]> git.saurik.com Git - apple/ld64.git/blame - src/other/machochecker.cpp
ld64-278.4.tar.gz
[apple/ld64.git] / src / other / machochecker.cpp
CommitLineData
d696c285
A
1/* -*- mode: C++; c-basic-offset: 4; tab-width: 4 -*-
2 *
a645023d 3 * Copyright (c) 2006-2010 Apple Inc. All rights reserved.
d696c285
A
4 *
5 * @APPLE_LICENSE_HEADER_START@
6 *
7 * This file contains Original Code and/or Modifications of Original Code
8 * as defined in and that are subject to the Apple Public Source License
9 * Version 2.0 (the 'License'). You may not use this file except in
10 * compliance with the License. Please obtain a copy of the License at
11 * http://www.opensource.apple.com/apsl/ and read it before using this
12 * file.
13 *
14 * The Original Code and all software distributed under the License are
15 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
16 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
17 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
19 * Please see the License for the specific language governing rights and
20 * limitations under the License.
21 *
22 * @APPLE_LICENSE_HEADER_END@
23 */
24
25#include <sys/types.h>
26#include <sys/stat.h>
27#include <sys/mman.h>
28#include <stdarg.h>
29#include <stdio.h>
ec29ba20 30#include <stdlib.h>
d696c285
A
31#include <fcntl.h>
32#include <unistd.h>
55e3d2f6 33#include <errno.h>
d696c285
A
34
35#include <vector>
a61fdf0a 36#include <set>
d425e388
A
37#include <unordered_set>
38
39#include "configure.h"
d696c285
A
40
41#include "MachOFileAbstraction.hpp"
42#include "Architectures.hpp"
43
44
45 __attribute__((noreturn))
46void throwf(const char* format, ...)
47{
48 va_list list;
49 char* p;
50 va_start(list, format);
51 vasprintf(&p, format, list);
52 va_end(list);
53
54 const char* t = p;
55 throw t;
56}
57
a645023d
A
58static uint64_t read_uleb128(const uint8_t*& p, const uint8_t* end)
59{
60 uint64_t result = 0;
61 int bit = 0;
62 do {
63 if (p == end)
64 throwf("malformed uleb128");
65
66 uint64_t slice = *p & 0x7f;
67
68 if (bit >= 64 || slice << bit >> bit != slice)
69 throwf("uleb128 too big");
70 else {
71 result |= (slice << bit);
72 bit += 7;
73 }
74 }
75 while (*p++ & 0x80);
76 return result;
77}
d696c285 78
afe874b1
A
79
80static int64_t read_sleb128(const uint8_t*& p, const uint8_t* end)
81{
82 int64_t result = 0;
83 int bit = 0;
84 uint8_t byte;
85 do {
86 if (p == end)
87 throwf("malformed sleb128");
88 byte = *p++;
89 result |= ((byte & 0x7f) << bit);
90 bit += 7;
91 } while (byte & 0x80);
92 // sign extend negative numbers
93 if ( (byte & 0x40) != 0 )
94 result |= (-1LL) << bit;
95 return result;
96}
97
98
d696c285
A
99template <typename A>
100class MachOChecker
101{
102public:
103 static bool validFile(const uint8_t* fileContent);
ec29ba20
A
104 static MachOChecker<A>* make(const uint8_t* fileContent, uint32_t fileLength, const char* path, const char* verifierDstRoot)
105 { return new MachOChecker<A>(fileContent, fileLength, path, verifierDstRoot); }
d696c285
A
106 virtual ~MachOChecker() {}
107
108
109private:
110 typedef typename A::P P;
111 typedef typename A::P::E E;
112 typedef typename A::P::uint_t pint_t;
113
d425e388
A
114 // utility classes for using std::unordered_map with c-strings
115 struct CStringHash {
116 size_t operator()(const char* __s) const {
117 size_t __h = 0;
118 for ( ; *__s; ++__s)
119 __h = 5 * __h + *__s;
120 return __h;
121 };
122 };
123 struct CStringEquals
a61fdf0a 124 {
a61fdf0a
A
125 bool operator()(const char* left, const char* right) const { return (strcmp(left, right) == 0); }
126 };
127
d425e388 128 typedef std::unordered_set<const char*, CStringHash, CStringEquals> StringSet;
a61fdf0a 129
ec29ba20 130 MachOChecker(const uint8_t* fileContent, uint32_t fileLength, const char* path, const char* verifierDstRoot);
d696c285
A
131 void checkMachHeader();
132 void checkLoadCommands();
133 void checkSection(const macho_segment_command<P>* segCmd, const macho_section<P>* sect);
134 uint8_t loadCommandSizeMask();
a61fdf0a 135 void checkSymbolTable();
afe874b1 136 void checkInitTerms();
d696c285 137 void checkIndirectSymbolTable();
69a49097
A
138 void checkRelocations();
139 void checkExternalReloation(const macho_relocation_info<P>* reloc);
140 void checkLocalReloation(const macho_relocation_info<P>* reloc);
ec29ba20
A
141 void verify();
142 void verifyInstallName();
143 void verifyNoRpaths();
144 void verifyNoFlatLookups();
145
69a49097
A
146 pint_t relocBase();
147 bool addressInWritableSegment(pint_t address);
a645023d
A
148 bool hasTextRelocInRange(pint_t start, pint_t end);
149 pint_t segStartAddress(uint8_t segIndex);
afe874b1
A
150 bool addressIsRebaseSite(pint_t addr);
151 bool addressIsBindingSite(pint_t addr);
152 pint_t getInitialStackPointer(const macho_thread_command<P>*);
153 pint_t getEntryPoint(const macho_thread_command<P>*);
ec29ba20 154 const char* archName();
afe874b1
A
155
156
d696c285 157 const char* fPath;
ec29ba20 158 const char* fDstRoot;
d696c285
A
159 const macho_header<P>* fHeader;
160 uint32_t fLength;
ec29ba20 161 const char* fInstallName;
d696c285
A
162 const char* fStrings;
163 const char* fStringsEnd;
164 const macho_nlist<P>* fSymbols;
165 uint32_t fSymbolCount;
a61fdf0a 166 const macho_dysymtab_command<P>* fDynamicSymbolTable;
d696c285
A
167 const uint32_t* fIndirectTable;
168 uint32_t fIndirectTableCount;
69a49097
A
169 const macho_relocation_info<P>* fLocalRelocations;
170 uint32_t fLocalRelocationsCount;
171 const macho_relocation_info<P>* fExternalRelocations;
172 uint32_t fExternalRelocationsCount;
69a49097 173 bool fWriteableSegmentWithAddrOver4G;
afe874b1 174 bool fSlidableImage;
ec29ba20 175 bool fHasLC_RPATH;
69a49097
A
176 const macho_segment_command<P>* fFirstSegment;
177 const macho_segment_command<P>* fFirstWritableSegment;
afe874b1 178 const macho_segment_command<P>* fTEXTSegment;
a645023d 179 const macho_dyld_info_command<P>* fDyldInfo;
c211e7c9 180 uint32_t fSectionCount;
a645023d 181 std::vector<const macho_segment_command<P>*>fSegments;
d696c285
A
182};
183
184
d696c285
A
185template <>
186bool MachOChecker<x86>::validFile(const uint8_t* fileContent)
187{
188 const macho_header<P>* header = (const macho_header<P>*)fileContent;
189 if ( header->magic() != MH_MAGIC )
190 return false;
191 if ( header->cputype() != CPU_TYPE_I386 )
192 return false;
193 switch (header->filetype()) {
194 case MH_EXECUTE:
195 case MH_DYLIB:
196 case MH_BUNDLE:
197 case MH_DYLINKER:
198 return true;
199 }
200 return false;
201}
202
69a49097
A
203template <>
204bool MachOChecker<x86_64>::validFile(const uint8_t* fileContent)
205{
206 const macho_header<P>* header = (const macho_header<P>*)fileContent;
207 if ( header->magic() != MH_MAGIC_64 )
208 return false;
209 if ( header->cputype() != CPU_TYPE_X86_64 )
210 return false;
211 switch (header->filetype()) {
212 case MH_EXECUTE:
213 case MH_DYLIB:
214 case MH_BUNDLE:
215 case MH_DYLINKER:
216 return true;
217 }
218 return false;
219}
d696c285 220
ec29ba20 221#if SUPPORT_ARCH_arm_any
2f2f92e4
A
222template <>
223bool MachOChecker<arm>::validFile(const uint8_t* fileContent)
224{
225 const macho_header<P>* header = (const macho_header<P>*)fileContent;
226 if ( header->magic() != MH_MAGIC )
227 return false;
228 if ( header->cputype() != CPU_TYPE_ARM )
229 return false;
230 switch (header->filetype()) {
231 case MH_EXECUTE:
232 case MH_DYLIB:
233 case MH_BUNDLE:
234 case MH_DYLINKER:
235 return true;
236 }
237 return false;
238}
ec29ba20 239#endif
d696c285 240
f80fe69f
A
241#if SUPPORT_ARCH_arm64
242template <>
243bool MachOChecker<arm64>::validFile(const uint8_t* fileContent)
244{
245 const macho_header<P>* header = (const macho_header<P>*)fileContent;
246 if ( header->magic() != MH_MAGIC_64 )
247 return false;
248 if ( header->cputype() != CPU_TYPE_ARM64 )
249 return false;
250 switch (header->filetype()) {
251 case MH_EXECUTE:
252 case MH_DYLIB:
253 case MH_BUNDLE:
254 case MH_DYLINKER:
255 return true;
256 }
257 return false;
258}
259#endif
260
0a8dc3df
A
261
262template <> uint8_t MachOChecker<ppc>::loadCommandSizeMask() { return 0x03; }
263template <> uint8_t MachOChecker<ppc64>::loadCommandSizeMask() { return 0x07; }
d696c285 264template <> uint8_t MachOChecker<x86>::loadCommandSizeMask() { return 0x03; }
69a49097 265template <> uint8_t MachOChecker<x86_64>::loadCommandSizeMask() { return 0x07; }
2f2f92e4 266template <> uint8_t MachOChecker<arm>::loadCommandSizeMask() { return 0x03; }
f80fe69f 267template <> uint8_t MachOChecker<arm64>::loadCommandSizeMask() { return 0x07; }
d696c285 268
afe874b1 269
afe874b1
A
270template <>
271x86::P::uint_t MachOChecker<x86>::getInitialStackPointer(const macho_thread_command<x86::P>* threadInfo)
272{
273 return threadInfo->thread_register(7);
