blob: 233515a23f205bfbf3ebb05e68f7a37c44391a52 [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;
259 if (!data_from_file)
Greg Claytonba812f42012-05-10 02:52:23 +0000260 {
Greg Clayton3faf47c2013-03-28 23:42:53 +0000261 if (exe_ctx)
262 {
263 Target *target = exe_ctx->GetTargetPtr();
264 if (target)
265 {
266 const lldb::addr_t load_addr = m_address.GetLoadAddress(target);
267 if (load_addr != LLDB_INVALID_ADDRESS)
268 {
269 pc = load_addr;
270 m_using_file_addr = false;
271 }
272 }
273 }
Greg Claytonba812f42012-05-10 02:52:23 +0000274 }
275
Greg Clayton3faf47c2013-03-28 23:42:53 +0000276 llvm_disasm.Lock(this, exe_ctx);
Greg Claytonba812f42012-05-10 02:52:23 +0000277
Greg Clayton3faf47c2013-03-28 23:42:53 +0000278 const uint8_t *opcode_data = data.GetDataStart();
Greg Claytonba812f42012-05-10 02:52:23 +0000279 const size_t opcode_data_len = data.GetByteSize();
Jim Ingham0f063ba2013-03-02 00:26:47 +0000280 llvm::MCInst inst;
Greg Clayton3faf47c2013-03-28 23:42:53 +0000281 size_t inst_size = mc_disasm_ptr->GetMCInst (opcode_data,
282 opcode_data_len,
283 pc,
284 inst);
Jim Ingham0f063ba2013-03-02 00:26:47 +0000285
286 if (inst_size > 0)
287 mc_disasm_ptr->PrintMCInst(inst, out_string, sizeof(out_string));
Greg Claytonba812f42012-05-10 02:52:23 +0000288
Greg Clayton3faf47c2013-03-28 23:42:53 +0000289 llvm_disasm.Unlock();
Greg Claytonba812f42012-05-10 02:52:23 +0000290
291 if (inst_size == 0)
292 {
293 m_comment.assign ("unknown opcode");
294 inst_size = m_opcode.GetByteSize();
295 StreamString mnemonic_strm;
Greg Claytonc7bece562013-01-25 18:06:21 +0000296 lldb::offset_t offset = 0;
Greg Claytonba812f42012-05-10 02:52:23 +0000297 switch (inst_size)
298 {
299 case 1:
300 {
301 const uint8_t uval8 = data.GetU8 (&offset);
302 m_opcode.SetOpcode8 (uval8);
303 m_opcode_name.assign (".byte");
304 mnemonic_strm.Printf("0x%2.2x", uval8);
305 }
306 break;
307 case 2:
308 {
309 const uint16_t uval16 = data.GetU16(&offset);
310 m_opcode.SetOpcode16(uval16);
311 m_opcode_name.assign (".short");
312 mnemonic_strm.Printf("0x%4.4x", uval16);
313 }
314 break;
315 case 4:
316 {
317 const uint32_t uval32 = data.GetU32(&offset);
318 m_opcode.SetOpcode32(uval32);
319 m_opcode_name.assign (".long");
320 mnemonic_strm.Printf("0x%8.8x", uval32);
321 }
322 break;
323 case 8:
324 {
325 const uint64_t uval64 = data.GetU64(&offset);
326 m_opcode.SetOpcode64(uval64);
327 m_opcode_name.assign (".quad");
Daniel Malead01b2952012-11-29 21:49:15 +0000328 mnemonic_strm.Printf("0x%16.16" PRIx64, uval64);
Greg Claytonba812f42012-05-10 02:52:23 +0000329 }
330 break;
331 default:
332 if (inst_size == 0)
333 return;
334 else
335 {
336 const uint8_t *bytes = data.PeekData(offset, inst_size);
337 if (bytes == NULL)
338 return;
339 m_opcode_name.assign (".byte");
340 m_opcode.SetOpcodeBytes(bytes, inst_size);
341 mnemonic_strm.Printf("0x%2.2x", bytes[0]);
342 for (uint32_t i=1; i<inst_size; ++i)
343 mnemonic_strm.Printf(" 0x%2.2x", bytes[i]);
344 }
345 break;
346 }
Jim Ingham0f063ba2013-03-02 00:26:47 +0000347 m_mnemonics.swap(mnemonic_strm.GetString());
Greg Claytonba812f42012-05-10 02:52:23 +0000348 return;
349 }
350 else
351 {
