1 // Copyright (c) 2011 The Chromium Authors. All rights reserved.
2 // Use of this source code is governed by a BSD-style license that can be
3 // found in the LICENSE file.
4
5 #include "courgette/encoded_program.h"
6
7 #include <algorithm>
8 #include <map>
9 #include <string>
10 #include <vector>
11
12 #include "base/environment.h"
13 #include "base/logging.h"
14 #include "base/memory/scoped_ptr.h"
15 #include "base/strings/string_util.h"
16 #include "base/strings/utf_string_conversions.h"
17 #include "courgette/courgette.h"
18 #include "courgette/disassembler_elf_32_arm.h"
19 #include "courgette/streams.h"
20 #include "courgette/types_elf.h"
21
22 namespace courgette {
23
24 // Stream indexes.
25 const int kStreamMisc = 0;
26 const int kStreamOps = 1;
27 const int kStreamBytes = 2;
28 const int kStreamAbs32Indexes = 3;
29 const int kStreamRel32Indexes = 4;
30 const int kStreamAbs32Addresses = 5;
31 const int kStreamRel32Addresses = 6;
32 const int kStreamCopyCounts = 7;
33 const int kStreamOriginAddresses = kStreamMisc;
34
35 const int kStreamLimit = 9;
36
37 // Constructor is here rather than in the header. Although the constructor
38 // appears to do nothing it is fact quite large because of the implicit calls to
39 // field constructors. Ditto for the destructor.
EncodedProgram()40 EncodedProgram::EncodedProgram() : image_base_(0) {}
~EncodedProgram()41 EncodedProgram::~EncodedProgram() {}
42
43 // Serializes a vector of integral values using Varint32 coding.
44 template<typename V>
WriteVector(const V & items,SinkStream * buffer)45 CheckBool WriteVector(const V& items, SinkStream* buffer) {
46 size_t count = items.size();
47 bool ok = buffer->WriteSizeVarint32(count);
48 for (size_t i = 0; ok && i < count; ++i) {
49 COMPILE_ASSERT(sizeof(items[0]) <= sizeof(uint32), // NOLINT
50 T_must_fit_in_uint32);
51 ok = buffer->WriteSizeVarint32(items[i]);
52 }
53 return ok;
54 }
55
56 template<typename V>
ReadVector(V * items,SourceStream * buffer)57 bool ReadVector(V* items, SourceStream* buffer) {
58 uint32 count;
59 if (!buffer->ReadVarint32(&count))
60 return false;
61
62 items->clear();
63
64 bool ok = items->reserve(count);
65 for (size_t i = 0; ok && i < count; ++i) {
66 uint32 item;
67 ok = buffer->ReadVarint32(&item);
68 if (ok)
69 ok = items->push_back(static_cast<typename V::value_type>(item));
70 }
71
72 return ok;
73 }
74
75 // Serializes a vector, using delta coding followed by Varint32 coding.
76 template<typename V>
WriteU32Delta(const V & set,SinkStream * buffer)77 CheckBool WriteU32Delta(const V& set, SinkStream* buffer) {
78 size_t count = set.size();
79 bool ok = buffer->WriteSizeVarint32(count);
80 uint32 prev = 0;
81 for (size_t i = 0; ok && i < count; ++i) {
82 uint32 current = set[i];
83 uint32 delta = current - prev;
84 ok = buffer->WriteVarint32(delta);
85 prev = current;
86 }
87 return ok;
88 }
89
90 template <typename V>
ReadU32Delta(V * set,SourceStream * buffer)91 static CheckBool ReadU32Delta(V* set, SourceStream* buffer) {
92 uint32 count;
93
94 if (!buffer->ReadVarint32(&count))
95 return false;
96
97 set->clear();
98 bool ok = set->reserve(count);
99 uint32 prev = 0;
100
101 for (size_t i = 0; ok && i < count; ++i) {
102 uint32 delta;
103 ok = buffer->ReadVarint32(&delta);
104 if (ok) {
105 uint32 current = prev + delta;
106 ok = set->push_back(current);
107 prev = current;
108 }
109 }
110
111 return ok;
112 }
113
114 // Write a vector as the byte representation of the contents.
