This source file includes following definitions.
- number_of_data_directories_
- ParseHeader
- ParseRelocs
- RVAToSection
- RVAToFileOffset
- RVAToPointer
- SectionName
- ParseFile
- ParseAbs32Relocs
- ParseRel32RelocsFromSections
- ParseRel32RelocsFromSection
- ParseNonSectionFileRegion
- ParseFileRegion
- HistogramTargets
- DescribeRVA
- FindNextSection
- FileOffsetToRVA
- ReadDataDirectory
#include "courgette/disassembler_win32_x86.h"
#include <algorithm>
#include <string>
#include <vector>
#include "base/basictypes.h"
#include "base/logging.h"
#include "courgette/assembly_program.h"
#include "courgette/courgette.h"
#include "courgette/encoded_program.h"
namespace courgette {
DisassemblerWin32X86::DisassemblerWin32X86(const void* start, size_t length)
: Disassembler(start, length),
incomplete_disassembly_(false),
is_PE32_plus_(false),
optional_header_(NULL),
size_of_optional_header_(0),
offset_of_data_directories_(0),
machine_type_(0),
number_of_sections_(0),
sections_(NULL),
has_text_section_(false),
size_of_code_(0),
size_of_initialized_data_(0),
size_of_uninitialized_data_(0),
base_of_code_(0),
base_of_data_(0),
image_base_(0),
size_of_image_(0),
number_of_data_directories_(0) {
}
bool DisassemblerWin32X86::ParseHeader() {
if (length() < kOffsetOfFileAddressOfNewExeHeader + 4 )
return Bad("Too small");
if (start()[0] != 'M' || start()[1] != 'Z')
return Bad("Not MZ");
uint32 offset = ReadU32(start(),
kOffsetOfFileAddressOfNewExeHeader);
if (offset >= length())
return Bad("Bad offset to PE header");
const uint8* const pe_header = OffsetToPointer(offset);
const size_t kMinPEHeaderSize = 4 + kSizeOfCoffHeader;
if (pe_header <= start() ||
pe_header >= end() - kMinPEHeaderSize)
return Bad("Bad offset to PE header");
if (offset % 8 != 0)
return Bad("Misaligned PE header");
if (!(pe_header[0] == 'P' &&
pe_header[1] == 'E' &&
pe_header[2] == 0 &&
pe_header[3] == 0))
return Bad("no PE signature");
const uint8* const coff_header = pe_header + 4;
machine_type_ = ReadU16(coff_header, 0);
number_of_sections_ = ReadU16(coff_header, 2);
size_of_optional_header_ = ReadU16(coff_header, 16);
const uint8* const optional_header = coff_header + kSizeOfCoffHeader;
optional_header_ = optional_header;
if (optional_header + size_of_optional_header_ >= end())
return Bad("optional header past end of file");
if (size_of_optional_header_ < 2)
return Bad("optional header no magic");
uint16 magic = ReadU16(optional_header, 0);
if (magic == kImageNtOptionalHdr32Magic) {
is_PE32_plus_ = false;
offset_of_data_directories_ =
kOffsetOfDataDirectoryFromImageOptionalHeader32;
} else if (magic == kImageNtOptionalHdr64Magic) {
is_PE32_plus_ = true;
offset_of_data_directories_ =
kOffsetOfDataDirectoryFromImageOptionalHeader64;
} else {
return Bad("unrecognized magic");
}
if (size_of_optional_header_ < offset_of_data_directories_)
return Bad("optional header too short");
size_of_code_ = ReadU32(optional_header, 4);
size_of_initialized_data_ = ReadU32(optional_header, 8);
size_of_uninitialized_data_ = ReadU32(optional_header, 12);
base_of_code_ = ReadU32(optional_header, 20);
if (is_PE32_plus_) {
base_of_data_ = 0;
image_base_ = ReadU64(optional_header, 24);
} else {
base_of_data_ = ReadU32(optional_header, 24);
image_base_ = ReadU32(optional_header, 28);
}
size_of_image_ = ReadU32(optional_header, 56);