352 if (m_does_branch == eLazyBoolCalculate)
353 {
Greg Clayton3faf47c2013-03-28 23:42:53 +0000354 const bool can_branch = mc_disasm_ptr->CanBranch(inst);
Jim Ingham0f063ba2013-03-02 00:26:47 +0000355 if (can_branch)
Greg Claytonba812f42012-05-10 02:52:23 +0000356 m_does_branch = eLazyBoolYes;
357 else
358 m_does_branch = eLazyBoolNo;
Jim Ingham0f063ba2013-03-02 00:26:47 +0000359
Greg Claytonba812f42012-05-10 02:52:23 +0000360 }
361 }
362
363 if (!s_regex_compiled)
364 {
365 ::regcomp(&s_regex, "[ \t]*([^ ^\t]+)[ \t]*([^ ^\t].*)?", REG_EXTENDED);
366 s_regex_compiled = true;
367 }
368
369 ::regmatch_t matches[3];
370
371 if (!::regexec(&s_regex, out_string, sizeof(matches) / sizeof(::regmatch_t), matches, 0))
372 {
373 if (matches[1].rm_so != -1)
374 m_opcode_name.assign(out_string + matches[1].rm_so, matches[1].rm_eo - matches[1].rm_so);
375 if (matches[2].rm_so != -1)
Jim Ingham0f063ba2013-03-02 00:26:47 +0000376 m_mnemonics.assign(out_string + matches[2].rm_so, matches[2].rm_eo - matches[2].rm_so);
Greg Claytonba812f42012-05-10 02:52:23 +0000377 }
378 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000379 }
380
381 bool
Greg Clayton3faf47c2013-03-28 23:42:53 +0000382 IsValid () const
Sean Callanan95e5c632012-02-17 00:53:45 +0000383 {
384 return m_is_valid;
385 }
386
Greg Clayton3faf47c2013-03-28 23:42:53 +0000387 bool
388 UsingFileAddress() const
389 {
390 return m_using_file_addr;
391 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000392 size_t
Greg Clayton3faf47c2013-03-28 23:42:53 +0000393 GetByteSize () const
Sean Callanan95e5c632012-02-17 00:53:45 +0000394 {
395 return m_opcode.GetByteSize();
396 }
Greg Clayton3faf47c2013-03-28 23:42:53 +0000397
398 DisassemblerLLVMC &
399 GetDisassemblerLLVMC ()
400 {
401 return *(DisassemblerLLVMC *)m_disasm_sp.get();
402 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000403protected:
Sean Callanan95e5c632012-02-17 00:53:45 +0000404
Sean Callanan7e6d4e52012-08-01 18:50:59 +0000405 DisassemblerSP m_disasm_sp; // for ownership
Sean Callanan7725a462012-03-02 23:22:53 +0000406 LazyBool m_does_branch;
Greg Clayton3faf47c2013-03-28 23:42:53 +0000407 bool m_is_valid;
408 bool m_using_file_addr;
Sean Callanan95e5c632012-02-17 00:53:45 +0000409
410 static bool s_regex_compiled;
411 static ::regex_t s_regex;
412};
413
414bool InstructionLLVMC::s_regex_compiled = false;
415::regex_t InstructionLLVMC::s_regex;
416
Jim Ingham0f063ba2013-03-02 00:26:47 +0000417DisassemblerLLVMC::LLVMCDisassembler::LLVMCDisassembler (const char *triple, unsigned flavor, DisassemblerLLVMC &owner):
418 m_is_valid(true)
419{
420 std::string Error;
421 const llvm::Target *curr_target = llvm::TargetRegistry::lookupTarget(triple, Error);
422 if (!curr_target)
423 {
424 m_is_valid = false;
425 return;
426 }
427
428 m_instr_info_ap.reset(curr_target->createMCInstrInfo());
429 m_reg_info_ap.reset (curr_target->createMCRegInfo(triple));
430
431 std::string features_str;
432
433 m_subtarget_info_ap.reset(curr_target->createMCSubtargetInfo(triple, "",
434 features_str));
435
Sylvestre Ledru7a89d6f2013-05-13 13:41:13 +0000436 m_asm_info_ap.reset(curr_target->createMCAsmInfo(*curr_target->createMCRegInfo(triple), triple));
