| OLD | NEW |
| (Empty) |
| 1 /* | |
| 2 * Copyright (c) 2012 The Native Client Authors. All rights reserved. | |
| 3 * Use of this source code is governed by a BSD-style license that can be | |
| 4 * found in the LICENSE file. | |
| 5 */ | |
| 6 | |
| 7 /* | |
| 8 * nacl_tester.c | |
| 9 * Uses the NaCl x86 validator/decoder to implement a NaClEnumeratorDecoder. | |
| 10 */ | |
| 11 #ifndef NACL_TRUSTED_BUT_NOT_TCB | |
| 12 #error("This file is not meant for use in the TCB.") | |
| 13 #endif | |
| 14 | |
| 15 #include "native_client/src/trusted/validator/x86/testing/enuminsts/enuminsts.h" | |
| 16 | |
| 17 #include <ctype.h> | |
| 18 #include <string.h> | |
| 19 #include "native_client/src/trusted/validator_x86/ncenuminsts.h" | |
| 20 #include "native_client/src/trusted/validator/x86/testing/enuminsts/str_utils.h" | |
| 21 | |
| 22 #define kBufferSize 1024 | |
| 23 | |
| 24 /* Defines the virtual table for the nacl decoder. */ | |
| 25 struct { | |
| 26 /* The virtual table that implements this decoder. */ | |
| 27 NaClEnumeratorDecoder _base; | |
| 28 /* Defines the NaCl state to use to parse instructions. */ | |
| 29 NaClInstStruct *_inst; | |
| 30 /* Defines if we are ignoring the instruction (i.e. an instruction | |
| 31 * that isn't xed implemented. | |
| 32 */ | |
| 33 Bool _ignore_instruction; | |
| 34 /* Defines the pc address associated with the instruction. */ | |
| 35 NaClPcAddress _pc_address; | |
| 36 /* If non-empty, the corresponding disassembly. */ | |
| 37 char _disassembly[kBufferSize]; | |
| 38 /* If non-empty, the simplified corresponding assembly, where | |
| 39 * the opcode byte sequence has been removed. | |
| 40 */ | |
| 41 char _simplified_disassembly[kBufferSize]; | |
| 42 /* If non-empty, the corresponding instruction mnemonic. */ | |
| 43 char _mnemonic[kBufferSize]; | |
| 44 /* if non-empty, the lowercase mnemonic. */ | |
| 45 char _mnemonic_lower[kBufferSize]; | |
| 46 /* If non-empty, stores the operands of the corresponding instruction. */ | |
| 47 char _operands[kBufferSize]; | |
| 48 /* True if we should translate special opcodes to matching xed nops. */ | |
| 49 Bool _translate_to_xed_nops; | |
| 50 /* True if we shouldn't accept instructions that aren't also implemented | |
| 51 * in xed. | |
| 52 */ | |
| 53 Bool _ignore_instructions_not_xed_implemented; | |
| 54 } nacl_decoder; | |
| 55 | |
| 56 static Bool IsInstLegal(const NaClEnumerator *enumerator) { | |
| 57 UNREFERENCED_PARAMETER(enumerator); | |
| 58 #if 0 | |
| 59 /* This forces an incorrect legality report for 00 24 c2, useful for */ | |
| 60 /* testing that related problems are being reported. */ | |
| 61 if (enumerator->_itext[0] == 0 && | |
| 62 enumerator->_itext[1] == 0x24 && | |
| 63 enumerator->_itext[2] == 0xc2) return 0; | |
| 64 #endif | |
| 65 return !nacl_decoder._ignore_instruction && | |
| 66 NaClInstDecodesCorrectly(nacl_decoder._inst); | |
| 67 } | |
| 68 | |
| 69 /* Instructions we assume that the NaCl validator accept and are valid, but | |
| 70 * are not legal according to xed. | |
| 71 */ | |
| 72 static Bool NaClIsntXedImplemented(const NaClEnumerator *enumerator) { | |
| 73 static const char* nacl_but_not_xed[] = { | |
| 74 "Pf2iw", | |
| 75 "Pf2id" | |
| 76 }; | |
| 77 const char* name = NaClOpcodeName(nacl_decoder._inst); | |
| 78 size_t i; | |
| 79 UNREFERENCED_PARAMETER(enumerator); | |
| 80 for (i = 0; i < NACL_ARRAY_SIZE(nacl_but_not_xed); ++i) { | |
| 81 if (0 == strcmp(name, nacl_but_not_xed[i])) { | |
| 82 return TRUE; | |
| 83 } | |
| 84 } | |
| 85 return FALSE; | |
| 86 } | |
| 87 | |
