?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Clean up linearizer output that got uglified by the show_instruction()
[smatch.git] / linearize.c
blob8824d445e704b81e8fe8457b59a7c7214c2605ac
1 /*
2 * Linearize - walk the statement tree (but _not_ the expressions)
3 * to generate a linear version of it and the basic blocks.
4 *
5 * NOTE! We're not interested in the actual sub-expressions yet,
6 * even though they can generate conditional branches and
7 * subroutine calls. That's all "local" behaviour.
8 *
9 * Copyright (C) 2004 Linus Torvalds
10 * Copyright (C) 2004 Christopher Li
11 */
12
13 #include <string.h>
14 #include <stdarg.h>
15 #include <stdlib.h>
16 #include <stdio.h>
17 #include <assert.h>
18
19 #include "parse.h"
20 #include "expression.h"
21 #include "linearize.h"
22 #include "flow.h"
23 #include "target.h"
24
25 pseudo_t linearize_statement(struct entrypoint *ep, struct statement *stmt);
26 pseudo_t linearize_expression(struct entrypoint *ep, struct expression *expr);
27
28 static pseudo_t add_binary_op(struct entrypoint *ep, struct symbol *ctype, int op, pseudo_t left, pseudo_t right);
29 static pseudo_t add_setval(struct entrypoint *ep, struct symbol *ctype, struct expression *val);
30 static void linearize_one_symbol(struct entrypoint *ep, struct symbol *sym);
31
32 struct access_data;
33 static pseudo_t add_load(struct entrypoint *ep, struct access_data *);
34 pseudo_t linearize_initializer(struct entrypoint *ep, struct expression *initializer, struct access_data *);
35
36 struct pseudo void_pseudo = {};
37
38 static struct instruction *alloc_instruction(int opcode, int size)
39 {
40 struct instruction * insn = __alloc_instruction(0);
41 insn->opcode = opcode;
42 insn->size = size;
43 return insn;
44 }
45
46 static inline int type_size(struct symbol *type)
47 {
48 return type ? type->bit_size > 0 ? type->bit_size : 0 : 0;
49 }
50
51 static struct instruction *alloc_typed_instruction(int opcode, struct symbol *type)
52 {
53 return alloc_instruction(opcode, type_size(type));
54 }
55
56 static struct entrypoint *alloc_entrypoint(void)
57 {
58 return __alloc_entrypoint(0);
59 }
60
61 static struct basic_block *alloc_basic_block(struct entrypoint *ep, struct position pos)
62 {
63 struct basic_block *bb = __alloc_basic_block(0);
64 bb->context = -1;
65 bb->pos = pos;
66 bb->ep = ep;
67 return bb;
68 }
69
70 static struct multijmp* alloc_multijmp(struct basic_block *target, int begin, int end)
71 {
72 struct multijmp *multijmp = __alloc_multijmp(0);
73 multijmp->target = target;
74 multijmp->begin = begin;
75 multijmp->end = end;
76 return multijmp;
77 }
78
79 static inline int regno(pseudo_t n)
80 {
81 int retval = -1;
82 if (n && n->type == PSEUDO_REG)
83 retval = n->nr;
84 return retval;
85 }
86
87 const char *show_pseudo(pseudo_t pseudo)
88 {
89 static int n;
90 static char buffer[4][64];
91 char *buf;
92 int i;
93
94 if (!pseudo)
95 return "no pseudo";
96 if (pseudo == VOID)
97 return "VOID";
98 buf = buffer[3 & ++n];
99 switch(pseudo->type) {
100 case PSEUDO_SYM: {
101 struct symbol *sym = pseudo->sym;
102 struct expression *expr;
103
104 if (sym->bb_target) {
105 snprintf(buf, 64, ".L%p", sym->bb_target);
106 break;
107 }
108 if (sym->ident) {
109 snprintf(buf, 64, "%s", show_ident(sym->ident));
110 break;
111 }
112 expr = sym->initializer;
113 snprintf(buf, 64, "<anon symbol:%p>", sym);
114 switch (expr->type) {
115 case EXPR_VALUE:
116 snprintf(buf, 64, "<symbol value: %lld>", expr->value);
117 break;
118 case EXPR_STRING:
119 return show_string(expr->string);
120 default:
121 break;
122 }
123 break;
124 }
125 case PSEUDO_REG:
126 i = snprintf(buf, 64, "%%r%d", pseudo->nr);
127 if (pseudo->ident)
128 sprintf(buf+i, "(%s)", show_ident(pseudo->ident));
129 break;
130 case PSEUDO_VAL: {
131 long long value = pseudo->value;
132 if (value > 1000 || value < -1000)
133 snprintf(buf, 64, "$%#llx", value);
134 else
135 snprintf(buf, 64, "$%lld", value);
136 break;
137 }
138 case PSEUDO_ARG:
139 snprintf(buf, 64, "%%arg%d", pseudo->nr);
140 break;
141 case PSEUDO_PHI:
142 i = snprintf(buf, 64, "%%phi%d", pseudo->nr);
143 if (pseudo->ident)
144 sprintf(buf+i, "(%s)", show_ident(pseudo->ident));
145 break;
146 default:
147 snprintf(buf, 64, "<bad pseudo type %d>", pseudo->type);
148 }
149 return buf;
150 }
151
152 static const char* opcodes[] = {
153 [OP_BADOP] = "bad_op",
154
155 /* Fn entrypoint */
156 [OP_ENTRY] = "<entry-point>",
157
158 /* Terminator */
159 [OP_RET] = "ret",
160 [OP_BR] = "br",
161 [OP_SWITCH] = "switch",
162 [OP_INVOKE] = "invoke",
163 [OP_COMPUTEDGOTO] = "jmp *",
164 [OP_UNWIND] = "unwind",
165
166 /* Binary */
167 [OP_ADD] = "add",
168 [OP_SUB] = "sub",
169 [OP_MUL] = "mul",
170 [OP_DIV] = "div",
171 [OP_MOD] = "mod",
172 [OP_SHL] = "shl",
173 [OP_SHR] = "shr",
174
175 /* Logical */
176 [OP_AND] = "and",
177 [OP_OR] = "or",
178 [OP_XOR] = "xor",
179 [OP_AND_BOOL] = "and-bool",
180 [OP_OR_BOOL] = "or-bool",
181
182 /* Binary comparison */
183 [OP_SET_EQ] = "seteq",
184 [OP_SET_NE] = "setne",
185 [OP_SET_LE] = "setle",
186 [OP_SET_GE] = "setge",
187 [OP_SET_LT] = "setlt",
188 [OP_SET_GT] = "setgt",
189 [OP_SET_B] = "setb",
190 [OP_SET_A] = "seta",
191 [OP_SET_BE] = "setbe",
192 [OP_SET_AE] = "setae",
193
194 /* Uni */
195 [OP_NOT] = "not",
196 [OP_NEG] = "neg",
197
198 /* Special three-input */
199 [OP_SEL] = "select",
200
201 /* Memory */
202 [OP_MALLOC] = "malloc",
203 [OP_FREE] = "free",
204 [OP_ALLOCA] = "alloca",
205 [OP_LOAD] = "load",
206 [OP_STORE] = "store",
207 [OP_SETVAL] = "set",
208 [OP_GET_ELEMENT_PTR] = "getelem",
209
210 /* Other */
211 [OP_PHI] = "phi",
212 [OP_PHISOURCE] = "phisrc",
213 [OP_CAST] = "cast",
214 [OP_PTRCAST] = "ptrcast",
215 [OP_CALL] = "call",
216 [OP_VANEXT] = "va_next",
217 [OP_VAARG] = "va_arg",
218 [OP_SLICE] = "slice",
219 [OP_SNOP] = "snop",
220 [OP_LNOP] = "lnop",
221 [OP_NOP] = "nop",
222 [OP_DEATHNOTE] = "dead",
223 [OP_ASM] = "asm",
224
225 /* Sparse tagging (line numbers, context, whatever) */
226 [OP_CONTEXT] = "context",
227 };
228
229 const char *show_instruction(struct instruction *insn)
230 {
231 int opcode = insn->opcode;
232 static char buffer[1024];
233 char *buf;
234
235 buf = buffer;
236 if (!insn->bb)
237 buf += sprintf(buf, "# ");
238
239 if (opcode < sizeof(opcodes)/sizeof(char *)) {
240 const char *op = opcodes[opcode];
241 if (!op)
242 buf += sprintf(buf, "opcode:%d", opcode);
243 else
244 buf += sprintf(buf, "%s", op);
245 if (insn->size)
246 buf += sprintf(buf, ".%d", insn->size);
247 memset(buf, ' ', 20);
248 buf++;
249 }
250
251 if (buf < buffer + 12)
252 buf = buffer + 12;
253 switch (opcode) {
254 case OP_RET:
255 if (insn->src && insn->src != VOID)
256 buf += sprintf(buf, "%s", show_pseudo(insn->src));
257 break;
258 case OP_BR:
259 if (insn->bb_true && insn->bb_false) {
260 buf += sprintf(buf, "%s, .L%p, .L%p", show_pseudo(insn->cond), insn->bb_true, insn->bb_false);
261 break;
262 }
263 buf += sprintf(buf, ".L%p", insn->bb_true ? insn->bb_true : insn->bb_false);
264 break;
265
266 case OP_SETVAL: {
267 struct expression *expr = insn->val;
268 pseudo_t pseudo = insn->symbol;