274}
275
276template <>
277x86_64::P::uint_t MachOChecker<x86_64>::getInitialStackPointer(const macho_thread_command<x86_64::P>* threadInfo)
278{
279 return threadInfo->thread_register(7);
280}
281
282template <>
283arm::P::uint_t MachOChecker<arm>::getInitialStackPointer(const macho_thread_command<arm::P>* threadInfo)
284{
285 return threadInfo->thread_register(13);
286}
287
f80fe69f
A
288template <>
289arm64::P::uint_t MachOChecker<arm64>::getInitialStackPointer(const macho_thread_command<arm64::P>* threadInfo)
290{
291 throw "LC_UNIXTHREAD not supported for arm64";
292}
afe874b1 293
afe874b1 294
0a8dc3df
A
295template <>
296ppc::P::uint_t MachOChecker<ppc>::getEntryPoint(const macho_thread_command<ppc::P>* threadInfo)
297{
298 return threadInfo->thread_register(0);
299}
300
301
afe874b1
A
302template <>
303x86::P::uint_t MachOChecker<x86>::getEntryPoint(const macho_thread_command<x86::P>* threadInfo)
304{
305 return threadInfo->thread_register(10);
306}
307
308template <>
309x86_64::P::uint_t MachOChecker<x86_64>::getEntryPoint(const macho_thread_command<x86_64::P>* threadInfo)
310{
311 return threadInfo->thread_register(16);
312}
313
314template <>
315arm::P::uint_t MachOChecker<arm>::getEntryPoint(const macho_thread_command<arm::P>* threadInfo)
316{
317 return threadInfo->thread_register(15);
318}
319
f80fe69f
A
320template <>
321arm64::P::uint_t MachOChecker<arm64>::getEntryPoint(const macho_thread_command<arm64::P>* threadInfo)
322{
323 throw "LC_UNIXTHREAD not supported for arm64";
324}
ec29ba20
A
325
326
327template <typename A>
328const char* MachOChecker<A>::archName()
329{
330 switch ( fHeader->cputype() ) {
331 case CPU_TYPE_I386:
332 return "i386";
333 case CPU_TYPE_X86_64:
334 if ( fHeader->cpusubtype() == CPU_SUBTYPE_X86_64_H )
335 return "x86_64h";
336 else
337 return "x86_64";
338 case CPU_TYPE_ARM:
339 switch ( fHeader->cpusubtype() ) {
340 case CPU_SUBTYPE_ARM_V7:
341 return "armv7";
342 case CPU_SUBTYPE_ARM_V7S:
343 return "armv7s";
344 case CPU_SUBTYPE_ARM_V7K:
345 return "armv7k";
346 }
347 return "arm";
348 case CPU_TYPE_ARM64:
349 return "arm64";
350 }
351 return "unknown";
352}
353
354
d696c285 355template <typename A>
ec29ba20
A
356MachOChecker<A>::MachOChecker(const uint8_t* fileContent, uint32_t fileLength, const char* path, const char* verifierDstRoot)
357 : fHeader(NULL), fLength(fileLength), fInstallName(NULL), fStrings(NULL), fSymbols(NULL), fSymbolCount(0), fDynamicSymbolTable(NULL), fIndirectTableCount(0),
69a49097 358 fLocalRelocations(NULL), fLocalRelocationsCount(0), fExternalRelocations(NULL), fExternalRelocationsCount(0),
ec29ba20 359 fWriteableSegmentWithAddrOver4G(false), fSlidableImage(false), fHasLC_RPATH(false), fFirstSegment(NULL), fFirstWritableSegment(NULL),
afe874b1 360 fTEXTSegment(NULL), fDyldInfo(NULL), fSectionCount(0)
d696c285
A
361{
362 // sanity check
363 if ( ! validFile(fileContent) )
364 throw "not a mach-o file that can be checked";
365
366 fPath = strdup(path);
ec29ba20 367 fDstRoot = verifierDstRoot ? strdup(verifierDstRoot) : NULL;
d696c285
A
368 fHeader = (const macho_header<P>*)fileContent;
369
370 // sanity check header
371 checkMachHeader();
372
373 // check load commands
374 checkLoadCommands();
375
376 checkIndirectSymbolTable();
377
69a49097 378 checkRelocations();
a61fdf0a
A
379
380 checkSymbolTable();
afe874b1
A
381
382 checkInitTerms();
ec29ba20
A
383
384 if ( verifierDstRoot != NULL )
385 verify();
d696c285
A
386}
387
388
389template <typename A>
390void MachOChecker<A>::checkMachHeader()
391{
392 if ( (fHeader->sizeofcmds() + sizeof(macho_header<P>)) > fLength )
393 throw "sizeofcmds in mach_header is larger than file";
394
395 uint32_t flags = fHeader->flags();
ec29ba20 396 const uint32_t invalidBits = MH_INCRLINK | MH_LAZY_INIT | 0xFC000000;
d696c285
A
397 if ( flags & invalidBits )
398 throw "invalid bits in mach_header flags";
a61fdf0a
A
399 if ( (flags & MH_NO_REEXPORTED_DYLIBS) && (fHeader->filetype() != MH_DYLIB) )
400 throw "MH_NO_REEXPORTED_DYLIBS bit of mach_header flags only valid for dylibs";
afe874b1
A
401
402 switch ( fHeader->filetype() ) {
403 case MH_EXECUTE:
404 fSlidableImage = ( flags & MH_PIE );
405 break;
406 case MH_DYLIB:
407 case MH_BUNDLE:
408 fSlidableImage = true;
409 break;
410 default:
411 throw "not a mach-o file type supported by this tool";
412 }
d696c285
A
413}
414
415template <typename A>
416void MachOChecker<A>::checkLoadCommands()
417{
418 // check that all load commands fit within the load command space file
2f2f92e4 419 const macho_encryption_info_command<P>* encryption_info = NULL;
afe874b1 420 const macho_thread_command<P>* threadInfo = NULL;
ebf6f434 421 const macho_entry_point_command<P>* entryPoint = NULL;
d696c285
A
422 const uint8_t* const endOfFile = (uint8_t*)fHeader + fLength;
423 const uint8_t* const endOfLoadCommands = (uint8_t*)fHeader + sizeof(macho_header<P>) + fHeader->sizeofcmds();
424 const uint32_t cmd_count = fHeader->ncmds();
425 const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)fHeader + sizeof(macho_header<P>));
426 const macho_load_command<P>* cmd = cmds;
ec29ba20 427 const macho_dylib_command<P>* dylibID;
d696c285
A
428 for (uint32_t i = 0; i < cmd_count; ++i) {
429 uint32_t size = cmd->cmdsize();
430 if ( (size & this->loadCommandSizeMask()) != 0 )
431 throwf("load command #%d has a unaligned size", i);
432 const uint8_t* endOfCmd = ((uint8_t*)cmd)+cmd->cmdsize();
433 if ( endOfCmd > endOfLoadCommands )
434 throwf("load command #%d extends beyond the end of the load commands", i);
435 if ( endOfCmd > endOfFile )
436 throwf("load command #%d extends beyond the end of the file", i);
437 switch ( cmd->cmd() ) {
438 case macho_segment_command<P>::CMD:
439 case LC_SYMTAB:
d696c285
A
440 case LC_DYSYMTAB:
441 case LC_LOAD_DYLIB:
d696c285 442 case LC_ID_DYLINKER:
ec29ba20 443 case LC_LOAD_DYLINKER:
d696c285
A
444 case macho_routines_command<P>::CMD:
445 case LC_SUB_FRAMEWORK:
d696c285
A
446 case LC_SUB_CLIENT:
447 case LC_TWOLEVEL_HINTS:
448 case LC_PREBIND_CKSUM:
449 case LC_LOAD_WEAK_DYLIB:
2f2f92e4 450 case LC_LAZY_LOAD_DYLIB:
d696c285 451 case LC_UUID:
a61fdf0a
A
452 case LC_REEXPORT_DYLIB:
453 case LC_SEGMENT_SPLIT_INFO:
2f2f92e4 454 case LC_CODE_SIGNATURE:
a645023d
A
455 case LC_LOAD_UPWARD_DYLIB:
456 case LC_VERSION_MIN_MACOSX:
457 case LC_VERSION_MIN_IPHONEOS:
82b4b32b
A
458 case LC_VERSION_MIN_TVOS:
459 case LC_VERSION_MIN_WATCHOS:
ebf6f434
A
460 case LC_FUNCTION_STARTS:
461 case LC_DYLD_ENVIRONMENT:
462 case LC_DATA_IN_CODE:
463 case LC_DYLIB_CODE_SIGN_DRS:
464 case LC_SOURCE_VERSION:
a645023d 465 break;
ec29ba20
A
466 case LC_RPATH:
467 fHasLC_RPATH = true;
468 break;
469 case LC_ID_DYLIB:
470 dylibID = (macho_dylib_command<P>*)cmd;
471 if ( dylibID->name_offset() > size )
472 throwf("malformed mach-o: LC_ID_DYLIB load command has offset (%u) outside its size (%u)", dylibID->name_offset(), size);
473 if ( (dylibID->name_offset() + strlen(dylibID->name()) + 1) > size )
474 throwf("malformed mach-o: LC_ID_DYLIB load command string extends beyond end of load command");
475 fInstallName = dylibID->name();
476 break;
55e3d2f6
A
477 case LC_DYLD_INFO:
478 case LC_DYLD_INFO_ONLY:
a645023d 479 fDyldInfo = (macho_dyld_info_command<P>*)cmd;
2f2f92e4
A
480 break;
481 case LC_ENCRYPTION_INFO:
f80fe69f 482 case LC_ENCRYPTION_INFO_64:
2f2f92e4 483 encryption_info = (macho_encryption_info_command<P>*)cmd;
a61fdf0a
A
484 break;
485 case LC_SUB_UMBRELLA:
486 case LC_SUB_LIBRARY:
487 if ( fHeader->flags() & MH_NO_REEXPORTED_DYLIBS )
488 throw "MH_NO_REEXPORTED_DYLIBS bit of mach_header flags should not be set in an image with LC_SUB_LIBRARY or LC_SUB_UMBRELLA";
d696c285 489 break;
ebf6f434
A
490 case LC_MAIN:
491 if ( fHeader->filetype() != MH_EXECUTE )
492 throw "LC_MAIN can only be used in MH_EXECUTE file types";
493 entryPoint = (macho_entry_point_command<P>*)cmd;
494 break;
afe874b1
A
495 case LC_UNIXTHREAD:
496 if ( fHeader->filetype() != MH_EXECUTE )
497 throw "LC_UNIXTHREAD can only be used in MH_EXECUTE file types";
498 threadInfo = (macho_thread_command<P>*)cmd;
499 break;
d696c285
A
500 default:
501 throwf("load command #%d is an unknown kind 0x%X", i, cmd->cmd());
502 }
503 cmd = (const macho_load_command<P>*)endOfCmd;
504 }
505
506 // check segments
507 cmd = cmds;
508 std::vector<std::pair<pint_t, pint_t> > segmentAddressRanges;
509 std::vector<std::pair<pint_t, pint_t> > segmentFileOffsetRanges;
510 const macho_segment_command<P>* linkEditSegment = NULL;
afe874b1 511 const macho_segment_command<P>* stackSegment = NULL;
d696c285
A
512 for (uint32_t i = 0; i < cmd_count; ++i) {
513 if ( cmd->cmd() == macho_segment_command<P>::CMD ) {
514 const macho_segment_command<P>* segCmd = (const macho_segment_command<P>*)cmd;