437
Jim Ingham0f063ba2013-03-02 00:26:47 +0000438 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)
439 {
Matt Kopec787d1622013-03-12 17:45:38 +0000440 m_is_valid = false;
Jim Ingham0f063ba2013-03-02 00:26:47 +0000441 return;
442 }
443
Bill Wendlingd63d0a82013-06-18 07:50:43 +0000444 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 +0000445
446 m_disasm_ap.reset(curr_target->createMCDisassembler(*m_subtarget_info_ap.get()));
Ashok Thirumurthi78059352013-05-24 15:55:54 +0000447 if (m_disasm_ap.get() && m_context_ap.get())
Jim Ingham0f063ba2013-03-02 00:26:47 +0000448 {
Ashok Thirumurthi78059352013-05-24 15:55:54 +0000449 llvm::OwningPtr<llvm::MCRelocationInfo> RelInfo(curr_target->createMCRelocationInfo(triple, *m_context_ap.get()));
450 if (!RelInfo)
451 {
452 m_is_valid = false;
453 return;
454 }
Jim Ingham0f063ba2013-03-02 00:26:47 +0000455 m_disasm_ap->setupForSymbolicDisassembly(NULL,
Ashok Thirumurthi78059352013-05-24 15:55:54 +0000456 DisassemblerLLVMC::SymbolLookupCallback,
457 (void *) &owner,
458 m_context_ap.get(),
459 RelInfo);
Jim Ingham0f063ba2013-03-02 00:26:47 +0000460
461 unsigned asm_printer_variant;
462 if (flavor == ~0U)
463 asm_printer_variant = m_asm_info_ap->getAssemblerDialect();
464 else
465 {
466 asm_printer_variant = flavor;
467 }
468
469 m_instr_printer_ap.reset(curr_target->createMCInstPrinter(asm_printer_variant,
470 *m_asm_info_ap.get(),
471 *m_instr_info_ap.get(),
472 *m_reg_info_ap.get(),
473 *m_subtarget_info_ap.get()));
474 if (m_instr_printer_ap.get() == NULL)
475 {
476 m_disasm_ap.reset();
477 m_is_valid = false;
478 }
479 }
480 else
481 m_is_valid = false;
482}
483
Greg Claytone01e07b2013-04-18 18:10:51 +0000484DisassemblerLLVMC::LLVMCDisassembler::~LLVMCDisassembler()
485{
486}
487
Jim Ingham0f063ba2013-03-02 00:26:47 +0000488namespace {
489 // This is the memory object we use in GetInstruction.
490 class LLDBDisasmMemoryObject : public llvm::MemoryObject {
Greg Clayton3faf47c2013-03-28 23:42:53 +0000491 const uint8_t *m_bytes;
Jim Ingham0f063ba2013-03-02 00:26:47 +0000492 uint64_t m_size;
493 uint64_t m_base_PC;
494 public:
Greg Clayton3faf47c2013-03-28 23:42:53 +0000495 LLDBDisasmMemoryObject(const uint8_t *bytes, uint64_t size, uint64_t basePC) :
Jim Ingham0f063ba2013-03-02 00:26:47 +0000496 m_bytes(bytes), m_size(size), m_base_PC(basePC) {}
497
498 uint64_t getBase() const { return m_base_PC; }
499 uint64_t getExtent() const { return m_size; }
500
501 int readByte(uint64_t addr, uint8_t *byte) const {
502 if (addr - m_base_PC >= m_size)
503 return -1;
504 *byte = m_bytes[addr - m_base_PC];
505 return 0;
506 }
507 };
508} // End Anonymous Namespace
509
510uint64_t
Greg Clayton3faf47c2013-03-28 23:42:53 +0000511DisassemblerLLVMC::LLVMCDisassembler::GetMCInst (const uint8_t *opcode_data,
512 size_t opcode_data_len,
513 lldb::addr_t pc,
514 llvm::MCInst &mc_inst)
Jim Ingham0f063ba2013-03-02 00:26:47 +0000515{
516 LLDBDisasmMemoryObject memory_object (opcode_data, opcode_data_len, pc);
Jim Ingham0f063ba2013-03-02 00:26:47 +0000517 llvm::MCDisassembler::DecodeStatus status;
518
519 uint64_t new_inst_size;