269 buf += sprintf(buf, "%s <- ", show_pseudo(insn->target));
270 if (pseudo) {
271 struct symbol *sym = pseudo->sym;
272 if (!sym) {
273 buf += sprintf(buf, "%s", show_pseudo(pseudo));
274 break;
275 }
276 if (sym->bb_target) {
277 buf += sprintf(buf, ".L%p", sym->bb_target);
278 break;
279 }
280 if (sym->ident) {
281 buf += sprintf(buf, "%s", show_ident(sym->ident));
282 break;
283 }
284 buf += sprintf(buf, "<anon symbol:%p>", sym);
285 break;
286 }
287
288 if (!expr) {
289 buf += sprintf(buf, "%s", "<none>");
290 break;
291 }
292
293 switch (expr->type) {
294 case EXPR_VALUE:
295 buf += sprintf(buf, "%lld", expr->value);
296 break;
297 case EXPR_FVALUE:
298 buf += sprintf(buf, "%Lf", expr->fvalue);
299 break;
300 case EXPR_STRING:
301 buf += sprintf(buf, "%.40s", show_string(expr->string));
302 break;
303 case EXPR_SYMBOL:
304 buf += sprintf(buf, "%s", show_ident(expr->symbol->ident));
305 break;
306 case EXPR_LABEL:
307 buf += sprintf(buf, ".L%p", expr->symbol->bb_target);
308 break;
309 default:
310 buf += sprintf(buf, "SETVAL EXPR TYPE %d", expr->type);
311 }
312 break;
313 }
314 case OP_SWITCH: {
315 struct multijmp *jmp;
316 buf += sprintf(buf, "%s", show_pseudo(insn->target));
317 FOR_EACH_PTR(insn->multijmp_list, jmp) {
318 if (jmp->begin == jmp->end)
319 buf += sprintf(buf, ", %d -> .L%p", jmp->begin, jmp->target);
320 else if (jmp->begin < jmp->end)
321 buf += sprintf(buf, ", %d ... %d -> .L%p", jmp->begin, jmp->end, jmp->target);
322 else
323 buf += sprintf(buf, ", default -> .L%p", jmp->target);
324 } END_FOR_EACH_PTR(jmp);
325 break;
326 }
327 case OP_COMPUTEDGOTO: {
328 struct multijmp *jmp;
329 buf += sprintf(buf, "%s", show_pseudo(insn->target));
330 FOR_EACH_PTR(insn->multijmp_list, jmp) {
331 buf += sprintf(buf, ", .L%p", jmp->target);
332 } END_FOR_EACH_PTR(jmp);
333 break;
334 }
335
336 case OP_PHISOURCE: {
337 struct instruction *phi;
338 buf += sprintf(buf, "%s <- %s ", show_pseudo(insn->target), show_pseudo(insn->phi_src));
339 FOR_EACH_PTR(insn->phi_users, phi) {
340 buf += sprintf(buf, " (%s)", show_pseudo(phi->target));
341 } END_FOR_EACH_PTR(phi);
342 break;
343 }
344
345 case OP_PHI: {
346 pseudo_t phi;
347 const char *s = " <-";
348 buf += sprintf(buf, "%s", show_pseudo(insn->target));
349 FOR_EACH_PTR(insn->phi_list, phi) {
350 buf += sprintf(buf, "%s %s", s, show_pseudo(phi));
351 s = ",";
352 } END_FOR_EACH_PTR(phi);
353 break;
354 }
355 case OP_LOAD: case OP_LNOP:
356 buf += sprintf(buf, "%s <- %d[%s]", show_pseudo(insn->target), insn->offset, show_pseudo(insn->src));
357 break;
358 case OP_STORE: case OP_SNOP:
359 buf += sprintf(buf, "%s -> %d[%s]", show_pseudo(insn->target), insn->offset, show_pseudo(insn->src));
360 break;
361 case OP_CALL: {
362 struct pseudo *arg;
363 if (insn->target && insn->target != VOID)
364 buf += sprintf(buf, "%s <- ", show_pseudo(insn->target));
365 buf += sprintf(buf, "%s", show_pseudo(insn->func));
366 FOR_EACH_PTR(insn->arguments, arg) {
367 buf += sprintf(buf, ", %s", show_pseudo(arg));
368 } END_FOR_EACH_PTR(arg);
369 break;
370 }
371 case OP_CAST:
372 case OP_PTRCAST:
373 buf += sprintf(buf, "%s <- (%d) %s",
374 show_pseudo(insn->target),
375 type_size(insn->orig_type),
376 show_pseudo(insn->src));
377 break;
378 case OP_BINARY ... OP_BINARY_END:
379 case OP_BINCMP ... OP_BINCMP_END:
380 buf += sprintf(buf, "%s <- %s, %s", show_pseudo(insn->target), show_pseudo(insn->src1), show_pseudo(insn->src2));
381 break;
382
383 case OP_SEL:
384 buf += sprintf(buf, "%s <- %s, %s, %s", show_pseudo(insn->target),
385 show_pseudo(insn->src1), show_pseudo(insn->src2), show_pseudo(insn->src3));
386 break;
387
388 case OP_SLICE:
389 buf += sprintf(buf, "%s <- %s, %d, %d", show_pseudo(insn->target), show_pseudo(insn->base), insn->from, insn->len);
390 break;
391
392 case OP_NOT: case OP_NEG:
393 buf += sprintf(buf, "%s <- %s", show_pseudo(insn->target), show_pseudo(insn->src1));
394 break;
395
396 case OP_CONTEXT:
397 buf += sprintf(buf, "%d", insn->increment);
398 break;
399 case OP_NOP:
400 buf += sprintf(buf, "%s <- %s", show_pseudo(insn->target), show_pseudo(insn->src1));
401 break;
402 case OP_DEATHNOTE:
403 buf += sprintf(buf, "%s", show_pseudo(insn->target));
404 break;
405 case OP_ASM:
406 buf += sprintf(buf, "\"%s\"", insn->string);
407 if (insn->outputs) {
408 pseudo_t pseudo;
409 buf += sprintf(buf, " (");
410 FOR_EACH_PTR(insn->outputs, pseudo) {
411 buf += sprintf(buf, " %s", show_pseudo(pseudo));
412 } END_FOR_EACH_PTR(pseudo);
413 buf += sprintf(buf, " ) <-");
414 }
415 if (insn->inputs) {
416 pseudo_t pseudo;
417 buf += sprintf(buf, " (");
418 FOR_EACH_PTR(insn->inputs, pseudo) {
419 buf += sprintf(buf, " %s", show_pseudo(pseudo));
420 } END_FOR_EACH_PTR(pseudo);
421 buf += sprintf(buf, " )");
422 }
423 break;
424 default:
425 break;
426 }
427 do { --buf; } while (*buf == ' ');
428 *++buf = 0;
429 return buffer;
430 }
431
432 void show_bb(struct basic_block *bb)
433 {
434 struct instruction *insn;
435
436 printf(".L%p:\n", bb);
437 if (verbose) {
438 pseudo_t needs, defines;
439 printf("%s:%d\n", stream_name(bb->pos.stream), bb->pos.line);
440
441 FOR_EACH_PTR(bb->needs, needs) {
442 struct instruction *def = needs->def;
443 if (def->opcode != OP_PHI) {
444 printf(" **uses %s (from .L%p)**\n", show_pseudo(needs), def->bb);
445 } else {
446 pseudo_t phi;
447 const char *sep = " ";
448 printf(" **uses %s (from", show_pseudo(needs));
449 FOR_EACH_PTR(def->phi_list, phi) {
450 if (phi == VOID)
451 continue;
452 printf("%s(%s:.L%p)", sep, show_pseudo(phi), phi->def->bb);
453 sep = ", ";
454 } END_FOR_EACH_PTR(phi);
455 printf(")**\n");
456 }
457 } END_FOR_EACH_PTR(needs);
458
459 FOR_EACH_PTR(bb->defines, defines) {
460 printf(" **defines %s **\n", show_pseudo(defines));
461 } END_FOR_EACH_PTR(defines);
462
463 if (bb->parents) {
464 struct basic_block *from;
465 FOR_EACH_PTR(bb->parents, from) {
466 printf(" **from %p (%s:%d:%d)**\n", from,
467 stream_name(from->pos.stream), from->pos.line, from->pos.pos);
468 } END_FOR_EACH_PTR(from);
469 }
470
471 if (bb->children) {
472 struct basic_block *to;
473 FOR_EACH_PTR(bb->children, to) {
474 printf(" **to %p (%s:%d:%d)**\n", to,
475 stream_name(to->pos.stream), to->pos.line, to->pos.pos);
476 } END_FOR_EACH_PTR(to);
477 }
478 }
479
480 FOR_EACH_PTR(bb->insns, insn) {
481 if (!insn->bb && verbose < 2)
482 continue;
483 printf("\t%s\n", show_instruction(insn));
484 } END_FOR_EACH_PTR(insn);
485 if (!bb_terminated(bb))
486 printf("\tEND\n");
487 }
488
489 static void show_symbol_usage(pseudo_t pseudo)
490 {
491 if (pseudo) {
492 pseudo_t *pp;
493 FOR_EACH_PTR(pseudo->users, pp) {
494 struct instruction *insn = container(pp, struct instruction, src);
495 printf("\t%s\n", show_instruction(insn));
496 } END_FOR_EACH_PTR(pp);
497 }
498 }
499
500 void show_entry(struct entrypoint *ep)
501 {
502 struct symbol *sym;
503 struct basic_block *bb;
504
505 printf("%s:\n", show_ident(ep->name->ident));
506
507 if (verbose) {
508 printf("ep %p: %s\n", ep, show_ident(ep->name->ident));
509
510 FOR_EACH_PTR(ep->syms, sym) {
511 if (!sym->pseudo)
512 continue;
513 if (!sym->pseudo->users)
514 continue;
