blob: 6a3f53820092cb90f7973de3abe423015c3e11ce [file] [log] [blame]
Carl Shapiro12eb78e2011-06-24 14:51:06 -07001// Copyright 2011 Google Inc. All Rights Reserved.
2
Brian Carlstrom578bbdc2011-07-21 14:07:47 -07003#include "dex_instruction.h"
Carl Shapiro12eb78e2011-06-24 14:51:06 -07004
Ian Rogersd81871c2011-10-03 13:57:23 -07005#include "dex_file.h"
6#include <iomanip>
7
Carl Shapiro12eb78e2011-06-24 14:51:06 -07008namespace art {
9
Carl Shapiroe4c1ce42011-07-09 02:31:57 -070010const char* const Instruction::kInstructionNames[] = {
jeffhaoba5ebb92011-08-25 17:24:37 -070011#define INSTRUCTION_NAME(o, c, pname, f, r, i, a, v) pname,
Brian Carlstrom578bbdc2011-07-21 14:07:47 -070012#include "dex_instruction_list.h"
Carl Shapiroe4c1ce42011-07-09 02:31:57 -070013 DEX_INSTRUCTION_LIST(INSTRUCTION_NAME)
14#undef DEX_INSTRUCTION_LIST
15#undef INSTRUCTION_NAME
16};
17
18Instruction::InstructionFormat const Instruction::kInstructionFormats[] = {
jeffhaoba5ebb92011-08-25 17:24:37 -070019#define INSTRUCTION_FORMAT(o, c, p, format, r, i, a, v) format,
Brian Carlstrom578bbdc2011-07-21 14:07:47 -070020#include "dex_instruction_list.h"
Carl Shapiroe4c1ce42011-07-09 02:31:57 -070021 DEX_INSTRUCTION_LIST(INSTRUCTION_FORMAT)
22#undef DEX_INSTRUCTION_LIST
23#undef INSTRUCTION_FORMAT
24};
25
26int const Instruction::kInstructionFlags[] = {
jeffhaoba5ebb92011-08-25 17:24:37 -070027#define INSTRUCTION_FLAGS(o, c, p, f, r, i, flags, v) flags,
Brian Carlstrom578bbdc2011-07-21 14:07:47 -070028#include "dex_instruction_list.h"
Carl Shapiroe4c1ce42011-07-09 02:31:57 -070029 DEX_INSTRUCTION_LIST(INSTRUCTION_FLAGS)
30#undef DEX_INSTRUCTION_LIST
31#undef INSTRUCTION_FLAGS
32};
33
jeffhaoba5ebb92011-08-25 17:24:37 -070034int const Instruction::kInstructionVerifyFlags[] = {
35#define INSTRUCTION_VERIFY_FLAGS(o, c, p, f, r, i, a, vflags) vflags,
36#include "dex_instruction_list.h"
37 DEX_INSTRUCTION_LIST(INSTRUCTION_VERIFY_FLAGS)
38#undef DEX_INSTRUCTION_LIST
39#undef INSTRUCTION_VERIFY_FLAGS
40};
41
42/*
43 * Handy macros for helping decode instructions.
44 */
45#define FETCH(_offset) (insns[(_offset)])
46#define FETCH_u4(_offset) (fetch_u4_impl((_offset), insns))
47#define INST_A(_insn) (((uint16_t)(_insn) >> 8) & 0x0f)
48#define INST_B(_insn) ((uint16_t)(_insn) >> 12)
49#define INST_AA(_insn) ((_insn) >> 8)
50
51/* Helper for FETCH_u4, above. */
52static inline uint32_t fetch_u4_impl(uint32_t offset, const uint16_t* insns) {
53 return insns[offset] | ((uint32_t) insns[offset+1] << 16);
54}
55
56void Instruction::Decode(uint32_t &vA, uint32_t &vB, uint64_t &vB_wide, uint32_t &vC, uint32_t arg[]) const {
57 const uint16_t* insns = reinterpret_cast<const uint16_t*>(this);
58 uint16_t insn = *insns;
59 int opcode = insn & 0xFF;
60
61 switch (Format()) {
62 case k10x: // op
63 /* nothing to do; copy the AA bits out for the verifier */
64 vA = INST_AA(insn);
65 break;
66 case k12x: // op vA, vB
67 vA = INST_A(insn);
68 vB = INST_B(insn);
69 break;
70 case k11n: // op vA, #+B
71 vA = INST_A(insn);
72 vB = (int32_t) (INST_B(insn) << 28) >> 28; // sign extend 4-bit value
73 break;
74 case k11x: // op vAA
75 vA = INST_AA(insn);
76 break;
77 case k10t: // op +AA
78 vA = (int8_t) INST_AA(insn); // sign-extend 8-bit value
79 break;
jeffhaoe0cfb6f2011-09-22 16:42:56 -070080 case k20bc: // op AA, kind@BBBB
81 break;
jeffhaoba5ebb92011-08-25 17:24:37 -070082 case k20t: // op +AAAA
83 vA = (int16_t) FETCH(1); // sign-extend 16-bit value
84 break;
85 case k21c: // op vAA, thing@BBBB
86 case k22x: // op vAA, vBBBB
87 vA = INST_AA(insn);
88 vB = FETCH(1);
89 break;
90 case k21s: // op vAA, #+BBBB
91 case k21t: // op vAA, +BBBB
92 vA = INST_AA(insn);
93 vB = (int16_t) FETCH(1); // sign-extend 16-bit value
94 break;
95 case k21h: // op vAA, #+BBBB0000[00000000]
96 vA = INST_AA(insn);
97 /*
98 * The value should be treated as right-zero-extended, but we don't
99 * actually do that here. Among other things, we don't know if it's
100 * the top bits of a 32- or 64-bit value.