| 88 /* Defines the funcdtion to parse the first instruction. */ | |
| 89 static void ParseInst(const NaClEnumerator* enumerator, | |
| 90 const int pc_address) { | |
| 91 nacl_decoder._inst = NULL; | |
| 92 nacl_decoder._ignore_instruction = FALSE; | |
| 93 nacl_decoder._pc_address = pc_address; | |
| 94 nacl_decoder._disassembly[0] = 0; | |
| 95 nacl_decoder._simplified_disassembly[0] = 0; | |
| 96 nacl_decoder._mnemonic[0] = 0; | |
| 97 nacl_decoder._mnemonic_lower[0] = 0; | |
| 98 nacl_decoder._operands[0] = 0; | |
| 99 nacl_decoder._inst = | |
| 100 NaClParseInst((uint8_t *)enumerator->_itext, | |
| 101 enumerator->_num_bytes, pc_address); | |
| 102 if (nacl_decoder._ignore_instructions_not_xed_implemented && | |
| 103 IsInstLegal(enumerator)) { | |
| 104 nacl_decoder._ignore_instruction = NaClIsntXedImplemented(enumerator); | |
| 105 } | |
| 106 } | |
| 107 | |
| 108 | |
| 109 /* Returns the disassembled instruction. */ | |
| 110 static const char* Disassemble(const NaClEnumerator* enumerator) { | |
| 111 char* stmp; | |
| 112 | |
| 113 UNREFERENCED_PARAMETER(enumerator); | |
| 114 /* First see if we have cached it. If so, return it. */ | |
| 115 if (nacl_decoder._disassembly[0] != 0) return nacl_decoder._disassembly; | |
| 116 | |
| 117 stmp = NaClInstToStr(nacl_decoder._inst); | |
| 118 cstrncpy(nacl_decoder._disassembly, stmp, kBufferSize); | |
| 119 free(stmp); | |
| 120 return nacl_decoder._disassembly; | |
| 121 } | |
| 122 | |
| 123 /* Returns the lower case name of the instruction. */ | |
| 124 static const char* GetInstMnemonicLower(const NaClEnumerator* enumerator) { | |
| 125 char mnemonic[kBufferSize]; | |
| 126 size_t i; | |
| 127 | |
| 128 UNREFERENCED_PARAMETER(enumerator); | |
| 129 if (nacl_decoder._mnemonic_lower[0] != 0) | |
| 130 return nacl_decoder._mnemonic_lower; | |
| 131 | |
| 132 cstrncpy(mnemonic, NaClOpcodeName(nacl_decoder._inst), kBufferSize); | |
| 133 for (i = 0; i < kBufferSize; ++i) { | |
| 134 mnemonic[i] = tolower(mnemonic[i]); | |
| 135 if (mnemonic[i] == '\0') break; | |
| 136 } | |
| 137 cstrncpy(nacl_decoder._mnemonic_lower, mnemonic, kBufferSize); | |
| 138 return nacl_decoder._mnemonic_lower; | |
| 139 } | |
| 140 | |
| 141 /* Defines nacl/xed mnemonic renaming pairs. */ | |
| 142 typedef struct { | |
| 143 const char* nacl_name; | |
| 144 const char* xed_name; | |
| 145 } NaClToXedPairs; | |
| 146 | |
| 147 /* Returns the disassembled instruction with the opcode byte sequence | |
| 148 * removed. | |
| 149 */ | |
| 150 static const char* SimplifiedDisassembly(const NaClEnumerator *enumerator) { | |
| 151 const char* disassembly; | |
| 152 const char* mnemonic; | |
| 153 const char* start; | |
| 154 | |
| 155 /* First see if we have cache it. If so, return it. */ | |
| 156 if (nacl_decoder._simplified_disassembly[0] != 0) | |
| 157 return nacl_decoder._simplified_disassembly; | |
| 158 | |
| 159 /* Take first guess of simplified assembly. */ | |
| 160 /* Find start of instruction mnemonic, and define as start of simplified | |
| 161 * disassembly. | |
| 162 */ | |
| 163 disassembly = Disassemble(enumerator); | |
| 164 mnemonic = GetInstMnemonicLower(enumerator); | |
| 165 start = strfind(disassembly, mnemonic); | |
| 166 if (NULL == start) { | |
| 167 /* Don't know how to simplify, give up and just use disassembly. */ | |
| 168 cstrncpy(nacl_decoder._simplified_disassembly, disassembly, kBufferSize); | |
| 169 return nacl_decoder._simplified_disassembly; | |
| 170 } | |
| 171 cstrncpy(nacl_decoder._simplified_disassembly, start, kBufferSize); | |