115 //
116 // (This only really makes sense for a type T that has sizeof(T)==1, otherwise
117 // serialized representation is not endian-agnostic. But it is useful to keep
118 // the possibility of a greater size for experiments comparing Varint32 encoding
119 // of a vector of larger integrals vs a plain form.)
120 //
121 template<typename V>
WriteVectorU8(const V & items,SinkStream * buffer)122 CheckBool WriteVectorU8(const V& items, SinkStream* buffer) {
123 size_t count = items.size();
124 bool ok = buffer->WriteSizeVarint32(count);
125 if (count != 0 && ok) {
126 size_t byte_count = count * sizeof(typename V::value_type);
127 ok = buffer->Write(static_cast<const void*>(&items[0]), byte_count);
128 }
129 return ok;
130 }
131
132 template<typename V>
ReadVectorU8(V * items,SourceStream * buffer)133 bool ReadVectorU8(V* items, SourceStream* buffer) {
134 uint32 count;
135 if (!buffer->ReadVarint32(&count))
136 return false;
137
138 items->clear();
139 bool ok = items->resize(count, 0);
140 if (ok && count != 0) {
141 size_t byte_count = count * sizeof(typename V::value_type);
142 return buffer->Read(static_cast<void*>(&((*items)[0])), byte_count);
143 }
144 return ok;
145 }
146
147 ////////////////////////////////////////////////////////////////////////////////
148
DefineRel32Label(int index,RVA value)149 CheckBool EncodedProgram::DefineRel32Label(int index, RVA value) {
150 return DefineLabelCommon(&rel32_rva_, index, value);
151 }
152
DefineAbs32Label(int index,RVA value)153 CheckBool EncodedProgram::DefineAbs32Label(int index, RVA value) {
154 return DefineLabelCommon(&abs32_rva_, index, value);
155 }
156
157 static const RVA kUnassignedRVA = static_cast<RVA>(-1);
158
DefineLabelCommon(RvaVector * rvas,int index,RVA rva)159 CheckBool EncodedProgram::DefineLabelCommon(RvaVector* rvas,
160 int index,
161 RVA rva) {
162 bool ok = true;
163 if (static_cast<int>(rvas->size()) <= index)
164 ok = rvas->resize(index + 1, kUnassignedRVA);
165
166 if (ok) {
167 DCHECK_EQ((*rvas)[index], kUnassignedRVA)
168 << "DefineLabel double assigned " << index;
169 (*rvas)[index] = rva;
170 }
171
172 return ok;
173 }
174
EndLabels()175 void EncodedProgram::EndLabels() {
176 FinishLabelsCommon(&abs32_rva_);
177 FinishLabelsCommon(&rel32_rva_);
178 }
179
FinishLabelsCommon(RvaVector * rvas)180 void EncodedProgram::FinishLabelsCommon(RvaVector* rvas) {
181 // Replace all unassigned slots with the value at the previous index so they
182 // delta-encode to zero. (There might be better values than zero. The way to
183 // get that is have the higher level assembly program assign the unassigned
184 // slots.)
185 RVA previous = 0;
186 size_t size = rvas->size();
187 for (size_t i = 0; i < size; ++i) {
188 if ((*rvas)[i] == kUnassignedRVA)
189 (*rvas)[i] = previous;
190 else
191 previous = (*rvas)[i];
192 }
193 }
194
AddOrigin(RVA origin)195 CheckBool EncodedProgram::AddOrigin(RVA origin) {
196 return ops_.push_back(ORIGIN) && origins_.push_back(origin);
197 }
198
AddCopy(uint32 count,const void * bytes)199 CheckBool EncodedProgram::AddCopy(uint32 count, const void* bytes) {
200 const uint8* source = static_cast<const uint8*>(bytes);
201
202 bool ok = true;
203
204 // Fold adjacent COPY instructions into one. This nearly halves the size of
205 // an EncodedProgram with only COPY1 instructions since there are approx plain
206 // 16 bytes per reloc. This has a working-set benefit during decompression.