number_of_data_directories_ =
ReadU32(optional_header, (is_PE32_plus_ ? 108 : 92));
if (size_of_code_ >= length() ||
size_of_initialized_data_ >= length() ||
size_of_code_ + size_of_initialized_data_ >= length()) {
}
bool b = true;
b &= ReadDataDirectory(0, &export_table_);
b &= ReadDataDirectory(1, &import_table_);
b &= ReadDataDirectory(2, &resource_table_);
b &= ReadDataDirectory(3, &exception_table_);
b &= ReadDataDirectory(5, &base_relocation_table_);
b &= ReadDataDirectory(11, &bound_import_table_);
b &= ReadDataDirectory(12, &import_address_table_);
b &= ReadDataDirectory(13, &delay_import_descriptor_);
b &= ReadDataDirectory(14, &clr_runtime_header_);
if (!b) {
return Bad("malformed data directory");
}
sections_ =
reinterpret_cast<const Section*>(optional_header +
size_of_optional_header_);
size_t detected_length = 0;
for (int i = 0; i < number_of_sections_; ++i) {
const Section* section = §ions_[i];
if (memcmp(section->name, ".text", 6) == 0)
has_text_section_ = true;
uint32 section_end =
section->file_offset_of_raw_data + section->size_of_raw_data;
if (section_end > detected_length)
detected_length = section_end;
}
ReduceLength(detected_length);
if (!is_32bit()) {
return Bad("64 bit executables are not supported by this disassembler");
}
if (!has_text_section()) {
return Bad("Resource-only executables are not yet supported");
}
return Good();
}
bool DisassemblerWin32X86::Disassemble(AssemblyProgram* target) {
if (!ok())
return false;
target->set_image_base(image_base());
if (!ParseAbs32Relocs())
return false;
ParseRel32RelocsFromSections();
if (!ParseFile(target))
return false;
target->DefaultAssignIndexes();
return true;
}
bool DisassemblerWin32X86::ParseRelocs(std::vector<RVA> *relocs) {
relocs->clear();
size_t relocs_size = base_relocation_table_.size_;
if (relocs_size == 0)
return true;
const uint8* relocs_start = RVAToPointer(base_relocation_table_.address_);
const uint8* relocs_end = relocs_start + relocs_size;
if (relocs_start < start() ||
relocs_start >= end() ||
relocs_end <= start() ||
relocs_end > end()) {
return Bad(".relocs outside image");
}
const uint8* block = relocs_start;
while (block + 8 < relocs_end) {
RVA page_rva = ReadU32(block, 0);
uint32 size = ReadU32(block, 4);
if (size < 8 ||
size % 4 != 0)
return Bad("unreasonable relocs block");
const uint8* end_entries = block + size;
if (end_entries <= block ||
end_entries <= start() ||
end_entries > end())
return Bad(".relocs block outside image");
for (const uint8* p = block + 8; p < end_entries; p += 2) {
uint16 entry = ReadU16(p, 0);
int type = entry >> 12;
int offset = entry & 0xFFF;
RVA rva = page_rva + offset;
if (type == 3) {
relocs->push_back(rva);
} else if (type == 0) {
} else {
return Bad("unknown type of reloc");
}
}
block += size;
}
std::sort(relocs->begin(), relocs->end());
return true;
}
const Section* DisassemblerWin32X86::RVAToSection(RVA rva) const {
for (int i = 0; i < number_of_sections_; i++) {
const Section* section = §ions_[i];
uint32 offset = rva - section->virtual_address;
if (offset < section->virtual_size) {
return section;
}
}
return NULL;
}
int DisassemblerWin32X86::RVAToFileOffset(RVA rva) const {
const Section* section = RVAToSection(rva);
if (section) {
uint32 offset = rva - section->virtual_address;