515 printf(" sym: %p %s\n", sym, show_ident(sym->ident));
516 if (sym->ctype.modifiers & (MOD_EXTERN | MOD_STATIC | MOD_ADDRESSABLE))
517 printf("\texternal visibility\n");
518 show_symbol_usage(sym->pseudo);
519 } END_FOR_EACH_PTR(sym);
520
521 printf("\n");
522 }
523
524 FOR_EACH_PTR(ep->bbs, bb) {
525 if (!bb)
526 continue;
527 if (!bb->parents && !bb->children && !bb->insns && verbose < 2)
528 continue;
529 show_bb(bb);
530 printf("\n");
531 } END_FOR_EACH_PTR(bb);
532
533 printf("\n");
534 }
535
536 static void bind_label(struct symbol *label, struct basic_block *bb, struct position pos)
537 {
538 if (label->bb_target)
539 warning(pos, "label '%s' already bound", show_ident(label->ident));
540 label->bb_target = bb;
541 }
542
543 static struct basic_block * get_bound_block(struct entrypoint *ep, struct symbol *label)
544 {
545 struct basic_block *bb = label->bb_target;
546
547 if (!bb) {
548 bb = alloc_basic_block(ep, label->pos);
549 label->bb_target = bb;
550 }
551 return bb;
552 }
553
554 static void finish_block(struct entrypoint *ep)
555 {
556 struct basic_block *src = ep->active;
557 if (bb_reachable(src))
558 ep->active = NULL;
559 }
560
561 static void add_goto(struct entrypoint *ep, struct basic_block *dst)
562 {
563 struct basic_block *src = ep->active;
564 if (bb_reachable(src)) {
565 struct instruction *br = alloc_instruction(OP_BR, 0);
566 br->bb_true = dst;
567 add_bb(&dst->parents, src);
568 add_bb(&src->children, dst);
569 br->bb = src;
570 add_instruction(&src->insns, br);
571 ep->active = NULL;
572 }
573 }
574
575 static void add_one_insn(struct entrypoint *ep, struct instruction *insn)
576 {
577 struct basic_block *bb = ep->active;
578
579 if (bb_reachable(bb)) {
580 insn->bb = bb;
581 add_instruction(&bb->insns, insn);
582 }
583 }
584
585 static void set_activeblock(struct entrypoint *ep, struct basic_block *bb)
586 {
587 if (!bb_terminated(ep->active))
588 add_goto(ep, bb);
589
590 ep->active = bb;
591 if (bb_reachable(bb))
592 add_bb(&ep->bbs, bb);
593 }
594
595 static void remove_parent(struct basic_block *child, struct basic_block *parent)
596 {
597 remove_bb_from_list(&child->parents, parent, 1);
598 if (!child->parents)
599 kill_bb(child);
600 }
601
602 /* Change a "switch" into a branch */
603 void insert_branch(struct basic_block *bb, struct instruction *jmp, struct basic_block *target)
604 {
605 struct instruction *br, *old;
606 struct basic_block *child;
607
608 /* Remove the switch */
609 old = delete_last_instruction(&bb->insns);
610 assert(old == jmp);
611
612 br = alloc_instruction(OP_BR, 0);
613 br->bb = bb;
614 br->bb_true = target;
615 add_instruction(&bb->insns, br);
616
617 FOR_EACH_PTR(bb->children, child) {
618 if (child == target) {
619 target = NULL; /* Trigger just once */
620 continue;
621 }
622 DELETE_CURRENT_PTR(child);
623 remove_parent(child, bb);
624 } END_FOR_EACH_PTR(child);
625 PACK_PTR_LIST(&bb->children);
626 }
627
628
629 void insert_select(struct basic_block *bb, struct instruction *br, struct instruction *phi_node, pseudo_t true, pseudo_t false)
630 {
631 pseudo_t target;
632 struct instruction *select;
633
634 /* Remove the 'br' */
635 delete_last_instruction(&bb->insns);
636
637 select = alloc_instruction(OP_SEL, phi_node->size);
638 select->bb = bb;
639
640 assert(br->cond);
641 use_pseudo(br->cond, &select->src1);
642
643 target = phi_node->target;
644 assert(target->def == phi_node);
645 select->target = target;
646 target->def = select;
647
648 use_pseudo(true, &select->src2);
649 use_pseudo(false, &select->src3);
650
651 add_instruction(&bb->insns, select);
652 add_instruction(&bb->insns, br);
653 }
654
655 static inline int bb_empty(struct basic_block *bb)
656 {
657 return !bb->insns;
658 }
659
660 /* Add a label to the currently active block, return new active block */
661 static struct basic_block * add_label(struct entrypoint *ep, struct symbol *label)
662 {
663 struct basic_block *bb = label->bb_target;
664
665 if (bb) {
666 set_activeblock(ep, bb);
667 return bb;
668 }
669 bb = ep->active;
670 if (!bb_reachable(bb) || !bb_empty(bb)) {
671 bb = alloc_basic_block(ep, label->pos);
672 set_activeblock(ep, bb);
673 }
674 label->bb_target = bb;
675 return bb;
676 }
677
678 static void add_branch(struct entrypoint *ep, struct expression *expr, pseudo_t cond, struct basic_block *bb_true, struct basic_block *bb_false)
679 {
680 struct basic_block *bb = ep->active;
681 struct instruction *br;
682
683 if (bb_reachable(bb)) {
684 br = alloc_instruction(OP_BR, 0);
685 use_pseudo(cond, &br->cond);
686 br->bb_true = bb_true;
687 br->bb_false = bb_false;
688 add_bb(&bb_true->parents, bb);
689 add_bb(&bb_false->parents, bb);
690 add_bb(&bb->children, bb_true);
691 add_bb(&bb->children, bb_false);
692 add_one_insn(ep, br);
693 }
694 }
695
696 /* Dummy pseudo allocator */
697 pseudo_t alloc_pseudo(struct instruction *def)
698 {
699 static int nr = 0;
700 struct pseudo * pseudo = __alloc_pseudo(0);
701 pseudo->type = PSEUDO_REG;
702 pseudo->nr = ++nr;
703 pseudo->def = def;
704 return pseudo;
705 }
706
707 static void clear_symbol_pseudos(struct entrypoint *ep)
708 {
709 struct symbol *sym;
710
711 FOR_EACH_PTR(ep->accesses, sym) {
712 sym->pseudo = NULL;
713 } END_FOR_EACH_PTR(sym);
714 }
715
716 static pseudo_t symbol_pseudo(struct entrypoint *ep, struct symbol *sym)
717 {
718 pseudo_t pseudo;
719
720 if (!sym)
721 return VOID;
722
723 pseudo = sym->pseudo;
724 if (!pseudo) {
725 pseudo = __alloc_pseudo(0);
726 pseudo->type = PSEUDO_SYM;
727 pseudo->sym = sym;
728 pseudo->ident = sym->ident;
729 sym->pseudo = pseudo;
730 add_symbol(&ep->accesses, sym);
731 }
732 /* Symbol pseudos have neither nr, usage nor def */
733 return pseudo;
734 }
735
736 pseudo_t value_pseudo(long long val)
737 {
738 #define MAX_VAL_HASH 64
739 static struct pseudo_list *prev[MAX_VAL_HASH];
740 int hash = val & (MAX_VAL_HASH-1);
741 struct pseudo_list **list = prev + hash;
742 pseudo_t pseudo;
743
744 FOR_EACH_PTR(*list, pseudo) {
745 if (pseudo->value == val)
746 return pseudo;
747 } END_FOR_EACH_PTR(pseudo);
748
749 pseudo = __alloc_pseudo(0);
750 pseudo->type = PSEUDO_VAL;
751 pseudo->value = val;
752 add_pseudo(list, pseudo);
753
754 /* Value pseudos have neither nr, usage nor def */
755 return pseudo;
756 }
757
758 static pseudo_t argument_pseudo(struct entrypoint *ep, int nr)
759 {
760 pseudo_t pseudo = __alloc_pseudo(0);
761 pseudo->type = PSEUDO_ARG;
762 pseudo->nr = nr;
763 pseudo->def = ep->entry;
764 /* Argument pseudos have neither usage nor def */
765 return pseudo;
766 }
767
768 pseudo_t alloc_phi(struct basic_block *source, pseudo_t pseudo, int size)
769 {
770 struct instruction *insn = alloc_instruction(OP_PHISOURCE, size);
771 pseudo_t phi = __alloc_pseudo(0);
772 static int nr = 0;
773
774 phi->type = PSEUDO_PHI;
775 phi->nr = ++nr;
776 phi->def = insn;
777
778 use_pseudo(pseudo, &insn->phi_src);
779 insn->bb = source;
780 insn->target = phi;
781 add_instruction(&source->insns, insn);
782 return phi;
783 }
784
785 /*
786 * We carry the "access_data" structure around for any accesses,
787 * which simplifies things a lot. It contains all the access
788 * information in one place.