520 status = m_disasm_ap->getInstruction(mc_inst,
521 new_inst_size,
522 memory_object,
523 pc,
524 llvm::nulls(),
525 llvm::nulls());
526 if (status == llvm::MCDisassembler::Success)
527 return new_inst_size;
528 else
529 return 0;
530}
531
532uint64_t
Greg Clayton3faf47c2013-03-28 23:42:53 +0000533DisassemblerLLVMC::LLVMCDisassembler::PrintMCInst (llvm::MCInst &mc_inst,
534 char *dst,
535 size_t dst_len)
Jim Ingham0f063ba2013-03-02 00:26:47 +0000536{
537 llvm::StringRef unused_annotations;
538 llvm::SmallString<64> inst_string;
539 llvm::raw_svector_ostream inst_stream(inst_string);
540 m_instr_printer_ap->printInst (&mc_inst, inst_stream, unused_annotations);
541 inst_stream.flush();
Greg Clayton3faf47c2013-03-28 23:42:53 +0000542 const size_t output_size = std::min(dst_len - 1, inst_string.size());
543 std::memcpy(dst, inst_string.data(), output_size);
544 dst[output_size] = '\0';
Jim Ingham0f063ba2013-03-02 00:26:47 +0000545
546 return output_size;
547}
548
549bool
550DisassemblerLLVMC::LLVMCDisassembler::CanBranch (llvm::MCInst &mc_inst)
551{
552 return m_instr_info_ap->get(mc_inst.getOpcode()).mayAffectControlFlow(mc_inst, *m_reg_info_ap.get());
553}
554
555bool
556DisassemblerLLVMC::FlavorValidForArchSpec (const lldb_private::ArchSpec &arch, const char *flavor)
557{
558 llvm::Triple triple = arch.GetTriple();
559 if (flavor == NULL || strcmp (flavor, "default") == 0)
560 return true;
561
562 if (triple.getArch() == llvm::Triple::x86 || triple.getArch() == llvm::Triple::x86_64)
563 {
564 if (strcmp (flavor, "intel") == 0 || strcmp (flavor, "att") == 0)
565 return true;
566 else
567 return false;
568 }
569 else
570 return false;
571}
572
573
Sean Callanan95e5c632012-02-17 00:53:45 +0000574Disassembler *
Jim Ingham0f063ba2013-03-02 00:26:47 +0000575DisassemblerLLVMC::CreateInstance (const ArchSpec &arch, const char *flavor)
Sean Callanan95e5c632012-02-17 00:53:45 +0000576{
Greg Clayton9e6cffc2012-09-19 22:25:17 +0000577 if (arch.GetTriple().getArch() != llvm::Triple::UnknownArch)
578 {
Greg Clayton7b0992d2013-04-18 22:45:39 +0000579 std::unique_ptr<DisassemblerLLVMC> disasm_ap (new DisassemblerLLVMC(arch, flavor));
Sean Callanan95e5c632012-02-17 00:53:45 +0000580
Greg Clayton9e6cffc2012-09-19 22:25:17 +0000581 if (disasm_ap.get() && disasm_ap->IsValid())
582 return disasm_ap.release();
583 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000584 return NULL;
585}
586
Jim Ingham0f063ba2013-03-02 00:26:47 +0000587DisassemblerLLVMC::DisassemblerLLVMC (const ArchSpec &arch, const char *flavor_string) :
588 Disassembler(arch, flavor_string),
Greg Claytonba812f42012-05-10 02:52:23 +0000589 m_exe_ctx (NULL),
Greg Clayton3faf47c2013-03-28 23:42:53 +0000590 m_inst (NULL),
591 m_data_from_file (false)
Sean Callanan95e5c632012-02-17 00:53:45 +0000592{
Jim Ingham0f063ba2013-03-02 00:26:47 +0000593 if (!FlavorValidForArchSpec (arch, m_flavor.c_str()))
594 {
595 m_flavor.assign("default");
596 }
597
598 const char *triple = arch.GetTriple().getTriple().c_str();
599 unsigned flavor = ~0U;
600
601 // So far the only supported flavor is "intel" on x86. The base class will set this
602 // correctly coming in.