207 // For compression of files with large differences this makes a small (4%)
208 // improvement in size. For files with small differences this degrades the
209 // compressed size by 1.3%
210 if (!ops_.empty()) {
211 if (ops_.back() == COPY1) {
212 ops_.back() = COPY;
213 ok = copy_counts_.push_back(1);
214 }
215 if (ok && ops_.back() == COPY) {
216 copy_counts_.back() += count;
217 for (uint32 i = 0; ok && i < count; ++i) {
218 ok = copy_bytes_.push_back(source[i]);
219 }
220 return ok;
221 }
222 }
223
224 if (ok) {
225 if (count == 1) {
226 ok = ops_.push_back(COPY1) && copy_bytes_.push_back(source[0]);
227 } else {
228 ok = ops_.push_back(COPY) && copy_counts_.push_back(count);
229 for (uint32 i = 0; ok && i < count; ++i) {
230 ok = copy_bytes_.push_back(source[i]);
231 }
232 }
233 }
234
235 return ok;
236 }
237
AddAbs32(int label_index)238 CheckBool EncodedProgram::AddAbs32(int label_index) {
239 return ops_.push_back(ABS32) && abs32_ix_.push_back(label_index);
240 }
241
AddRel32(int label_index)242 CheckBool EncodedProgram::AddRel32(int label_index) {
243 return ops_.push_back(REL32) && rel32_ix_.push_back(label_index);
244 }
245
AddRel32ARM(uint16 op,int label_index)246 CheckBool EncodedProgram::AddRel32ARM(uint16 op, int label_index) {
247 return ops_.push_back(static_cast<OP>(op)) &&
248 rel32_ix_.push_back(label_index);
249 }
250
AddPeMakeRelocs(ExecutableType kind)251 CheckBool EncodedProgram::AddPeMakeRelocs(ExecutableType kind) {
252 if (kind == EXE_WIN_32_X86)
253 return ops_.push_back(MAKE_PE_RELOCATION_TABLE);
254 return ops_.push_back(MAKE_PE64_RELOCATION_TABLE);
255 }
256
AddElfMakeRelocs()257 CheckBool EncodedProgram::AddElfMakeRelocs() {
258 return ops_.push_back(MAKE_ELF_RELOCATION_TABLE);
259 }
260
AddElfARMMakeRelocs()261 CheckBool EncodedProgram::AddElfARMMakeRelocs() {
262 return ops_.push_back(MAKE_ELF_ARM_RELOCATION_TABLE);
263 }
264
DebuggingSummary()265 void EncodedProgram::DebuggingSummary() {
266 VLOG(1) << "EncodedProgram Summary"
267 << "\n image base " << image_base_
268 << "\n abs32 rvas " << abs32_rva_.size()
269 << "\n rel32 rvas " << rel32_rva_.size()
270 << "\n ops " << ops_.size()
271 << "\n origins " << origins_.size()
272 << "\n copy_counts " << copy_counts_.size()
273 << "\n copy_bytes " << copy_bytes_.size()
274 << "\n abs32_ix " << abs32_ix_.size()
275 << "\n rel32_ix " << rel32_ix_.size();
276 }
277
278 ////////////////////////////////////////////////////////////////////////////////
279
280 // For algorithm refinement purposes it is useful to write subsets of the file
281 // format. This gives us the ability to estimate the entropy of the
282 // differential compression of the individual streams, which can provide
283 // invaluable insights. The default, of course, is to include all the streams.