789 */
790 struct access_data {
791 struct symbol *result_type; // result ctype
792 struct symbol *source_type; // source ctype
793 pseudo_t address; // pseudo containing address ..
794 pseudo_t origval; // pseudo for original value ..
795 unsigned int offset, alignment; // byte offset
796 unsigned int bit_size, bit_offset; // which bits
797 struct position pos;
798 };
799
800 static void finish_address_gen(struct entrypoint *ep, struct access_data *ad)
801 {
802 }
803
804 static int linearize_simple_address(struct entrypoint *ep,
805 struct expression *addr,
806 struct access_data *ad)
807 {
808 if (addr->type == EXPR_SYMBOL) {
809 linearize_one_symbol(ep, addr->symbol);
810 ad->address = symbol_pseudo(ep, addr->symbol);
811 return 1;
812 }
813 if (addr->type == EXPR_BINOP) {
814 if (addr->right->type == EXPR_VALUE) {
815 if (addr->op == '+') {
816 ad->offset += get_expression_value(addr->right);
817 return linearize_simple_address(ep, addr->left, ad);
818 }
819 }
820 }
821 ad->address = linearize_expression(ep, addr);
822 return 1;
823 }
824
825 static struct symbol *base_type(struct symbol *sym)
826 {
827 struct symbol *base = sym;
828
829 if (sym) {
830 if (sym->type == SYM_NODE)
831 base = base->ctype.base_type;
832 if (base->type == SYM_BITFIELD)
833 return base->ctype.base_type;
834 }
835 return sym;
836 }
837
838 static int linearize_address_gen(struct entrypoint *ep,
839 struct expression *expr,
840 struct access_data *ad)
841 {
842 struct symbol *ctype = expr->ctype;
843
844 if (!ctype)
845 return 0;
846 ad->pos = expr->pos;
847 ad->result_type = ctype;
848 ad->source_type = base_type(ctype);
849 ad->bit_size = ctype->bit_size;
850 ad->alignment = ctype->ctype.alignment;
851 ad->bit_offset = ctype->bit_offset;
852 if (expr->type == EXPR_PREOP && expr->op == '*')
853 return linearize_simple_address(ep, expr->unop, ad);
854
855 warning(expr->pos, "generating address of non-lvalue (%d)", expr->type);
856 return 0;
857 }
858
859 static pseudo_t add_load(struct entrypoint *ep, struct access_data *ad)
860 {
861 struct instruction *insn;
862 pseudo_t new;
863
864 new = ad->origval;
865 if (0 && new)
866 return new;
867
868 insn = alloc_typed_instruction(OP_LOAD, ad->source_type);
869 new = alloc_pseudo(insn);
870 ad->origval = new;
871
872 insn->target = new;
873 insn->offset = ad->offset;
874 use_pseudo(ad->address, &insn->src);
875 add_one_insn(ep, insn);
876 return new;
877 }
878
879 static void add_store(struct entrypoint *ep, struct access_data *ad, pseudo_t value)
880 {
881 struct basic_block *bb = ep->active;
882
883 if (bb_reachable(bb)) {
884 struct instruction *store = alloc_typed_instruction(OP_STORE, ad->source_type);
885 store->offset = ad->offset;
886 use_pseudo(value, &store->target);
887 use_pseudo(ad->address, &store->src);
888 add_one_insn(ep, store);
889 }
890 }
891
892 static pseudo_t linearize_store_gen(struct entrypoint *ep,
893 pseudo_t value,
894 struct access_data *ad)
895 {
896 pseudo_t store = value;
897
898 if (type_size(ad->source_type) != type_size(ad->result_type)) {
899 pseudo_t orig = add_load(ep, ad);
900 int shift = ad->bit_offset;
901 unsigned long long mask = (1ULL << ad->bit_size)-1;
902
903 if (shift) {
904 store = add_binary_op(ep, ad->source_type, OP_SHL, value, value_pseudo(shift));
905 mask <<= shift;
906 }
907 orig = add_binary_op(ep, ad->source_type, OP_AND, orig, value_pseudo(~mask));
908 store = add_binary_op(ep, ad->source_type, OP_OR, orig, store);
909 }
910 add_store(ep, ad, store);
911 return value;
912 }
913
914 static pseudo_t add_binary_op(struct entrypoint *ep, struct symbol *ctype, int op, pseudo_t left, pseudo_t right)
915 {
916 struct instruction *insn = alloc_typed_instruction(op, ctype);
917 pseudo_t target = alloc_pseudo(insn);
918 insn->target = target;
919 use_pseudo(left, &insn->src1);
920 use_pseudo(right, &insn->src2);
921 add_one_insn(ep, insn);
922 return target;
923 }
924
925 static pseudo_t add_setval(struct entrypoint *ep, struct symbol *ctype, struct expression *val)
926 {
927 struct instruction *insn = alloc_typed_instruction(OP_SETVAL, ctype);
928 pseudo_t target = alloc_pseudo(insn);
929 insn->target = target;
930 insn->val = val;
931 if (!val) {
932 pseudo_t addr = symbol_pseudo(ep, ctype);
933 use_pseudo(addr, &insn->symbol);
934 insn->size = bits_in_pointer;
935 }
936 add_one_insn(ep, insn);
937 return target;
938 }
939
940 static pseudo_t linearize_load_gen(struct entrypoint *ep, struct access_data *ad)
941 {
942 pseudo_t new = add_load(ep, ad);
943
944 if (ad->bit_offset) {
945 pseudo_t shift = value_pseudo(ad->bit_offset);
946 pseudo_t newval = add_binary_op(ep, ad->source_type, OP_SHR, new, shift);
947 new = newval;
948 }
949
950 return new;
951 }
952
953 static pseudo_t linearize_access(struct entrypoint *ep, struct expression *expr)
954 {
955 struct access_data ad = { NULL, };
956 pseudo_t value;
957
958 if (!linearize_address_gen(ep, expr, &ad))
959 return VOID;
960 value = linearize_load_gen(ep, &ad);
961 finish_address_gen(ep, &ad);
962 return value;
963 }
964
965 /* FIXME: FP */
966 static pseudo_t linearize_inc_dec(struct entrypoint *ep, struct expression *expr, int postop)
967 {
968 struct access_data ad = { NULL, };
969 pseudo_t old, new, one;
970 int op = expr->op == SPECIAL_INCREMENT ? OP_ADD : OP_SUB;
971
972 if (!linearize_address_gen(ep, expr->unop, &ad))
973 return VOID;
974
975 old = linearize_load_gen(ep, &ad);
976 one = value_pseudo(expr->op_value);
977 new = add_binary_op(ep, expr->ctype, op, old, one);
978 linearize_store_gen(ep, new, &ad);
979 finish_address_gen(ep, &ad);
980 return postop ? old : new;
981 }
982
983 static pseudo_t add_uniop(struct entrypoint *ep, struct expression *expr, int op, pseudo_t src)
984 {
985 struct instruction *insn = alloc_typed_instruction(op, expr->ctype);
986 pseudo_t new = alloc_pseudo(insn);
987
988 insn->target = new;
989 use_pseudo(src, &insn->src1);
990 add_one_insn(ep, insn);
991 return new;
992 }
993
994 static pseudo_t linearize_slice(struct entrypoint *ep, struct expression *expr)
995 {
996 pseudo_t pre = linearize_expression(ep, expr->base);
997 struct instruction *insn = alloc_typed_instruction(OP_SLICE, expr->ctype);
998 pseudo_t new = alloc_pseudo(insn);
999
1000 insn->target = new;
1001 insn->from = expr->r_bitpos;
1002 insn->len = expr->r_nrbits;
1003 use_pseudo(pre, &insn->base);
1004 add_one_insn(ep, insn);
1005 return new;
1006 }
1007
1008 static pseudo_t linearize_regular_preop(struct entrypoint *ep, struct expression *expr)
1009 {
1010 pseudo_t pre = linearize_expression(ep, expr->unop);
1011 switch (expr->op) {
1012 case '+':
1013 return pre;
1014 case '!': {
1015 pseudo_t zero = value_pseudo(0);
1016 return add_binary_op(ep, expr->ctype, OP_SET_EQ, pre, zero);
1017 }
1018 case '~':
1019 return add_uniop(ep, expr, OP_NOT, pre);
1020 case '-':
1021 return add_uniop(ep, expr, OP_NEG, pre);
1022 }
1023 return VOID;
1024 }
1025
1026 static pseudo_t linearize_preop(struct entrypoint *ep, struct expression *expr)
1027 {
1028 /*
1029 * '*' is an lvalue access, and is fundamentally different
1030 * from an arithmetic operation. Maybe it should have an
1031 * expression type of its own..