603 if (arch.GetTriple().getArch() == llvm::Triple::x86
604 || arch.GetTriple().getArch() == llvm::Triple::x86_64)
605 {
606 if (m_flavor == "intel")
607 {
608 flavor = 1;
609 }
610 else if (m_flavor == "att")
611 {
612 flavor = 0;
613 }
614 }
615
Jason Molenda01aa5342013-07-22 22:11:53 +0000616 ArchSpec thumb_arch(arch);
Sean Callanan95e5c632012-02-17 00:53:45 +0000617 if (arch.GetTriple().getArch() == llvm::Triple::arm)
618 {
Greg Claytona80ea122013-05-03 01:05:04 +0000619 std::string thumb_arch_name (thumb_arch.GetTriple().getArchName().str());
620 // Replace "arm" with "thumb" so we get all thumb variants correct
621 if (thumb_arch_name.size() > 3)
622 {
623 thumb_arch_name.erase(0,3);
624 thumb_arch_name.insert(0, "thumb");
625 }
626 else
627 {
628 thumb_arch_name = "thumbv7";
629 }
630 thumb_arch.GetTriple().setArchName(llvm::StringRef(thumb_arch_name.c_str()));
Jason Molenda01aa5342013-07-22 22:11:53 +0000631 }
632
633 // Cortex-M3 devices (e.g. armv7m) can only execute thumb (T2) instructions,
634 // so hardcode the primary disassembler to thumb mode.
635 if (arch.GetTriple().getArch() == llvm::Triple::arm
636 && (arch.GetCore() == ArchSpec::Core::eCore_arm_armv7m || arch.GetCore() == ArchSpec::Core::eCore_arm_armv7em))
637 {
638 triple = thumb_arch.GetTriple().getTriple().c_str();
639 }
640
641 m_disasm_ap.reset (new LLVMCDisassembler(triple, flavor, *this));
642 if (!m_disasm_ap->IsValid())
643 {
644 // We use m_disasm_ap.get() to tell whether we are valid or not, so if this isn't good for some reason,
645 // we reset it, and then we won't be valid and FindPlugin will fail and we won't get used.
646 m_disasm_ap.reset();
647 }
648
649 // For arm CPUs that can execute arm or thumb instructions, also create a thumb instruction disassembler.
650 if (arch.GetTriple().getArch() == llvm::Triple::arm)
651 {
Greg Claytonba812f42012-05-10 02:52:23 +0000652 std::string thumb_triple(thumb_arch.GetTriple().getTriple());
Jim Ingham0f063ba2013-03-02 00:26:47 +0000653 m_alternate_disasm_ap.reset(new LLVMCDisassembler(thumb_triple.c_str(), flavor, *this));
654 if (!m_alternate_disasm_ap->IsValid())
655 {
656 m_disasm_ap.reset();
657 m_alternate_disasm_ap.reset();
658 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000659 }
660}
661
662DisassemblerLLVMC::~DisassemblerLLVMC()
663{
664}
665
666size_t
667DisassemblerLLVMC::DecodeInstructions (const Address &base_addr,
668 const DataExtractor& data,
Greg Claytonc7bece562013-01-25 18:06:21 +0000669 lldb::offset_t data_offset,
670 size_t num_instructions,
Greg Clayton3faf47c2013-03-28 23:42:53 +0000671 bool append,
672 bool data_from_file)
Sean Callanan95e5c632012-02-17 00:53:45 +0000673{
674 if (!append)
675 m_instruction_list.Clear();
676
677 if (!IsValid())
678 return 0;
679
Greg Clayton3faf47c2013-03-28 23:42:53 +0000680 m_data_from_file = data_from_file;
Sean Callanan95e5c632012-02-17 00:53:45 +0000681 uint32_t data_cursor = data_offset;
Greg Claytonba812f42012-05-10 02:52:23 +0000682 const size_t data_byte_size = data.GetByteSize();
Sean Callanan95e5c632012-02-17 00:53:45 +0000683 uint32_t instructions_parsed = 0;