a645023d 515 fSegments.push_back(segCmd);
d696c285
A
516 if ( segCmd->cmdsize() != (sizeof(macho_segment_command<P>) + segCmd->nsects() * sizeof(macho_section_content<P>)) )
517 throw "invalid segment load command size";
518
519 // see if this overlaps another segment address range
520 uint64_t startAddr = segCmd->vmaddr();
521 uint64_t endAddr = startAddr + segCmd->vmsize();
522 for (typename std::vector<std::pair<pint_t, pint_t> >::iterator it = segmentAddressRanges.begin(); it != segmentAddressRanges.end(); ++it) {
523 if ( it->first < startAddr ) {
524 if ( it->second > startAddr )
525 throw "overlapping segment vm addresses";
526 }
527 else if ( it->first > startAddr ) {
528 if ( it->first < endAddr )
529 throw "overlapping segment vm addresses";
530 }
531 else {
532 throw "overlapping segment vm addresses";
533 }
ec29ba20 534 segmentAddressRanges.push_back(std::make_pair(startAddr, endAddr));
d696c285
A
535 }
536 // see if this overlaps another segment file offset range
537 uint64_t startOffset = segCmd->fileoff();
538 uint64_t endOffset = startOffset + segCmd->filesize();
539 for (typename std::vector<std::pair<pint_t, pint_t> >::iterator it = segmentFileOffsetRanges.begin(); it != segmentFileOffsetRanges.end(); ++it) {
540 if ( it->first < startOffset ) {
541 if ( it->second > startOffset )
542 throw "overlapping segment file data";
543 }
544 else if ( it->first > startOffset ) {
545 if ( it->first < endOffset )
546 throw "overlapping segment file data";
547 }
548 else {
549 throw "overlapping segment file data";
550 }
ec29ba20 551 segmentFileOffsetRanges.push_back(std::make_pair(startOffset, endOffset));
d696c285
A
552 // check is within file bounds
553 if ( (startOffset > fLength) || (endOffset > fLength) )
554 throw "segment file data is past end of file";
555 }
556 // verify it fits in file
557 if ( startOffset > fLength )
558 throw "segment fileoff does not fit in file";
559 if ( endOffset > fLength )
560 throw "segment fileoff+filesize does not fit in file";
561
afe874b1 562 // record special segments
d696c285
A
563 if ( strcmp(segCmd->segname(), "__LINKEDIT") == 0 )
564 linkEditSegment = segCmd;
afe874b1
A
565 else if ( strcmp(segCmd->segname(), "__UNIXSTACK") == 0 )
566 stackSegment = segCmd;
69a49097
A
567
568 // cache interesting segments
569 if ( fFirstSegment == NULL )
570 fFirstSegment = segCmd;
afe874b1
A
571 if ( (fTEXTSegment == NULL) && (strcmp(segCmd->segname(), "__TEXT") == 0) )
572 fTEXTSegment = segCmd;
a61fdf0a
A
573 if ( (segCmd->initprot() & VM_PROT_WRITE) != 0 ) {
574 if ( fFirstWritableSegment == NULL )
575 fFirstWritableSegment = segCmd;
576 if ( segCmd->vmaddr() > 0x100000000ULL )
577 fWriteableSegmentWithAddrOver4G = true;
578 }
afe874b1 579
d696c285
A
580 // check section ranges
581 const macho_section<P>* const sectionsStart = (macho_section<P>*)((char*)segCmd + sizeof(macho_segment_command<P>));
582 const macho_section<P>* const sectionsEnd = &sectionsStart[segCmd->nsects()];
583 for(const macho_section<P>* sect = sectionsStart; sect < sectionsEnd; ++sect) {
afe874b1 584 // check all non-zero sized sections are within segment
d696c285
A
585 if ( sect->addr() < startAddr )
586 throwf("section %s vm address not within segment", sect->sectname());
587 if ( (sect->addr()+sect->size()) > endAddr )
588 throwf("section %s vm address not within segment", sect->sectname());
a645023d
A
589 if ( ((sect->flags() & SECTION_TYPE) != S_ZEROFILL)
590 && ((sect->flags() & SECTION_TYPE) != S_THREAD_LOCAL_ZEROFILL)
afe874b1
A
591 && (segCmd->filesize() != 0)
592 && (sect->size() != 0) ) {
d696c285
A
593 if ( sect->offset() < startOffset )
594 throwf("section %s file offset not within segment", sect->sectname());
595 if ( (sect->offset()+sect->size()) > endOffset )
596 throwf("section %s file offset not within segment", sect->sectname());
597 }
598 checkSection(segCmd, sect);
c211e7c9 599 ++fSectionCount;
d696c285
A
600 }
601 }
602 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
603 }
604
605 // verify there was a LINKEDIT segment
606 if ( linkEditSegment == NULL )
607 throw "no __LINKEDIT segment";
608
afe874b1
A
609 // verify there was an executable __TEXT segment and load commands are in it
610 if ( fTEXTSegment == NULL )
611 throw "no __TEXT segment";
612 if ( fTEXTSegment->initprot() != (VM_PROT_READ|VM_PROT_EXECUTE) )
613 throw "__TEXT segment does not have r-x init permissions";
614 //if ( fTEXTSegment->maxprot() != (VM_PROT_READ|VM_PROT_EXECUTE|VM_PROT_WRITE) )
615 // throw "__TEXT segment does not have rwx max permissions";
616 if ( fTEXTSegment->fileoff() != 0 )
617 throw "__TEXT segment does not start at mach_header";
618 if ( fTEXTSegment->filesize() < (sizeof(macho_header<P>)+fHeader->sizeofcmds()) )
619 throw "__TEXT segment smaller than load commands";
620
621 // verify if custom stack used, that stack is in __UNIXSTACK segment
622 if ( threadInfo != NULL ) {
623 pint_t initialSP = getInitialStackPointer(threadInfo);
624 if ( initialSP != 0 ) {
625 if ( stackSegment == NULL )
626 throw "LC_UNIXTHREAD specifics custom initial stack pointer, but no __UNIXSTACK segment";
627 if ( (initialSP < stackSegment->vmaddr()) || (initialSP > (stackSegment->vmaddr()+stackSegment->vmsize())) )
628 throw "LC_UNIXTHREAD specifics custom initial stack pointer which does not point into __UNIXSTACK segment";
629 }
630 }
631
632 // verify __UNIXSTACK is zero fill
633 if ( stackSegment != NULL ) {
634 if ( (stackSegment->filesize() != 0) || (stackSegment->fileoff() != 0) )
635 throw "__UNIXSTACK is not a zero-fill segment";
636 if ( stackSegment->vmsize() < 4096 )
637 throw "__UNIXSTACK segment is too small";
638 }
639
640 // verify entry point is in __TEXT segment
641 if ( threadInfo != NULL ) {
642 pint_t initialPC = getEntryPoint(threadInfo);
643 if ( (initialPC < fTEXTSegment->vmaddr()) || (initialPC >= (fTEXTSegment->vmaddr()+fTEXTSegment->vmsize())) )
644 throwf("entry point 0x%0llX is outside __TEXT segment", (long long)initialPC);
645 }
ebf6f434
A
646 else if ( entryPoint != NULL ) {
647 pint_t initialOffset = entryPoint->entryoff();
648 if ( (initialOffset < fTEXTSegment->fileoff()) || (initialOffset >= (fTEXTSegment->fileoff()+fTEXTSegment->filesize())) )
649 throwf("entry point 0x%0llX is outside __TEXT segment", (long long)initialOffset);
650 }
afe874b1 651
d696c285
A
652 // checks for executables
653 bool isStaticExecutable = false;
654 if ( fHeader->filetype() == MH_EXECUTE ) {
655 isStaticExecutable = true;
656 cmd = cmds;
657 for (uint32_t i = 0; i < cmd_count; ++i) {
658 switch ( cmd->cmd() ) {
659 case LC_LOAD_DYLINKER:
660 // the existence of a dyld load command makes a executable dynamic
661 isStaticExecutable = false;
662 break;
663 }
664 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
665 }
666 if ( isStaticExecutable ) {
ebf6f434 667 if ( (fHeader->flags() != MH_NOUNDEFS) && (fHeader->flags() != (MH_NOUNDEFS|MH_PIE)) )
d696c285
A
668 throw "invalid bits in mach_header flags for static executable";
669 }
670 }
671
2f2f92e4
A
672 // verify encryption info
673 if ( encryption_info != NULL ) {
0a8dc3df
A
674 switch ( fHeader->filetype() ) {
675 case MH_EXECUTE: case MH_DYLIB: case MH_BUNDLE:
676 break; // okay
677 default:
678 throw "LC_ENCRYPTION_INFO load command is not allowed in this file type";
679 }
2f2f92e4
A
680 if ( encryption_info->cryptoff() < (sizeof(macho_header<P>) + fHeader->sizeofcmds()) )
681 throw "LC_ENCRYPTION_INFO load command has cryptoff covers some load commands";
682 if ( (encryption_info->cryptoff() % 4096) != 0 )
683 throw "LC_ENCRYPTION_INFO load command has cryptoff which is not page aligned";
684 if ( (encryption_info->cryptsize() % 4096) != 0 )
685 throw "LC_ENCRYPTION_INFO load command has cryptsize which is not page sized";
686 for (typename std::vector<std::pair<pint_t, pint_t> >::iterator it = segmentFileOffsetRanges.begin();
687 it != segmentFileOffsetRanges.end(); ++it) {
688 if ( (it->first <= encryption_info->cryptoff()) && (encryption_info->cryptoff() < it->second) ) {
689 if ( (encryption_info->cryptoff() + encryption_info->cryptsize()) > it->second )
690 throw "LC_ENCRYPTION_INFO load command is not contained within one segment";
691 }
692 }
693 }
694
ec29ba20
A
695 // verify dylib has LC_ID_DYLIB
696 if ( fHeader->filetype() == MH_DYLIB ) {
697 if ( fInstallName == NULL )
698 throw "MH_DYLIB missing LC_ID_DYLIB";
699 }
700 else {
701 if ( fInstallName != NULL )
702 throw "LC_ID_DYLIB found but file type is not MH_DYLIB";
703 }
704
a61fdf0a 705 // check LC_SYMTAB, LC_DYSYMTAB, and LC_SEGMENT_SPLIT_INFO
d696c285
A
706 cmd = cmds;
707 bool foundDynamicSymTab = false;
708 for (uint32_t i = 0; i < cmd_count; ++i) {
709 switch ( cmd->cmd() ) {