1032 */
1033 if (expr->op == '*')
1034 return linearize_access(ep, expr);
1035 if (expr->op == SPECIAL_INCREMENT || expr->op == SPECIAL_DECREMENT)
1036 return linearize_inc_dec(ep, expr, 0);
1037 return linearize_regular_preop(ep, expr);
1038 }
1039
1040 static pseudo_t linearize_postop(struct entrypoint *ep, struct expression *expr)
1041 {
1042 return linearize_inc_dec(ep, expr, 1);
1043 }
1044
1045 static pseudo_t linearize_assignment(struct entrypoint *ep, struct expression *expr)
1046 {
1047 struct access_data ad = { NULL, };
1048 struct expression *target = expr->left;
1049 pseudo_t value;
1050
1051 value = linearize_expression(ep, expr->right);
1052 if (!linearize_address_gen(ep, target, &ad))
1053 return VOID;
1054 if (expr->op != '=') {
1055 pseudo_t oldvalue = linearize_load_gen(ep, &ad);
1056 pseudo_t dst;
1057 static const int op_trans[] = {
1058 [SPECIAL_ADD_ASSIGN - SPECIAL_BASE] = OP_ADD,
1059 [SPECIAL_SUB_ASSIGN - SPECIAL_BASE] = OP_SUB,
1060 [SPECIAL_MUL_ASSIGN - SPECIAL_BASE] = OP_MUL,
1061 [SPECIAL_DIV_ASSIGN - SPECIAL_BASE] = OP_DIV,
1062 [SPECIAL_MOD_ASSIGN - SPECIAL_BASE] = OP_MOD,
1063 [SPECIAL_SHL_ASSIGN - SPECIAL_BASE] = OP_SHL,
1064 [SPECIAL_SHR_ASSIGN - SPECIAL_BASE] = OP_SHR,
1065 [SPECIAL_AND_ASSIGN - SPECIAL_BASE] = OP_AND,
1066 [SPECIAL_OR_ASSIGN - SPECIAL_BASE] = OP_OR,
1067 [SPECIAL_XOR_ASSIGN - SPECIAL_BASE] = OP_XOR
1068 };
1069 dst = add_binary_op(ep, expr->ctype, op_trans[expr->op - SPECIAL_BASE], oldvalue, value);
1070 value = dst;
1071 }
1072 value = linearize_store_gen(ep, value, &ad);
1073 finish_address_gen(ep, &ad);
1074 return value;
1075 }
1076
1077 static pseudo_t linearize_call_expression(struct entrypoint *ep, struct expression *expr)
1078 {
1079 struct expression *arg, *fn;
1080 struct instruction *insn = alloc_typed_instruction(OP_CALL, expr->ctype);
1081 pseudo_t retval, call;
1082 int context_diff;
1083
1084 if (!expr->ctype) {
1085 warning(expr->pos, "call with no type!");
1086 return VOID;
1087 }
1088
1089 FOR_EACH_PTR(expr->args, arg) {
1090 pseudo_t new = linearize_expression(ep, arg);
1091 use_pseudo(new, add_pseudo(&insn->arguments, new));
1092 } END_FOR_EACH_PTR(arg);
1093
1094 fn = expr->fn;
1095
1096 context_diff = 0;
1097 if (fn->ctype) {
1098 int in = fn->ctype->ctype.in_context;
1099 int out = fn->ctype->ctype.out_context;
1100 if (in < 0 || out < 0)
1101 in = out = 0;
1102 context_diff = out - in;
1103 }
1104
1105 if (fn->type == EXPR_PREOP) {
1106 if (fn->unop->type == EXPR_SYMBOL) {
1107 struct symbol *sym = fn->unop->symbol;
1108 if (sym->ctype.base_type->type == SYM_FN)
1109 fn = fn->unop;
1110 }
1111 }
1112 if (fn->type == EXPR_SYMBOL) {
1113 call = symbol_pseudo(ep, fn->symbol);
1114 } else {
1115 call = linearize_expression(ep, fn);
1116 }
1117 use_pseudo(call, &insn->func);
1118 retval = VOID;
1119 if (expr->ctype != &void_ctype)
1120 retval = alloc_pseudo(insn);
1121 insn->target = retval;
1122 add_one_insn(ep, insn);
1123
1124 if (context_diff) {
1125 insn = alloc_instruction(OP_CONTEXT, 0);
1126 insn->increment = context_diff;
1127 add_one_insn(ep, insn);
1128 }
1129
1130 return retval;
1131 }
1132
1133 static pseudo_t linearize_binop(struct entrypoint *ep, struct expression *expr)
1134 {
1135 pseudo_t src1, src2, dst;
1136 static const int opcode[] = {
1137 ['+'] = OP_ADD, ['-'] = OP_SUB,
1138 ['*'] = OP_MUL, ['/'] = OP_DIV,
1139 ['%'] = OP_MOD, ['&'] = OP_AND,
1140 ['|'] = OP_OR, ['^'] = OP_XOR,
1141 [SPECIAL_LEFTSHIFT] = OP_SHL,
1142 [SPECIAL_RIGHTSHIFT] = OP_SHR,
1143 [SPECIAL_LOGICAL_AND] = OP_AND_BOOL,
1144 [SPECIAL_LOGICAL_OR] = OP_OR_BOOL,
1145 };
1146
1147 src1 = linearize_expression(ep, expr->left);
1148 src2 = linearize_expression(ep, expr->right);
1149 dst = add_binary_op(ep, expr->ctype, opcode[expr->op], src1, src2);
1150 return dst;
1151 }
1152
1153 static pseudo_t linearize_logical_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false);
1154
1155 pseudo_t linearize_cond_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false);
1156
1157 static pseudo_t linearize_select(struct entrypoint *ep, struct expression *expr)
1158 {
1159 pseudo_t cond, true, false, res;
1160 struct instruction *insn;
1161
1162 true = linearize_expression(ep, expr->cond_true);
1163 false = linearize_expression(ep, expr->cond_false);
1164 cond = linearize_expression(ep, expr->conditional);
1165
1166 insn = alloc_typed_instruction(OP_SEL, expr->ctype);
1167 if (!expr->cond_true)
1168 true = cond;
1169 use_pseudo(cond, &insn->src1);
1170 use_pseudo(true, &insn->src2);
1171 use_pseudo(false, &insn->src3);
1172
1173 res = alloc_pseudo(insn);
1174 insn->target = res;
1175 add_one_insn(ep, insn);
1176 return res;
1177 }
1178
1179 static pseudo_t add_join_conditional(struct entrypoint *ep, struct expression *expr,
1180 pseudo_t phi1, pseudo_t phi2)
1181 {
1182 pseudo_t target;
1183 struct instruction *phi_node;
1184
1185 if (phi1 == VOID)
1186 return phi2;
1187 if (phi2 == VOID)
1188 return phi1;
1189
1190 phi_node = alloc_typed_instruction(OP_PHI, expr->ctype);
1191 use_pseudo(phi1, add_pseudo(&phi_node->phi_list, phi1));
1192 use_pseudo(phi2, add_pseudo(&phi_node->phi_list, phi2));
1193 phi_node->target = target = alloc_pseudo(phi_node);
1194 add_one_insn(ep, phi_node);
1195 return target;
1196 }
1197
1198 static pseudo_t linearize_short_conditional(struct entrypoint *ep, struct expression *expr,
1199 struct expression *cond,
1200 struct expression *expr_false)
1201 {
1202 pseudo_t src1, src2;
1203 struct basic_block *bb_false = alloc_basic_block(ep, expr_false->pos);
1204 struct basic_block *merge = alloc_basic_block(ep, expr->pos);
1205 pseudo_t phi1, phi2;
1206 int size = type_size(expr->ctype);
1207
1208 src1 = linearize_expression(ep, cond);
1209 phi1 = alloc_phi(ep->active, src1, size);
1210 add_branch(ep, expr, src1, merge, bb_false);
1211
1212 set_activeblock(ep, bb_false);
1213 src2 = linearize_expression(ep, expr_false);
1214 phi2 = alloc_phi(ep->active, src2, size);
1215 set_activeblock(ep, merge);
1216
1217 return add_join_conditional(ep, expr, phi1, phi2);
1218 }
1219
1220 static pseudo_t linearize_conditional(struct entrypoint *ep, struct expression *expr,
1221 struct expression *cond,
1222 struct expression *expr_true,
1223 struct expression *expr_false)
1224 {
1225 pseudo_t src1, src2;
1226 pseudo_t phi1, phi2;
1227 struct basic_block *bb_true = alloc_basic_block(ep, expr_true->pos);
1228 struct basic_block *bb_false = alloc_basic_block(ep, expr_false->pos);
1229 struct basic_block *merge = alloc_basic_block(ep, expr->pos);
1230 int size = type_size(expr->ctype);
1231
1232 linearize_cond_branch(ep, cond, bb_true, bb_false);
1233
1234 set_activeblock(ep, bb_true);
1235 src1 = linearize_expression(ep, expr_true);
1236 phi1 = alloc_phi(ep->active, src1, size);
1237 add_goto(ep, merge);
1238
1239 set_activeblock(ep, bb_false);
1240 src2 = linearize_expression(ep, expr_false);
1241 phi2 = alloc_phi(ep->active, src2, size);
1242 set_activeblock(ep, merge);