284 //
285 enum FieldSelect {
286 INCLUDE_ABS32_ADDRESSES = 0x0001,
287 INCLUDE_REL32_ADDRESSES = 0x0002,
288 INCLUDE_ABS32_INDEXES = 0x0010,
289 INCLUDE_REL32_INDEXES = 0x0020,
290 INCLUDE_OPS = 0x0100,
291 INCLUDE_BYTES = 0x0200,
292 INCLUDE_COPY_COUNTS = 0x0400,
293 INCLUDE_MISC = 0x1000
294 };
295
GetFieldSelect()296 static FieldSelect GetFieldSelect() {
297 #if 1
298 // TODO(sra): Use better configuration.
299 scoped_ptr<base::Environment> env(base::Environment::Create());
300 std::string s;
301 env->GetVar("A_FIELDS", &s);
302 if (!s.empty()) {
303 return static_cast<FieldSelect>(
304 wcstoul(base::ASCIIToWide(s).c_str(), 0, 0));
305 }
306 #endif
307 return static_cast<FieldSelect>(~0);
308 }
309
WriteTo(SinkStreamSet * streams)310 CheckBool EncodedProgram::WriteTo(SinkStreamSet* streams) {
311 FieldSelect select = GetFieldSelect();
312
313 // The order of fields must be consistent in WriteTo and ReadFrom, regardless
314 // of the streams used. The code can be configured with all kStreamXXX
315 // constants the same.
316 //
317 // If we change the code to pipeline reading with assembly (to avoid temporary
318 // storage vectors by consuming operands directly from the stream) then we
319 // need to read the base address and the random access address tables first,
320 // the rest can be interleaved.
321
322 if (select & INCLUDE_MISC) {
323 // TODO(sra): write 64 bits.
324 if (!streams->stream(kStreamMisc)->WriteVarint32(
325 static_cast<uint32>(image_base_))) {
326 return false;
327 }
328 }
329
330 bool success = true;
331
332 if (select & INCLUDE_ABS32_ADDRESSES) {
333 success &= WriteU32Delta(abs32_rva_,
334 streams->stream(kStreamAbs32Addresses));
335 }
336
337 if (select & INCLUDE_REL32_ADDRESSES) {
338 success &= WriteU32Delta(rel32_rva_,
339 streams->stream(kStreamRel32Addresses));
340 }
341
342 if (select & INCLUDE_MISC)
343 success &= WriteVector(origins_, streams->stream(kStreamOriginAddresses));
344
345 if (select & INCLUDE_OPS) {
346 // 5 for length.
347 success &= streams->stream(kStreamOps)->Reserve(ops_.size() + 5);
348 success &= WriteVector(ops_, streams->stream(kStreamOps));
349 }
350
351 if (select & INCLUDE_COPY_COUNTS)
352 success &= WriteVector(copy_counts_, streams->stream(kStreamCopyCounts));
353
354 if (select & INCLUDE_BYTES)
355 success &= WriteVectorU8(copy_bytes_, streams->stream(kStreamBytes));
356
357 if (select & INCLUDE_ABS32_INDEXES)
358 success &= WriteVector(abs32_ix_, streams->stream(kStreamAbs32Indexes));
359
360 if (select & INCLUDE_REL32_INDEXES)
361 success &= WriteVector(rel32_ix_, streams->stream(kStreamRel32Indexes));
362
363 return success;
364 }
365
ReadFrom(SourceStreamSet * streams)366 bool EncodedProgram::ReadFrom(SourceStreamSet* streams) {
367 // TODO(sra): read 64 bits.