101 */
102 vB = FETCH(1);
103 break;
104 case k23x: // op vAA, vBB, vCC
105 vA = INST_AA(insn);
106 vB = FETCH(1) & 0xff;
107 vC = FETCH(1) >> 8;
108 break;
109 case k22b: // op vAA, vBB, #+CC
110 vA = INST_AA(insn);
111 vB = FETCH(1) & 0xff;
112 vC = (int8_t) (FETCH(1) >> 8); // sign-extend 8-bit value
113 break;
114 case k22s: // op vA, vB, #+CCCC
115 case k22t: // op vA, vB, +CCCC
116 vA = INST_A(insn);
117 vB = INST_B(insn);
118 vC = (int16_t) FETCH(1); // sign-extend 16-bit value
119 break;
120 case k22c: // op vA, vB, thing@CCCC
121 vA = INST_A(insn);
122 vB = INST_B(insn);
123 vC = FETCH(1);
124 break;
125 case k30t: // op +AAAAAAAA
126 vA = FETCH_u4(1); // signed 32-bit value
127 break;
128 case k31t: // op vAA, +BBBBBBBB
129 case k31c: // op vAA, string@BBBBBBBB
130 vA = INST_AA(insn);
131 vB = FETCH_u4(1); // 32-bit value
132 break;
133 case k32x: // op vAAAA, vBBBB
134 vA = FETCH(1);
135 vB = FETCH(2);
136 break;
137 case k31i: // op vAA, #+BBBBBBBB
138 vA = INST_AA(insn);
139 vB = FETCH_u4(1); // signed 32-bit value
140 break;
141 case k35c: // op {vC, vD, vE, vF, vG}, thing@BBBB
142 {
143 /*
144 * Note that the fields mentioned in the spec don't appear in
145 * their "usual" positions here compared to most formats. This
146 * was done so that the field names for the argument count and
147 * reference index match between this format and the corresponding
148 * range formats (3rc and friends).
149 *
150 * Bottom line: The argument count is always in vA, and the
151 * method constant (or equivalent) is always in vB.
152 */
153 uint16_t regList;
154 int count;
155
156 vA = INST_B(insn); // This is labeled A in the spec.
157 vB = FETCH(1);
158 regList = FETCH(2);
159
160 count = vA;
161
162 /*
163 * Copy the argument registers into the arg[] array, and
164 * also copy the first argument (if any) into vC. (The
165 * DecodedInstruction structure doesn't have separate
166 * fields for {vD, vE, vF, vG}, so there's no need to make
167 * copies of those.) Note that cases 5..2 fall through.
168 */
169 switch (count) {
170 case 5: arg[4] = INST_A(insn);
171 case 4: arg[3] = (regList >> 12) & 0x0f;
172 case 3: arg[2] = (regList >> 8) & 0x0f;
173 case 2: arg[1] = (regList >> 4) & 0x0f;
174 case 1: vC = arg[0] = regList & 0x0f; break;
175 case 0: break; // Valid, but no need to do anything.
