blob: e920d70cd596c3b1b4282a2dccdeb59f31119a20 [file] [log] [blame]
Sean Callanan95e5c632012-02-17 00:53:45 +00001//===-- DisassemblerLLVMC.cpp -----------------------------------*- C++ -*-===//
2//
3// The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9
10#include "DisassemblerLLVMC.h"
11
12#include "llvm-c/Disassembler.h"
Ashok Thirumurthi78059352013-05-24 15:55:54 +000013#include "llvm/ADT/OwningPtr.h"
Jim Ingham0f063ba2013-03-02 00:26:47 +000014#include "llvm/MC/MCAsmInfo.h"
15#include "llvm/MC/MCContext.h"
16#include "llvm/MC/MCDisassembler.h"
17#include "llvm/MC/MCInst.h"
18#include "llvm/MC/MCInstPrinter.h"
19#include "llvm/MC/MCInstrInfo.h"
20#include "llvm/MC/MCRegisterInfo.h"
Ashok Thirumurthi78059352013-05-24 15:55:54 +000021#include "llvm/MC/MCRelocationInfo.h"
Jim Ingham0f063ba2013-03-02 00:26:47 +000022#include "llvm/MC/MCSubtargetInfo.h"
23#include "llvm/Support/ErrorHandling.h"
24#include "llvm/Support/MemoryObject.h"
25#include "llvm/Support/TargetRegistry.h"
Sean Callanan95e5c632012-02-17 00:53:45 +000026#include "llvm/Support/TargetSelect.h"
Jim Ingham0f063ba2013-03-02 00:26:47 +000027#include "llvm/ADT/SmallString.h"
28
Sean Callanan95e5c632012-02-17 00:53:45 +000029
30#include "lldb/Core/Address.h"
31#include "lldb/Core/DataExtractor.h"
Greg Clayton1f746072012-08-29 21:13:06 +000032#include "lldb/Core/Module.h"
Sean Callanan95e5c632012-02-17 00:53:45 +000033#include "lldb/Core/Stream.h"
34#include "lldb/Symbol/SymbolContext.h"
35#include "lldb/Target/ExecutionContext.h"
36#include "lldb/Target/Process.h"
37#include "lldb/Target/RegisterContext.h"
38#include "lldb/Target/Target.h"
39#include "lldb/Target/StackFrame.h"
40
41#include <regex.h>
42
43using namespace lldb;
44using namespace lldb_private;
45
46class InstructionLLVMC : public lldb_private::Instruction
47{
48public:
49 InstructionLLVMC (DisassemblerLLVMC &disasm,
50 const lldb_private::Address &address,
Greg Claytonc8e0c242012-04-13 00:07:34 +000051 AddressClass addr_class) :
Greg Clayton3faf47c2013-03-28 23:42:53 +000052 Instruction (address, addr_class),
53 m_disasm_sp (disasm.shared_from_this()),
54 m_does_branch (eLazyBoolCalculate),
55 m_is_valid (false),
56 m_using_file_addr (false)
Sean Callanan95e5c632012-02-17 00:53:45 +000057 {
58 }
59
60 virtual
61 ~InstructionLLVMC ()
62 {
63 }
64
Sean Callanan95e5c632012-02-17 00:53:45 +000065 virtual bool
Jim Ingham32ce20c2013-03-13 01:55:16 +000066 DoesBranch ()
Sean Callanan95e5c632012-02-17 00:53:45 +000067 {
Jim Ingham32ce20c2013-03-13 01:55:16 +000068 if (m_does_branch == eLazyBoolCalculate)
69 {
Greg Clayton3faf47c2013-03-28 23:42:53 +000070 GetDisassemblerLLVMC().Lock(this, NULL);
Jim Ingham32ce20c2013-03-13 01:55:16 +000071 DataExtractor data;
72 if (m_opcode.GetData(data))
73 {
74 bool is_alternate_isa;
75 lldb::addr_t pc = m_address.GetFileAddress();
76
77 DisassemblerLLVMC::LLVMCDisassembler *mc_disasm_ptr = GetDisasmToUse (is_alternate_isa);
Greg Clayton3faf47c2013-03-28 23:42:53 +000078 const uint8_t *opcode_data = data.GetDataStart();
Jim Ingham32ce20c2013-03-13 01:55:16 +000079 const size_t opcode_data_len = data.GetByteSize();
80 llvm::MCInst inst;
Greg Clayton3faf47c2013-03-28 23:42:53 +000081 const size_t inst_size = mc_disasm_ptr->GetMCInst (opcode_data,
82 opcode_data_len,
83 pc,
84 inst);
Jim Ingham32ce20c2013-03-13 01:55:16 +000085 // Be conservative, if we didn't understand the instruction, say it might branch...