Greg Claytonba812f42012-05-10 02:52:23 +0000684 Address inst_addr(base_addr);
Sean Callanan95e5c632012-02-17 00:53:45 +0000685
Greg Claytonba812f42012-05-10 02:52:23 +0000686 while (data_cursor < data_byte_size && instructions_parsed < num_instructions)
Sean Callanan95e5c632012-02-17 00:53:45 +0000687 {
Sean Callanan95e5c632012-02-17 00:53:45 +0000688
Greg Claytonba812f42012-05-10 02:52:23 +0000689 AddressClass address_class = eAddressClassCode;
Sean Callanan95e5c632012-02-17 00:53:45 +0000690
Jim Ingham0f063ba2013-03-02 00:26:47 +0000691 if (m_alternate_disasm_ap.get() != NULL)
Greg Claytonba812f42012-05-10 02:52:23 +0000692 address_class = inst_addr.GetAddressClass ();
Sean Callanan95e5c632012-02-17 00:53:45 +0000693
694 InstructionSP inst_sp(new InstructionLLVMC(*this,
Greg Claytonba812f42012-05-10 02:52:23 +0000695 inst_addr,
Sean Callanan95e5c632012-02-17 00:53:45 +0000696 address_class));
697
698 if (!inst_sp)
Greg Claytonba812f42012-05-10 02:52:23 +0000699 break;
700
Sean Callanan95e5c632012-02-17 00:53:45 +0000701 uint32_t inst_size = inst_sp->Decode(*this, data, data_cursor);
702
Greg Claytonba812f42012-05-10 02:52:23 +0000703 if (inst_size == 0)
704 break;
705
Sean Callanan95e5c632012-02-17 00:53:45 +0000706 m_instruction_list.Append(inst_sp);
Sean Callanan95e5c632012-02-17 00:53:45 +0000707 data_cursor += inst_size;
Greg Claytonba812f42012-05-10 02:52:23 +0000708 inst_addr.Slide(inst_size);
Sean Callanan95e5c632012-02-17 00:53:45 +0000709 instructions_parsed++;
710 }
711
712 return data_cursor - data_offset;
713}
714
715void
716DisassemblerLLVMC::Initialize()
717{
718 PluginManager::RegisterPlugin (GetPluginNameStatic(),
Greg Clayton57abc5d2013-05-10 21:47:16 +0000719 "Disassembler that uses LLVM MC to disassemble i386, x86_64 and ARM.",
Sean Callanan95e5c632012-02-17 00:53:45 +0000720 CreateInstance);
721
722 llvm::InitializeAllTargetInfos();
723 llvm::InitializeAllTargetMCs();
724 llvm::InitializeAllAsmParsers();
725 llvm::InitializeAllDisassemblers();
726}
727
728void
729DisassemblerLLVMC::Terminate()
730{
731 PluginManager::UnregisterPlugin (CreateInstance);
732}
733
734
Greg Clayton57abc5d2013-05-10 21:47:16 +0000735ConstString
Sean Callanan95e5c632012-02-17 00:53:45 +0000736DisassemblerLLVMC::GetPluginNameStatic()
737{
Greg Clayton57abc5d2013-05-10 21:47:16 +0000738 static ConstString g_name("llvm-mc");
739 return g_name;
Sean Callanan95e5c632012-02-17 00:53:45 +0000740}
741
Greg Claytonba812f42012-05-10 02:52:23 +0000742int DisassemblerLLVMC::OpInfoCallback (void *disassembler,
743 uint64_t pc,
744 uint64_t offset,
745 uint64_t size,
746 int tag_type,
747 void *tag_bug)
Sean Callanan95e5c632012-02-17 00:53:45 +0000748{
Greg Claytonba812f42012-05-10 02:52:23 +0000749 return static_cast<DisassemblerLLVMC*>(disassembler)->OpInfo (pc,
750 offset,
751 size,
752 tag_type,
753 tag_bug);
Sean Callanan95e5c632012-02-17 00:53:45 +0000754}
755
Greg Claytonba812f42012-05-10 02:52:23 +0000756const char *DisassemblerLLVMC::SymbolLookupCallback (void *disassembler,
757 uint64_t value,
758 uint64_t *type,
759 uint64_t pc,
760 const char **name)
Sean Callanan95e5c632012-02-17 00:53:45 +0000761{