1243
1244 return add_join_conditional(ep, expr, phi1, phi2);
1245 }
1246
1247 static pseudo_t linearize_logical(struct entrypoint *ep, struct expression *expr)
1248 {
1249 struct expression *shortcut;
1250
1251 shortcut = alloc_const_expression(expr->pos, expr->op == SPECIAL_LOGICAL_OR);
1252 shortcut->ctype = expr->ctype;
1253 return linearize_conditional(ep, expr, expr->left, shortcut, expr->right);
1254 }
1255
1256 static pseudo_t linearize_compare(struct entrypoint *ep, struct expression *expr)
1257 {
1258 static const int cmpop[] = {
1259 ['>'] = OP_SET_GT, ['<'] = OP_SET_LT,
1260 [SPECIAL_EQUAL] = OP_SET_EQ,
1261 [SPECIAL_NOTEQUAL] = OP_SET_NE,
1262 [SPECIAL_GTE] = OP_SET_GE,
1263 [SPECIAL_LTE] = OP_SET_LE,
1264 [SPECIAL_UNSIGNED_LT] = OP_SET_B,
1265 [SPECIAL_UNSIGNED_GT] = OP_SET_A,
1266 [SPECIAL_UNSIGNED_LTE] = OP_SET_BE,
1267 [SPECIAL_UNSIGNED_GTE] = OP_SET_AE,
1268 };
1269
1270 pseudo_t src1 = linearize_expression(ep, expr->left);
1271 pseudo_t src2 = linearize_expression(ep, expr->right);
1272 pseudo_t dst = add_binary_op(ep, expr->ctype, cmpop[expr->op], src1, src2);
1273 return dst;
1274 }
1275
1276
1277 pseudo_t linearize_cond_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false)
1278 {
1279 pseudo_t cond;
1280
1281 if (!expr || !bb_reachable(ep->active))
1282 return VOID;
1283
1284 switch (expr->type) {
1285
1286 case EXPR_STRING:
1287 case EXPR_VALUE:
1288 add_goto(ep, expr->value ? bb_true : bb_false);
1289 return VOID;
1290
1291 case EXPR_FVALUE:
1292 add_goto(ep, expr->fvalue ? bb_true : bb_false);
1293 return VOID;
1294
1295 case EXPR_LOGICAL:
1296 linearize_logical_branch(ep, expr, bb_true, bb_false);
1297 return VOID;
1298
1299 case EXPR_COMPARE:
1300 cond = linearize_compare(ep, expr);
1301 add_branch(ep, expr, cond, bb_true, bb_false);
1302 break;
1303
1304 case EXPR_PREOP:
1305 if (expr->op == '!')
1306 return linearize_cond_branch(ep, expr->unop, bb_false, bb_true);
1307 /* fall through */
1308 default: {
1309 cond = linearize_expression(ep, expr);
1310 add_branch(ep, expr, cond, bb_true, bb_false);
1311
1312 return VOID;
1313 }
1314 }
1315 return VOID;
1316 }
1317
1318
1319
1320 static pseudo_t linearize_logical_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false)
1321 {
1322 struct basic_block *next = alloc_basic_block(ep, expr->pos);
1323
1324 if (expr->op == SPECIAL_LOGICAL_OR)
1325 linearize_cond_branch(ep, expr->left, bb_true, next);
1326 else
1327 linearize_cond_branch(ep, expr->left, next, bb_false);
1328 set_activeblock(ep, next);
1329 linearize_cond_branch(ep, expr->right, bb_true, bb_false);
1330 return VOID;
1331 }
1332
1333 /*
1334 * Casts to pointers are "less safe" than other casts, since
1335 * they imply type-unsafe accesses. "void *" is a special
1336 * case, since you can't access through it anyway without another
1337 * cast.
1338 */
1339 static struct instruction *alloc_cast_instruction(struct symbol *ctype)
1340 {
1341 int opcode = OP_CAST;
1342 struct symbol *base = ctype;
1343
1344 if (base->type == SYM_NODE)
1345 base = base->ctype.base_type;
1346 if (base->type == SYM_PTR) {
1347 base = base->ctype.base_type;
1348 if (base != &void_ctype)
1349 opcode = OP_PTRCAST;
1350 }
1351 return alloc_typed_instruction(opcode, ctype);
1352 }
1353
1354 pseudo_t linearize_cast(struct entrypoint *ep, struct expression *expr)
1355 {
1356 pseudo_t src, result;
1357 struct instruction *insn;
1358
1359 src = linearize_expression(ep, expr->cast_expression);
1360 if (src == VOID)
1361 return VOID;
1362 if (!expr->ctype)
1363 return VOID;
1364 if (expr->ctype->bit_size < 0)
1365 return VOID;
1366
1367 insn = alloc_cast_instruction(expr->ctype);
1368 result = alloc_pseudo(insn);
1369 insn->target = result;
1370 insn->orig_type = expr->cast_expression->ctype;
1371 use_pseudo(src, &insn->src);
1372 add_one_insn(ep, insn);
1373 return result;
1374 }
1375
1376 pseudo_t linearize_position(struct entrypoint *ep, struct expression *pos, struct access_data *ad)
1377 {
1378 struct expression *init_expr = pos->init_expr;
1379 pseudo_t value = linearize_expression(ep, init_expr);
1380
1381 ad->offset = pos->init_offset;
1382 ad->source_type = base_type(init_expr->ctype);
1383 ad->result_type = init_expr->ctype;
1384 linearize_store_gen(ep, value, ad);
1385 return VOID;
1386 }
1387
1388 pseudo_t linearize_initializer(struct entrypoint *ep, struct expression *initializer, struct access_data *ad)
1389 {
1390 switch (initializer->type) {
1391 case EXPR_INITIALIZER: {
1392 struct expression *expr;
1393 FOR_EACH_PTR(initializer->expr_list, expr) {
1394 linearize_initializer(ep, expr, ad);
1395 } END_FOR_EACH_PTR(expr);
1396 break;
1397 }
1398 case EXPR_POS:
1399 linearize_position(ep, initializer, ad);
1400 break;
1401 default: {
1402 pseudo_t value = linearize_expression(ep, initializer);
1403 ad->source_type = base_type(initializer->ctype);
1404 ad->result_type = initializer->ctype;
1405 linearize_store_gen(ep, value, ad);
1406 }
1407 }
1408
1409 return VOID;
1410 }
1411
1412 void linearize_argument(struct entrypoint *ep, struct symbol *arg, int nr)
1413 {
1414 struct access_data ad = { NULL, };
1415
1416 ad.source_type = arg;
1417 ad.result_type = arg;
1418 ad.address = symbol_pseudo(ep, arg);
1419 linearize_store_gen(ep, argument_pseudo(ep, nr), &ad);
1420 finish_address_gen(ep, &ad);
1421 }
1422
1423 pseudo_t linearize_expression(struct entrypoint *ep, struct expression *expr)
1424 {
1425 if (!expr)
1426 return VOID;
1427
1428 switch (expr->type) {
1429 case EXPR_SYMBOL:
1430 linearize_one_symbol(ep, expr->symbol);
1431 return add_setval(ep, expr->symbol, NULL);
1432
1433 case EXPR_VALUE:
1434 return value_pseudo(expr->value);
1435
1436 case EXPR_STRING: case EXPR_FVALUE: case EXPR_LABEL:
1437 return add_setval(ep, expr->ctype, expr);
1438
1439 case EXPR_STATEMENT:
1440 return linearize_statement(ep, expr->statement);
1441
1442 case EXPR_CALL:
1443 return linearize_call_expression(ep, expr);
1444
1445 case EXPR_BINOP:
1446 return linearize_binop(ep, expr);
1447
1448 case EXPR_LOGICAL:
1449 return linearize_logical(ep, expr);
1450
1451 case EXPR_COMPARE:
1452 return linearize_compare(ep, expr);
1453
1454 case EXPR_SELECT:
1455 return linearize_select(ep, expr);
1456
1457 case EXPR_CONDITIONAL:
1458 if (!expr->cond_true)
1459 return linearize_short_conditional(ep, expr, expr->conditional, expr->cond_false);
1460
1461 return linearize_conditional(ep, expr, expr->conditional,
1462 expr->cond_true, expr->cond_false);
1463
1464 case EXPR_COMMA:
1465 linearize_expression(ep, expr->left);
1466 return linearize_expression(ep, expr->right);
1467
1468 case EXPR_ASSIGNMENT:
1469 return linearize_assignment(ep, expr);
1470
1471 case EXPR_PREOP:
1472 return linearize_preop(ep, expr);
1473
1474 case EXPR_POSTOP:
1475 return linearize_postop(ep, expr);
1476
1477 case EXPR_CAST:
1478 case EXPR_IMPLIED_CAST:
1479 return linearize_cast(ep, expr);
1480
1481 case EXPR_SLICE:
1482 return linearize_slice(ep, expr);