86 if (inst_size == 0)
87 m_does_branch = eLazyBoolYes;
88 else
89 {
Greg Clayton3faf47c2013-03-28 23:42:53 +000090 const bool can_branch = mc_disasm_ptr->CanBranch(inst);
Jim Ingham32ce20c2013-03-13 01:55:16 +000091 if (can_branch)
92 m_does_branch = eLazyBoolYes;
93 else
94 m_does_branch = eLazyBoolNo;
95 }
96 }
Greg Clayton3faf47c2013-03-28 23:42:53 +000097 GetDisassemblerLLVMC().Unlock();
Jim Ingham32ce20c2013-03-13 01:55:16 +000098 }
Sean Callanan7725a462012-03-02 23:22:53 +000099 return m_does_branch == eLazyBoolYes;
Sean Callanan95e5c632012-02-17 00:53:45 +0000100 }
101
Jim Ingham32ce20c2013-03-13 01:55:16 +0000102 DisassemblerLLVMC::LLVMCDisassembler *
103 GetDisasmToUse (bool &is_alternate_isa)
104 {
Jim Ingham32ce20c2013-03-13 01:55:16 +0000105 is_alternate_isa = false;
Greg Clayton3faf47c2013-03-28 23:42:53 +0000106 DisassemblerLLVMC &llvm_disasm = GetDisassemblerLLVMC();
107 if (llvm_disasm.m_alternate_disasm_ap.get() != NULL)
Jim Ingham32ce20c2013-03-13 01:55:16 +0000108 {
109 const AddressClass address_class = GetAddressClass ();
110
111 if (address_class == eAddressClassCodeAlternateISA)
112 {
Jim Ingham32ce20c2013-03-13 01:55:16 +0000113 is_alternate_isa = true;
Greg Clayton3faf47c2013-03-28 23:42:53 +0000114 return llvm_disasm.m_alternate_disasm_ap.get();
Jim Ingham32ce20c2013-03-13 01:55:16 +0000115 }
116 }
Greg Clayton3faf47c2013-03-28 23:42:53 +0000117 return llvm_disasm.m_disasm_ap.get();
Jim Ingham32ce20c2013-03-13 01:55:16 +0000118 }
119
Sean Callanan95e5c632012-02-17 00:53:45 +0000120 virtual size_t
121 Decode (const lldb_private::Disassembler &disassembler,
122 const lldb_private::DataExtractor &data,
Greg Claytonc7bece562013-01-25 18:06:21 +0000123 lldb::offset_t data_offset)
Sean Callanan95e5c632012-02-17 00:53:45 +0000124 {
Greg Claytonba812f42012-05-10 02:52:23 +0000125 // All we have to do is read the opcode which can be easy for some
Michael Sartaincc791bb2013-07-11 16:40:56 +0000126 // architectures
Greg Claytonba812f42012-05-10 02:52:23 +0000127 bool got_op = false;
Greg Clayton3faf47c2013-03-28 23:42:53 +0000128 DisassemblerLLVMC &llvm_disasm = GetDisassemblerLLVMC();
129 const ArchSpec &arch = llvm_disasm.GetArchitecture();
Sean Callanan95e5c632012-02-17 00:53:45 +0000130
Greg Claytonba812f42012-05-10 02:52:23 +0000131 const uint32_t min_op_byte_size = arch.GetMinimumOpcodeByteSize();
132 const uint32_t max_op_byte_size = arch.GetMaximumOpcodeByteSize();
133 if (min_op_byte_size == max_op_byte_size)
134 {
135 // Fixed size instructions, just read that amount of data.
136 if (!data.ValidOffsetForDataOfSize(data_offset, min_op_byte_size))
137 return false;
138
139 switch (min_op_byte_size)
140 {
141 case 1:
142 m_opcode.SetOpcode8 (data.GetU8 (&data_offset));
143 got_op = true;
144 break;
145
146 case 2:
147 m_opcode.SetOpcode16 (data.GetU16 (&data_offset));
148 got_op = true;
149 break;
150
151 case 4:
152 m_opcode.SetOpcode32 (data.GetU32 (&data_offset));
153 got_op = true;
154 break;
155
156 case 8:
157 m_opcode.SetOpcode64 (data.GetU64 (&data_offset));
158 got_op = true;
159 break;
160
161 default:
162 m_opcode.SetOpcodeBytes(data.PeekData(data_offset, min_op_byte_size), min_op_byte_size);
163 got_op = true;
164 break;
165 }
166 }
167 if (!got_op)
168 {
Jim Ingham32ce20c2013-03-13 01:55:16 +0000169 bool is_alternate_isa = false;
170 DisassemblerLLVMC::LLVMCDisassembler *mc_disasm_ptr = GetDisasmToUse (is_alternate_isa);
Greg Claytonba812f42012-05-10 02:52:23 +0000171
Greg Claytonba812f42012-05-10 02:52:23 +0000172 const llvm::Triple::ArchType machine = arch.GetMachine();
173 if (machine == llvm::Triple::arm || machine == llvm::Triple::thumb)
174 {
Jim Ingham32ce20c2013-03-13 01:55:16 +0000175 if (machine == llvm::Triple::thumb || is_alternate_isa)
Greg Claytonba812f42012-05-10 02:52:23 +0000176 {
Greg Clayton79101b52012-08-07 01:29:29 +0000177 uint32_t thumb_opcode = data.GetU16(&data_offset);
Greg Claytonba812f42012-05-10 02:52:23 +0000178 if ((thumb_opcode & 0xe000) != 0xe000 || ((thumb_opcode & 0x1800u) == 0))
179 {
180 m_opcode.SetOpcode16 (thumb_opcode);
Sean Callanan5c97c2f2012-08-06 23:42:52 +0000181 m_is_valid = true;
Greg Claytonba812f42012-05-10 02:52:23 +0000182 }
183 else
184 {
Greg Clayton79101b52012-08-07 01:29:29 +0000185 thumb_opcode <<= 16;
186 thumb_opcode |= data.GetU16(&data_offset);
187 m_opcode.SetOpcode16_2 (thumb_opcode);
Greg Claytonba812f42012-05-10 02:52:23 +0000188 m_is_valid = true;
189 }
190 }
191 else
192 {
193 m_opcode.SetOpcode32 (data.GetU32(&data_offset));
Sean Callanan5c97c2f2012-08-06 23:42:52 +0000194 m_is_valid = true;
Greg Claytonba812f42012-05-10 02:52:23 +0000195 }
196 }
197 else
198 {
199 // The opcode isn't evenly sized, so we need to actually use the llvm
200 // disassembler to parse it and get the size.