710 case LC_SYMTAB:
711 {
712 const macho_symtab_command<P>* symtab = (macho_symtab_command<P>*)cmd;
713 fSymbolCount = symtab->nsyms();
714 fSymbols = (const macho_nlist<P>*)((char*)fHeader + symtab->symoff());
715 if ( symtab->symoff() < linkEditSegment->fileoff() )
716 throw "symbol table not in __LINKEDIT";
ec29ba20
A
717 if ( (symtab->symoff() + fSymbolCount*sizeof(macho_nlist<P>*)) > symtab->stroff() )
718 throw "symbol table overlaps string pool";
a61fdf0a
A
719 if ( (symtab->symoff() % sizeof(pint_t)) != 0 )
720 throw "symbol table start not pointer aligned";
d696c285
A
721 fStrings = (char*)fHeader + symtab->stroff();
722 fStringsEnd = fStrings + symtab->strsize();
723 if ( symtab->stroff() < linkEditSegment->fileoff() )
724 throw "string pool not in __LINKEDIT";
725 if ( (symtab->stroff()+symtab->strsize()) > (linkEditSegment->fileoff()+linkEditSegment->filesize()) )
726 throw "string pool extends beyond __LINKEDIT";
a61fdf0a
A
727 if ( (symtab->stroff() % 4) != 0 ) // work around until rdar://problem/4737991 is fixed
728 throw "string pool start not pointer aligned";
d696c285
A
729 }
730 break;
731 case LC_DYSYMTAB:
732 {
ebf6f434 733 if ( isStaticExecutable &&! fSlidableImage )
d696c285
A
734 throw "LC_DYSYMTAB should not be used in static executable";
735 foundDynamicSymTab = true;
a645023d 736 fDynamicSymbolTable = (macho_dysymtab_command<P>*)cmd;
a61fdf0a
A
737 fIndirectTable = (uint32_t*)((char*)fHeader + fDynamicSymbolTable->indirectsymoff());
738 fIndirectTableCount = fDynamicSymbolTable->nindirectsyms();
74cfe461 739 if ( fIndirectTableCount != 0 ) {
a61fdf0a 740 if ( fDynamicSymbolTable->indirectsymoff() < linkEditSegment->fileoff() )
74cfe461 741 throw "indirect symbol table not in __LINKEDIT";
a61fdf0a 742 if ( (fDynamicSymbolTable->indirectsymoff()+fIndirectTableCount*8) > (linkEditSegment->fileoff()+linkEditSegment->filesize()) )
74cfe461 743 throw "indirect symbol table not in __LINKEDIT";
a61fdf0a
A
744 if ( (fDynamicSymbolTable->indirectsymoff() % sizeof(pint_t)) != 0 )
745 throw "indirect symbol table not pointer aligned";
74cfe461 746 }
a61fdf0a 747 fLocalRelocationsCount = fDynamicSymbolTable->nlocrel();
69a49097 748 if ( fLocalRelocationsCount != 0 ) {
a61fdf0a
A
749 fLocalRelocations = (const macho_relocation_info<P>*)((char*)fHeader + fDynamicSymbolTable->locreloff());
750 if ( fDynamicSymbolTable->locreloff() < linkEditSegment->fileoff() )
69a49097 751 throw "local relocations not in __LINKEDIT";
a61fdf0a 752 if ( (fDynamicSymbolTable->locreloff()+fLocalRelocationsCount*sizeof(macho_relocation_info<P>)) > (linkEditSegment->fileoff()+linkEditSegment->filesize()) )
69a49097 753 throw "local relocations not in __LINKEDIT";
a61fdf0a
A
754 if ( (fDynamicSymbolTable->locreloff() % sizeof(pint_t)) != 0 )
755 throw "local relocations table not pointer aligned";
69a49097 756 }
a61fdf0a 757 fExternalRelocationsCount = fDynamicSymbolTable->nextrel();
69a49097 758 if ( fExternalRelocationsCount != 0 ) {
a61fdf0a
A
759 fExternalRelocations = (const macho_relocation_info<P>*)((char*)fHeader + fDynamicSymbolTable->extreloff());
760 if ( fDynamicSymbolTable->extreloff() < linkEditSegment->fileoff() )
761 throw "external relocations not in __LINKEDIT";
762 if ( (fDynamicSymbolTable->extreloff()+fExternalRelocationsCount*sizeof(macho_relocation_info<P>)) > (linkEditSegment->fileoff()+linkEditSegment->filesize()) )
763 throw "external relocations not in __LINKEDIT";
764 if ( (fDynamicSymbolTable->extreloff() % sizeof(pint_t)) != 0 )
765 throw "external relocations table not pointer aligned";
69a49097 766 }
d696c285
A
767 }
768 break;
a61fdf0a
A
769 case LC_SEGMENT_SPLIT_INFO:
770 {
771 if ( isStaticExecutable )
772 throw "LC_SEGMENT_SPLIT_INFO should not be used in static executable";
a645023d 773 const macho_linkedit_data_command<P>* info = (macho_linkedit_data_command<P>*)cmd;
a61fdf0a
A
774 if ( info->dataoff() < linkEditSegment->fileoff() )
775 throw "split seg info not in __LINKEDIT";
776 if ( (info->dataoff()+info->datasize()) > (linkEditSegment->fileoff()+linkEditSegment->filesize()) )
777 throw "split seg info not in __LINKEDIT";
778 if ( (info->dataoff() % sizeof(pint_t)) != 0 )
779 throw "split seg info table not pointer aligned";
780 if ( (info->datasize() % sizeof(pint_t)) != 0 )
781 throw "split seg info size not a multiple of pointer size";
782 }
783 break;
a645023d
A
784 case LC_FUNCTION_STARTS:
785 {
786 const macho_linkedit_data_command<P>* info = (macho_linkedit_data_command<P>*)cmd;
787 if ( info->dataoff() < linkEditSegment->fileoff() )
788 throw "function starts data not in __LINKEDIT";
789 if ( (info->dataoff()+info->datasize()) > (linkEditSegment->fileoff()+linkEditSegment->filesize()) )
790 throw "function starts data not in __LINKEDIT";
791 if ( (info->dataoff() % sizeof(pint_t)) != 0 )
792 throw "function starts data table not pointer aligned";
793 if ( (info->datasize() % sizeof(pint_t)) != 0 )
794 throw "function starts data size not a multiple of pointer size";
795 }
796 break;
ebf6f434
A
797 case LC_DATA_IN_CODE:
798 {
799 const macho_linkedit_data_command<P>* info = (macho_linkedit_data_command<P>*)cmd;
800 if ( info->dataoff() < linkEditSegment->fileoff() )
801 throw "data-in-code data not in __LINKEDIT";
802 if ( (info->dataoff()+info->datasize()) > (linkEditSegment->fileoff()+linkEditSegment->filesize()) )
803 throw "data-in-code data not in __LINKEDIT";
804 if ( (info->dataoff() % sizeof(pint_t)) != 0 )
805 throw "data-in-code data table not pointer aligned";
806 if ( (info->datasize() % sizeof(pint_t)) != 0 )
807 throw "data-in-code data size not a multiple of pointer size";
808 }
809 break;
810 case LC_DYLIB_CODE_SIGN_DRS:
811 {
812 const macho_linkedit_data_command<P>* info = (macho_linkedit_data_command<P>*)cmd;
813 if ( info->dataoff() < linkEditSegment->fileoff() )
814 throw "dependent dylib DR data not in __LINKEDIT";
815 if ( (info->dataoff()+info->datasize()) > (linkEditSegment->fileoff()+linkEditSegment->filesize()) )
816 throw "dependent dylib DR data not in __LINKEDIT";
817 if ( (info->dataoff() % sizeof(pint_t)) != 0 )
818 throw "dependent dylib DR data table not pointer aligned";
819 if ( (info->datasize() % sizeof(pint_t)) != 0 )
820 throw "dependent dylib DR data size not a multiple of pointer size";
821 }
822 break;
d696c285
A
823 }
824 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
825 }
826 if ( !isStaticExecutable && !foundDynamicSymTab )
827 throw "missing dynamic symbol table";
828 if ( fStrings == NULL )
829 throw "missing symbol table";
a61fdf0a 830
d696c285
A
831}
832
833template <typename A>
834void MachOChecker<A>::checkSection(const macho_segment_command<P>* segCmd, const macho_section<P>* sect)
835{
836 uint8_t sectionType = (sect->flags() & SECTION_TYPE);
837 if ( sectionType == S_ZEROFILL ) {
838 if ( sect->offset() != 0 )
839 throwf("section offset should be zero for zero-fill section %s", sect->sectname());
840 }
841
afe874b1
A
842 // check section's segment name matches segment
843// if ( strncmp(sect->segname(), segCmd->segname(), 16) != 0 )
844// throwf("section %s in segment %s has wrong segment name", sect->sectname(), segCmd->segname());
845
d696c285
A
846 // more section tests here
847}
848
afe874b1 849
ec29ba20
A
850template <typename A>
851void MachOChecker<A>::verify()
852{
853 bool sharedCacheCandidate = false;
854 if ( fInstallName != NULL ) {
855 if ( (strncmp(fInstallName, "/usr/lib/", 9) == 0) || (strncmp(fInstallName, "/System/Library/", 16) == 0) ) {
856 sharedCacheCandidate = true;
857 verifyInstallName();
858 verifyNoRpaths();
859 }
860 }
861 verifyNoFlatLookups();
862}
863
864
865template <typename A>
866void MachOChecker<A>::verifyInstallName()
867{
868 // Don't allow @rpath to be used as -install_name for OS dylibs
869 if ( strncmp(fInstallName, "@rpath/", 7) == 0 ) {
870 printf("os_dylib_rpath_install_name\tfatal\t-install_name uses @rpath in arch %s\n", archName());
871 }
872 else {
873 // Verify -install_name match actual path of dylib
874 const char* installPathWithinDstRoot = &fPath[strlen(fDstRoot)];
875 if ( strcmp(installPathWithinDstRoot, fInstallName) != 0 ) {
876 // see if install name is a symlink to actual file
877 bool symlinkToDylib = false;
878 char absDstPath[PATH_MAX];
879 if ( realpath(fDstRoot, absDstPath) != NULL ) {
880 char fullInstallNamePath[PATH_MAX];
881 strlcpy(fullInstallNamePath, absDstPath, PATH_MAX);
882 strlcat(fullInstallNamePath, fInstallName, PATH_MAX);
883 char absInstallNamePath[PATH_MAX];
884 if ( realpath(fullInstallNamePath, absInstallNamePath) != NULL ) {
885 char absFPath[PATH_MAX];
886 if ( realpath(fPath, absFPath) != NULL ) {
887 if ( strcmp(absInstallNamePath, absFPath) == 0 )