1483
1484 case EXPR_INITIALIZER:
1485 case EXPR_POS:
1486 warning(expr->pos, "unexpected initializer expression (%d %d)", expr->type, expr->op);
1487 return VOID;
1488 default:
1489 warning(expr->pos, "unknown expression (%d %d)", expr->type, expr->op);
1490 return VOID;
1491 }
1492 return VOID;
1493 }
1494
1495 static void linearize_one_symbol(struct entrypoint *ep, struct symbol *sym)
1496 {
1497 struct access_data ad = { NULL, };
1498
1499 if (!sym || !sym->initializer || sym->initialized)
1500 return;
1501
1502 /* We need to output these puppies some day too.. */
1503 if (sym->ctype.modifiers & (MOD_STATIC | MOD_TOPLEVEL))
1504 return;
1505
1506 sym->initialized = 1;
1507 ad.address = symbol_pseudo(ep, sym);
1508 linearize_initializer(ep, sym->initializer, &ad);
1509 finish_address_gen(ep, &ad);
1510 }
1511
1512 static pseudo_t linearize_compound_statement(struct entrypoint *ep, struct statement *stmt)
1513 {
1514 pseudo_t pseudo;
1515 struct statement *s;
1516 struct symbol *sym;
1517 struct symbol *ret = stmt->ret;
1518
1519 concat_symbol_list(stmt->syms, &ep->syms);
1520
1521 FOR_EACH_PTR(stmt->syms, sym) {
1522 linearize_one_symbol(ep, sym);
1523 } END_FOR_EACH_PTR(sym);
1524
1525 pseudo = VOID;
1526 FOR_EACH_PTR(stmt->stmts, s) {
1527 pseudo = linearize_statement(ep, s);
1528 } END_FOR_EACH_PTR(s);
1529
1530 if (ret) {
1531 struct basic_block *bb = add_label(ep, ret);
1532 struct instruction *phi_node = first_instruction(bb->insns);
1533
1534 if (!phi_node)
1535 return pseudo;
1536
1537 if (pseudo_list_size(phi_node->phi_list)==1) {
1538 pseudo = first_pseudo(phi_node->phi_list);
1539 assert(pseudo->type == PSEUDO_PHI);
1540 return pseudo->def->src1;
1541 }
1542 return phi_node->target;
1543 }
1544 return pseudo;
1545 }
1546
1547 pseudo_t linearize_internal(struct entrypoint *ep, struct statement *stmt)
1548 {
1549 struct instruction *insn = alloc_instruction(OP_CONTEXT, 0);
1550 struct expression *expr = stmt->expression;
1551 int value = 0;
1552
1553 if (expr->type == EXPR_VALUE)
1554 value = expr->value;
1555
1556 insn->increment = value;
1557 add_one_insn(ep, insn);
1558 return VOID;
1559 }
1560
1561 static void add_asm_input(struct entrypoint *ep, struct instruction *insn, struct expression *expr)
1562 {
1563 pseudo_t pseudo = linearize_expression(ep, expr);
1564
1565 use_pseudo(pseudo, add_pseudo(&insn->inputs, pseudo));
1566 }
1567
1568 static void add_asm_output(struct entrypoint *ep, struct instruction *insn, struct expression *expr)
1569 {
1570 struct access_data ad = { NULL, };
1571 pseudo_t pseudo = alloc_pseudo(insn);
1572
1573 if (!linearize_address_gen(ep, expr, &ad))
1574 return;
1575 linearize_store_gen(ep, pseudo, &ad);
1576 finish_address_gen(ep, &ad);
1577 add_pseudo(&insn->outputs, pseudo);
1578 }
1579
1580 pseudo_t linearize_asm_statement(struct entrypoint *ep, struct statement *stmt)
1581 {
1582 int even_odd;
1583 struct expression *expr;
1584 struct instruction *insn;
1585
1586 insn = alloc_instruction(OP_ASM, 0);
1587 expr = stmt->asm_string;
1588 if (!expr || expr->type != EXPR_STRING) {
1589 warning(stmt->pos, "expected string in inline asm");
1590 return VOID;
1591 }
1592 insn->string = expr->string->data;
1593
1594 /* Gather the inputs.. */
1595 even_odd = 0;
1596 FOR_EACH_PTR(stmt->asm_inputs, expr) {
1597 even_odd = 1 - even_odd;
1598
1599 /* FIXME! We ignore the constraints for now.. */
1600 if (even_odd)
1601 continue;
1602 add_asm_input(ep, insn, expr);
1603 } END_FOR_EACH_PTR(expr);
1604
1605 add_one_insn(ep, insn);
1606
1607 /* Assign the outputs */
1608 even_odd = 0;
1609 FOR_EACH_PTR(stmt->asm_outputs, expr) {
1610 even_odd = 1 - even_odd;
1611
1612 /* FIXME! We ignore the constraints for now.. */
1613 if (even_odd)
1614 continue;
1615 add_asm_output(ep, insn, expr);
1616 } END_FOR_EACH_PTR(expr);
1617
1618 return VOID;
1619 }
1620
1621 pseudo_t linearize_statement(struct entrypoint *ep, struct statement *stmt)
1622 {
1623 struct basic_block *bb;
1624
1625 if (!stmt)
1626 return VOID;
1627
1628 bb = ep->active;
1629 if (bb && !bb->insns)
1630 bb->pos = stmt->pos;
1631
1632 switch (stmt->type) {
1633 case STMT_NONE:
1634 break;
1635
1636 case STMT_INTERNAL:
1637 return linearize_internal(ep, stmt);
1638
1639 case STMT_EXPRESSION:
1640 return linearize_expression(ep, stmt->expression);
1641
1642 case STMT_ASM:
1643 return linearize_asm_statement(ep, stmt);
1644
1645 case STMT_RETURN: {
1646 struct expression *expr = stmt->expression;
1647 struct basic_block *bb_return = get_bound_block(ep, stmt->ret_target);
1648 struct basic_block *active;
1649 pseudo_t src = linearize_expression(ep, expr);
1650 active = ep->active;
1651 if (active && src != &void_pseudo) {
1652 struct instruction *phi_node = first_instruction(bb_return->insns);
1653 pseudo_t phi;
1654 if (!phi_node) {
1655 phi_node = alloc_typed_instruction(OP_PHI, expr->ctype);
1656 phi_node->target = alloc_pseudo(phi_node);
1657 phi_node->bb = bb_return;
1658 add_instruction(&bb_return->insns, phi_node);
1659 }
1660 phi = alloc_phi(active, src, type_size(expr->ctype));
1661 phi->ident = &return_ident;
1662 use_pseudo(phi, add_pseudo(&phi_node->phi_list, phi));
1663 }
1664 add_goto(ep, bb_return);
1665 return VOID;
1666 }
1667
1668 case STMT_CASE: {
1669 add_label(ep, stmt->case_label);
1670 linearize_statement(ep, stmt->case_statement);
1671 break;
1672 }
1673
1674 case STMT_LABEL: {
1675 struct symbol *label = stmt->label_identifier;
1676
1677 if (label->used) {
1678 add_label(ep, label);
1679 linearize_statement(ep, stmt->label_statement);
1680 }
1681 break;
1682 }
1683
1684 case STMT_GOTO: {
1685 struct symbol *sym;
1686 struct expression *expr;
1687 struct instruction *goto_ins;
1688 struct basic_block *active;
1689 pseudo_t pseudo;
1690
1691 active = ep->active;
1692 if (!bb_reachable(active))
1693 break;
1694
1695 if (stmt->goto_label) {
1696 add_goto(ep, get_bound_block(ep, stmt->goto_label));
1697 break;
1698 }
1699
1700 expr = stmt->goto_expression;
1701 if (!expr)
1702 break;
1703
1704 /* This can happen as part of simplification */
1705 if (expr->type == EXPR_LABEL) {
1706 add_goto(ep, get_bound_block(ep, expr->label_symbol));
1707 break;
1708 }
1709
1710 pseudo = linearize_expression(ep, expr);
1711 goto_ins = alloc_instruction(OP_COMPUTEDGOTO, 0);
1712 use_pseudo(pseudo, &goto_ins->target);
1713 add_one_insn(ep, goto_ins);
1714
1715 FOR_EACH_PTR(stmt->target_list, sym) {
1716 struct basic_block *bb_computed = get_bound_block(ep, sym);
1717 struct multijmp *jmp = alloc_multijmp(bb_computed, 1, 0);
1718 add_multijmp(&goto_ins->multijmp_list, jmp);
1719 add_bb(&bb_computed->parents, ep->active);
1720 add_bb(&active->children, bb_computed);
1721 } END_FOR_EACH_PTR(sym);
1722
1723 finish_block(ep);
1724 break;
1725 }
1726
1727 case STMT_COMPOUND:
1728 return linearize_compound_statement(ep, stmt);
1729
1730 /*
1731 * This could take 'likely/unlikely' into account, and
1732 * switch the arms around appropriately..