176 default:
177 LOG(ERROR) << "Invalid arg count in 35c (" << count << ")";
178 return;
179 }
180 }
181 break;
182 case k3rc: // op {vCCCC .. v(CCCC+AA-1)}, meth@BBBB
183 vA = INST_AA(insn);
184 vB = FETCH(1);
185 vC = FETCH(2);
186 break;
187 case k51l: // op vAA, #+BBBBBBBBBBBBBBBB
188 vA = INST_AA(insn);
189 vB_wide = FETCH_u4(1) | ((uint64_t) FETCH_u4(3) << 32);
190 break;
191 default:
192 LOG(ERROR) << "Can't decode unexpected format " << (int) Format() << " (op=" << opcode << ")";
193 return;
194 }
195}
196
Ian Rogersd81871c2011-10-03 13:57:23 -0700197size_t Instruction::SizeInCodeUnits() const {
Carl Shapiro12eb78e2011-06-24 14:51:06 -0700198 const uint16_t* insns = reinterpret_cast<const uint16_t*>(this);
Carl Shapiroe4c1ce42011-07-09 02:31:57 -0700199 if (*insns == kPackedSwitchSignature) {
jeffhaoba5ebb92011-08-25 17:24:37 -0700200 return (4 + insns[1] * 2);
Carl Shapiroe4c1ce42011-07-09 02:31:57 -0700201 } else if (*insns == kSparseSwitchSignature) {
jeffhaoba5ebb92011-08-25 17:24:37 -0700202 return (2 + insns[1] * 4);
Carl Shapiroe4c1ce42011-07-09 02:31:57 -0700203 } else if (*insns == kArrayDataSignature) {
204 uint16_t element_size = insns[1];
205 uint32_t length = insns[2] | (((uint32_t)insns[3]) << 16);
206 // The plus 1 is to round up for odd size and width.
jeffhaoba5ebb92011-08-25 17:24:37 -0700207 return (4 + (element_size * length + 1) / 2);
Carl Shapiroe4c1ce42011-07-09 02:31:57 -0700208 } else {
209 switch (Format()) {
210 case k10x:
211 case k12x:
212 case k11n:
213 case k11x:
214 case k10t:
jeffhaoba5ebb92011-08-25 17:24:37 -0700215 return 1;
Carl Shapiroe4c1ce42011-07-09 02:31:57 -0700216 case k20t:
217 case k22x:
218 case k21t:
219 case k21s:
220 case k21h:
221 case k21c:
222 case k23x:
223 case k22b:
224 case k22t:
225 case k22s:
226 case k22c:
jeffhaoba5ebb92011-08-25 17:24:37 -0700227 return 2;
Carl Shapiroe4c1ce42011-07-09 02:31:57 -0700228 case k32x:
229 case k30t:
230 case k31t:
231 case k31i:
232 case k31c:
233 case k35c:
234 case k3rc:
jeffhaoba5ebb92011-08-25 17:24:37 -0700235 return 3;
Carl Shapiroe4c1ce42011-07-09 02:31:57 -0700236 case k51l:
jeffhaoba5ebb92011-08-25 17:24:37 -0700237 return 5;
Carl Shapiroe4c1ce42011-07-09 02:31:57 -0700238 default:
239 LOG(FATAL) << "Unreachable";
240 }
241 }
jeffhaoba5ebb92011-08-25 17:24:37 -0700242 return 0;
Carl Shapiro12eb78e2011-06-24 14:51:06 -0700243}
244
Carl Shapiroe4c1ce42011-07-09 02:31:57 -0700245Instruction::Code Instruction::Opcode() const {
Carl Shapiro12eb78e2011-06-24 14:51:06 -0700246 const uint16_t* insns = reinterpret_cast<const uint16_t*>(this);
Carl Shapiroe4c1ce42011-07-09 02:31:57 -0700247 int opcode = *insns & 0xFF;
248 return static_cast<Code>(opcode);
Carl Shapiro12eb78e2011-06-24 14:51:06 -0700249}
250
Carl Shapiroe4c1ce42011-07-09 02:31:57 -0700251const Instruction* Instruction::Next() const {
Ian Rogersd81871c2011-10-03 13:57:23 -0700252 size_t current_size_in_bytes = SizeInCodeUnits() * sizeof(uint16_t);
Carl Shapiro12eb78e2011-06-24 14:51:06 -0700253 const uint8_t* ptr = reinterpret_cast<const uint8_t*>(this);
Ian Rogersd81871c2011-10-03 13:57:23 -0700254 return reinterpret_cast<const Instruction*>(ptr + current_size_in_bytes);
255}
256
Ian Rogers2c8a8572011-10-24 17:11:36 -0700257std::string Instruction::DumpHex(size_t code_units) const {
Ian Rogersd81871c2011-10-03 13:57:23 -0700258 size_t inst_length = SizeInCodeUnits();
259 if (inst_length > code_units) {
260 inst_length = code_units;
261 }
Ian Rogers2c8a8572011-10-24 17:11:36 -0700262 std::ostringstream os;
Ian Rogersd81871c2011-10-03 13:57:23 -0700263 const uint16_t* insn = reinterpret_cast<const uint16_t*>(this);
264 for (size_t i = 0; i < inst_length; i++) {
