?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
fix boolean context for OP_AND_BOOL & OP_OR_BOOL
[smatch.git] / linearize.c
blob236d37a319703febf70b336467a1dbbe79e606a8
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 pseudo_t 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 static pseudo_t linearize_initializer(struct entrypoint *ep, struct expression *initializer, struct access_data *);
35 static pseudo_t cast_pseudo(struct entrypoint *ep, pseudo_t src, struct symbol *from, struct symbol *to);
36
37 struct pseudo void_pseudo = {};
38
39 static struct position current_pos;
40
41 ALLOCATOR(pseudo_user, "pseudo_user");
42
43 static struct instruction *alloc_instruction(int opcode, int size)
44 {
45 struct instruction * insn = __alloc_instruction(0);
46 insn->opcode = opcode;
47 insn->size = size;
48 insn->pos = current_pos;
49 return insn;
50 }
51
52 static inline int type_size(struct symbol *type)
53 {
54 return type ? type->bit_size > 0 ? type->bit_size : 0 : 0;
55 }
56
57 static struct instruction *alloc_typed_instruction(int opcode, struct symbol *type)
58 {
59 struct instruction *insn = alloc_instruction(opcode, type_size(type));
60 insn->type = type;
61 return insn;
62 }
63
64 static struct entrypoint *alloc_entrypoint(void)
65 {
66 return __alloc_entrypoint(0);
67 }
68
69 static struct basic_block *alloc_basic_block(struct entrypoint *ep, struct position pos)
70 {
71 static int nr;
72 struct basic_block *bb = __alloc_basic_block(0);
73 bb->context = -1;
74 bb->pos = pos;
75 bb->ep = ep;
76 bb->nr = nr++;
77 return bb;
78 }
79
80 static struct multijmp *alloc_multijmp(struct basic_block *target, int begin, int end)
81 {
82 struct multijmp *multijmp = __alloc_multijmp(0);
83 multijmp->target = target;
84 multijmp->begin = begin;
85 multijmp->end = end;
86 return multijmp;
87 }
88
89 static inline int regno(pseudo_t n)
90 {
91 int retval = -1;
92 if (n && n->type == PSEUDO_REG)
93 retval = n->nr;
94 return retval;
95 }
96
97 const char *show_pseudo(pseudo_t pseudo)
98 {
99 static int n;
100 static char buffer[4][64];
101 char *buf;
102 int i;
103
104 if (!pseudo)
105 return "no pseudo";
106 if (pseudo == VOID)
107 return "VOID";
108 buf = buffer[3 & ++n];
109 switch(pseudo->type) {
110 case PSEUDO_SYM: {
111 struct symbol *sym = pseudo->sym;
112 struct expression *expr;
113
114 if (sym->bb_target) {
115 snprintf(buf, 64, ".L%u", sym->bb_target->nr);
116 break;
117 }
118 if (sym->ident) {
119 snprintf(buf, 64, "%s", show_ident(sym->ident));
120 break;
121 }
122 expr = sym->initializer;
123 snprintf(buf, 64, "<anon symbol:%p>", sym);
124 if (expr) {
125 switch (expr->type) {
126 case EXPR_VALUE:
127 snprintf(buf, 64, "<symbol value: %lld>", expr->value);
128 break;
129 case EXPR_STRING:
130 return show_string(expr->string);
131 default:
132 break;
133 }
134 }
135 break;
136 }
137 case PSEUDO_REG:
138 i = snprintf(buf, 64, "%%r%d", pseudo->nr);
139 if (pseudo->ident)
140 sprintf(buf+i, "(%s)", show_ident(pseudo->ident));
141 break;
142 case PSEUDO_VAL: {
143 long long value = pseudo->value;
144 if (value > 1000 || value < -1000)
145 snprintf(buf, 64, "$%#llx", value);
146 else
147 snprintf(buf, 64, "$%lld", value);
148 break;
149 }