1733 */
1734 case STMT_IF: {
1735 struct basic_block *bb_true, *bb_false, *endif;
1736 struct expression *cond = stmt->if_conditional;
1737
1738 bb_true = alloc_basic_block(ep, stmt->pos);
1739 bb_false = endif = alloc_basic_block(ep, stmt->pos);
1740
1741 linearize_cond_branch(ep, cond, bb_true, bb_false);
1742
1743 set_activeblock(ep, bb_true);
1744 linearize_statement(ep, stmt->if_true);
1745
1746 if (stmt->if_false) {
1747 endif = alloc_basic_block(ep, stmt->pos);
1748 add_goto(ep, endif);
1749 set_activeblock(ep, bb_false);
1750 linearize_statement(ep, stmt->if_false);
1751 }
1752 set_activeblock(ep, endif);
1753 break;
1754 }
1755
1756 case STMT_SWITCH: {
1757 struct symbol *sym;
1758 struct instruction *switch_ins;
1759 struct basic_block *switch_end = alloc_basic_block(ep, stmt->pos);
1760 struct basic_block *active, *default_case;
1761 struct multijmp *jmp;
1762 pseudo_t pseudo;
1763
1764 pseudo = linearize_expression(ep, stmt->switch_expression);
1765
1766 active = ep->active;
1767 if (!bb_reachable(active))
1768 break;
1769
1770 switch_ins = alloc_instruction(OP_SWITCH, 0);
1771 use_pseudo(pseudo, &switch_ins->cond);
1772 add_one_insn(ep, switch_ins);
1773 finish_block(ep);
1774
1775 default_case = NULL;
1776 FOR_EACH_PTR(stmt->switch_case->symbol_list, sym) {
1777 struct statement *case_stmt = sym->stmt;
1778 struct basic_block *bb_case = get_bound_block(ep, sym);
1779
1780 if (!case_stmt->case_expression) {
1781 default_case = bb_case;
1782 continue;
1783 } else {
1784 int begin, end;
1785
1786 begin = end = case_stmt->case_expression->value;
1787 if (case_stmt->case_to)
1788 end = case_stmt->case_to->value;
1789 if (begin > end)
1790 jmp = alloc_multijmp(bb_case, end, begin);
1791 else
1792 jmp = alloc_multijmp(bb_case, begin, end);
1793
1794 }
1795 add_multijmp(&switch_ins->multijmp_list, jmp);
1796 add_bb(&bb_case->parents, active);
1797 add_bb(&active->children, bb_case);
1798 } END_FOR_EACH_PTR(sym);
1799
1800 bind_label(stmt->switch_break, switch_end, stmt->pos);
1801
1802 /* And linearize the actual statement */
1803 linearize_statement(ep, stmt->switch_statement);
1804 set_activeblock(ep, switch_end);
1805
1806 if (!default_case)
1807 default_case = switch_end;
1808
1809 jmp = alloc_multijmp(default_case, 1, 0);
1810 add_multijmp(&switch_ins->multijmp_list, jmp);
1811 add_bb(&default_case->parents, active);
1812 add_bb(&active->children, default_case);
1813
1814 break;
1815 }
1816
1817 case STMT_ITERATOR: {
1818 struct statement *pre_statement = stmt->iterator_pre_statement;
1819 struct expression *pre_condition = stmt->iterator_pre_condition;
1820 struct statement *statement = stmt->iterator_statement;
1821 struct statement *post_statement = stmt->iterator_post_statement;
1822 struct expression *post_condition = stmt->iterator_post_condition;
1823 struct basic_block *loop_top, *loop_body, *loop_continue, *loop_end;
1824
1825 concat_symbol_list(stmt->iterator_syms, &ep->syms);
1826 linearize_statement(ep, pre_statement);
1827
1828 loop_body = loop_top = alloc_basic_block(ep, stmt->pos);
1829 loop_continue = alloc_basic_block(ep, stmt->pos);
1830 loop_end = alloc_basic_block(ep, stmt->pos);
1831
1832 if (pre_condition == post_condition) {
1833 loop_top = alloc_basic_block(ep, stmt->pos);
1834 set_activeblock(ep, loop_top);
1835 }
1836
1837 if (pre_condition)
1838 linearize_cond_branch(ep, pre_condition, loop_body, loop_end);
1839
1840 bind_label(stmt->iterator_continue, loop_continue, stmt->pos);
1841 bind_label(stmt->iterator_break, loop_end, stmt->pos);
1842
1843 set_activeblock(ep, loop_body);
1844 linearize_statement(ep, statement);
1845 add_goto(ep, loop_continue);
1846
1847 set_activeblock(ep, loop_continue);
1848 linearize_statement(ep, post_statement);
1849 if (!post_condition || pre_condition == post_condition)
1850 add_goto(ep, loop_top);
1851 else
1852 linearize_cond_branch(ep, post_condition, loop_top, loop_end);
1853 set_activeblock(ep, loop_end);
1854 break;
1855 }
1856
1857 default:
1858 break;
1859 }
1860 return VOID;
1861 }
1862
1863 static struct entrypoint *linearize_fn(struct symbol *sym, struct symbol *base_type)
1864 {
1865 struct entrypoint *ep;
1866 struct basic_block *bb;
1867 struct symbol *arg;
1868 struct instruction *entry;
1869 pseudo_t result;
1870 int i;
1871
1872 if (!base_type->stmt)
1873 return NULL;
1874
1875 ep = alloc_entrypoint();
1876 bb = alloc_basic_block(ep, sym->pos);
1877
1878 ep->name = sym;
1879 set_activeblock(ep, bb);
1880
1881 entry = alloc_instruction(OP_ENTRY, 0);
1882 add_one_insn(ep, entry);
1883 ep->entry = entry;
1884
1885 concat_symbol_list(base_type->arguments, &ep->syms);
1886
1887 /* FIXME!! We should do something else about varargs.. */
1888 i = 0;
1889 FOR_EACH_PTR(base_type->arguments, arg) {
1890 linearize_argument(ep, arg, ++i);
1891 } END_FOR_EACH_PTR(arg);
1892
1893 result = linearize_statement(ep, base_type->stmt);
1894 if (bb_reachable(ep->active) && !bb_terminated(ep->active)) {
1895 struct symbol *ret_type = base_type->ctype.base_type;
1896 struct instruction *insn = alloc_typed_instruction(OP_RET, ret_type);
1897
1898 if (type_size(ret_type) > 0)
1899 use_pseudo(result, &insn->src);
1900 add_one_insn(ep, insn);
1901 }
1902
1903 /*
1904 * Do trivial flow simplification - branches to
1905 * branches, kill dead basicblocks etc
1906 */
1907 kill_unreachable_bbs(ep);
1908
1909 /*
1910 * Turn symbols into pseudos
1911 */
1912 simplify_symbol_usage(ep);
1913
1914 repeat:
1915 /*
1916 * Remove trivial instructions, and try to CSE
1917 * the rest.
1918 */
1919 do {
1920 cleanup_and_cse(ep);
1921 pack_basic_blocks(ep);
1922 } while (repeat_phase & REPEAT_CSE);
1923
1924 kill_unreachable_bbs(ep);
1925 vrfy_flow(ep);
1926
1927 /* Cleanup */
1928 clear_symbol_pseudos(ep);
1929
1930 /* And track pseudo register usage */
1931 track_pseudo_liveness(ep);
1932
1933 /*
1934 * Some flow optimizations can only effectively
1935 * be done when we've done liveness analysis. But
1936 * if they trigger, we need to start all over
1937 * again
1938 */
1939 if (simplify_flow(ep)) {
1940 clear_liveness(ep);
1941 goto repeat;
1942 }
1943
1944 /* Finally, add deathnotes to pseudos now that we have them */
1945 track_pseudo_death(ep);
1946
1947 return ep;
1948 }
1949
1950 struct entrypoint *linearize_symbol(struct symbol *sym)
1951 {
1952 struct symbol *base_type;
1953
1954 if (!sym)
1955 return NULL;
1956 base_type = sym->ctype.base_type;
1957 if (!base_type)
1958 return NULL;
1959 if (base_type->type == SYM_FN)
1960 return linearize_fn(sym, base_type);
1961 return NULL;
1962 }
Static analysis for C