368 uint32 temp;
369 if (!streams->stream(kStreamMisc)->ReadVarint32(&temp))
370 return false;
371 image_base_ = temp;
372
373 if (!ReadU32Delta(&abs32_rva_, streams->stream(kStreamAbs32Addresses)))
374 return false;
375 if (!ReadU32Delta(&rel32_rva_, streams->stream(kStreamRel32Addresses)))
376 return false;
377 if (!ReadVector(&origins_, streams->stream(kStreamOriginAddresses)))
378 return false;
379 if (!ReadVector(&ops_, streams->stream(kStreamOps)))
380 return false;
381 if (!ReadVector(©_counts_, streams->stream(kStreamCopyCounts)))
382 return false;
383 if (!ReadVectorU8(©_bytes_, streams->stream(kStreamBytes)))
384 return false;
385 if (!ReadVector(&abs32_ix_, streams->stream(kStreamAbs32Indexes)))
386 return false;
387 if (!ReadVector(&rel32_ix_, streams->stream(kStreamRel32Indexes)))
388 return false;
389
390 // Check that streams have been completely consumed.
391 for (int i = 0; i < kStreamLimit; ++i) {
392 if (streams->stream(i)->Remaining() > 0)
393 return false;
394 }
395
396 return true;
397 }
398
399 // Safe, non-throwing version of std::vector::at(). Returns 'true' for success,
400 // 'false' for out-of-bounds index error.
401 template<typename V, typename T>
VectorAt(const V & v,size_t index,T * output)402 bool VectorAt(const V& v, size_t index, T* output) {
403 if (index >= v.size())
404 return false;
405 *output = v[index];
406 return true;
407 }
408
EvaluateRel32ARM(OP op,size_t & ix_rel32_ix,RVA & current_rva,SinkStream * output)409 CheckBool EncodedProgram::EvaluateRel32ARM(OP op,
410 size_t& ix_rel32_ix,
411 RVA& current_rva,
412 SinkStream* output) {
413 switch (op & 0x0000F000) {
414 case REL32ARM8: {
415 uint32 index;
416 if (!VectorAt(rel32_ix_, ix_rel32_ix, &index))
417 return false;
418 ++ix_rel32_ix;
419 RVA rva;
420 if (!VectorAt(rel32_rva_, index, &rva))
421 return false;
422 uint32 decompressed_op;
423 if (!DisassemblerElf32ARM::Decompress(ARM_OFF8,
424 static_cast<uint16>(op),
425 static_cast<uint32>(rva -
426 current_rva),
427 &decompressed_op)) {
428 return false;
429 }
430 uint16 op16 = decompressed_op;
431 if (!output->Write(&op16, 2))
432 return false;
433 current_rva += 2;
434 break;
435 }
436 case REL32ARM11: {
437 uint32 index;
438 if (!VectorAt(rel32_ix_, ix_rel32_ix, &index))
439 return false;
440 ++ix_rel32_ix;
441 RVA rva;
442 if (!VectorAt(rel32_rva_, index, &rva))
443 return false;
444 uint32 decompressed_op;
445 if (!DisassemblerElf32ARM::Decompress(ARM_OFF11, (uint16) op,
446 (uint32) (rva - current_rva),
447 &decompressed_op)) {
448 return false;
449 }
450 uint16 op16 = decompressed_op;
451 if (!output->Write(&op16, 2))
452 return false;
453 current_rva += 2;
454 break;
455 }
456 case REL32ARM24: {
457 uint32 index;
458 if (!VectorAt(rel32_ix_, ix_rel32_ix, &index))
459 return false;
460 ++ix_rel32_ix;
461 RVA rva;
462 if (!VectorAt(rel32_rva_, index, &rva))
463 return false;
464 uint32 decompressed_op;
465 if (!DisassemblerElf32ARM::Decompress(ARM_OFF24, (uint16) op,
466 (uint32) (rva - current_rva),
467 &decompressed_op)) {
468 return false;
469 }
470 if (!output->Write(&decompressed_op, 4))
471 return false;
472 current_rva += 4;
473 break;
474 }
475 case REL32ARM25: {
476 uint32 index;
477 if (!VectorAt(rel32_ix_, ix_rel32_ix, &index))
478 return false;
479 ++ix_rel32_ix;
480 RVA rva;
481 if (!VectorAt(rel32_rva_, index, &rva))
482 return false;
483 uint32 decompressed_op;
484 if (!DisassemblerElf32ARM::Decompress(ARM_OFF25, (uint16) op,
485 (uint32) (rva - current_rva),