Greg Claytonba812f42012-05-10 02:52:23 +0000201 uint8_t *opcode_data = const_cast<uint8_t *>(data.PeekData (data_offset, 1));
Greg Clayton3faf47c2013-03-28 23:42:53 +0000202 const size_t opcode_data_len = data.BytesLeft(data_offset);
Greg Claytonba812f42012-05-10 02:52:23 +0000203 const addr_t pc = m_address.GetFileAddress();
Jim Ingham0f063ba2013-03-02 00:26:47 +0000204 llvm::MCInst inst;
205
Greg Clayton3faf47c2013-03-28 23:42:53 +0000206 llvm_disasm.Lock(this, NULL);
Jim Ingham0f063ba2013-03-02 00:26:47 +0000207 const size_t inst_size = mc_disasm_ptr->GetMCInst(opcode_data,
Greg Clayton3faf47c2013-03-28 23:42:53 +0000208 opcode_data_len,
209 pc,
210 inst);
211 llvm_disasm.Unlock();
Greg Claytonba812f42012-05-10 02:52:23 +0000212 if (inst_size == 0)
213 m_opcode.Clear();
214 else
215 {
216 m_opcode.SetOpcodeBytes(opcode_data, inst_size);
217 m_is_valid = true;
218 }
219 }
220 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000221 return m_opcode.GetByteSize();
222 }
223
224 void
Greg Claytonba812f42012-05-10 02:52:23 +0000225 AppendComment (std::string &description)
Sean Callanan95e5c632012-02-17 00:53:45 +0000226 {
Greg Claytonba812f42012-05-10 02:52:23 +0000227 if (m_comment.empty())
228 m_comment.swap (description);
Sean Callanan95e5c632012-02-17 00:53:45 +0000229 else
Greg Claytonba812f42012-05-10 02:52:23 +0000230 {
231 m_comment.append(", ");
232 m_comment.append(description);
233 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000234 }
235
236 virtual void
Greg Claytonba812f42012-05-10 02:52:23 +0000237 CalculateMnemonicOperandsAndComment (const lldb_private::ExecutionContext *exe_ctx)
Sean Callanan95e5c632012-02-17 00:53:45 +0000238 {
Greg Claytonba812f42012-05-10 02:52:23 +0000239 DataExtractor data;
240 const AddressClass address_class = GetAddressClass ();
241
Sean Callanancd4ae1a2012-08-07 01:44:58 +0000242 if (m_opcode.GetData(data))
Greg Claytonba812f42012-05-10 02:52:23 +0000243 {
244 char out_string[512];
245
Greg Clayton3faf47c2013-03-28 23:42:53 +0000246 DisassemblerLLVMC &llvm_disasm = GetDisassemblerLLVMC();
247
Jim Ingham0f063ba2013-03-02 00:26:47 +0000248 DisassemblerLLVMC::LLVMCDisassembler *mc_disasm_ptr;
Greg Claytonba812f42012-05-10 02:52:23 +0000249
250 if (address_class == eAddressClassCodeAlternateISA)
Greg Clayton3faf47c2013-03-28 23:42:53 +0000251 mc_disasm_ptr = llvm_disasm.m_alternate_disasm_ap.get();
Greg Claytonba812f42012-05-10 02:52:23 +0000252 else
Greg Clayton3faf47c2013-03-28 23:42:53 +0000253 mc_disasm_ptr = llvm_disasm.m_disasm_ap.get();
Greg Claytonba812f42012-05-10 02:52:23 +0000254
Greg Clayton3faf47c2013-03-28 23:42:53 +0000255 lldb::addr_t pc = m_address.GetFileAddress();
256 m_using_file_addr = true;
Greg Claytonba812f42012-05-10 02:52:23 +0000257
Greg Clayton3faf47c2013-03-28 23:42:53 +0000258 const bool data_from_file = GetDisassemblerLLVMC().m_data_from_file;
Daniel Malead79ae052013-08-07 21:54:09 +0000259 bool use_hex_immediates = true;
260 Disassembler::HexImmediateStyle hex_style = Disassembler::eHexStyleC;
261
262 if (exe_ctx)
Greg Claytonba812f42012-05-10 02:52:23 +0000263 {
Daniel Malead79ae052013-08-07 21:54:09 +0000264 Target *target = exe_ctx->GetTargetPtr();
265 if (target)
Greg Clayton3faf47c2013-03-28 23:42:53 +0000266 {
Daniel Malead79ae052013-08-07 21:54:09 +0000267 use_hex_immediates = target->GetUseHexImmediates();
268 hex_style = target->GetHexImmediateStyle();
269
270 if (!data_from_file)
Greg Clayton3faf47c2013-03-28 23:42:53 +0000271 {
272 const lldb::addr_t load_addr = m_address.GetLoadAddress(target);
273 if (load_addr != LLDB_INVALID_ADDRESS)
274 {
275 pc = load_addr;
276 m_using_file_addr = false;
277 }
278 }
279 }
Greg Claytonba812f42012-05-10 02:52:23 +0000280 }
281
Greg Clayton3faf47c2013-03-28 23:42:53 +0000282 llvm_disasm.Lock(this, exe_ctx);
Greg Claytonba812f42012-05-10 02:52:23 +0000283
Greg Clayton3faf47c2013-03-28 23:42:53 +0000284 const uint8_t *opcode_data = data.GetDataStart();
