https://github.com/akkartik/mu/blob/master/subx/031check_operands.cc
1
2
3
4 :(scenario check_missing_imm8_operand)
5 % Hide_errors = true;
6 == 0x1
7 cd
8 +error: 'cd' (software interrupt): missing imm8 operand
9
10 :(before "Pack Operands(segment code)")
11 check_operands(code);
12 if (trace_contains_errors()) return;
13
14 :(code)
15 void check_operands(const segment& code) {
16 trace(99, "transform") << "-- check operands" << end();
17 for (int i = 0; i < SIZE(code.lines); ++i) {
18 check_operands(code.lines.at(i));
19 if (trace_contains_errors()) return;
20 }
21 }
22
23 void check_operands(const line& inst) {
24 word op = preprocess_op(inst.words.at(0));
25 if (op.data == "0f") {
26 check_operands_0f(inst);
27 return;
28 }
29 if (op.data == "f3") {
30 check_operands_f3(inst);
31 return;
32 }
33 check_operands(inst, op);
34 }
35
36 word preprocess_op(word op) {
37 op.data = tolower(op.data.c_str());
38
39 if (starts_with(op.data, "0x"))
40 op.data = op.data.substr(2);
41 if (SIZE(op.data) == 1)
42 op.data = string("0")+op.data;
43 return op;
44 }
45
46 void test_preprocess_op() {
47 word w1; w1.data = "0xf";
48 word w2; w2.data = "0f";
49 CHECK_EQ(preprocess_op(w1).data, preprocess_op(w2).data);
50 }
51
52
53
54
55
56
57
58
59
60
61 :(before "End Types")
62 enum expected_operand_type {
63
64 MODRM,
65 SUBOP,
66 DISP8,
67 DISP16,
68 DISP32,
69 IMM8,
70 IMM32,
71 NUM_OPERAND_TYPES
72 };
73 :(before "End Globals")
74 vector<string> Operand_type_name;
75 map<string, expected_operand_type> Operand_type;
76 :(before "End One-time Setup")
77 init_op_types();
78 :(code)
79 void init_op_types() {
80 assert(NUM_OPERAND_TYPES <= 8);
81 Operand_type_name.resize(NUM_OPERAND_TYPES);
82
83 DEF(MODRM);
84 DEF(SUBOP);
85 DEF(DISP8);
86 DEF(DISP16);
87 DEF(DISP32);
88 DEF(IMM8);
89 DEF(IMM32);
90
91 }
92
93 :(before "End Globals")
94 map<string, uint8_t> Permitted_operands;
95 const uint8_t INVALID_OPERANDS = 0xff;
96 :(before "End One-time Setup")
97 init_permitted_operands();
98 :(code)
99 void init_permitted_operands() {
100
101
102 put(Permitted_operands, "f4", 0x00);
103
104 put(Permitted_operands, "40", 0x00);
105 put(Permitted_operands, "41", 0x00);
106 put(Permitted_operands, "42", 0x00);
107 put(Permitted_operands, "43", 0x00);
108 put(Permitted_operands, "44", 0x00);
109 put(Permitted_operands, "45", 0x00);
110 put(Permitted_operands, "46", 0x00);
111 put(Permitted_operands, "47", 0x00);
112
113 put(Permitted_operands, "48", 0x00);
114 put(Permitted_operands, "49", 0x00);
115 put(Permitted_operands, "4a", 0x00);
116 put(Permitted_operands, "4b", 0x00);
117 put(Permitted_operands, "4c", 0x00);
118 put(Permitted_operands, "4d", 0x00);
119 put(Permitted_operands, "4e", 0x00);
120 put(Permitted_operands, "4f", 0x00);
121
122 put(Permitted_operands, "50", 0x00);
123 put(Permitted_operands, "51", 0x00);
124 put(Permitted_operands, "52", 0x00);
125 put(Permitted_operands, "53", 0x00);
126 put(Permitted_operands, "54", 0x00);
127 put(Permitted_operands, "55", 0x00);
128 put(Permitted_operands, "56", 0x00);
129 put(Permitted_operands, "57", 0x00);
130
131 put(Permitted_operands, "58", 0x00);
132 put(Permitted_operands, "59", 0x00);
133 put(Permitted_operands, "5a", 0x00);
134 put(Permitted_operands, "5b", 0x00);
135 put(Permitted_operands, "5c", 0x00);
136 put(Permitted_operands, "5d", 0x00);
137 put(Permitted_operands, "5e", 0x00);
138 put(Permitted_operands, "5f", 0x00);