486 &decompressed_op)) {
487 return false;
488 }
489 uint32 words = (decompressed_op << 16) | (decompressed_op >> 16);
490 if (!output->Write(&words, 4))
491 return false;
492 current_rva += 4;
493 break;
494 }
495 case REL32ARM21: {
496 uint32 index;
497 if (!VectorAt(rel32_ix_, ix_rel32_ix, &index))
498 return false;
499 ++ix_rel32_ix;
500 RVA rva;
501 if (!VectorAt(rel32_rva_, index, &rva))
502 return false;
503 uint32 decompressed_op;
504 if (!DisassemblerElf32ARM::Decompress(ARM_OFF21, (uint16) op,
505 (uint32) (rva - current_rva),
506 &decompressed_op)) {
507 return false;
508 }
509 uint32 words = (decompressed_op << 16) | (decompressed_op >> 16);
510 if (!output->Write(&words, 4))
511 return false;
512 current_rva += 4;
513 break;
514 }
515 default:
516 return false;
517 }
518
519 return true;
520 }
521
AssembleTo(SinkStream * final_buffer)522 CheckBool EncodedProgram::AssembleTo(SinkStream* final_buffer) {
523 // For the most part, the assembly process walks the various tables.
524 // ix_mumble is the index into the mumble table.
525 size_t ix_origins = 0;
526 size_t ix_copy_counts = 0;
527 size_t ix_copy_bytes = 0;
528 size_t ix_abs32_ix = 0;
529 size_t ix_rel32_ix = 0;
530
531 RVA current_rva = 0;
532
533 bool pending_pe_relocation_table = false;
534 uint8 pending_pe_relocation_table_type = 0x03; // IMAGE_REL_BASED_HIGHLOW
535 Elf32_Word pending_elf_relocation_table_type = 0;
536 SinkStream bytes_following_relocation_table;
537
538 SinkStream* output = final_buffer;
539
540 for (size_t ix_ops = 0; ix_ops < ops_.size(); ++ix_ops) {
541 OP op = ops_[ix_ops];
542
543 switch (op) {
544 default:
545 if (!EvaluateRel32ARM(op, ix_rel32_ix, current_rva, output))
546 return false;
547 break;
548
549 case ORIGIN: {
550 RVA section_rva;
551 if (!VectorAt(origins_, ix_origins, §ion_rva))
552 return false;
553 ++ix_origins;
554 current_rva = section_rva;
555 break;
556 }
557
558 case COPY: {
559 uint32 count;
560 if (!VectorAt(copy_counts_, ix_copy_counts, &count))
561 return false;
562 ++ix_copy_counts;
563 for (uint32 i = 0; i < count; ++i) {
564 uint8 b;
565 if (!VectorAt(copy_bytes_, ix_copy_bytes, &b))
566 return false;
567 ++ix_copy_bytes;
568 if (!output->Write(&b, 1))
569 return false;
570 }
571 current_rva += count;
572 break;
573 }
574
575 case COPY1: {
576 uint8 b;
577 if (!VectorAt(copy_bytes_, ix_copy_bytes, &b))
578 return false;
579 ++ix_copy_bytes;
580 if (!output->Write(&b, 1))
581 return false;
582 current_rva += 1;
583 break;
584 }
585
586 case REL32: {
587 uint32 index;
588 if (!VectorAt(rel32_ix_, ix_rel32_ix, &index))
589 return false;
590 ++ix_rel32_ix;
591 RVA rva;
592 if (!VectorAt(rel32_rva_, index, &rva))
593 return false;
594 uint32 offset = (rva - (current_rva + 4));
595 if (!output->Write(&offset, 4))
596 return false;
597 current_rva += 4;
598 break;
599 }
600
601 case ABS32: {
602 uint32 index;
603 if (!VectorAt(abs32_ix_, ix_abs32_ix, &index))
604 return false;
605 ++ix_abs32_ix;
606 RVA rva;
607 if (!VectorAt(abs32_rva_, index, &rva))
608 return false;
609 uint32 abs32 = static_cast<uint32>(rva + image_base_);
610 if (!abs32_relocs_.push_back(current_rva) || !output->Write(&abs32, 4))
611 return false;
612 current_rva += 4;
613 break;
614 }
615
616 case MAKE_PE_RELOCATION_TABLE: {
617 // We can see the base relocation anywhere, but we only have the
618 // information to generate it at the very end. So we divert the bytes
619 // we are generating to a temporary stream.