Greg Claytonba812f42012-05-10 02:52:23 +0000285 const size_t opcode_data_len = data.GetByteSize();
Jim Ingham0f063ba2013-03-02 00:26:47 +0000286 llvm::MCInst inst;
Greg Clayton3faf47c2013-03-28 23:42:53 +0000287 size_t inst_size = mc_disasm_ptr->GetMCInst (opcode_data,
288 opcode_data_len,
289 pc,
290 inst);
Daniel Malead79ae052013-08-07 21:54:09 +0000291
Jim Ingham0f063ba2013-03-02 00:26:47 +0000292 if (inst_size > 0)
Daniel Malead79ae052013-08-07 21:54:09 +0000293 {
294 mc_disasm_ptr->SetStyle(use_hex_immediates, hex_style);
Jim Ingham0f063ba2013-03-02 00:26:47 +0000295 mc_disasm_ptr->PrintMCInst(inst, out_string, sizeof(out_string));
Daniel Malead79ae052013-08-07 21:54:09 +0000296 }
297
Greg Clayton3faf47c2013-03-28 23:42:53 +0000298 llvm_disasm.Unlock();
Greg Claytonba812f42012-05-10 02:52:23 +0000299
300 if (inst_size == 0)
301 {
302 m_comment.assign ("unknown opcode");
303 inst_size = m_opcode.GetByteSize();
304 StreamString mnemonic_strm;
Greg Claytonc7bece562013-01-25 18:06:21 +0000305 lldb::offset_t offset = 0;
Greg Claytonba812f42012-05-10 02:52:23 +0000306 switch (inst_size)
307 {
308 case 1:
309 {
310 const uint8_t uval8 = data.GetU8 (&offset);
311 m_opcode.SetOpcode8 (uval8);
312 m_opcode_name.assign (".byte");
313 mnemonic_strm.Printf("0x%2.2x", uval8);
314 }
315 break;
316 case 2:
317 {
318 const uint16_t uval16 = data.GetU16(&offset);
319 m_opcode.SetOpcode16(uval16);
320 m_opcode_name.assign (".short");
321 mnemonic_strm.Printf("0x%4.4x", uval16);
322 }
323 break;
324 case 4:
325 {
326 const uint32_t uval32 = data.GetU32(&offset);
327 m_opcode.SetOpcode32(uval32);
328 m_opcode_name.assign (".long");
329 mnemonic_strm.Printf("0x%8.8x", uval32);
330 }
331 break;
332 case 8:
333 {
334 const uint64_t uval64 = data.GetU64(&offset);
335 m_opcode.SetOpcode64(uval64);
336 m_opcode_name.assign (".quad");
Daniel Malead01b2952012-11-29 21:49:15 +0000337 mnemonic_strm.Printf("0x%16.16" PRIx64, uval64);
Greg Claytonba812f42012-05-10 02:52:23 +0000338 }
339 break;
340 default:
341 if (inst_size == 0)
342 return;
343 else
344 {
345 const uint8_t *bytes = data.PeekData(offset, inst_size);
346 if (bytes == NULL)
347 return;
348 m_opcode_name.assign (".byte");
349 m_opcode.SetOpcodeBytes(bytes, inst_size);
350 mnemonic_strm.Printf("0x%2.2x", bytes[0]);
351 for (uint32_t i=1; i<inst_size; ++i)
352 mnemonic_strm.Printf(" 0x%2.2x", bytes[i]);
353 }
354 break;
355 }
Jim Ingham0f063ba2013-03-02 00:26:47 +0000356 m_mnemonics.swap(mnemonic_strm.GetString());
Greg Claytonba812f42012-05-10 02:52:23 +0000357 return;
358 }
359 else
360 {
361 if (m_does_branch == eLazyBoolCalculate)
362 {
Greg Clayton3faf47c2013-03-28 23:42:53 +0000363 const bool can_branch = mc_disasm_ptr->CanBranch(inst);
Jim Ingham0f063ba2013-03-02 00:26:47 +0000364 if (can_branch)
Greg Claytonba812f42012-05-10 02:52:23 +0000365 m_does_branch = eLazyBoolYes;
366 else
367 m_does_branch = eLazyBoolNo;
Jim Ingham0f063ba2013-03-02 00:26:47 +0000368
Greg Claytonba812f42012-05-10 02:52:23 +0000369 }
370 }
371
372 if (!s_regex_compiled)
373 {
374 ::regcomp(&s_regex, "[ \t]*([^ ^\t]+)[ \t]*([^ ^\t].*)?", REG_EXTENDED);
375 s_regex_compiled = true;
376 }
377
378 ::regmatch_t matches[3];
379
380 if (!::regexec(&s_regex, out_string, sizeof(matches) / sizeof(::regmatch_t), matches, 0))
381 {
382 if (matches[1].rm_so != -1)
383 m_opcode_name.assign(out_string + matches[1].rm_so, matches[1].rm_eo - matches[1].rm_so);
384 if (matches[2].rm_so != -1)
Jim Ingham0f063ba2013-03-02 00:26:47 +0000385 m_mnemonics.assign(out_string + matches[2].rm_so, matches[2].rm_eo - matches[2].rm_so);
Greg Claytonba812f42012-05-10 02:52:23 +0000386 }
387 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000388 }
389
390 bool
Greg Clayton3faf47c2013-03-28 23:42:53 +0000391 IsValid () const
Sean Callanan95e5c632012-02-17 00:53:45 +0000392 {
393 return m_is_valid;
394 }
395