888 symlinkToDylib = true;
889 }
890 }
891 }
892 if ( !symlinkToDylib )
893 printf("os_dylib_bad_install_name\twarn\t-install_name does not match install location in arch %s\n", archName());
894 }
895 }
afe874b1 896
ec29ba20
A
897}
898
899template <typename A>
900void MachOChecker<A>::verifyNoRpaths()
901{
902 // Don't allow OS dylibs to add rpaths
903 if ( fHasLC_RPATH ) {
904 printf("os_dylib_rpath\twarn\tcontains LC_RPATH load command in arch %s\n", archName());
905 }
906}
907
908
909template <typename A>
910void MachOChecker<A>::verifyNoFlatLookups()
911{
912 if ( (fHeader->flags() & MH_TWOLEVEL) == 0 ) {
913 printf("os_dylib_flat_namespace\twarn\tbuilt with -flat_namespace in arch %s\n", archName());
914 return;
915 }
916
917 if ( fDynamicSymbolTable != NULL ) {
918 const macho_nlist<P>* const undefinesStart = &fSymbols[fDynamicSymbolTable->iundefsym()];
919 const macho_nlist<P>* const undefinesEnd = &undefinesStart[fDynamicSymbolTable->nundefsym()];
920 for(const macho_nlist<P>* sym = undefinesStart; sym < undefinesEnd; ++sym) {
921 //printf("0x%04X %s\n", sym->n_desc(), &fStrings[sym->n_strx()]);
922 if ( GET_LIBRARY_ORDINAL(sym->n_desc()) == DYNAMIC_LOOKUP_ORDINAL ) {
923 const char* symName = &fStrings[sym->n_strx()];
924 printf("os_dylib_undefined_dynamic_lookup\twarn\tbuilt with -undefined dynamic_lookup for symbol %s in arch %s\n", symName, archName());
925 }
926 }
927 }
928}
afe874b1 929
d696c285
A
930template <typename A>
931void MachOChecker<A>::checkIndirectSymbolTable()
932{
a645023d
A
933 // static executables don't have indirect symbol table
934 if ( fDynamicSymbolTable == NULL )
935 return;
d696c285
A
936 const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)fHeader + sizeof(macho_header<P>));
937 const uint32_t cmd_count = fHeader->ncmds();
938 const macho_load_command<P>* cmd = cmds;
939 for (uint32_t i = 0; i < cmd_count; ++i) {
940 if ( cmd->cmd() == macho_segment_command<P>::CMD ) {
941 const macho_segment_command<P>* segCmd = (const macho_segment_command<P>*)cmd;
942 const macho_section<P>* const sectionsStart = (macho_section<P>*)((char*)segCmd + sizeof(macho_segment_command<P>));
943 const macho_section<P>* const sectionsEnd = &sectionsStart[segCmd->nsects()];
944 for(const macho_section<P>* sect = sectionsStart; sect < sectionsEnd; ++sect) {
945 // make sure all magic sections that use indirect symbol table fit within it
946 uint32_t start = 0;
947 uint32_t elementSize = 0;
948 switch ( sect->flags() & SECTION_TYPE ) {
949 case S_SYMBOL_STUBS:
950 elementSize = sect->reserved2();
951 start = sect->reserved1();
952 break;
953 case S_LAZY_SYMBOL_POINTERS:
954 case S_NON_LAZY_SYMBOL_POINTERS:
955 elementSize = sizeof(pint_t);
956 start = sect->reserved1();
957 break;
958 }
959 if ( elementSize != 0 ) {
960 uint32_t count = sect->size() / elementSize;
961 if ( (count*elementSize) != sect->size() )
962 throwf("%s section size is not an even multiple of element size", sect->sectname());
963 if ( (start+count) > fIndirectTableCount )
964 throwf("%s section references beyond end of indirect symbol table (%d > %d)", sect->sectname(), start+count, fIndirectTableCount );
965 }
966 }
967 }
968 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
969 }
ec29ba20
A
970
971
972 if ( fDynamicSymbolTable->ilocalsym() != 0 )
973 throwf("start of local symbols (%d) not at start of symbol table", fDynamicSymbolTable->ilocalsym());
974
975 if ( fDynamicSymbolTable->ilocalsym() > fSymbolCount )
976 throwf("start of local symbols out of range (%d > %d) in indirect symbol table", fDynamicSymbolTable->ilocalsym(), fSymbolCount);
977 if ( fDynamicSymbolTable->ilocalsym() + fDynamicSymbolTable->nlocalsym() > fSymbolCount ) {
978 throwf("local symbols out of range (%d+%d > %d) in indirect symbol table",
979 fDynamicSymbolTable->ilocalsym(), fDynamicSymbolTable->nlocalsym(), fSymbolCount);
980 }
981
982 if ( fDynamicSymbolTable->iextdefsym() > fSymbolCount )
983 throwf("start of extern symbols out of range (%d > %d) in indirect symbol table", fDynamicSymbolTable->iextdefsym(), fSymbolCount);
984 if ( fDynamicSymbolTable->iextdefsym() != fDynamicSymbolTable->ilocalsym() + fDynamicSymbolTable->nlocalsym() ) {
985 throwf("start of extern symbols (%d) not contiguous to local symbols (%d+%d) in indirect symbol table",
986 fDynamicSymbolTable->iextdefsym(), fDynamicSymbolTable->ilocalsym(), fDynamicSymbolTable->nlocalsym() );
987 }
988 if ( fDynamicSymbolTable->iextdefsym() + fDynamicSymbolTable->nextdefsym() > fSymbolCount ) {
989 throwf("extern symbols out of range (%d+%d > %d) in indirect symbol table",
990 fDynamicSymbolTable->iextdefsym(), fDynamicSymbolTable->nextdefsym(), fSymbolCount);
991 }
992
993 if ( fDynamicSymbolTable->iundefsym() > fSymbolCount )
994 throwf("start of undefined symbols out of range (%d > %d) in indirect symbol table", fDynamicSymbolTable->iundefsym(), fSymbolCount);
995 if ( fDynamicSymbolTable->iundefsym() != fDynamicSymbolTable->iextdefsym() + fDynamicSymbolTable->nextdefsym() ) {
996 throwf("start of undefined symbols (%d) not contiguous to extern symbols (%d+%d) in indirect symbol table",
997 fDynamicSymbolTable->iundefsym(), fDynamicSymbolTable->iextdefsym(), fDynamicSymbolTable->nextdefsym());
998 }
999 if ( fDynamicSymbolTable->iundefsym() + fDynamicSymbolTable->nundefsym() > fSymbolCount ) {
1000 throwf("undefined symbols out of range (%d+%d > %d) in indirect symbol table",
1001 fDynamicSymbolTable->iundefsym(), fDynamicSymbolTable->nundefsym(), fSymbolCount);
1002 }
1003
1004 if ( fDynamicSymbolTable->iundefsym() + fDynamicSymbolTable->nundefsym() != fSymbolCount ) {
1005 throwf("end undefined symbols (%d+%d) not at end of all symbols (%d) in indirect symbol table",
1006 fDynamicSymbolTable->iundefsym(), fDynamicSymbolTable->nundefsym(), fSymbolCount );
1007 }
d696c285
A
1008}
1009
1010
afe874b1
A
1011
1012
a61fdf0a
A
1013template <typename A>
1014void MachOChecker<A>::checkSymbolTable()
1015{
1016 // verify no duplicate external symbol names
1017 if ( fDynamicSymbolTable != NULL ) {
1018 StringSet externalNames;
1019 const macho_nlist<P>* const exportedStart = &fSymbols[fDynamicSymbolTable->iextdefsym()];
1020 const macho_nlist<P>* const exportedEnd = &exportedStart[fDynamicSymbolTable->nextdefsym()];
55e3d2f6
A
1021 int i = fDynamicSymbolTable->iextdefsym();
1022 for(const macho_nlist<P>* p = exportedStart; p < exportedEnd; ++p, ++i) {
a61fdf0a 1023 const char* symName = &fStrings[p->n_strx()];
55e3d2f6
A
1024 if ( symName > fStringsEnd )
1025 throw "string index out of range";
1026 //fprintf(stderr, "sym[%d] = %s\n", i, symName);
a61fdf0a
A
1027 if ( externalNames.find(symName) != externalNames.end() )
1028 throwf("duplicate external symbol: %s", symName);
a645023d
A
1029 if ( (p->n_type() & N_EXT) == 0 )
1030 throwf("non-external symbol in external symbol range: %s", symName);
1031 // don't add N_INDR to externalNames because there is likely an undefine with same name
1032 if ( (p->n_type() & N_INDR) == 0 )
1033 externalNames.insert(symName);
a61fdf0a 1034 }
55e3d2f6
A
1035 // verify no undefines with same name as an external symbol
1036 const macho_nlist<P>* const undefinesStart = &fSymbols[fDynamicSymbolTable->iundefsym()];
1037 const macho_nlist<P>* const undefinesEnd = &undefinesStart[fDynamicSymbolTable->nundefsym()];
1038 for(const macho_nlist<P>* p = undefinesStart; p < undefinesEnd; ++p) {
1039 const char* symName = &fStrings[p->n_strx()];
1040 if ( symName > fStringsEnd )
1041 throw "string index out of range";
1042 if ( externalNames.find(symName) != externalNames.end() )
1043 throwf("undefine with same name as external symbol: %s", symName);
1044 }
c211e7c9
A
1045 // verify all N_SECT values are valid
1046 for(const macho_nlist<P>* p = fSymbols; p < &fSymbols[fSymbolCount]; ++p) {
1047 uint8_t type = p->n_type();
1048 if ( ((type & N_STAB) == 0) && ((type & N_TYPE) == N_SECT) ) {
1049 if ( p->n_sect() > fSectionCount ) {
1050 throwf("symbol '%s' has n_sect=%d which is too large", &fStrings[p->n_strx()], p->n_sect());
1051 }
1052 }
1053 }
a61fdf0a
A
1054 }
1055}
1056
69a49097 1057
afe874b1
A
1058template <typename A>
1059void MachOChecker<A>::checkInitTerms()
1060{
1061 const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)fHeader + sizeof(macho_header<P>));
1062 const uint32_t cmd_count = fHeader->ncmds();
1063 const macho_load_command<P>* cmd = cmds;
1064 for (uint32_t i = 0; i < cmd_count; ++i) {
1065 if ( cmd->cmd() == macho_segment_command<P>::CMD ) {
1066 const macho_segment_command<P>* segCmd = (const macho_segment_command<P>*)cmd;
1067 const macho_section<P>* const sectionsStart = (macho_section<P>*)((char*)segCmd + sizeof(macho_segment_command<P>));