139
140 put(Permitted_operands, "c3", 0x00);
141
142
143
144
145
146
147 put(Permitted_operands, "eb", 0x04);
148 put(Permitted_operands, "74", 0x04);
149 put(Permitted_operands, "75", 0x04);
150 put(Permitted_operands, "7c", 0x04);
151 put(Permitted_operands, "7d", 0x04);
152 put(Permitted_operands, "7e", 0x04);
153 put(Permitted_operands, "7f", 0x04);
154
155
156
157
158 put(Permitted_operands, "e8", 0x10);
159 put(Permitted_operands, "e9", 0x10);
160
161
162
163
164 put(Permitted_operands, "cd", 0x20);
165
166
167
168
169 put(Permitted_operands, "05", 0x40);
170 put(Permitted_operands, "2d", 0x40);
171 put(Permitted_operands, "25", 0x40);
172 put(Permitted_operands, "0d", 0x40);
173 put(Permitted_operands, "35", 0x40);
174 put(Permitted_operands, "3d", 0x40);
175 put(Permitted_operands, "68", 0x40);
176
177 put(Permitted_operands, "b8", 0x40);
178 put(Permitted_operands, "b9", 0x40);
179 put(Permitted_operands, "ba", 0x40);
180 put(Permitted_operands, "bb", 0x40);
181 put(Permitted_operands, "bc", 0x40);
182 put(Permitted_operands, "bd", 0x40);
183 put(Permitted_operands, "be", 0x40);
184 put(Permitted_operands, "bf", 0x40);
185
186
187
188
189
190
191 put(Permitted_operands, "01", 0x01);
192 put(Permitted_operands, "03", 0x01);
193
194 put(Permitted_operands, "29", 0x01);
195 put(Permitted_operands, "2b", 0x01);
196
197 put(Permitted_operands, "21", 0x01);
198 put(Permitted_operands, "23", 0x01);
199
200 put(Permitted_operands, "09", 0x01);
201 put(Permitted_operands, "0b", 0x01);
202
203 put(Permitted_operands, "31", 0x01);
204 put(Permitted_operands, "33", 0x01);
205
206 put(Permitted_operands, "39", 0x01);
207 put(Permitted_operands, "3b", 0x01);
208
209 put(Permitted_operands, "88", 0x01);
210 put(Permitted_operands, "89", 0x01);
211 put(Permitted_operands, "8a", 0x01);
212 put(Permitted_operands, "8b", 0x01);
213
214 put(Permitted_operands, "87", 0x01);
215
216 put(Permitted_operands, "8d", 0x01);
217
218 put(Permitted_operands, "8f", 0x01);
219
220
221
222
223 put(Permitted_operands, "d3", 0x03);
224 put(Permitted_operands, "f7", 0x03);
225 put(Permitted_operands, "ff", 0x03);
226
227
228
229
230 put(Permitted_operands, "c1", 0x23);
231
232
233
234
235 put(Permitted_operands, "81", 0x43);
236 put(Permitted_operands, "c7", 0x43);
237
238
239 }
240
241 :(code)
242
243
244
245
246 void check_operands(const line& inst, const word& op) {
247 if (!is_hex_byte(op)) return;
248 uint8_t expected_bitvector = get(Permitted_operands, op.data);
249 if (HAS(expected_bitvector, MODRM)) {
250 check_operands_modrm(inst, op);
251 compare_bitvector_modrm(inst, expected_bitvector, op);
252 }
253 else {
254 compare_bitvector(inst, expected_bitvector, op);
255 }
256 }
257
258
259
260 void compare_bitvector(const line& inst, uint8_t expected, const word& op) {
261 if (all_hex_bytes(inst) && has_operands(inst)) return;
262 uint8_t bitvector = computed_expected_operand_bitvector(inst);
263 if (trace_contains_errors()) return;
264 if (bitvector == expected) return;
265 for (int i = 0; i < NUM_OPERAND_TYPES; ++i, bitvector >>= 1, expected >>= 1) {
266
267 if ((bitvector & 0x1) == (expected & 0x1)) continue;
268 const string& optype = Operand_type_name.at(i);
269 if ((bitvector & 0x1) > (expected & 0x1))
270 raise << "'" << to_string(inst) << "'" << maybe_name(op) << ": unexpected " << optype << " operand\n" << end();
271 else