if (offset < section->size_of_raw_data) {
return section->file_offset_of_raw_data + offset;
} else {
return kNoOffset;
}
}
if (rva == 0 || rva == 2)
return rva;
NOTREACHED();
return kNoOffset;
}
const uint8* DisassemblerWin32X86::RVAToPointer(RVA rva) const {
int file_offset = RVAToFileOffset(rva);
if (file_offset == kNoOffset)
return NULL;
else
return OffsetToPointer(file_offset);
}
std::string DisassemblerWin32X86::SectionName(const Section* section) {
if (section == NULL)
return "<none>";
char name[9];
memcpy(name, section->name, 8);
name[8] = '\0';
return name;
}
CheckBool DisassemblerWin32X86::ParseFile(AssemblyProgram* program) {
uint32 file_offset = 0;
while (file_offset < length()) {
const Section* section = FindNextSection(file_offset);
if (section == NULL) {
break;
}
if (file_offset < section->file_offset_of_raw_data) {
uint32 section_start_offset = section->file_offset_of_raw_data;
if(!ParseNonSectionFileRegion(file_offset, section_start_offset,
program))
return false;
file_offset = section_start_offset;
}
uint32 end = file_offset + section->size_of_raw_data;
if (!ParseFileRegion(section, file_offset, end, program))
return false;
file_offset = end;
}
#if COURGETTE_HISTOGRAM_TARGETS
HistogramTargets("abs32 relocs", abs32_target_rvas_);
HistogramTargets("rel32 relocs", rel32_target_rvas_);
#endif
return true;
}
bool DisassemblerWin32X86::ParseAbs32Relocs() {
abs32_locations_.clear();
if (!ParseRelocs(&abs32_locations_))
return false;
std::sort(abs32_locations_.begin(), abs32_locations_.end());
#if COURGETTE_HISTOGRAM_TARGETS
for (size_t i = 0; i < abs32_locations_.size(); ++i) {
RVA rva = abs32_locations_[i];
uint32 target_address = Read32LittleEndian(RVAToPointer(rva));
++abs32_target_rvas_[target_address - image_base()];
}
#endif
return true;
}
void DisassemblerWin32X86::ParseRel32RelocsFromSections() {
uint32 file_offset = 0;
while (file_offset < length()) {
const Section* section = FindNextSection(file_offset);
if (section == NULL)
break;
if (file_offset < section->file_offset_of_raw_data)
file_offset = section->file_offset_of_raw_data;
ParseRel32RelocsFromSection(section);
file_offset += section->size_of_raw_data;
}
std::sort(rel32_locations_.begin(), rel32_locations_.end());
#if COURGETTE_HISTOGRAM_TARGETS
VLOG(1) << "abs32_locations_ " << abs32_locations_.size()
<< "\nrel32_locations_ " << rel32_locations_.size()
<< "\nabs32_target_rvas_ " << abs32_target_rvas_.size()
<< "\nrel32_target_rvas_ " << rel32_target_rvas_.size();
int common = 0;
std::map<RVA, int>::iterator abs32_iter = abs32_target_rvas_.begin();
std::map<RVA, int>::iterator rel32_iter = rel32_target_rvas_.begin();
while (abs32_iter != abs32_target_rvas_.end() &&
rel32_iter != rel32_target_rvas_.end()) {
if (abs32_iter->first < rel32_iter->first)
++abs32_iter;
else if (rel32_iter->first < abs32_iter->first)
++rel32_iter;
else {
++common;
++abs32_iter;
++rel32_iter;
}
}
VLOG(1) << "common " << common;
#endif
}
void DisassemblerWin32X86::ParseRel32RelocsFromSection(const Section* section) {
bool isCode = strcmp(section->name, ".text") == 0;
if (!isCode)
return;
uint32 start_file_offset = section->file_offset_of_raw_data;
uint32 end_file_offset = start_file_offset + section->size_of_raw_data;
RVA relocs_start_rva = base_relocation_table().address_;
const uint8* start_pointer = OffsetToPointer(start_file_offset);