1068 const macho_section<P>* const sectionsEnd = &sectionsStart[segCmd->nsects()];
1069 for(const macho_section<P>* sect = sectionsStart; sect < sectionsEnd; ++sect) {
1070 // make sure all magic sections that use indirect symbol table fit within it
1071 uint32_t count;
1072 pint_t* arrayStart;
1073 pint_t* arrayEnd;
1074 const char* kind = "initializer";
1075 switch ( sect->flags() & SECTION_TYPE ) {
1076 case S_MOD_TERM_FUNC_POINTERS:
1077 kind = "terminator";
1078 // fall through
1079 case S_MOD_INIT_FUNC_POINTERS:
1080 count = sect->size() / sizeof(pint_t);
1081 if ( (count*sizeof(pint_t)) != sect->size() )
1082 throwf("%s section size is not an even multiple of element size", sect->sectname());
1083 if ( (sect->addr() % sizeof(pint_t)) != 0 )
1084 throwf("%s section size is not pointer size aligned", sect->sectname());
1085 // check each pointer in array points within TEXT
1086 arrayStart = (pint_t*)((char*)fHeader + sect->offset());
1087 arrayEnd = (pint_t*)((char*)fHeader + sect->offset() + sect->size());
1088 for (pint_t* p=arrayStart; p < arrayEnd; ++p) {
1089 pint_t pointer = P::getP(*p);
1090 if ( (pointer < fTEXTSegment->vmaddr()) || (pointer >= (fTEXTSegment->vmaddr()+fTEXTSegment->vmsize())) )
1091 throwf("%s 0x%08llX points outside __TEXT segment", kind, (long long)pointer);
1092 }
1093 // check each pointer in array will be rebased and not bound
1094 if ( fSlidableImage ) {
1095 pint_t sectionBeginAddr = sect->addr();
1096 pint_t sectionEndddr = sect->addr() + sect->size();
1097 for(pint_t addr = sectionBeginAddr; addr < sectionEndddr; addr += sizeof(pint_t)) {
1098 if ( addressIsBindingSite(addr) )
1099 throwf("%s at 0x%0llX has binding to external symbol", kind, (long long)addr);
1100 if ( ! addressIsRebaseSite(addr) )
1101 throwf("%s at 0x%0llX is not rebased", kind, (long long)addr);
1102 }
1103 }
1104 break;
1105 }
1106 }
1107 }
1108 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
1109 }
1110
1111}
1112
1113
69a49097
A
1114
1115template <>
1116x86::P::uint_t MachOChecker<x86>::relocBase()
1117{
1118 if ( fHeader->flags() & MH_SPLIT_SEGS )
1119 return fFirstWritableSegment->vmaddr();
1120 else
1121 return fFirstSegment->vmaddr();
1122}
1123
1124template <>
1125x86_64::P::uint_t MachOChecker<x86_64>::relocBase()
1126{
1127 // check for split-seg
1128 return fFirstWritableSegment->vmaddr();
1129}
1130
2f2f92e4
A
1131template <>
1132arm::P::uint_t MachOChecker<arm>::relocBase()
1133{
1134 if ( fHeader->flags() & MH_SPLIT_SEGS )
1135 return fFirstWritableSegment->vmaddr();
1136 else
1137 return fFirstSegment->vmaddr();
1138}
1139
f80fe69f
A
1140template <>
1141arm64::P::uint_t MachOChecker<arm64>::relocBase()
1142{
1143 return fFirstWritableSegment->vmaddr();
1144}
69a49097 1145
0a8dc3df 1146
69a49097
A
1147template <typename A>
1148bool MachOChecker<A>::addressInWritableSegment(pint_t address)
1149{
1150 const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)fHeader + sizeof(macho_header<P>));
1151 const uint32_t cmd_count = fHeader->ncmds();
1152 const macho_load_command<P>* cmd = cmds;
1153 for (uint32_t i = 0; i < cmd_count; ++i) {
1154 if ( cmd->cmd() == macho_segment_command<P>::CMD ) {
1155 const macho_segment_command<P>* segCmd = (const macho_segment_command<P>*)cmd;
2f2f92e4
A
1156 if ( (address >= segCmd->vmaddr()) && (address < segCmd->vmaddr()+segCmd->vmsize()) ) {
1157 // if segment is writable, we are fine
1158 if ( (segCmd->initprot() & VM_PROT_WRITE) != 0 )
69a49097 1159 return true;
2f2f92e4
A
1160 // could be a text reloc, make sure section bit is set
1161 const macho_section<P>* const sectionsStart = (macho_section<P>*)((char*)segCmd + sizeof(macho_segment_command<P>));
1162 const macho_section<P>* const sectionsEnd = &sectionsStart[segCmd->nsects()];
1163 for(const macho_section<P>* sect = sectionsStart; sect < sectionsEnd; ++sect) {
1164 if ( (sect->addr() <= address) && (address < (sect->addr()+sect->size())) ) {
1165 // found section for this address, if has relocs we are fine
1166 return ( (sect->flags() & (S_ATTR_EXT_RELOC|S_ATTR_LOC_RELOC)) != 0 );
1167 }
1168 }
69a49097
A
1169 }
1170 }
1171 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
1172 }
1173 return false;
1174}
1175
1176
69a49097
A
1177
1178template <>
1179void MachOChecker<x86>::checkExternalReloation(const macho_relocation_info<P>* reloc)
1180{
a61fdf0a
A
1181 if ( reloc->r_length() != 2 )
1182 throw "bad external relocation length";
1183 if ( reloc->r_type() != GENERIC_RELOC_VANILLA )
1184 throw "unknown external relocation type";
1185 if ( reloc->r_pcrel() != 0 )
1186 throw "bad external relocation pc_rel";
1187 if ( reloc->r_extern() == 0 )
1188 throw "local relocation found with external relocations";
1189 if ( ! this->addressInWritableSegment(reloc->r_address() + this->relocBase()) )
1190 throw "external relocation address not in writable segment";
1191 // FIX: check r_symbol
69a49097
A
1192}
1193
1194
1195template <>
1196void MachOChecker<x86_64>::checkExternalReloation(const macho_relocation_info<P>* reloc)
1197{
1198 if ( reloc->r_length() != 3 )
1199 throw "bad external relocation length";
1200 if ( reloc->r_type() != X86_64_RELOC_UNSIGNED )
1201 throw "unknown external relocation type";
1202 if ( reloc->r_pcrel() != 0 )
1203 throw "bad external relocation pc_rel";
1204 if ( reloc->r_extern() == 0 )
1205 throw "local relocation found with external relocations";
1206 if ( ! this->addressInWritableSegment(reloc->r_address() + this->relocBase()) )
1207 throw "exernal relocation address not in writable segment";
1208 // FIX: check r_symbol
1209}
1210
ebf6f434 1211#if SUPPORT_ARCH_arm_any
2f2f92e4
A
1212template <>
1213void MachOChecker<arm>::checkExternalReloation(const macho_relocation_info<P>* reloc)
1214{
1215 if ( reloc->r_length() != 2 )
1216 throw "bad external relocation length";
ec29ba20 1217 if ( reloc->r_type() != ARM_RELOC_VANILLA )
2f2f92e4
A
1218 throw "unknown external relocation type";
1219 if ( reloc->r_pcrel() != 0 )
1220 throw "bad external relocation pc_rel";
1221 if ( reloc->r_extern() == 0 )
1222 throw "local relocation found with external relocations";
1223 if ( ! this->addressInWritableSegment(reloc->r_address() + this->relocBase()) )
1224 throw "external relocation address not in writable segment";
1225 // FIX: check r_symbol
1226}
ebf6f434 1227#endif
2f2f92e4 1228
f80fe69f
A
1229#if SUPPORT_ARCH_arm64
1230template <>
1231void MachOChecker<arm64>::checkExternalReloation(const macho_relocation_info<P>* reloc)
1232{
1233 throw "external relocations not used for arm64";
1234}
1235#endif
1236
2f2f92e4 1237
69a49097
A
1238template <>
1239void MachOChecker<x86>::checkLocalReloation(const macho_relocation_info<P>* reloc)
1240{
1241 // FIX
1242}
1243
1244template <>
1245void MachOChecker<x86_64>::checkLocalReloation(const macho_relocation_info<P>* reloc)
1246{
1247 if ( reloc->r_length() != 3 )
1248 throw "bad local relocation length";
1249 if ( reloc->r_type() != X86_64_RELOC_UNSIGNED )
1250 throw "unknown local relocation type";
1251 if ( reloc->r_pcrel() != 0 )
1252 throw "bad local relocation pc_rel";
1253 if ( reloc->r_extern() != 0 )
1254 throw "external relocation found with local relocations";
1255 if ( ! this->addressInWritableSegment(reloc->r_address() + this->relocBase()) )
1256 throw "local relocation address not in writable segment";
1257}
1258
ebf6f434 1259#if SUPPORT_ARCH_arm_any
2f2f92e4
A
1260template <>
1261void MachOChecker<arm>::checkLocalReloation(const macho_relocation_info<P>* reloc)
1262{
1263 if ( reloc->r_address() & R_SCATTERED ) {
1264 // scattered
1265 const macho_scattered_relocation_info<P>* sreloc = (const macho_scattered_relocation_info<P>*)reloc;
1266 if ( sreloc->r_length() != 2 )
1267 throw "bad local scattered relocation length";
1268 if ( sreloc->r_type() != ARM_RELOC_PB_LA_PTR )
1269 throw "bad local scattered relocation type";
1270 }
1271 else {
1272 if ( reloc->r_length() != 2 )
1273 throw "bad local relocation length";
1274 if ( reloc->r_extern() != 0 )
1275 throw "external relocation found with local relocations";
1276 if ( ! this->addressInWritableSegment(reloc->r_address() + this->relocBase()) )
1277 throw "local relocation address not in writable segment";
1278 }
1279}
ebf6f434 1280#endif
69a49097 1281
f80fe69f
A
1282#if SUPPORT_ARCH_arm64
1283template <>
1284void MachOChecker<arm64>::checkLocalReloation(const macho_relocation_info<P>* reloc)
1285{
1286 throw "local relocations not used for arm64";
1287}
1288#endif
1289
0a8dc3df 1290
69a49097
A
1291template <typename A>
1292void MachOChecker<A>::checkRelocations()
1293{
a61fdf0a
A
1294 // external relocations should be sorted to minimize dyld symbol lookups
1295 // therefore every reloc with the same r_symbolnum value should be contiguous
