]> git.siccegge.de Git - frida/frida.git/blob - src/disassembler/llvm/LLVMDisassembler.cxx
Add interface to iterate over functions
[frida/frida.git] / src / disassembler / llvm / LLVMDisassembler.cxx
1 #include "disassembler/llvm/LLVMDisassembler.hxx"
2 #include "disassembler/llvm/LLVMBasicBlock.hxx"
3 #include "disassembler/llvm/LLVMFunction.hxx"
4
5 #include <stack>
6 #include <algorithm>
7
8 using namespace llvm;
9 using namespace llvm::object;
10
11 /*
12 * TODO: fallback code falls die Datei kein ELF/PE/COFF/MacO/.. binary
13 * ist sondern z.B. einfach nur Instruktionen oder ein Bootsektor oder
14 * foo
15 */
16 LLVMDisassembler::LLVMDisassembler(const std::string& filename)
17 : Disassembler(filename)
18 , logger(log4cxx::Logger::getLogger("LLVMDisassembler"))
19 , triple("unknown-unknown-unknown")
20 {
21 LOG4CXX_DEBUG(logger, "Handling file" << filename);
22 auto result = createBinary(filename);
23
24 error_code ec;
25 if ((ec = result.getError())) {
26 LOG4CXX_ERROR(logger, "Failed to load Binary" << ec.message());
27 binary = NULL;
28 return;
29 }
30
31 binary.reset(result.get());
32
33 o = dyn_cast<ObjectFile>(binary.get());
34
35 triple.setArch(Triple::ArchType(o->getArch()));
36 std::string tripleName(triple.getTriple());
37
38 LOG4CXX_INFO(logger, "Architecture " << tripleName);
39
40
41 std::string es;
42 target = TargetRegistry::lookupTarget("", triple, es);
43 if (!target) {
44 LOG4CXX_ERROR(logger, es);
45 return;
46 }
47
48 LOG4CXX_INFO(logger, "Target " << target->getName());
49
50 MRI.reset(target->createMCRegInfo(tripleName));
51 if (!MRI) {
52 LOG4CXX_ERROR(logger, "no register info for target " << tripleName);
53 return;
54 }
55
56 // Set up disassembler.
57 AsmInfo.reset(target->createMCAsmInfo(*MRI, tripleName));
58 if (!AsmInfo) {
59 LOG4CXX_ERROR(logger, "no assembly info for target " << tripleName);
60 return;
61 }
62
63 STI.reset(target->createMCSubtargetInfo(tripleName, "", ""));
64 if (!STI) {
65 LOG4CXX_ERROR(logger, "no subtarget info for target " << tripleName);
66 return;
67 }
68
69 MII.reset(target->createMCInstrInfo());
70 if (!MII) {
71 LOG4CXX_ERROR(logger, "no instruction info for target " << tripleName);
72 return;
73 }
74
75 DisAsm.reset(target->createMCDisassembler(*STI));
76 if (!DisAsm) {
77 LOG4CXX_ERROR(logger, "no disassembler for target " << tripleName);
78 return;
79 }
80
81 MOFI.reset(new MCObjectFileInfo);
82 Ctx.reset(new MCContext(AsmInfo.get(), MRI.get(), MOFI.get()));
83 RelInfo.reset(
84 target->createMCRelocationInfo(tripleName, *Ctx.get()));
85 if (RelInfo) {
86 Symzer.reset(
87 MCObjectSymbolizer::createObjectSymbolizer(*Ctx.get(), RelInfo, o));
88 if (Symzer)
89 DisAsm->setSymbolizer(Symzer);
90 }
91
92 MIA.reset(target->createMCInstrAnalysis(MII.get()));
93
94 int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
95 IP.reset(target->createMCInstPrinter(AsmPrinterVariant, *AsmInfo, *MII, *MRI, *STI));
96 if (!IP) {
97 LOG4CXX_ERROR(logger, "no instruction printer for target " << tripleName);
98 return;
99 }
100
101 IP->setPrintImmHex(llvm::HexStyle::C);
102 IP->setPrintImmHex(true);
103
104 OwningPtr<MCObjectDisassembler> OD(
105 new MCObjectDisassembler(*o, *DisAsm, *MIA));
106 Mod.reset(OD->buildModule(false));
107
108 readSymbols();
109 readSections();
110 disassemble();
111 }
112
113
114 void LLVMDisassembler::disassemble() {
115 std::stack<LLVMFunction*> remaining_functions;
116 std::stack<LLVMBasicBlock*> remaining_blocks;
117 SectionRef text_section = sections[".text"];
118
119 for (auto x = symbols.begin(); x != symbols.end(); ++x) {
120 uint64_t result;
121 bool contains;
122 SymbolRef::Type symbol_type;
123
124 /*
125 * TODO: If we jump into some Basic Block we need to split it there into two
126 */
127
128 if (text_section.containsSymbol(x->second, contains) || !contains)
129 continue;
130
131 if (x->second.getType(symbol_type)
132 || SymbolRef::ST_Function != symbol_type)
133 continue;
134
135 if (!x->second.getAddress(result)) {
136 LLVMFunction * fun = new LLVMFunction(x->first, result);
137 remaining_functions.push(fun);
138 functions.insert(std::make_pair(result, fun));
139 LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
140 }
141 }
142