const uint8* end_pointer = OffsetToPointer(end_file_offset);
RVA start_rva = FileOffsetToRVA(start_file_offset);
RVA end_rva = start_rva + section->virtual_size;
const uint8* const adjust_pointer_to_rva = start_pointer - start_rva;
std::vector<RVA>::iterator abs32_pos = abs32_locations_.begin();
const uint8* p = start_pointer;
while (p < end_pointer) {
RVA current_rva = static_cast<RVA>(p - adjust_pointer_to_rva);
if (current_rva == relocs_start_rva) {
uint32 relocs_size = base_relocation_table().size_;
if (relocs_size) {
p += relocs_size;
continue;
}
}
const uint8* rel32 = NULL;
if (p + 5 <= end_pointer) {
if (*p == 0xE8 || *p == 0xE9) {
rel32 = p + 1;
}
}
if (p + 6 <= end_pointer) {
if (*p == 0x0F && (*(p+1) & 0xF0) == 0x80) {
if (p[1] != 0x8A && p[1] != 0x8B)
rel32 = p + 2;
}
}
if (rel32) {
RVA rel32_rva = static_cast<RVA>(rel32 - adjust_pointer_to_rva);
while (abs32_pos != abs32_locations_.end() && *abs32_pos < rel32_rva - 3)
++abs32_pos;
if (abs32_pos != abs32_locations_.end()) {
if (*abs32_pos < rel32_rva + 4) {
p += (*abs32_pos + 4) - current_rva;
continue;
}
}
RVA target_rva = rel32_rva + 4 + Read32LittleEndian(rel32);
if (IsValidRVA(target_rva) &&
start_rva <= target_rva && target_rva < end_rva) {
rel32_locations_.push_back(rel32_rva);
#if COURGETTE_HISTOGRAM_TARGETS
++rel32_target_rvas_[target_rva];
#endif
p = rel32 + 4;
continue;
}
}
p += 1;
}
}
CheckBool DisassemblerWin32X86::ParseNonSectionFileRegion(
uint32 start_file_offset,
uint32 end_file_offset,
AssemblyProgram* program) {
if (incomplete_disassembly_)
return true;
const uint8* start = OffsetToPointer(start_file_offset);
const uint8* end = OffsetToPointer(end_file_offset);
const uint8* p = start;
while (p < end) {
if (!program->EmitByteInstruction(*p))
return false;
++p;
}
return true;
}
CheckBool DisassemblerWin32X86::ParseFileRegion(
const Section* section,
uint32 start_file_offset, uint32 end_file_offset,
AssemblyProgram* program) {
RVA relocs_start_rva = base_relocation_table().address_;
const uint8* start_pointer = OffsetToPointer(start_file_offset);
const uint8* end_pointer = OffsetToPointer(end_file_offset);
RVA start_rva = FileOffsetToRVA(start_file_offset);
RVA end_rva = start_rva + section->virtual_size;
const uint8* const adjust_pointer_to_rva = start_pointer - start_rva;
std::vector<RVA>::iterator rel32_pos = rel32_locations_.begin();
std::vector<RVA>::iterator abs32_pos = abs32_locations_.begin();
if (!program->EmitOriginInstruction(start_rva))
return false;
const uint8* p = start_pointer;
while (p < end_pointer) {
RVA current_rva = static_cast<RVA>(p - adjust_pointer_to_rva);
if (current_rva == relocs_start_rva) {
if (!program->EmitPeRelocsInstruction())
return false;
uint32 relocs_size = base_relocation_table().size_;
if (relocs_size) {
p += relocs_size;
continue;
}
}
while (abs32_pos != abs32_locations_.end() && *abs32_pos < current_rva)
++abs32_pos;
if (abs32_pos != abs32_locations_.end() && *abs32_pos == current_rva) {
uint32 target_address = Read32LittleEndian(p);
RVA target_rva = target_address - image_base();
if (!program->EmitAbs32(program->FindOrMakeAbs32Label(target_rva)))
return false;
p += 4;
continue;
}
while (rel32_pos != rel32_locations_.end() && *rel32_pos < current_rva)
++rel32_pos;