272 raise << "'" << to_string(inst) << "'" << maybe_name(op) << ": missing " << optype << " operand\n" << end();
273
274 }
275
276 }
277
278 string maybe_name(const word& op) {
279 if (!is_hex_byte(op)) return "";
280 if (!contains_key(Name, op.data)) return "";
281
282 const string& s = get(Name, op.data);
283 return " ("+s.substr(0, s.find(" ("))+')';
284 }
285
286 uint32_t computed_expected_operand_bitvector(const line& inst) {
287 uint32_t bitvector = 0;
288 for (int i = 1; i < SIZE(inst.words); ++i) {
289 bitvector = bitvector | expected_bit_for_received_operand(inst.words.at(i));
290 if (trace_contains_errors()) return INVALID_OPERANDS;
291 }
292 return bitvector;
293 }
294
295 bool has_operands(const line& inst) {
296 return SIZE(inst.words) > first_operand(inst);
297 }
298
299 int first_operand(const line& inst) {
300 if (inst.words.at(0).data == "0f") return 2;
301 if (inst.words.at(0).data == "f2" || inst.words.at(0).data == "f3") {
302 if (inst.words.at(1).data == "0f")
303 return 3;
304 else
305 return 2;
306 }
307 return 1;
308 }
309
310
311
312 uint32_t expected_bit_for_received_operand(const word& w) {
313 uint32_t bv = 0;
314 bool found = false;
315 for (int i = 0; i < SIZE(w.metadata); ++i) {
316 string curr = w.metadata.at(i);
317 if (curr == "mod" || curr == "rm32" || curr == "r32" || curr == "scale" || curr == "index" || curr == "base")
318 curr = "modrm";
319 else if (!contains_key(Operand_type, curr)) continue;
320 if (found) {
321 raise << "'" << w.original << "' has conflicting operand types; it should have only one\n" << end();
322 return INVALID_OPERANDS;
323 }
324 bv = (1 << get(Operand_type, curr));
325 found = true;
326 }
327 return bv;
328 }
329
330 :(scenario conflicting_operand_type)
331 % Hide_errors = true;
332 == 0x1
333 cd/software-interrupt 80/imm8/imm32
334 +error: '80/imm8/imm32' has conflicting operand types; it should have only one
335
336
337
338
339 :(scenario check_missing_mod_operand)
340 % Hide_errors = true;
341 == 0x1
342 81 0/add/subop 3/rm32/ebx 1/imm32
343 +error: '81 0/add/subop 3/rm32/ebx 1/imm32' (combine rm32 with imm32 based on subop): missing mod operand
344
345 :(code)
346 void check_operands_modrm(const line& inst, const word& op) {
347 if (all_hex_bytes(inst)) return;
348 check_operand_metadata_present(inst, "mod", op);
349 check_operand_metadata_present(inst, "rm32", op);
350
351 if (op.data == "81" || op.data == "8f" || op.data == "ff") {
352 check_operand_metadata_present(inst, "subop", op);
353 check_operand_metadata_absent(inst, "r32", op, "should be replaced by subop");
354 }
355 if (trace_contains_errors()) return;
356 if (metadata(inst, "rm32").data != "4") return;
357
358 uint8_t mod = hex_byte(metadata(inst, "mod").data);
359 if (mod != 3) {
360 check_operand_metadata_present(inst, "base", op);
361 check_operand_metadata_present(inst, "index", op);
362 }
363 else {
364 check_operand_metadata_absent(inst, "base", op, "direct mode");
365 check_operand_metadata_absent(inst, "index", op, "direct mode");
366 }
367
368 }
369
370
371
372 void compare_bitvector_modrm(const line& inst, uint8_t expected, const word& op) {
373 if (all_hex_bytes(inst) && has_operands(inst)) return;
374 uint8_t bitvector = computed_expected_operand_bitvector(inst);
375 if (trace_contains_errors()) return;
376 if (bitvector == expected) return;
377 for (int i = 0; i < NUM_OPERAND_TYPES; ++i, bitvector >>= 1, expected >>= 1) {
378