| 172 rstrip(nacl_decoder._simplified_disassembly); | |
| 173 | |
| 174 /* Now handle special cases where we should treat the nacl instruction as a | |
| 175 * nop, so that they will match xed instructions. | |
| 176 */ | |
| 177 if (nacl_decoder._translate_to_xed_nops) { | |
| 178 static const NaClToXedPairs pairs[] = { | |
| 179 { "xchg %eax, %eax" , "nop" }, | |
| 180 { "xchg %rax, %rax" , "nop" }, | |
| 181 { "xchg %ax, %ax" , "nop" }, | |
| 182 }; | |
| 183 size_t i; | |
| 184 char buf[kBufferSize]; | |
| 185 const char* desc = strfind(nacl_decoder._simplified_disassembly, mnemonic); | |
| 186 if (NULL == desc) return nacl_decoder._simplified_disassembly; | |
| 187 cstrncpy(buf, desc, kBufferSize); | |
| 188 rstrip(buf); | |
| 189 for (i = 0; i < NACL_ARRAY_SIZE(pairs); ++i) { | |
| 190 if (0 == strcmp(nacl_decoder._simplified_disassembly, | |
| 191 pairs[i].nacl_name)) { | |
| 192 cstrncpy(nacl_decoder._simplified_disassembly, | |
| 193 pairs[i].xed_name, kBufferSize); | |
| 194 cstrncpy(nacl_decoder._mnemonic, pairs[i].xed_name, kBufferSize); | |
| 195 } | |
| 196 } | |
| 197 } | |
| 198 return nacl_decoder._simplified_disassembly; | |
| 199 } | |
| 200 | |
| 201 /* Returns the mnemonic name for the disassembled instruction. */ | |
| 202 static const char* GetInstMnemonic(const NaClEnumerator* enumerator) { | |
| 203 char mnemonic[kBufferSize]; | |
| 204 const char* disassembly; | |
| 205 | |
| 206 /* First see if we have cached it. If so, return it. */ | |
| 207 if (nacl_decoder._mnemonic[0] != 0) return nacl_decoder._mnemonic; | |
| 208 | |
| 209 /* Force simplifications if needed. Use mnemonic if defined. */ | |
| 210 disassembly = SimplifiedDisassembly(enumerator); | |
| 211 (void) disassembly; /* TODO(bradchen): use or remove this variable */ | |
| 212 if (nacl_decoder._mnemonic[0] != 0) return nacl_decoder._mnemonic; | |
| 213 | |
| 214 /* If reached, we haven't cached it, so find the name from the | |
| 215 * disassembled instruction and cache it. | |
| 216 */ | |
| 217 cstrncpy(mnemonic, GetInstMnemonicLower(enumerator), kBufferSize); | |
| 218 | |
| 219 /* Now fix mnemonic to corresponding xed name if needed. */ | |
| 220 if ((mnemonic[0] == 'p') && (mnemonic[1] == 'f') && (mnemonic[2] == 'r')) { | |
| 221 static const NaClToXedPairs pairs[] = { | |
| 222 { "pfrsqrt", "pfsqrt" }, | |
| 223 { "pfrcpit1", "pfcpit1" } | |
| 224 }; | |
| 225 size_t i; | |
| 226 for (i = 0; i < NACL_ARRAY_SIZE(pairs); ++i) { | |
| 227 if (0 == strcmp(mnemonic, pairs[i].nacl_name)) { | |
| 228 const char* start = | |
| 229 strfind(nacl_decoder._simplified_disassembly, mnemonic); | |
| 230 if (NULL != start) { | |
| 231 /* replace nacl_name with xed name in simplified disassembly. */ | |
| 232 cstrncpy(mnemonic, pairs[i].xed_name, kBufferSize); | |
| 233 } | |
| 234 } | |
| 235 } | |
| 236 } | |
| 237 | |
| 238 /* Install mnemonic and return. */ | |
| 239 cstrncpy(nacl_decoder._mnemonic, mnemonic, kBufferSize); | |
| 240 return nacl_decoder._mnemonic; | |
| 241 } | |
| 242 | |
| 243 /* Returns the text for the operands. To be used by the driver | |
| 244 * to compare accross decoders. | |
| 245 */ | |
| 246 static const char* GetInstOperandsText(const NaClEnumerator* enumerator) { | |
| 247 char operands[kBufferSize]; | |
| 248 const char* disassembly; | |
| 249 const char* after_mnemonic; | |
| 250 | |
| 251 /* First see if we have cached it. If so, return it. */ | |
| 252 if (nacl_decoder._operands[0] != 0) return nacl_decoder._operands; | |
| 253 | |
| 254 disassembly = SimplifiedDisassembly(enumerator); | |
| 255 after_mnemonic = strskip(disassembly, GetInstMnemonicLower(enumerator)); | |