if (rel32_pos != rel32_locations_.end() && *rel32_pos == current_rva) {
RVA target_rva = current_rva + 4 + Read32LittleEndian(p);
if (!program->EmitRel32(program->FindOrMakeRel32Label(target_rva)))
return false;
p += 4;
continue;
}
if (incomplete_disassembly_) {
if ((abs32_pos == abs32_locations_.end() || end_rva <= *abs32_pos) &&
(rel32_pos == rel32_locations_.end() || end_rva <= *rel32_pos) &&
(end_rva <= relocs_start_rva || current_rva >= relocs_start_rva)) {
break;
}
}
if (!program->EmitByteInstruction(*p))
return false;
p += 1;
}
return true;
}
#if COURGETTE_HISTOGRAM_TARGETS
void DisassemblerWin32X86::HistogramTargets(const char* kind,
const std::map<RVA, int>& map) {
int total = 0;
std::map<int, std::vector<RVA> > h;
for (std::map<RVA, int>::const_iterator p = map.begin();
p != map.end();
++p) {
h[p->second].push_back(p->first);
total += p->second;
}
std::cout << total << " " << kind << " to "
<< map.size() << " unique targets" << std::endl;
std::cout << "indegree: #targets-with-indegree (example)" << std::endl;
const int kFirstN = 15;
bool someSkipped = false;
int index = 0;
for (std::map<int, std::vector<RVA> >::reverse_iterator p = h.rbegin();
p != h.rend();
++p) {
++index;
if (index <= kFirstN || p->first <= 3) {
if (someSkipped) {
std::cout << "..." << std::endl;
}
size_t count = p->second.size();
std::cout << std::dec << p->first << ": " << count;
if (count <= 2) {
for (size_t i = 0; i < count; ++i)
std::cout << " " << DescribeRVA(p->second[i]);
}
std::cout << std::endl;
someSkipped = false;
} else {
someSkipped = true;
}
}
}
#endif
std::string DisassemblerWin32X86::DescribeRVA(RVA rva) const {
const Section* section = RVAToSection(rva);
std::ostringstream s;
s << std::hex << rva;
if (section) {
s << " (";
s << SectionName(section) << "+"
<< std::hex << (rva - section->virtual_address)
<< ")";
}
return s.str();
}
const Section* DisassemblerWin32X86::FindNextSection(uint32 fileOffset) const {
const Section* best = 0;
for (int i = 0; i < number_of_sections_; i++) {
const Section* section = §ions_[i];
if (section->size_of_raw_data > 0) {
if (fileOffset <= section->file_offset_of_raw_data) {
if (best == 0 ||
section->file_offset_of_raw_data < best->file_offset_of_raw_data) {
best = section;
}
}
}
}
return best;
}
RVA DisassemblerWin32X86::FileOffsetToRVA(uint32 file_offset) const {
for (int i = 0; i < number_of_sections_; i++) {
const Section* section = §ions_[i];
uint32 offset = file_offset - section->file_offset_of_raw_data;
if (offset < section->size_of_raw_data) {
return section->virtual_address + offset;
}
}
return 0;
}
bool DisassemblerWin32X86::ReadDataDirectory(
int index,
ImageDataDirectory* directory) {
if (index < number_of_data_directories_) {
size_t offset = index * 8 + offset_of_data_directories_;
if (offset >= size_of_optional_header_)
return Bad("number of data directories inconsistent");
const uint8* data_directory = optional_header_ + offset;
if (data_directory < start() ||
data_directory + 8 >= end())
return Bad("data directory outside image");
RVA rva = ReadU32(data_directory, 0);
size_t size = ReadU32(data_directory, 4);
if (size > size_of_image_)
return Bad("data directory size too big");
directory->address_ = rva;
directory->size_ = static_cast<uint32>(size);
return true;
} else {
directory->address_ = 0;
directory->size_ = 0;
return true;
}
}
}