Greg Clayton3faf47c2013-03-28 23:42:53 +0000396 bool
397 UsingFileAddress() const
398 {
399 return m_using_file_addr;
400 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000401 size_t
Greg Clayton3faf47c2013-03-28 23:42:53 +0000402 GetByteSize () const
Sean Callanan95e5c632012-02-17 00:53:45 +0000403 {
404 return m_opcode.GetByteSize();
405 }
Greg Clayton3faf47c2013-03-28 23:42:53 +0000406
407 DisassemblerLLVMC &
408 GetDisassemblerLLVMC ()
409 {
410 return *(DisassemblerLLVMC *)m_disasm_sp.get();
411 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000412protected:
Sean Callanan95e5c632012-02-17 00:53:45 +0000413
Sean Callanan7e6d4e52012-08-01 18:50:59 +0000414 DisassemblerSP m_disasm_sp; // for ownership
Sean Callanan7725a462012-03-02 23:22:53 +0000415 LazyBool m_does_branch;
Greg Clayton3faf47c2013-03-28 23:42:53 +0000416 bool m_is_valid;
417 bool m_using_file_addr;
Sean Callanan95e5c632012-02-17 00:53:45 +0000418
419 static bool s_regex_compiled;
420 static ::regex_t s_regex;
421};
422
423bool InstructionLLVMC::s_regex_compiled = false;
424::regex_t InstructionLLVMC::s_regex;
425
Jim Ingham0f063ba2013-03-02 00:26:47 +0000426DisassemblerLLVMC::LLVMCDisassembler::LLVMCDisassembler (const char *triple, unsigned flavor, DisassemblerLLVMC &owner):
427 m_is_valid(true)
428{
429 std::string Error;
430 const llvm::Target *curr_target = llvm::TargetRegistry::lookupTarget(triple, Error);
431 if (!curr_target)
432 {
433 m_is_valid = false;
434 return;
435 }
436
437 m_instr_info_ap.reset(curr_target->createMCInstrInfo());
438 m_reg_info_ap.reset (curr_target->createMCRegInfo(triple));
439
440 std::string features_str;
441
442 m_subtarget_info_ap.reset(curr_target->createMCSubtargetInfo(triple, "",
443 features_str));
444
Sylvestre Ledru7a89d6f2013-05-13 13:41:13 +0000445 m_asm_info_ap.reset(curr_target->createMCAsmInfo(*curr_target->createMCRegInfo(triple), triple));
446
Jim Ingham0f063ba2013-03-02 00:26:47 +0000447 if (m_instr_info_ap.get() == NULL || m_reg_info_ap.get() == NULL || m_subtarget_info_ap.get() == NULL || m_asm_info_ap.get() == NULL)
448 {
Matt Kopec787d1622013-03-12 17:45:38 +0000449 m_is_valid = false;
Jim Ingham0f063ba2013-03-02 00:26:47 +0000450 return;
451 }
452
Bill Wendlingd63d0a82013-06-18 07:50:43 +0000453 m_context_ap.reset(new llvm::MCContext(m_asm_info_ap.get(), m_reg_info_ap.get(), 0));
Jim Ingham0f063ba2013-03-02 00:26:47 +0000454
455 m_disasm_ap.reset(curr_target->createMCDisassembler(*m_subtarget_info_ap.get()));
Ashok Thirumurthi78059352013-05-24 15:55:54 +0000456 if (m_disasm_ap.get() && m_context_ap.get())
Jim Ingham0f063ba2013-03-02 00:26:47 +0000457 {
Ashok Thirumurthi78059352013-05-24 15:55:54 +0000458 llvm::OwningPtr<llvm::MCRelocationInfo> RelInfo(curr_target->createMCRelocationInfo(triple, *m_context_ap.get()));
459 if (!RelInfo)
460 {
461 m_is_valid = false;
462 return;
463 }
Jim Ingham0f063ba2013-03-02 00:26:47 +0000464 m_disasm_ap->setupForSymbolicDisassembly(NULL,
Ashok Thirumurthi78059352013-05-24 15:55:54 +0000465 DisassemblerLLVMC::SymbolLookupCallback,
466 (void *) &owner,
467 m_context_ap.get(),
468 RelInfo);
Jim Ingham0f063ba2013-03-02 00:26:47 +0000469
470 unsigned asm_printer_variant;
471 if (flavor == ~0U)
472 asm_printer_variant = m_asm_info_ap->getAssemblerDialect();
473 else
474 {
475 asm_printer_variant = flavor;
476 }
477
478 m_instr_printer_ap.reset(curr_target->createMCInstPrinter(asm_printer_variant,
479 *m_asm_info_ap.get(),
480 *m_instr_info_ap.get(),
481 *m_reg_info_ap.get(),
482 *m_subtarget_info_ap.get()));
483 if (m_instr_printer_ap.get() == NULL)
484 {
485 m_disasm_ap.reset();
486 m_is_valid = false;
487 }
488 }
489 else
490 m_is_valid = false;
491}
492
Greg Claytone01e07b2013-04-18 18:10:51 +0000493DisassemblerLLVMC::LLVMCDisassembler::~LLVMCDisassembler()
494{
495}
496
Jim Ingham0f063ba2013-03-02 00:26:47 +0000497namespace {
498 // This is the memory object we use in GetInstruction.