620 if (pending_pe_relocation_table)
621 return false; // Can't have two base relocation tables.
622
623 pending_pe_relocation_table = true;
624 output = &bytes_following_relocation_table;
625 break;
626 // There is a potential problem *if* the instruction stream contains
627 // some REL32 relocations following the base relocation and in the same
628 // section. We don't know the size of the table, so 'current_rva' will
629 // be wrong, causing REL32 offsets to be miscalculated. This never
630 // happens; the base relocation table is usually in a section of its
631 // own, a data-only section, and following everything else in the
632 // executable except some padding zero bytes. We could fix this by
633 // emitting an ORIGIN after the MAKE_BASE_RELOCATION_TABLE.
634 }
635
636 case MAKE_PE64_RELOCATION_TABLE: {
637 if (pending_pe_relocation_table)
638 return false; // Can't have two base relocation tables.
639
640 pending_pe_relocation_table = true;
641 pending_pe_relocation_table_type = 0x0A; // IMAGE_REL_BASED_DIR64
642 output = &bytes_following_relocation_table;
643 break;
644 }
645
646 case MAKE_ELF_ARM_RELOCATION_TABLE: {
647 // We can see the base relocation anywhere, but we only have the
648 // information to generate it at the very end. So we divert the bytes
649 // we are generating to a temporary stream.
650 if (pending_elf_relocation_table_type)
651 return false; // Can't have two base relocation tables.
652
653 pending_elf_relocation_table_type = R_ARM_RELATIVE;
654 output = &bytes_following_relocation_table;
655 break;
656 }
657
658 case MAKE_ELF_RELOCATION_TABLE: {
659 // We can see the base relocation anywhere, but we only have the
660 // information to generate it at the very end. So we divert the bytes
661 // we are generating to a temporary stream.
662 if (pending_elf_relocation_table_type)
663 return false; // Can't have two base relocation tables.
664
665 pending_elf_relocation_table_type = R_386_RELATIVE;
666 output = &bytes_following_relocation_table;
667 break;
668 }
669 }
670 }
671
672 if (pending_pe_relocation_table) {
673 if (!GeneratePeRelocations(final_buffer,
674 pending_pe_relocation_table_type) ||
675 !final_buffer->Append(&bytes_following_relocation_table))
676 return false;
677 }
678
679 if (pending_elf_relocation_table_type) {
680 if (!GenerateElfRelocations(pending_elf_relocation_table_type,
681 final_buffer) ||
682 !final_buffer->Append(&bytes_following_relocation_table))
683 return false;
684 }
685
686 // Final verification check: did we consume all lists?
687 if (ix_copy_counts != copy_counts_.size())
688 return false;
689 if (ix_copy_bytes != copy_bytes_.size())
690 return false;
691 if (ix_abs32_ix != abs32_ix_.size())
692 return false;
693 if (ix_rel32_ix != rel32_ix_.size())
694 return false;
695
696 return true;
697 }
698
699 // RelocBlock has the layout of a block of relocations in the base relocation
700 // table file format.
701 //
702 struct RelocBlockPOD {
703 uint32 page_rva;
704 uint32 block_size;
705 uint16 relocs[4096]; // Allow up to one relocation per byte of a 4k page.