150 case PSEUDO_ARG:
151 snprintf(buf, 64, "%%arg%d", pseudo->nr);
152 break;
153 case PSEUDO_PHI:
154 i = snprintf(buf, 64, "%%phi%d", pseudo->nr);
155 if (pseudo->ident)
156 sprintf(buf+i, "(%s)", show_ident(pseudo->ident));
157 break;
158 default:
159 snprintf(buf, 64, "<bad pseudo type %d>", pseudo->type);
160 }
161 return buf;
162 }
163
164 static const char *opcodes[] = {
165 [OP_BADOP] = "bad_op",
166
167 /* Fn entrypoint */
168 [OP_ENTRY] = "<entry-point>",
169
170 /* Terminator */
171 [OP_RET] = "ret",
172 [OP_BR] = "br",
173 [OP_CBR] = "cbr",
174 [OP_SWITCH] = "switch",
175 [OP_INVOKE] = "invoke",
176 [OP_COMPUTEDGOTO] = "jmp *",
177 [OP_UNWIND] = "unwind",
178
179 /* Binary */
180 [OP_ADD] = "add",
181 [OP_SUB] = "sub",
182 [OP_MULU] = "mulu",
183 [OP_MULS] = "muls",
184 [OP_DIVU] = "divu",
185 [OP_DIVS] = "divs",
186 [OP_MODU] = "modu",
187 [OP_MODS] = "mods",
188 [OP_SHL] = "shl",
189 [OP_LSR] = "lsr",
190 [OP_ASR] = "asr",
191
192 /* Logical */
193 [OP_AND] = "and",
194 [OP_OR] = "or",
195 [OP_XOR] = "xor",
196 [OP_AND_BOOL] = "and-bool",
197 [OP_OR_BOOL] = "or-bool",
198
199 /* Binary comparison */
200 [OP_SET_EQ] = "seteq",
201 [OP_SET_NE] = "setne",
202 [OP_SET_LE] = "setle",
203 [OP_SET_GE] = "setge",
204 [OP_SET_LT] = "setlt",
205 [OP_SET_GT] = "setgt",
206 [OP_SET_B] = "setb",
207 [OP_SET_A] = "seta",
208 [OP_SET_BE] = "setbe",
209 [OP_SET_AE] = "setae",
210
211 /* Uni */
212 [OP_NOT] = "not",
213 [OP_NEG] = "neg",
214
215 /* Special three-input */
216 [OP_SEL] = "select",
217
218 /* Memory */
219 [OP_MALLOC] = "malloc",
220 [OP_FREE] = "free",
221 [OP_ALLOCA] = "alloca",
222 [OP_LOAD] = "load",
223 [OP_STORE] = "store",
224 [OP_SETVAL] = "set",
225 [OP_SYMADDR] = "symaddr",
226 [OP_GET_ELEMENT_PTR] = "getelem",
227
228 /* Other */
229 [OP_PHI] = "phi",
230 [OP_PHISOURCE] = "phisrc",
231 [OP_CAST] = "cast",
232 [OP_SCAST] = "scast",
233 [OP_FPCAST] = "fpcast",
234 [OP_PTRCAST] = "ptrcast",
235 [OP_INLINED_CALL] = "# call",
236 [OP_CALL] = "call",
237 [OP_VANEXT] = "va_next",
238 [OP_VAARG] = "va_arg",
239 [OP_SLICE] = "slice",
240 [OP_SNOP] = "snop",
241 [OP_LNOP] = "lnop",
242 [OP_NOP] = "nop",
243 [OP_DEATHNOTE] = "dead",
244 [OP_ASM] = "asm",
245
246 /* Sparse tagging (line numbers, context, whatever) */
247 [OP_CONTEXT] = "context",
248 [OP_RANGE] = "range-check",
249
250 [OP_COPY] = "copy",
251 };
252
253 static char *show_asm_constraints(char *buf, const char *sep, struct asm_constraint_list *list)
254 {
255 struct asm_constraint *entry;
256
257 FOR_EACH_PTR(list, entry) {
258 buf += sprintf(buf, "%s\"%s\"", sep, entry->constraint);
259 if (entry->pseudo)
260 buf += sprintf(buf, " (%s)", show_pseudo(entry->pseudo));
261 if (entry->ident)
262 buf += sprintf(buf, " [%s]", show_ident(entry->ident));
263 sep = ", ";
264 } END_FOR_EACH_PTR(entry);
265 return buf;
266 }
267
268 static char *show_asm(char *buf, struct instruction *insn)
269 {
270 struct asm_rules *rules = insn->asm_rules;
271
272 buf += sprintf(buf, "\"%s\"", insn->string);
273 buf = show_asm_constraints(buf, "\n\t\tout: ", rules->outputs);
274 buf = show_asm_constraints(buf, "\n\t\tin: ", rules->inputs);
275 buf = show_asm_constraints(buf, "\n\t\tclobber: ", rules->clobbers);