1296 std::set<uint32_t> previouslySeenSymbolIndexes;
1297 uint32_t lastSymbolIndex = 0xFFFFFFFF;
69a49097
A
1298 const macho_relocation_info<P>* const externRelocsEnd = &fExternalRelocations[fExternalRelocationsCount];
1299 for (const macho_relocation_info<P>* reloc = fExternalRelocations; reloc < externRelocsEnd; ++reloc) {
1300 this->checkExternalReloation(reloc);
a61fdf0a
A
1301 if ( reloc->r_symbolnum() != lastSymbolIndex ) {
1302 if ( previouslySeenSymbolIndexes.count(reloc->r_symbolnum()) != 0 )
1303 throw "external relocations not sorted";
1304 previouslySeenSymbolIndexes.insert(lastSymbolIndex);
1305 lastSymbolIndex = reloc->r_symbolnum();
1306 }
69a49097
A
1307 }
1308
1309 const macho_relocation_info<P>* const localRelocsEnd = &fLocalRelocations[fLocalRelocationsCount];
1310 for (const macho_relocation_info<P>* reloc = fLocalRelocations; reloc < localRelocsEnd; ++reloc) {
1311 this->checkLocalReloation(reloc);
1312 }
a645023d
A
1313
1314 // verify any section with S_ATTR_LOC_RELOC bits set actually has text relocs
1315 const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)fHeader + sizeof(macho_header<P>));
1316 const uint32_t cmd_count = fHeader->ncmds();
1317 const macho_load_command<P>* cmd = cmds;
1318 for (uint32_t i = 0; i < cmd_count; ++i) {
1319 if ( cmd->cmd() == macho_segment_command<P>::CMD ) {
1320 const macho_segment_command<P>* segCmd = (const macho_segment_command<P>*)cmd;
1321 // if segment is writable, we are fine
1322 if ( (segCmd->initprot() & VM_PROT_WRITE) != 0 )
1323 continue;
1324 // look at sections that have text reloc bit set
1325 const macho_section<P>* const sectionsStart = (macho_section<P>*)((char*)segCmd + sizeof(macho_segment_command<P>));
1326 const macho_section<P>* const sectionsEnd = &sectionsStart[segCmd->nsects()];
1327 for(const macho_section<P>* sect = sectionsStart; sect < sectionsEnd; ++sect) {
1328 if ( (sect->flags() & S_ATTR_LOC_RELOC) != 0 ) {
1329 if ( ! hasTextRelocInRange(sect->addr(), sect->addr()+sect->size()) ) {
1330 throwf("section %s has attribute set that it has relocs, but it has none", sect->sectname());
1331 }
1332 }
1333 }
1334 }
1335 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
1336 }
69a49097
A
1337}
1338
a645023d
A
1339template <typename A>
1340typename A::P::uint_t MachOChecker<A>::segStartAddress(uint8_t segIndex)
1341{
1342 if ( segIndex > fSegments.size() )
1343 throw "segment index out of range";
1344 return fSegments[segIndex]->vmaddr();
1345}
1346
1347template <typename A>
1348bool MachOChecker<A>::hasTextRelocInRange(pint_t rangeStart, pint_t rangeEnd)
1349{
1350 // look at local relocs
1351 const macho_relocation_info<P>* const localRelocsEnd = &fLocalRelocations[fLocalRelocationsCount];
1352 for (const macho_relocation_info<P>* reloc = fLocalRelocations; reloc < localRelocsEnd; ++reloc) {
1353 pint_t relocAddress = reloc->r_address() + this->relocBase();
1354 if ( (rangeStart <= relocAddress) && (relocAddress < rangeEnd) )
1355 return true;
1356 }
1357 // look rebase info
1358 if ( fDyldInfo != NULL ) {
1359 const uint8_t* p = (uint8_t*)fHeader + fDyldInfo->rebase_off();
1360 const uint8_t* end = &p[fDyldInfo->rebase_size()];
1361
1362 uint8_t type = 0;
1363 uint64_t segOffset = 0;
1364 uint32_t count;
1365 uint32_t skip;
1366 int segIndex;
1367 pint_t segStartAddr = 0;
1368 pint_t addr;
1369 bool done = false;
1370 while ( !done && (p < end) ) {
1371 uint8_t immediate = *p & REBASE_IMMEDIATE_MASK;
1372 uint8_t opcode = *p & REBASE_OPCODE_MASK;
1373 ++p;
1374 switch (opcode) {
1375 case REBASE_OPCODE_DONE:
1376 done = true;
1377 break;
1378 case REBASE_OPCODE_SET_TYPE_IMM:
1379 type = immediate;
1380 break;
1381 case REBASE_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB:
1382 segIndex = immediate;
1383 segStartAddr = segStartAddress(segIndex);
1384 segOffset = read_uleb128(p, end);
1385 break;
1386 case REBASE_OPCODE_ADD_ADDR_ULEB:
1387 segOffset += read_uleb128(p, end);
1388 break;
1389 case REBASE_OPCODE_ADD_ADDR_IMM_SCALED:
1390 segOffset += immediate*sizeof(pint_t);
1391 break;
1392 case REBASE_OPCODE_DO_REBASE_IMM_TIMES:
1393 for (int i=0; i < immediate; ++i) {
1394 addr = segStartAddr+segOffset;
1395 if ( (rangeStart <= addr) && (addr < rangeEnd) )
1396 return true;
1397 //printf("%-7s %-16s 0x%08llX %s\n", segName, sectionName(segIndex, segStartAddr+segOffset), segStartAddr+segOffset, typeName);
1398 segOffset += sizeof(pint_t);
1399 }
1400 break;
1401 case REBASE_OPCODE_DO_REBASE_ULEB_TIMES:
1402 count = read_uleb128(p, end);
1403 for (uint32_t i=0; i < count; ++i) {
1404 addr = segStartAddr+segOffset;
1405 if ( (rangeStart <= addr) && (addr < rangeEnd) )
1406 return true;
1407 //printf("%-7s %-16s 0x%08llX %s\n", segName, sectionName(segIndex, segStartAddr+segOffset), segStartAddr+segOffset, typeName);
1408 segOffset += sizeof(pint_t);
1409 }
1410 break;
1411 case REBASE_OPCODE_DO_REBASE_ADD_ADDR_ULEB:
1412 addr = segStartAddr+segOffset;
1413 if ( (rangeStart <= addr) && (addr < rangeEnd) )
1414 return true;
1415 //printf("%-7s %-16s 0x%08llX %s\n", segName, sectionName(segIndex, segStartAddr+segOffset), segStartAddr+segOffset, typeName);
1416 segOffset += read_uleb128(p, end) + sizeof(pint_t);
1417 break;
1418 case REBASE_OPCODE_DO_REBASE_ULEB_TIMES_SKIPPING_ULEB:
1419 count = read_uleb128(p, end);
1420 skip = read_uleb128(p, end);
1421 for (uint32_t i=0; i < count; ++i) {
1422 addr = segStartAddr+segOffset;
1423 if ( (rangeStart <= addr) && (addr < rangeEnd) )
1424 return true;
1425 //printf("%-7s %-16s 0x%08llX %s\n", segName, sectionName(segIndex, segStartAddr+segOffset), segStartAddr+segOffset, typeName);
1426 segOffset += skip + sizeof(pint_t);
1427 }
1428 break;
1429 default:
1430 throwf("bad rebase opcode %d", *p);
1431 }
1432 }
1433 }
ebf6f434 1434 return false;
a645023d 1435}
69a49097 1436
afe874b1
A
1437template <typename A>
1438bool MachOChecker<A>::addressIsRebaseSite(pint_t targetAddr)
1439{
1440 // look at local relocs
1441 const macho_relocation_info<P>* const localRelocsEnd = &fLocalRelocations[fLocalRelocationsCount];
1442 for (const macho_relocation_info<P>* reloc = fLocalRelocations; reloc < localRelocsEnd; ++reloc) {
1443 pint_t relocAddress = reloc->r_address() + this->relocBase();
1444 if ( relocAddress == targetAddr )
1445 return true;
1446 }
1447 // look rebase info
1448 if ( fDyldInfo != NULL ) {
1449 const uint8_t* p = (uint8_t*)fHeader + fDyldInfo->rebase_off();
1450 const uint8_t* end = &p[fDyldInfo->rebase_size()];
1451
1452 uint8_t type = 0;
1453 uint64_t segOffset = 0;
1454 uint32_t count;
1455 uint32_t skip;
1456 int segIndex;
1457 pint_t segStartAddr = 0;
1458 pint_t addr;
1459 bool done = false;
1460 while ( !done && (p < end) ) {
1461 uint8_t immediate = *p & REBASE_IMMEDIATE_MASK;
1462 uint8_t opcode = *p & REBASE_OPCODE_MASK;
1463 ++p;
1464 switch (opcode) {
1465 case REBASE_OPCODE_DONE:
1466 done = true;
1467 break;
1468 case REBASE_OPCODE_SET_TYPE_IMM:
1469 type = immediate;
1470 break;
1471 case REBASE_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB:
1472 segIndex = immediate;
1473 segStartAddr = segStartAddress(segIndex);
1474 segOffset = read_uleb128(p, end);
1475 break;
1476 case REBASE_OPCODE_ADD_ADDR_ULEB:
1477 segOffset += read_uleb128(p, end);
1478 break;
1479 case REBASE_OPCODE_ADD_ADDR_IMM_SCALED:
1480 segOffset += immediate*sizeof(pint_t);
1481 break;
1482 case REBASE_OPCODE_DO_REBASE_IMM_TIMES:
1483 for (int i=0; i < immediate; ++i) {
1484 addr = segStartAddr+segOffset;
1485 if ( addr == targetAddr )
1486 return true;
1487 //printf("%-7s %-16s 0x%08llX %s\n", segName, sectionName(segIndex, segStartAddr+segOffset), segStartAddr+segOffset, typeName);
1488 segOffset += sizeof(pint_t);
1489 }
1490 break;
1491 case REBASE_OPCODE_DO_REBASE_ULEB_TIMES:
1492 count = read_uleb128(p, end);
1493 for (uint32_t i=0; i < count; ++i) {
1494 addr = segStartAddr+segOffset;
1495 if ( addr == targetAddr )
1496 return true;
1497 //printf("%-7s %-16s 0x%08llX %s\n", segName, sectionName(segIndex, segStartAddr+segOffset), segStartAddr+segOffset, typeName);
1498 segOffset += sizeof(pint_t);
1499 }
1500 break;
1501 case REBASE_OPCODE_DO_REBASE_ADD_ADDR_ULEB:
1502 addr = segStartAddr+segOffset;
1503 if ( addr == targetAddr )
1504 return true;
1505 //printf("%-7s %-16s 0x%08llX %s\n", segName, sectionName(segIndex, segStartAddr+segOffset), segStartAddr+segOffset, typeName);
1506 segOffset += read_uleb128(p, end) + sizeof(pint_t);
1507 break;
1508 case REBASE_OPCODE_DO_REBASE_ULEB_TIMES_SKIPPING_ULEB:
1509 count = read_uleb128(p, end);
1510 skip = read_uleb128(p, end);
1511 for (uint32_t i=0; i < count; ++i) {
1512 addr = segStartAddr+segOffset;
1513 if ( addr == targetAddr )
1514 return true;