143 StringRef bytes;
144 text_section.getContents(bytes);
145 StringRefMemoryObject ref(bytes);
146
147 while (remaining_functions.size()) {
148 LLVMFunction * current_function = remaining_functions.top();
149 remaining_functions.pop();
150
151 LOG4CXX_DEBUG(logger, "Handling function " << current_function->getName());
152
153 // if ("_start" != current_function->getName())
154 // continue;
155
156 LLVMBasicBlock * block = new LLVMBasicBlock(current_function->getStartAddress());
157 remaining_blocks.push(block);
158 blocks.insert(std::make_pair(block->getStartAddress(), block));
159
160 while (remaining_blocks.size()) {
161 LLVMBasicBlock * current_block = remaining_blocks.top();
162 remaining_blocks.pop();
163
164 LOG4CXX_DEBUG(logger, "Handling Block starting at " << std::hex << current_block->getStartAddress());
165
166 uint64_t inst_size;
167 uint64_t base_address;
168 text_section.getAddress(base_address);
169 uint64_t current_address = current_block->getStartAddress() - base_address;
170 while(true) {
171 MCInst inst;
172 std::string buf;
173 llvm::raw_string_ostream s(buf);
174
175 if(llvm::MCDisassembler::Success ==
176 DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
177
178 uint8_t bytes[inst_size+2];
179 ref.readBytes(current_address, inst_size, bytes);
180 s << '\t';
181 for(uint8_t* cur = bytes; cur < bytes + inst_size; ++cur) {
182 s.write_hex(*cur);
183 s << ' ';
184 }
185 s << '\t';
186
187 IP->printInst(&inst, s, "");
188
189 LOG4CXX_DEBUG(logger, std::hex << current_address + base_address << s.str());
190
191 uint64_t jmptarget;
192 if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
193 jmptarget += base_address;
194 if (!MIA->isIndirectBranch(inst)) {
195 if (MIA->isCall(inst)) {
196 if (functions.find(jmptarget) == functions.end()) {
197 std::stringstream s;
198 s << "<Unnamed 0x" << std::hex << jmptarget << ">";
199 LLVMFunction * fun = new LLVMFunction(s.str(), jmptarget);
200 functions.insert(std::make_pair(jmptarget, fun));
201 remaining_functions.push(fun);
202 }
203 } else {
204 if (blocks.find(jmptarget) == blocks.end()) {
205 LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget);
206 blocks.insert(std::make_pair(block->getStartAddress(), block));
207 remaining_blocks.push(block);
208 }
209 if (MIA->isConditionalBranch(inst)) {
210 jmptarget = base_address + current_address + inst_size;
211 if (blocks.find(jmptarget) == blocks.end()) {
212 LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget);
213 blocks.insert(std::make_pair(block->getStartAddress(), block));
214 remaining_blocks.push(new LLVMBasicBlock(jmptarget));
215 }
216 }
217 }
218 }
219 } else {
220 inst_size = 0;
221 }
222
223
224 if (inst_size == 0 || MIA->isTerminator(inst) || MIA->isBranch(inst)) {
225 current_block->setEndAddress(current_address + base_address);
226 LOG4CXX_DEBUG(logger, "Finished Block at " << std::hex <<
227 current_block->getEndAddress());
228 }
229 break;
230 }
231 current_address += inst_size;
232 }
233 }
234 LOG4CXX_DEBUG(logger, "Finished function " << current_function->getName());
235 }
236 }
237
238 void LLVMDisassembler::readSymbols() {
239 error_code ec;
240 symbol_iterator si(o->symbol_begin()), se(o->symbol_end());
241 for (; si != se; ++si) {
242 StringRef name;
243 if ((ec = si->getName(name))) {
244 LOG4CXX_ERROR(logger, ec.message());
245 break;
246 }
247 LOG4CXX_DEBUG(logger, "Added symbol " << name.str());
248 symbols.insert(make_pair(name.str(), *si));
249 }
250 }
251
252 void LLVMDisassembler::readSections() {
253 error_code ec;
254 section_iterator i(o->section_begin()), e(o->section_end());
255 for (; i != e; ++i) {
256 StringRef name;
257 if ((ec = i->getName(name))) {
258 LOG4CXX_ERROR(logger, ec.message());
259 break;
260 }
261 LOG4CXX_DEBUG(logger, "Added section " << name.str());
262 sections.insert(make_pair(name.str(), *i));
263 }
264
265 }
266
267 void LLVMDisassembler::forEachFunction(std::function<void (uint64_t, Function*)> callback) {
268 std::for_each(functions.begin(), functions.end(),
269 [&](std::pair<uint64_t, LLVMFunction*> x) {
270 callback(x.first, x.second);
271 });
272 }
273
274
275
276 void LLVMDisassembler::generateControlFlowGraph(uint64_t address) {
277
278 }