Ian Rogers2c8a8572011-10-24 17:11:36 -0700265 os << StringPrintf("0x%04x", insn[i]) << " ";
Ian Rogersd81871c2011-10-03 13:57:23 -0700266 }
267 for (size_t i = inst_length; i < code_units; i++) {
268 os << " ";
269 }
Ian Rogers2c8a8572011-10-24 17:11:36 -0700270 return os.str();
Ian Rogersd81871c2011-10-03 13:57:23 -0700271}
272
Ian Rogers2c8a8572011-10-24 17:11:36 -0700273std::string Instruction::DumpString(const DexFile* file) const {
Ian Rogersd81871c2011-10-03 13:57:23 -0700274 DecodedInstruction insn(this);
Ian Rogers2c8a8572011-10-24 17:11:36 -0700275 std::ostringstream os;
Ian Rogersd81871c2011-10-03 13:57:23 -0700276 const char* opcode = kInstructionNames[insn.opcode_];
277 switch (Format()) {
278 case k10x: os << opcode; break;
279 case k12x: os << opcode << " v" << insn.vA_ << ", v" << insn.vB_; break;
280 case k11n: os << opcode << " v" << insn.vA_ << ", #+" << insn.vB_; break;
281 case k11x: os << opcode << " v" << insn.vA_; break;
282 case k10t: os << opcode << " +" << (int)insn.vA_; break;
283 case k20bc: os << opcode << " " << insn.vA_ << ", kind@" << insn.vB_; break;
284 case k20t: os << opcode << " +" << (int)insn.vA_; break;
285 case k22x: os << opcode << " v" << insn.vA_ << ", v" << insn.vB_; break;
286 case k21t: os << opcode << " v" << insn.vA_ << ", +" << insn.vB_; break;
287 case k21s: os << opcode << " v" << insn.vA_ << ", #+" << insn.vB_; break;
288 case k21h: os << opcode << " v" << insn.vA_ << ", #+" << insn.vB_ << "00000[00000000]"; break;
289 case k21c: os << opcode << " " << insn.vA_ << ", thing@" << insn.vB_; break;
290 case k23x: os << opcode << " v" << insn.vA_ << ", v" << insn.vB_ << ", v" << insn.vC_; break;
291 case k22b: os << opcode << " v" << insn.vA_ << ", v" << insn.vB_ << ", #+" << insn.vC_; break;
292 case k22t: os << opcode << " v" << insn.vA_ << ", v" << insn.vB_ << ", +" << insn.vC_; break;
293 case k22s: os << opcode << " v" << insn.vA_ << ", v" << insn.vB_ << ", #+" << insn.vC_; break;
294 case k22c: os << opcode << " v" << insn.vA_ << ", v" << insn.vB_ << ", thing@" << insn.vC_; break;
295 case k32x: os << opcode << " v" << insn.vA_ << ", v" << insn.vB_; break;
296 case k30t: os << opcode << " +" << (int)insn.vA_; break;
297 case k31t: os << opcode << " v" << insn.vA_ << ", +" << insn.vB_; break;
298 case k31i: os << opcode << " v" << insn.vA_ << ", #+" << insn.vB_; break;
299 case k31c: os << opcode << " v" << insn.vA_ << ", thing@" << insn.vB_; break;
300 case k35c: {
301 switch (insn.opcode_) {
302 case INVOKE_VIRTUAL:
303 case INVOKE_SUPER:
304 case INVOKE_DIRECT:
305 case INVOKE_STATIC:
306 case INVOKE_INTERFACE:
307 if (file != NULL) {
308 const DexFile::MethodId& meth_id = file->GetMethodId(insn.vB_);
309 os << opcode << " {v" << insn.arg_[0] << ", v" << insn.arg_[1] << ", v" << insn.arg_[2]
310 << ", v" << insn.arg_[3] << ", v" << insn.arg_[4] << "}, "
311 << file->GetMethodName(meth_id) << file->GetMethodSignature(meth_id)
312 << " // method@" << insn.vB_;
313 break;
314 } // else fall-through
315 default:
316 os << opcode << " {v" << insn.arg_[0] << ", v" << insn.arg_[1] << ", v" << insn.arg_[2]
317 << ", v" << insn.arg_[3] << ", v" << insn.arg_[4] << "}, thing@" << insn.vB_;
318 break;
319 }
320 break;
321 }
322 case k3rc: os << opcode << " {v" << insn.vC_ << " .. v" << (insn.vC_+ insn.vA_ - 1) << "}, method@" << insn.vB_; break;
323 case k51l: os << opcode << " v" << insn.vA_ << ", #+" << insn.vB_; break;
Ian Rogers2c8a8572011-10-24 17:11:36 -0700324 default: os << " unknown format (" << DumpHex(5) << ")"; break;
Ian Rogersd81871c2011-10-03 13:57:23 -0700325 }
Ian Rogers2c8a8572011-10-24 17:11:36 -0700326 return os.str();
Carl Shapiro12eb78e2011-06-24 14:51:06 -0700327}
328
329} // namespace art