499 class LLDBDisasmMemoryObject : public llvm::MemoryObject {
Greg Clayton3faf47c2013-03-28 23:42:53 +0000500 const uint8_t *m_bytes;
Jim Ingham0f063ba2013-03-02 00:26:47 +0000501 uint64_t m_size;
502 uint64_t m_base_PC;
503 public:
Greg Clayton3faf47c2013-03-28 23:42:53 +0000504 LLDBDisasmMemoryObject(const uint8_t *bytes, uint64_t size, uint64_t basePC) :
Jim Ingham0f063ba2013-03-02 00:26:47 +0000505 m_bytes(bytes), m_size(size), m_base_PC(basePC) {}
506
507 uint64_t getBase() const { return m_base_PC; }
508 uint64_t getExtent() const { return m_size; }
509
510 int readByte(uint64_t addr, uint8_t *byte) const {
511 if (addr - m_base_PC >= m_size)
512 return -1;
513 *byte = m_bytes[addr - m_base_PC];
514 return 0;
515 }
516 };
517} // End Anonymous Namespace
518
519uint64_t
Greg Clayton3faf47c2013-03-28 23:42:53 +0000520DisassemblerLLVMC::LLVMCDisassembler::GetMCInst (const uint8_t *opcode_data,
521 size_t opcode_data_len,
522 lldb::addr_t pc,
523 llvm::MCInst &mc_inst)
Jim Ingham0f063ba2013-03-02 00:26:47 +0000524{
525 LLDBDisasmMemoryObject memory_object (opcode_data, opcode_data_len, pc);
Jim Ingham0f063ba2013-03-02 00:26:47 +0000526 llvm::MCDisassembler::DecodeStatus status;
527
528 uint64_t new_inst_size;
529 status = m_disasm_ap->getInstruction(mc_inst,
530 new_inst_size,
531 memory_object,
532 pc,
533 llvm::nulls(),
534 llvm::nulls());
535 if (status == llvm::MCDisassembler::Success)
536 return new_inst_size;
537 else
538 return 0;
539}
540
541uint64_t
Greg Clayton3faf47c2013-03-28 23:42:53 +0000542DisassemblerLLVMC::LLVMCDisassembler::PrintMCInst (llvm::MCInst &mc_inst,
543 char *dst,
544 size_t dst_len)
Jim Ingham0f063ba2013-03-02 00:26:47 +0000545{
546 llvm::StringRef unused_annotations;
547 llvm::SmallString<64> inst_string;
548 llvm::raw_svector_ostream inst_stream(inst_string);
549 m_instr_printer_ap->printInst (&mc_inst, inst_stream, unused_annotations);
550 inst_stream.flush();
Greg Clayton3faf47c2013-03-28 23:42:53 +0000551 const size_t output_size = std::min(dst_len - 1, inst_string.size());
552 std::memcpy(dst, inst_string.data(), output_size);
553 dst[output_size] = '\0';
Jim Ingham0f063ba2013-03-02 00:26:47 +0000554
555 return output_size;
556}
557
Daniel Malead79ae052013-08-07 21:54:09 +0000558void
559DisassemblerLLVMC::LLVMCDisassembler::SetStyle (bool use_hex_immed, HexImmediateStyle hex_style)
560{
561 m_instr_printer_ap->setPrintImmHex(use_hex_immed);
562 switch(hex_style)
563 {
564 case eHexStyleC: m_instr_printer_ap->setPrintImmHex(llvm::HexStyle::C); break;
565 case eHexStyleAsm: m_instr_printer_ap->setPrintImmHex(llvm::HexStyle::Asm); break;
566 }
567}
568
Jim Ingham0f063ba2013-03-02 00:26:47 +0000569bool
570DisassemblerLLVMC::LLVMCDisassembler::CanBranch (llvm::MCInst &mc_inst)
571{
572 return m_instr_info_ap->get(mc_inst.getOpcode()).mayAffectControlFlow(mc_inst, *m_reg_info_ap.get());
573}
574
575bool
576DisassemblerLLVMC::FlavorValidForArchSpec (const lldb_private::ArchSpec &arch, const char *flavor)
577{
578 llvm::Triple triple = arch.GetTriple();
579 if (flavor == NULL || strcmp (flavor, "default") == 0)
580 return true;
581
582 if (triple.getArch() == llvm::Triple::x86 || triple.getArch() == llvm::Triple::x86_64)
583 {
584 if (strcmp (flavor, "intel") == 0 || strcmp (flavor, "att") == 0)
585 return true;
586 else
587 return false;
588 }
589 else
590 return false;
591}
592
593
Sean Callanan95e5c632012-02-17 00:53:45 +0000594Disassembler *
Jim Ingham0f063ba2013-03-02 00:26:47 +0000595DisassemblerLLVMC::CreateInstance (const ArchSpec &arch, const char *flavor)
Sean Callanan95e5c632012-02-17 00:53:45 +0000596{
Greg Clayton9e6cffc2012-09-19 22:25:17 +0000597 if (arch.GetTriple().getArch() != llvm::Triple::UnknownArch)
598 {
Greg Clayton7b0992d2013-04-18 22:45:39 +0000599 std::unique_ptr<DisassemblerLLVMC> disasm_ap (new DisassemblerLLVMC(arch, flavor));
Sean Callanan95e5c632012-02-17 00:53:45 +0000600
Greg Clayton9e6cffc2012-09-19 22:25:17 +0000601 if (disasm_ap.get() && disasm_ap->IsValid())
602 return disasm_ap.release();
603 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000604 return NULL;
605}
606
Jim Ingham0f063ba2013-03-02 00:26:47 +0000607DisassemblerLLVMC::DisassemblerLLVMC (const ArchSpec &arch, const char *flavor_string) :
608 Disassembler(arch, flavor_string),
Greg Claytonba812f42012-05-10 02:52:23 +0000609 m_exe_ctx (NULL),
Greg Clayton3faf47c2013-03-28 23:42:53 +0000610 m_inst (NULL),
611 m_data_from_file (false)
Sean Callanan95e5c632012-02-17 00:53:45 +0000612{
Jim Ingham0f063ba2013-03-02 00:26:47 +0000613 if (!FlavorValidForArchSpec (arch, m_flavor.c_str()))
614 {
615 m_flavor.assign("default");
616 }
617
618 const char *triple = arch.GetTriple().getTriple().c_str();
619 unsigned flavor = ~0U;
620
621 // So far the only supported flavor is "intel" on x86. The base class will set this
622 // correctly coming in.
623 if (arch.GetTriple().getArch() == llvm::Triple::x86
624 || arch.GetTriple().getArch() == llvm::Triple::x86_64)
625 {
626 if (m_flavor == "intel")
627 {
628 flavor = 1;
629 }
630 else if (m_flavor == "att")
631 {
632 flavor = 0;
633 }
634 }
635
Jason Molenda01aa5342013-07-22 22:11:53 +0000636 ArchSpec thumb_arch(arch);
Sean Callanan95e5c632012-02-17 00:53:45 +0000637 if (arch.GetTriple().getArch() == llvm::Triple::arm)
638 {
Greg Claytona80ea122013-05-03 01:05:04 +0000639 std::string thumb_arch_name (thumb_arch.GetTriple().getArchName().str());
640 // Replace "arm" with "thumb" so we get all thumb variants correct
641 if (thumb_arch_name.size() > 3)
642 {
643 thumb_arch_name.erase(0,3);
644 thumb_arch_name.insert(0, "thumb");
645 }
646 else
647 {
648 thumb_arch_name = "thumbv7";
649 }
650 thumb_arch.GetTriple().setArchName(llvm::StringRef(thumb_arch_name.c_str()));
Jason Molenda01aa5342013-07-22 22:11:53 +0000651 }
652
653 // Cortex-M3 devices (e.g. armv7m) can only execute thumb (T2) instructions,
654 // so hardcode the primary disassembler to thumb mode.