1515 //printf("%-7s %-16s 0x%08llX %s\n", segName, sectionName(segIndex, segStartAddr+segOffset), segStartAddr+segOffset, typeName);
1516 segOffset += skip + sizeof(pint_t);
1517 }
1518 break;
1519 default:
1520 throwf("bad rebase opcode %d", *p);
1521 }
1522 }
1523 }
1524 return false;
1525}
1526
1527
1528template <typename A>
1529bool MachOChecker<A>::addressIsBindingSite(pint_t targetAddr)
1530{
1531 // look at external relocs
1532 const macho_relocation_info<P>* const externRelocsEnd = &fExternalRelocations[fExternalRelocationsCount];
1533 for (const macho_relocation_info<P>* reloc = fExternalRelocations; reloc < externRelocsEnd; ++reloc) {
1534 pint_t relocAddress = reloc->r_address() + this->relocBase();
1535 if ( relocAddress == targetAddr )
1536 return true;
1537 }
1538 // look bind info
1539 if ( fDyldInfo != NULL ) {
1540 const uint8_t* p = (uint8_t*)fHeader + fDyldInfo->bind_off();
1541 const uint8_t* end = &p[fDyldInfo->bind_size()];
1542
1543 uint8_t type = 0;
1544 uint64_t segOffset = 0;
1545 uint32_t count;
1546 uint32_t skip;
1547 uint8_t flags;
1548 const char* symbolName = NULL;
1549 int libraryOrdinal = 0;
1550 int segIndex;
1551 int64_t addend = 0;
1552 pint_t segStartAddr = 0;
1553 pint_t addr;
1554 bool done = false;
1555 while ( !done && (p < end) ) {
1556 uint8_t immediate = *p & BIND_IMMEDIATE_MASK;
1557 uint8_t opcode = *p & BIND_OPCODE_MASK;
1558 ++p;
1559 switch (opcode) {
1560 case BIND_OPCODE_DONE:
1561 done = true;
1562 break;
1563 case BIND_OPCODE_SET_DYLIB_ORDINAL_IMM:
1564 libraryOrdinal = immediate;
1565 break;
1566 case BIND_OPCODE_SET_DYLIB_ORDINAL_ULEB:
1567 libraryOrdinal = read_uleb128(p, end);
1568 break;
1569 case BIND_OPCODE_SET_DYLIB_SPECIAL_IMM:
1570 // the special ordinals are negative numbers
1571 if ( immediate == 0 )
1572 libraryOrdinal = 0;
1573 else {
1574 int8_t signExtended = BIND_OPCODE_MASK | immediate;
1575 libraryOrdinal = signExtended;
1576 }
1577 break;
1578 case BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM:
1579 symbolName = (char*)p;
1580 while (*p != '\0')
1581 ++p;
1582 ++p;
1583 break;
1584 case BIND_OPCODE_SET_TYPE_IMM:
1585 type = immediate;
1586 break;
1587 case BIND_OPCODE_SET_ADDEND_SLEB:
1588 addend = read_sleb128(p, end);
1589 break;
1590 case BIND_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB:
1591 segIndex = immediate;
1592 segStartAddr = segStartAddress(segIndex);
1593 segOffset = read_uleb128(p, end);
1594 break;
1595 case BIND_OPCODE_ADD_ADDR_ULEB:
1596 segOffset += read_uleb128(p, end);
1597 break;
1598 case BIND_OPCODE_DO_BIND:
1599 if ( (segStartAddr+segOffset) == targetAddr )
1600 return true;
1601 segOffset += sizeof(pint_t);
1602 break;
1603 case BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB:
1604 if ( (segStartAddr+segOffset) == targetAddr )
1605 return true;
1606 segOffset += read_uleb128(p, end) + sizeof(pint_t);
1607 break;
1608 case BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED:
1609 if ( (segStartAddr+segOffset) == targetAddr )
1610 return true;
1611 segOffset += immediate*sizeof(pint_t) + sizeof(pint_t);
1612 break;
1613 case BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB:
1614 count = read_uleb128(p, end);
1615 skip = read_uleb128(p, end);
1616 for (uint32_t i=0; i < count; ++i) {
1617 if ( (segStartAddr+segOffset) == targetAddr )
1618 return true;
1619 segOffset += skip + sizeof(pint_t);
1620 }
1621 break;
1622 default:
1623 throwf("bad bind opcode %d", *p);
1624 }
1625 }
1626 }
1627 return false;
1628}
1629
1630
ec29ba20 1631static void check(const char* path, const char* verifierDstRoot)
d696c285
A
1632{
1633 struct stat stat_buf;
1634
1635 try {
1636 int fd = ::open(path, O_RDONLY, 0);
1637 if ( fd == -1 )
1638 throw "cannot open file";
55e3d2f6
A
1639 if ( ::fstat(fd, &stat_buf) != 0 )
1640 throwf("fstat(%s) failed, errno=%d\n", path, errno);
d696c285 1641 uint32_t length = stat_buf.st_size;
69a49097
A
1642 uint8_t* p = (uint8_t*)::mmap(NULL, stat_buf.st_size, PROT_READ, MAP_FILE | MAP_PRIVATE, fd, 0);
1643 if ( p == ((uint8_t*)(-1)) )
1644 throw "cannot map file";
d696c285
A
1645 ::close(fd);
1646 const mach_header* mh = (mach_header*)p;
1647 if ( mh->magic == OSSwapBigToHostInt32(FAT_MAGIC) ) {
1648 const struct fat_header* fh = (struct fat_header*)p;
1649 const struct fat_arch* archs = (struct fat_arch*)(p + sizeof(struct fat_header));
a61fdf0a
A
1650 for (unsigned long i=0; i < OSSwapBigToHostInt32(fh->nfat_arch); ++i) {
1651 size_t offset = OSSwapBigToHostInt32(archs[i].offset);
1652 size_t size = OSSwapBigToHostInt32(archs[i].size);
1653 unsigned int cputype = OSSwapBigToHostInt32(archs[i].cputype);
1654
1655 switch(cputype) {
a61fdf0a
A
1656 case CPU_TYPE_I386:
1657 if ( MachOChecker<x86>::validFile(p + offset) )
ec29ba20 1658 MachOChecker<x86>::make(p + offset, size, path, verifierDstRoot);
d696c285
A
1659 else
1660 throw "in universal file, i386 slice does not contain i386 mach-o";
a61fdf0a 1661 break;
a61fdf0a
A
1662 case CPU_TYPE_X86_64:
1663 if ( MachOChecker<x86_64>::validFile(p + offset) )
ec29ba20 1664 MachOChecker<x86_64>::make(p + offset, size, path, verifierDstRoot);
69a49097
A
1665 else
1666 throw "in universal file, x86_64 slice does not contain x86_64 mach-o";
a61fdf0a 1667 break;
ebf6f434 1668#if SUPPORT_ARCH_arm_any
2f2f92e4
A
1669 case CPU_TYPE_ARM:
1670 if ( MachOChecker<arm>::validFile(p + offset) )
ec29ba20 1671 MachOChecker<arm>::make(p + offset, size, path, verifierDstRoot);
2f2f92e4
A
1672 else
1673 throw "in universal file, arm slice does not contain arm mach-o";
1674 break;
ec29ba20
A
1675#endif
1676#if SUPPORT_ARCH_arm64
1677 case CPU_TYPE_ARM64:
1678 if ( MachOChecker<arm64>::validFile(p + offset) )
1679 MachOChecker<arm64>::make(p + offset, size, path, verifierDstRoot);
1680 else
1681 throw "in universal file, arm64 slice does not contain arm mach-o";
1682 break;
ebf6f434 1683#endif
a61fdf0a
A
1684 default:
1685 throwf("in universal file, unknown architecture slice 0x%x\n", cputype);
d696c285
A
1686 }
1687 }
1688 }
1689 else if ( MachOChecker<x86>::validFile(p) ) {
ec29ba20 1690 MachOChecker<x86>::make(p, length, path, verifierDstRoot);
d696c285 1691 }
69a49097 1692 else if ( MachOChecker<x86_64>::validFile(p) ) {
ec29ba20 1693 MachOChecker<x86_64>::make(p, length, path, verifierDstRoot);
69a49097 1694 }
ebf6f434 1695#if SUPPORT_ARCH_arm_any
2f2f92e4 1696 else if ( MachOChecker<arm>::validFile(p) ) {
ec29ba20 1697 MachOChecker<arm>::make(p, length, path, verifierDstRoot);
2f2f92e4 1698 }
f80fe69f
A
1699#endif
1700#if SUPPORT_ARCH_arm64
1701 else if ( MachOChecker<arm64>::validFile(p) ) {
ec29ba20 1702 MachOChecker<arm64>::make(p, length, path, verifierDstRoot);
f80fe69f 1703 }
ebf6f434 1704#endif
d696c285
A
1705 else {
1706 throw "not a known file type";
1707 }
1708 }
1709 catch (const char* msg) {
1710 throwf("%s in %s", msg, path);
1711 }
1712}
1713
1714
1715int main(int argc, const char* argv[])
1716{
afe874b1 1717 bool progress = false;
ec29ba20 1718 const char* verifierDstRoot = NULL;
afe874b1
A
1719 int result = 0;
1720 for(int i=1; i < argc; ++i) {
1721 const char* arg = argv[i];
1722 if ( arg[0] == '-' ) {
1723 if ( strcmp(arg, "-progress") == 0 ) {
1724 progress = true;
d696c285 1725 }
ec29ba20
A
1726 else if ( strcmp(arg, "-verifier_dstroot") == 0 ) {
1727 verifierDstRoot = argv[++i];
1728 }
1729 else if ( strcmp(arg, "-verifier_error_list") == 0 ) {
1730 printf("os_dylib_rpath_install_name\tOS dylibs (those in /usr/lib/ or /System/Library/) must be built with -install_name that is an absolute path - not an @rpath\n");
1731 printf("os_dylib_bad_install_name\tOS dylibs (those in /usr/lib/ or /System/Library/) must be built with -install_name matching their file system location\n");
1732 printf("os_dylib_rpath\tOS dylibs should not contain LC_RPATH load commands (from -rpath linker option)\n");
1733 printf("os_dylib_flat_namespace\tOS dylibs should not be built with -flat_namespace\n");
1734 printf("os_dylib_undefined_dynamic_lookup\tOS dylibs should not be built with -undefined dynamic_lookup\n");
1735 printf("os_dylib_malformed\the mach-o is malformed\n");
1736 return 0;
1737 }
d696c285 1738 else {
afe874b1
A
1739 throwf("unknown option: %s\n", arg);
1740 }
1741 }
1742 else {
1743 bool success = true;
1744 try {
ec29ba20 1745 check(arg, verifierDstRoot);
d696c285 1746 }
afe874b1 1747 catch (const char* msg) {
ec29ba20
A
1748 if ( verifierDstRoot ) {
1749 printf("os_dylib_malformed\twarn\t%s\n", msg);
1750 }
1751 else {
1752 fprintf(stderr, "machocheck failed: %s\n", msg);
1753 result = 1;
1754 success = false;
1755 }
afe874b1
A
1756 }
1757 if ( success && progress )
1758 printf("ok: %s\n", arg);
d696c285
A
1759 }
1760 }
d696c285 1761
afe874b1 1762 return result;
d696c285
A
1763}
1764