379 if ((bitvector & 0x1) == (expected & 0x1)) continue;
380 const string& optype = Operand_type_name.at(i);
381 if (i == DISP8) {
382 int32_t mod = parse_int(metadata(inst, "mod").data);
383 if (mod != 1)
384 raise << "'" << to_string(inst) << "'" << maybe_name(op) << ": unexpected " << optype << " operand\n" << end();
385 continue;
386 }
387 if (i == DISP32) {
388 int32_t mod = parse_int(metadata(inst, "mod").data);
389 int32_t rm32 = parse_int(metadata(inst, "rm32").data);
390 if (mod == 0 && rm32 == 5)
391 ;
392 else if (mod != 2)
393 raise << "'" << to_string(inst) << "'" << maybe_name(op) << ": unexpected " << optype << " operand\n" << end();
394 continue;
395 }
396 if ((bitvector & 0x1) > (expected & 0x1))
397 raise << "'" << to_string(inst) << "'" << maybe_name(op) << ": unexpected " << optype << " operand\n" << end();
398 else
399 raise << "'" << to_string(inst) << "'" << maybe_name(op) << ": missing " << optype << " operand\n" << end();
400
401 }
402
403 }
404
405 void check_operand_metadata_present(const line& inst, const string& type, const word& op) {
406 if (!has_operand_metadata(inst, type))
407 raise << "'" << to_string(inst) << "'" << maybe_name(op) << ": missing " << type << " operand\n" << end();
408 }
409
410 void check_operand_metadata_absent(const line& inst, const string& type, const word& op, const string& msg) {
411 if (has_operand_metadata(inst, type))
412 raise << "'" << to_string(inst) << "'" << maybe_name(op) << ": unexpected " << type << " operand (" << msg << ")\n" << end();
413 }
414
415 :(scenarios transform)
416 :(scenario modrm_with_displacement)
417 % Reg[EAX].u = 0x1;
418 == 0x1
419
420 8b/copy 1/mod/lookup+disp8 0/rm32/EAX 2/r32/EDX 4/disp8
421 $error: 0
422 :(scenarios run)
423
424 :(scenario conflicting_operands_in_modrm_instruction)
425 % Hide_errors = true;
426 == 0x1
427 01/add 0/mod 3/mod
428 +error: '01/add 0/mod 3/mod' has conflicting mod operands
429
430 :(scenario conflicting_operand_type_modrm)
431 % Hide_errors = true;
432 == 0x1
433 01/add 0/mod 3/rm32/r32
434 +error: '3/rm32/r32' has conflicting operand types; it should have only one
435
436 :(scenario check_missing_rm32_operand)
437 % Hide_errors = true;
438 == 0x1
439 81 0/add/subop 0/mod 1/imm32
440 +error: '81 0/add/subop 0/mod 1/imm32' (combine rm32 with imm32 based on subop): missing rm32 operand
441
442 :(scenario check_missing_subop_operand)
443 % Hide_errors = true;
444 == 0x1
445 81 0/mod 3/rm32/ebx 1/imm32
446 +error: '81 0/mod 3/rm32/ebx 1/imm32' (combine rm32 with imm32 based on subop): missing subop operand
447
448 :(scenario check_missing_base_operand)
449 % Hide_errors = true;
450 == 0x1
451 81 0/add/subop 0/mod/indirect 4/rm32/use-sib 1/imm32
452 +error: '81 0/add/subop 0/mod/indirect 4/rm32/use-sib 1/imm32' (combine rm32 with imm32 based on subop): missing base operand
453
454 :(scenario check_missing_index_operand)
455 % Hide_errors = true;
456 == 0x1
457 81 0/add/subop 0/mod/indirect 4/rm32/use-sib 0/base 1/imm32
458 +error: '81 0/add/subop 0/mod/indirect 4/rm32/use-sib 0/base 1/imm32' (combine rm32 with imm32 based on subop): missing index operand
459
460 :(scenario check_missing_base_operand_2)
461 % Hide_errors = true;
462 == 0x1
463 81 0/add/subop 0/mod/indirect 4/rm32/use-sib 2/index 3/scale 1/imm32
464 +error: '81 0/add/subop 0/mod/indirect 4/rm32/use-sib 2/index 3/scale 1/imm32' (combine rm32 with imm32 based on subop): missing base operand
465