655 if (arch.GetTriple().getArch() == llvm::Triple::arm
656 && (arch.GetCore() == ArchSpec::Core::eCore_arm_armv7m || arch.GetCore() == ArchSpec::Core::eCore_arm_armv7em))
657 {
658 triple = thumb_arch.GetTriple().getTriple().c_str();
659 }
660
661 m_disasm_ap.reset (new LLVMCDisassembler(triple, flavor, *this));
662 if (!m_disasm_ap->IsValid())
663 {
664 // We use m_disasm_ap.get() to tell whether we are valid or not, so if this isn't good for some reason,
665 // we reset it, and then we won't be valid and FindPlugin will fail and we won't get used.
666 m_disasm_ap.reset();
667 }
668
669 // For arm CPUs that can execute arm or thumb instructions, also create a thumb instruction disassembler.
670 if (arch.GetTriple().getArch() == llvm::Triple::arm)
671 {
Greg Claytonba812f42012-05-10 02:52:23 +0000672 std::string thumb_triple(thumb_arch.GetTriple().getTriple());
Jim Ingham0f063ba2013-03-02 00:26:47 +0000673 m_alternate_disasm_ap.reset(new LLVMCDisassembler(thumb_triple.c_str(), flavor, *this));
674 if (!m_alternate_disasm_ap->IsValid())
675 {
676 m_disasm_ap.reset();
677 m_alternate_disasm_ap.reset();
678 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000679 }
680}
681
682DisassemblerLLVMC::~DisassemblerLLVMC()
683{
684}
685
686size_t
687DisassemblerLLVMC::DecodeInstructions (const Address &base_addr,
688 const DataExtractor& data,
Greg Claytonc7bece562013-01-25 18:06:21 +0000689 lldb::offset_t data_offset,
690 size_t num_instructions,
Greg Clayton3faf47c2013-03-28 23:42:53 +0000691 bool append,
692 bool data_from_file)
Sean Callanan95e5c632012-02-17 00:53:45 +0000693{
694 if (!append)
695 m_instruction_list.Clear();
696
697 if (!IsValid())
698 return 0;
699
Greg Clayton3faf47c2013-03-28 23:42:53 +0000700 m_data_from_file = data_from_file;
Sean Callanan95e5c632012-02-17 00:53:45 +0000701 uint32_t data_cursor = data_offset;
Greg Claytonba812f42012-05-10 02:52:23 +0000702 const size_t data_byte_size = data.GetByteSize();
Sean Callanan95e5c632012-02-17 00:53:45 +0000703 uint32_t instructions_parsed = 0;
Greg Claytonba812f42012-05-10 02:52:23 +0000704 Address inst_addr(base_addr);
Sean Callanan95e5c632012-02-17 00:53:45 +0000705
Greg Claytonba812f42012-05-10 02:52:23 +0000706 while (data_cursor < data_byte_size && instructions_parsed < num_instructions)
Sean Callanan95e5c632012-02-17 00:53:45 +0000707 {
Sean Callanan95e5c632012-02-17 00:53:45 +0000708
Greg Claytonba812f42012-05-10 02:52:23 +0000709 AddressClass address_class = eAddressClassCode;
Sean Callanan95e5c632012-02-17 00:53:45 +0000710
Jim Ingham0f063ba2013-03-02 00:26:47 +0000711 if (m_alternate_disasm_ap.get() != NULL)
Greg Claytonba812f42012-05-10 02:52:23 +0000712 address_class = inst_addr.GetAddressClass ();
Sean Callanan95e5c632012-02-17 00:53:45 +0000713
714 InstructionSP inst_sp(new InstructionLLVMC(*this,
Greg Claytonba812f42012-05-10 02:52:23 +0000715 inst_addr,
Sean Callanan95e5c632012-02-17 00:53:45 +0000716 address_class));
717
718 if (!inst_sp)
Greg Claytonba812f42012-05-10 02:52:23 +0000719 break;
720
Sean Callanan95e5c632012-02-17 00:53:45 +0000721 uint32_t inst_size = inst_sp->Decode(*this, data, data_cursor);
722
Greg Claytonba812f42012-05-10 02:52:23 +0000723 if (inst_size == 0)
724 break;
725
Sean Callanan95e5c632012-02-17 00:53:45 +0000726 m_instruction_list.Append(inst_sp);
Sean Callanan95e5c632012-02-17 00:53:45 +0000727 data_cursor += inst_size;
Greg Claytonba812f42012-05-10 02:52:23 +0000728 inst_addr.Slide(inst_size);
Sean Callanan95e5c632012-02-17 00:53:45 +0000729 instructions_parsed++;
730 }
731
732 return data_cursor - data_offset;
733}
734
735void
736DisassemblerLLVMC::Initialize()
737{
738 PluginManager::RegisterPlugin (GetPluginNameStatic(),
Greg Clayton57abc5d2013-05-10 21:47:16 +0000739 "Disassembler that uses LLVM MC to disassemble i386, x86_64 and ARM.",
Sean Callanan95e5c632012-02-17 00:53:45 +0000740 CreateInstance);
741
742 llvm::InitializeAllTargetInfos();
743 llvm::InitializeAllTargetMCs();
744 llvm::InitializeAllAsmParsers();
745 llvm::InitializeAllDisassemblers();
746}
747
748void
749DisassemblerLLVMC::Terminate()
750{
751 PluginManager::UnregisterPlugin (CreateInstance);
752}
753
754
Greg Clayton57abc5d2013-05-10 21:47:16 +0000755ConstString
