_type_size(struct_type, member):
 435     cur = con.cursor()
 436     cur.execute("select * from type_size where type like '(struct %s)->%s';" %(struct_type, member))
 437     print "type | size"
 438     for txt in cur:
 439         print "%-15s | %s" %(txt[0], txt[1])
 440
 441     cur.execute("select * from function_type_size where type like '(struct %s)->%s';" %(struct_type, member))
 442     print "file | function | type | size"
 443     for txt in cur:
 444         print "%-15s | %-15s | %-15s | %s" %(txt[0], txt[1], txt[2], txt[3])
 445
 446 def print_data_info(struct_type, member):
 447     cur = con.cursor()
 448     cur.execute("select * from data_info where data like '(struct %s)->%s';" %(struct_type, member))
 449     print "file | data | type | value"
 450     for txt in cur:
 451         print "%-15s | %-15s | %-15s | %s" %(txt[0], txt[1], type_to_str(txt[2]), txt[3])
 452
 453 def print_fn_ptrs(func):
 454     ptrs = get_function_pointers(func)
 455     if not ptrs:
 456         return
 457     print "%s = " %(func),
 458     i = 0
 459     for p in ptrs:
 460         if i > 0:
 461             print ",",
 462         i = i + 1
 463         print "'%s'" %(p),
 464     print ""
 465
 466 def print_functions(member):
 467     cur = con.cursor()
 468     cur.execute("select * from function_ptr where ptr like '%%->%s';" %(member))
 469     print "File | Pointer | Function | Static"
 470     for txt in cur:
 471         print "%-15s | %-15s | %-15s | %s" %(txt[0], txt[2], txt[1], txt[3])
 472
 473 def get_callers(func):
 474     ret = []
 475     cur = con.cursor()
 476     ptrs = get_function_pointers(func)
 477     for ptr in ptrs:
 478         cur.execute("select distinct caller from caller_info where function = '%s';" %(ptr))
 479         for row in cur:
 480             ret.append(row[0])
 481     return ret
 482
 483 printed_funcs = []
 484 def call_tree_helper(func, indent = 0):
 485     global printed_funcs
 486     if func in printed_funcs:
 487         return
 488     print "%s%s()" %(" " * indent, func)
 489     if func == "too common":
 490         return
 491     if indent > 6:
 492         return
 493     printed_funcs.append(func)
 494     callers = get_callers(func)
 495     if len(callers) >= 20:
 496         print "Over 20 callers for %s()" %(func)
 497         return
 498     for caller in callers:
 499         call_tree_helper(caller, indent + 2)
 500
 501 def print_call_tree(func):
 502     global printed_funcs
 503     printed_funcs = []
 504     call_tree_helper(func)
 505
 506 def function_type_value(struct_type, member):
 507     cur = con.cursor()
 508     cur.execute("select * from function_type_value where type like '(struct %s)->%s';" %(struct_type, member))
 509     for txt in cur:
 510         print "%-30s | %-30s | %s | %s" %(txt[0], txt[1], txt[2], txt[3])
 511
 512 def trace_callers(func, param):
 513     sources = []
 514     prev_type = 0
 515
 516     cur = con.cursor()
 517     ptrs = get_function_pointers(func)
 518     for ptr in ptrs:
 519         cur.execute("select type, caller, value from caller_info where function = '%s' and (type = 0 or type = 1014 or type = 1028) and (parameter = -1 or parameter = %d);" %(ptr, param))
 520         for row in cur:
 521             data_type = int(row[0])
 522             if data_type == 1014:
 523                 sources.append((row[1], row[2]))
 524             elif data_type == 1028:
 525                 sources.append(("%", row[2])) # hack...
 526             elif data_type == 0 and prev_type == 0:
 527                 sources.append((row[1], ""))
 528             prev_type = data_type
 529     return sources
 530
 531 def trace_param_helper(func, param, indent = 0):
 532     global printed_funcs
 533     if func in printed_funcs:
 534         return
 535     print "%s%s(param %d)" %(" " * indent, func, param)
 536     if func == "too common":
 537         return
 538     if indent > 20:
 539         return
 540     printed_funcs.append(func)
 541     sources = trace_callers(func, param)
 542     for path in sources:
 543
 544         if len(path[1]) and path[1][0] == 'p' and path[1][1] == ' ':
 545             p = int(path[1][2:])
 546             trace_param_helper(path[0], p, indent + 2)
 547         elif len(path[0]) and path[0][0] == '%':
 548             print "  %s%s" %(" " * indent, path[1])
 549         else:
 550             print "* %s%s %s" %(" " * (indent - 1), path[0], path[1])
 551
 552 def trace_param(func, param):
 553     global printed_funcs
 554     printed_funcs = []
 555     print "tracing %s %d" %(func, param)
 556     trace_param_helper(func, param)
 557
 558 if len(sys.argv) < 2:
 559     usage()
 560
 561 if len(sys.argv) == 2:
 562     func = sys.argv[1]
 563     print_caller_info(func)
 564 elif sys.argv[1] == "call_info":
 565     if len(sys.argv) != 4:
 566         usage()
 567     filename = sys.argv[2]
 568     func = sys.argv[3]
 569     caller_info_values(filename, func)
 570     print_caller_info(func)
 571 elif sys.argv[1] == "user_data":
 572     func = sys.argv[2]
 573     print_caller_info(func, "USER_DATA")
 574 elif sys.argv[1] == "param_value":
 575     func = sys.argv[2]
 576     print_caller_info(func, "PARAM_VALUE")
 577 elif sys.argv[1] == "function_ptr" or sys.argv[1] == "fn_ptr":
 578     func = sys.argv[2]
 579     print_fn_ptrs(func)
 580 elif sys.argv[1] == "return_states":
 581     func = sys.argv[2]
 582     print_return_states(func)
 583     print "================================================"
 584     print_call_implies(func)
 585 elif sys.argv[1] == "call_implies":
 586     func = sys.argv[2]
 587     print_call_implies(func)
 588 elif sys.argv[1] == "type_size" or sys.argv[1] == "buf_size":
 589     struct_type = sys.argv[2]
 590     member = sys.argv[3]
 591     print_type_size(struct_type, member)
 592 elif sys.argv[1] == "data_info":
 593     struct_type = sys.argv[2]
 594     member = sys.argv[3]
 595     print_data_info(struct_type, member)
 596 elif sys.argv[1] == "call_tree":
 597     func = sys.argv[2]
 598     print_call_tree(func)
 599 elif sys.argv[1] == "where":
 600     if len(sys.argv) == 3:
 601         struct_type = "%"
 602         member = sys.argv[2]
 603     elif len(sys.argv) == 4:
 604         struct_type = sys.argv[2]
 605         member = sys.argv[3]
 606     function_type_value(struct_type, member)
 607 elif sys.argv[1] == "local":
 608     filename = sys.argv[2]
 609     variable = ""
 610     if len(sys.argv) == 4:
 611         variable = sys.argv[3]
 612     local_values(filename, variable)
 613 elif sys.argv[1] == "functions":
 614     member = sys.argv[2]
 615     print_functions(member)
 616 elif sys.argv[1] == "trace_param":
 617     if len(sys.argv) != 4:
 618         usage()
 619     func = sys.argv[2]
 620     param = int(sys.argv[3])
 621     trace_param(func, param)
 622 else:
 623     usage()