466 :(scenario check_extra_displacement)
467 % Hide_errors = true;
468 == 0x1
469 89/copy 0/mod/indirect 0/rm32/EAX 1/r32/ECX 4/disp8
470 +error: '89/copy 0/mod/indirect 0/rm32/EAX 1/r32/ECX 4/disp8' (copy r32 to rm32): unexpected disp8 operand
471
472 :(scenario check_base_operand_not_needed_in_direct_mode)
473 == 0x1
474 81 0/add/subop 3/mod/indirect 4/rm32/use-sib 1/imm32
475 $error: 0
476
477 :(scenario extra_modrm)
478 % Hide_errors = true;
479 == 0x1
480 59/pop-to-ECX 3/mod/direct 1/rm32/ECX 4/r32/ESP
481 +error: '59/pop-to-ECX 3/mod/direct 1/rm32/ECX 4/r32/ESP' (pop top of stack to ECX): unexpected modrm operand
482
483
484
485 :(code)
486 void check_operands_0f(const line& inst) {
487 assert(inst.words.at(0).data == "0f");
488 if (SIZE(inst.words) == 1) {
489 raise << "opcode '0f' requires a second opcode\n" << end();
490 return;
491 }
492 word op = preprocess_op(inst.words.at(1));
493 if (!contains_key(Name_0f, op.data)) {
494 raise << "unknown 2-byte opcode '0f " << op.data << "'\n" << end();
495 return;
496 }
497 check_operands_0f(inst, op);
498 }
499
500 void check_operands_f3(const line& ) {
501 raise << "no supported opcodes starting with f3\n" << end();
502 }
503
504 :(scenario check_missing_disp32_operand)
505 % Hide_errors = true;
506 == 0x1
507
508
509
510 0f 84
511 +error: '0f 84' (jump disp32 bytes away if equal, if ZF is set): missing disp32 operand
512
513 :(before "End Globals")
514 map<string, uint8_t> Permitted_operands_0f;
515 :(before "End Init Permitted Operands")
516
517
518
519 put_new(Permitted_operands_0f, "84", 0x10);
520 put_new(Permitted_operands_0f, "85", 0x10);
521 put_new(Permitted_operands_0f, "8c", 0x10);
522 put_new(Permitted_operands_0f, "8d", 0x10);
523 put_new(Permitted_operands_0f, "8e", 0x10);
524 put_new(Permitted_operands_0f, "8f", 0x10);
525
526
527
528
529 put_new(Permitted_operands_0f, "af", 0x01);
530
531 :(code)
532 void check_operands_0f(const line& inst, const word& op) {
533 uint8_t expected_bitvector = get(Permitted_operands_0f, op.data);
534 if (HAS(expected_bitvector, MODRM))
535 check_operands_modrm(inst, op);
536 compare_bitvector_0f(inst, CLEAR(expected_bitvector, MODRM), op);
537 }
538
539 void compare_bitvector_0f(const line& inst, uint8_t expected, const word& op) {
540 if (all_hex_bytes(inst) && has_operands(inst)) return;
541 uint8_t bitvector = computed_expected_operand_bitvector(inst);
542 if (trace_contains_errors()) return;
543 if (bitvector == expected) return;
544 for (int i = 0; i < NUM_OPERAND_TYPES; ++i, bitvector >>= 1, expected >>= 1) {
545
546 if ((bitvector & 0x1) == (expected & 0x1)) continue;
547 const string& optype = Operand_type_name.at(i);
548 if ((bitvector & 0x1) > (expected & 0x1))
549 raise << "'" << to_string(inst) << "'" << maybe_name_0f(op) << ": unexpected " << optype << " operand\n" << end();
550 else
551 raise << "'" << to_string(inst) << "'" << maybe_name_0f(op) << ": missing " << optype << " operand\n" << end();
552
553 }
554
555 }
556
557 string maybe_name_0f(const word& op) {
558 if (!is_hex_byte(op)) return "";
559 if (!contains_key(Name_0f, op.data)) return "";
560
561 const string& s = get(Name_0f, op.data);
562 return " ("+s.substr(0, s.find(" ("))+')';
563 }
564
565 string tolower(const char* s) {
566 ostringstream out;
567 for (; *s; ++s)
568 out << static_cast<char>(tolower(*s));
569 return out.str();
570 }
571
572
573
574
575
576 :(before "End Includes")
577