| 256 if (NULL == after_mnemonic) after_mnemonic = disassembly; | |
| 257 cstrncpy(operands, after_mnemonic, kBufferSize); | |
| 258 strnzapchar(operands, '%'); | |
| 259 strnzapchar(operands, '\n'); | |
| 260 cstrncpy(nacl_decoder._operands, strip(operands), kBufferSize); | |
| 261 return nacl_decoder._operands; | |
| 262 } | |
| 263 | |
| 264 /* Prints out the disassembled instruction. */ | |
| 265 static void PrintInst(const NaClEnumerator* enumerator) { | |
| 266 printf(" NaCl: %s", Disassemble(enumerator)); | |
| 267 } | |
| 268 | |
| 269 /* Returns the number of bytes in the disassembled instruction. */ | |
| 270 static size_t InstLength(const NaClEnumerator* enumerator) { | |
| 271 UNREFERENCED_PARAMETER(enumerator); | |
| 272 #if 0 | |
| 273 /* This forces an incorrect length report for 00 24 c2, useful for */ | |
| 274 /* testing that length problems are being reported. */ | |
| 275 if (enumerator->_itext[0] == 0 && | |
| 276 enumerator->_itext[1] == 0x24 && | |
| 277 enumerator->_itext[2] == 0xc2) return 4; | |
| 278 #endif | |
| 279 return (size_t) NaClInstLength(nacl_decoder._inst); | |
| 280 } | |
| 281 | |
| 282 /* Returns true if the instruction decodes, and static (single instruction) | |
| 283 * validator tests pass. | |
| 284 */ | |
| 285 static Bool MaybeInstValidates(const NaClEnumerator *enumerator) { | |
| 286 return NaClInstValidates((uint8_t*) enumerator->_itext, | |
| 287 NaClInstLength(nacl_decoder._inst), | |
| 288 nacl_decoder._pc_address, | |
| 289 nacl_decoder._inst); | |
| 290 } | |
| 291 | |
| 292 /* Runs the validator on the given code segment. */ | |
| 293 static Bool SegmentValidates(const NaClEnumerator *enumerator, | |
| 294 const uint8_t* segment, | |
| 295 const size_t size, | |
| 296 const int pc_address) { | |
| 297 UNREFERENCED_PARAMETER(enumerator); | |
| 298 return NaClSegmentValidates((uint8_t*) segment, size, pc_address); | |
| 299 } | |
| 300 | |
| 301 /* Installs NaCl-specific flags. */ | |
| 302 static void InstallFlag(const NaClEnumerator* enumerator, | |
| 303 const char* flag_name, | |
| 304 const void* flag_address) { | |
| 305 UNREFERENCED_PARAMETER(enumerator); | |
| 306 if (0 == strcmp(flag_name, "--nops")) { | |
| 307 nacl_decoder._translate_to_xed_nops = *((Bool*) flag_address); | |
| 308 } else if (0 == strcmp(flag_name, "--xedimplemented")) { | |
| 309 nacl_decoder._ignore_instructions_not_xed_implemented = | |
| 310 *((Bool*) flag_address); | |
| 311 } | |
| 312 } | |
| 313 | |
| 314 /* Generates a decoder for the (sel_ldr) nacl validator. */ | |
| 315 NaClEnumeratorDecoder* RegisterNaClDecoder(void) { | |
| 316 nacl_decoder._base._id_name = "nacl"; | |
| 317 nacl_decoder._base._parse_inst_fn = ParseInst; | |
| 318 nacl_decoder._base._inst_length_fn = InstLength; | |
| 319 nacl_decoder._base._print_inst_fn = PrintInst; | |
| 320 nacl_decoder._base._get_inst_mnemonic_fn = GetInstMnemonic; | |
| 321 nacl_decoder._base._get_inst_num_operands_fn = NULL; | |
| 322 nacl_decoder._base._get_inst_operands_text_fn = GetInstOperandsText; | |
| 323 nacl_decoder._base._writes_to_reserved_reg_fn = NULL; | |
| 324 nacl_decoder._base._is_inst_legal_fn = IsInstLegal; | |
| 325 nacl_decoder._base._maybe_inst_validates_fn = MaybeInstValidates; | |
| 326 nacl_decoder._base._segment_validates_fn = SegmentValidates; | |
| 327 nacl_decoder._base._install_flag_fn = InstallFlag; | |
| 328 nacl_decoder._base._usage_message = | |
| 329 "Runs nacl decoder to decode instructions"; | |
| 330 return &nacl_decoder._base; | |
| 331 } | |
| OLD | NEW |