Greg Claytonba812f42012-05-10 02:52:23 +0000762 return static_cast<DisassemblerLLVMC*>(disassembler)->SymbolLookup(value,
763 type,
764 pc,
765 name);
Sean Callanan95e5c632012-02-17 00:53:45 +0000766}
767
768int DisassemblerLLVMC::OpInfo (uint64_t PC,
769 uint64_t Offset,
770 uint64_t Size,
Greg Claytonba812f42012-05-10 02:52:23 +0000771 int tag_type,
772 void *tag_bug)
Sean Callanan95e5c632012-02-17 00:53:45 +0000773{
Greg Claytonba812f42012-05-10 02:52:23 +0000774 switch (tag_type)
Sean Callanan95e5c632012-02-17 00:53:45 +0000775 {
776 default:
777 break;
778 case 1:
Greg Claytonba812f42012-05-10 02:52:23 +0000779 bzero (tag_bug, sizeof(::LLVMOpInfo1));
Sean Callanan95e5c632012-02-17 00:53:45 +0000780 break;
781 }
782 return 0;
783}
784
Greg Claytonba812f42012-05-10 02:52:23 +0000785const char *DisassemblerLLVMC::SymbolLookup (uint64_t value,
786 uint64_t *type_ptr,
787 uint64_t pc,
788 const char **name)
Sean Callanan95e5c632012-02-17 00:53:45 +0000789{
Greg Claytonba812f42012-05-10 02:52:23 +0000790 if (*type_ptr)
791 {
792 if (m_exe_ctx && m_inst)
793 {
794 //std::string remove_this_prior_to_checkin;
Greg Claytonba812f42012-05-10 02:52:23 +0000795 Target *target = m_exe_ctx ? m_exe_ctx->GetTargetPtr() : NULL;
Greg Clayton3faf47c2013-03-28 23:42:53 +0000796 Address value_so_addr;
797 if (m_inst->UsingFileAddress())
Greg Claytonba812f42012-05-10 02:52:23 +0000798 {
799 ModuleSP module_sp(m_inst->GetAddress().GetModule());
800 if (module_sp)
Greg Clayton3faf47c2013-03-28 23:42:53 +0000801 module_sp->ResolveFileAddress(value, value_so_addr);
Greg Claytonba812f42012-05-10 02:52:23 +0000802 }
Greg Clayton3faf47c2013-03-28 23:42:53 +0000803 else if (target && !target->GetSectionLoadList().IsEmpty())
804 {
805 target->GetSectionLoadList().ResolveLoadAddress(value, value_so_addr);
806 }
807
808 if (value_so_addr.IsValid() && value_so_addr.GetSection())
Sean Callanan95e5c632012-02-17 00:53:45 +0000809 {
Sean Callanan95e5c632012-02-17 00:53:45 +0000810 StreamString ss;
811
Greg Clayton3faf47c2013-03-28 23:42:53 +0000812 value_so_addr.Dump (&ss,
813 target,
814 Address::DumpStyleResolvedDescriptionNoModule,
815 Address::DumpStyleSectionNameOffset);
Sean Callanan95e5c632012-02-17 00:53:45 +0000816
Sean Callanan745af462012-03-22 20:04:23 +0000817 if (!ss.GetString().empty())
Greg Claytonba812f42012-05-10 02:52:23 +0000818 {
Greg Claytonba812f42012-05-10 02:52:23 +0000819 m_inst->AppendComment(ss.GetString());
820 }
Sean Callanan95e5c632012-02-17 00:53:45 +0000821 }
822 }
823 }
Greg Claytonba812f42012-05-10 02:52:23 +0000824
825 *type_ptr = LLVMDisassembler_ReferenceType_InOut_None;
826 *name = NULL;
827 return NULL;
Sean Callanan95e5c632012-02-17 00:53:45 +0000828}
829
830//------------------------------------------------------------------
831// PluginInterface protocol
832//------------------------------------------------------------------
Greg Clayton57abc5d2013-05-10 21:47:16 +0000833ConstString
Sean Callanan95e5c632012-02-17 00:53:45 +0000834DisassemblerLLVMC::GetPluginName()
835{
Sean Callanan95e5c632012-02-17 00:53:45 +0000836 return GetPluginNameStatic();
837}
838
839uint32_t
840DisassemblerLLVMC::GetPluginVersion()
841{
842 return 1;
843}
844