706 };
707
708 COMPILE_ASSERT(offsetof(RelocBlockPOD, relocs) == 8, reloc_block_header_size);
709
710 class RelocBlock {
711 public:
RelocBlock()712 RelocBlock() {
713 pod.page_rva = ~0;
714 pod.block_size = 8;
715 }
716
Add(uint16 item)717 void Add(uint16 item) {
718 pod.relocs[(pod.block_size-8)/2] = item;
719 pod.block_size += 2;
720 }
721
Flush(SinkStream * buffer)722 CheckBool Flush(SinkStream* buffer) WARN_UNUSED_RESULT {
723 bool ok = true;
724 if (pod.block_size != 8) {
725 if (pod.block_size % 4 != 0) { // Pad to make size multiple of 4 bytes.
726 Add(0);
727 }
728 ok = buffer->Write(&pod, pod.block_size);
729 pod.block_size = 8;
730 }
731 return ok;
732 }
733 RelocBlockPOD pod;
734 };
735
GeneratePeRelocations(SinkStream * buffer,uint8 type)736 CheckBool EncodedProgram::GeneratePeRelocations(SinkStream* buffer,
737 uint8 type) {
738 std::sort(abs32_relocs_.begin(), abs32_relocs_.end());
739
740 RelocBlock block;
741
742 bool ok = true;
743 for (size_t i = 0; ok && i < abs32_relocs_.size(); ++i) {
744 uint32 rva = abs32_relocs_[i];
745 uint32 page_rva = rva & ~0xFFF;
746 if (page_rva != block.pod.page_rva) {
747 ok &= block.Flush(buffer);
748 block.pod.page_rva = page_rva;
749 }
750 if (ok)
751 block.Add(((static_cast<uint16>(type)) << 12 ) | (rva & 0xFFF));
752 }
753 ok &= block.Flush(buffer);
754 return ok;
755 }
756
GenerateElfRelocations(Elf32_Word r_info,SinkStream * buffer)757 CheckBool EncodedProgram::GenerateElfRelocations(Elf32_Word r_info,
758 SinkStream* buffer) {
759 std::sort(abs32_relocs_.begin(), abs32_relocs_.end());
760
761 Elf32_Rel relocation_block;
762
763 relocation_block.r_info = r_info;
764
765 bool ok = true;
766 for (size_t i = 0; ok && i < abs32_relocs_.size(); ++i) {
767 relocation_block.r_offset = abs32_relocs_[i];
768 ok = buffer->Write(&relocation_block, sizeof(Elf32_Rel));
769 }
770
771 return ok;
772 }
773 ////////////////////////////////////////////////////////////////////////////////
774
WriteEncodedProgram(EncodedProgram * encoded,SinkStreamSet * sink)775 Status WriteEncodedProgram(EncodedProgram* encoded, SinkStreamSet* sink) {
776 if (!encoded->WriteTo(sink))
777 return C_STREAM_ERROR;
778 return C_OK;
779 }
780
ReadEncodedProgram(SourceStreamSet * streams,EncodedProgram ** output)781 Status ReadEncodedProgram(SourceStreamSet* streams, EncodedProgram** output) {
782 EncodedProgram* encoded = new EncodedProgram();
783 if (encoded->ReadFrom(streams)) {
784 *output = encoded;
785 return C_OK;
786 }
787 delete encoded;
788 return C_DESERIALIZATION_FAILED;
789 }
790
Assemble(EncodedProgram * encoded,SinkStream * buffer)791 Status Assemble(EncodedProgram* encoded, SinkStream* buffer) {
792 bool assembled = encoded->AssembleTo(buffer);
793 if (assembled)
794 return C_OK;
795 return C_ASSEMBLY_FAILED;
796 }
797
DeleteEncodedProgram(EncodedProgram * encoded)798 void DeleteEncodedProgram(EncodedProgram* encoded) {
799 delete encoded;
800 }
801
802 } // end namespace
803