Sean Callanan95e5c632012-02-17 00:53:45 +0000756DisassemblerLLVMC::GetPluginNameStatic()
757{
Greg Clayton57abc5d2013-05-10 21:47:16 +0000758 static ConstString g_name("llvm-mc");
759 return g_name;
Sean Callanan95e5c632012-02-17 00:53:45 +0000760}
761
Greg Claytonba812f42012-05-10 02:52:23 +0000762int DisassemblerLLVMC::OpInfoCallback (void *disassembler,
763 uint64_t pc,
764 uint64_t offset,
765 uint64_t size,
766 int tag_type,
767 void *tag_bug)
Sean Callanan95e5c632012-02-17 00:53:45 +0000768{
Greg Claytonba812f42012-05-10 02:52:23 +0000769 return static_cast<DisassemblerLLVMC*>(disassembler)->OpInfo (pc,
770 offset,
771 size,
772 tag_type,
773 tag_bug);
Sean Callanan95e5c632012-02-17 00:53:45 +0000774}
775
Greg Claytonba812f42012-05-10 02:52:23 +0000776const char *DisassemblerLLVMC::SymbolLookupCallback (void *disassembler,
777 uint64_t value,
778 uint64_t *type,
779 uint64_t pc,
780 const char **name)
Sean Callanan95e5c632012-02-17 00:53:45 +0000781{
Greg Claytonba812f42012-05-10 02:52:23 +0000782 return static_cast<DisassemblerLLVMC*>(disassembler)->SymbolLookup(value,
783 type,
784 pc,
785 name);
Sean Callanan95e5c632012-02-17 00:53:45 +0000786}
787
788int DisassemblerLLVMC::OpInfo (uint64_t PC,
789 uint64_t Offset,
790 uint64_t Size,
Greg Claytonba812f42012-05-10 02:52:23 +0000791 int tag_type,
792 void *tag_bug)
Sean Callanan95e5c632012-02-17 00:53:45 +0000793{
Greg Claytonba812f42012-05-10 02:52:23 +0000794 switch (tag_type)
Sean Callanan95e5c632012-02-17 00:53:45 +0000795 {
796 default:
797 break;
798 case 1:
Greg Claytonba812f42012-05-10 02:52:23 +0000799 bzero (tag_bug, sizeof(::LLVMOpInfo1));
Sean Callanan95e5c632012-02-17 00:53:45 +0000800 break;
801 }
802 return 0;
803}
804
Greg Claytonba812f42012-05-10 02:52:23 +0000805const char *DisassemblerLLVMC::SymbolLookup (uint64_t value,
806 uint64_t *type_ptr,
807 uint64_t pc,
808 const char **name)
Sean Callanan95e5c632012-02-17 00:53:45 +0000809{
Greg Claytonba812f42012-05-10 02:52:23 +0000810 if (*type_ptr)
811 {
812 if (m_exe_ctx && m_inst)
813 {
814 //std::string remove_this_prior_to_checkin;
Greg Claytonba812f42012-05-10 02:52:23 +0000815 Target *target = m_exe_ctx ? m_exe_ctx->GetTargetPtr() : NULL;
Greg Clayton3faf47c2013-03-28 23:42:53 +0000816 Address value_so_addr;
817 if (m_inst->UsingFileAddress())
Greg Claytonba812f42012-05-10 02:52:23 +0000818 {
819 ModuleSP module_sp(m_inst->GetAddress().GetModule());
820 if (module_sp)
Greg Clayton3faf47c2013-03-28 23:42:53 +0000821 module_sp->ResolveFileAddress(value, value_so_addr);
Greg Claytonba812f42012-05-10 02:52:23 +0000822 }
Greg Clayton3faf47c2013-03-28 23:42:53 +0000823 else if (target && !target->GetSectionLoadList().IsEmpty())
824 {
825 target->GetSectionLoadList().ResolveLoadAddress(value, value_so_addr);
826 }
827
828 if (value_so_addr.IsValid() && value_so_addr.GetSection())
Sean Callanan95e5c632012-02-17 00:53:45 +0000829 {
Sean Callanan95e5c632012-02-17 00:53:45 +0000830 StreamString ss;
831
Greg Clayton3faf47c2013-03-28 23:42:53 +0000832 value_so_addr.Dump (&ss,
833 target,
834 Address::DumpStyleResolvedDescriptionNoModule,
835 Address::DumpStyleSectionNameOffset);
Sean Callanan95e5c632012-02-17 00:53:45 +0000836
Sean Callanan745af462012-03-22 20:04:23 +0000837 if (!ss.GetString().empty())
Greg Claytonba812f42012-05-10 02:52:23 +0000838 {
Greg Claytonba812f42012-05-10 02:52:23 +0000839 m_inst->AppendComment(ss.GetString());
840 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000841 }
842 }
843 }
Greg Claytonba812f42012-05-10 02:52:23 +0000844
845 *type_ptr = LLVMDisassembler_ReferenceType_InOut_None;
846 *name = NULL;
847 return NULL;
Sean Callanan95e5c632012-02-17 00:53:45 +0000848}
849
850//------------------------------------------------------------------
851// PluginInterface protocol
852//------------------------------------------------------------------
Greg Clayton57abc5d2013-05-10 21:47:16 +0000853ConstString
Sean Callanan95e5c632012-02-17 00:53:45 +0000854DisassemblerLLVMC::GetPluginName()
855{
Sean Callanan95e5c632012-02-17 00:53:45 +0000856 return GetPluginNameStatic();
857}
858
859uint32_t
860DisassemblerLLVMC::GetPluginVersion()
861{
862 return 1;
863}
864