276 return buf;
277 }
278
279 const char *show_instruction(struct instruction *insn)
280 {
281 int opcode = insn->opcode;
282 static char buffer[4096];
283 char *buf;
284
285 buf = buffer;
286 if (!insn->bb)
287 buf += sprintf(buf, "# ");
288
289 if (opcode < ARRAY_SIZE(opcodes)) {
290 const char *op = opcodes[opcode];
291 if (!op)
292 buf += sprintf(buf, "opcode:%d", opcode);
293 else
294 buf += sprintf(buf, "%s", op);
295 if (insn->size)
296 buf += sprintf(buf, ".%d", insn->size);
297 memset(buf, ' ', 20);
298 buf++;
299 }
300
301 if (buf < buffer + 12)
302 buf = buffer + 12;
303 switch (opcode) {
304 case OP_RET:
305 if (insn->src && insn->src != VOID)
306 buf += sprintf(buf, "%s", show_pseudo(insn->src));
307 break;
308
309 case OP_CBR:
310 buf += sprintf(buf, "%s, .L%u, .L%u", show_pseudo(insn->cond), insn->bb_true->nr, insn->bb_false->nr);
311 break;
312
313 case OP_BR:
314 buf += sprintf(buf, ".L%u", insn->bb_true->nr);
315 break;
316
317 case OP_SYMADDR: {
318 struct symbol *sym = insn->symbol->sym;
319 buf += sprintf(buf, "%s <- ", show_pseudo(insn->target));
320
321 if (!insn->bb && !sym)
322 break;
323 if (sym->bb_target) {
324 buf += sprintf(buf, ".L%u", sym->bb_target->nr);
325 break;
326 }
327 if (sym->ident) {
328 buf += sprintf(buf, "%s", show_ident(sym->ident));
329 break;
330 }
331 buf += sprintf(buf, "<anon symbol:%p>", sym);
332 break;
333 }
334
335 case OP_SETVAL: {
336 struct expression *expr = insn->val;
337 buf += sprintf(buf, "%s <- ", show_pseudo(insn->target));
338
339 if (!expr) {
340 buf += sprintf(buf, "%s", "<none>");
341 break;
342 }
343
344 switch (expr->type) {
345 case EXPR_VALUE:
346 buf += sprintf(buf, "%lld", expr->value);
347 break;
348 case EXPR_FVALUE:
349 buf += sprintf(buf, "%Lf", expr->fvalue);
350 break;
351 case EXPR_STRING:
352 buf += sprintf(buf, "%.40s", show_string(expr->string));
353 break;
354 case EXPR_SYMBOL:
355 buf += sprintf(buf, "%s", show_ident(expr->symbol->ident));
356 break;
357 case EXPR_LABEL:
358 buf += sprintf(buf, ".L%u", expr->symbol->bb_target->nr);
359 break;
360 default:
361 buf += sprintf(buf, "SETVAL EXPR TYPE %d", expr->type);
362 }
363 break;
364 }
365 case OP_SWITCH: {
366 struct multijmp *jmp;
367 buf += sprintf(buf, "%s", show_pseudo(insn->cond));
368 FOR_EACH_PTR(insn->multijmp_list, jmp) {
369 if (jmp->begin == jmp->end)
370 buf += sprintf(buf, ", %d -> .L%u", jmp->begin, jmp->target->nr);
371 else if (jmp->begin < jmp->end)
372 buf += sprintf(buf, ", %d ... %d -> .L%u", jmp->begin, jmp->end, jmp->target->nr);
373 else
374 buf += sprintf(buf, ", default -> .L%u", jmp->target->nr);
375 } END_FOR_EACH_PTR(jmp);
376 break;
377 }
378 case OP_COMPUTEDGOTO: {
379 struct multijmp *jmp;
380 buf += sprintf(buf, "%s", show_pseudo(insn->target));
381 FOR_EACH_PTR(insn->multijmp_list, jmp) {
382 buf += sprintf(buf, ", .L%u", jmp->target->nr);
383 } END_FOR_EACH_PTR(jmp);
384 break;
385 }
386
387 case OP_PHISOURCE: {
388 struct instruction *phi;
389 buf += sprintf(buf, "%s <- %s ", show_pseudo(insn->target), show_pseudo(insn->phi_src));
390 FOR_EACH_PTR(insn->phi_users, phi) {
391 buf += sprintf(buf, " (%s)", show_pseudo(phi->target));
392 } END_FOR_EACH_PTR(phi);
393 break;
394 }
395
396 case OP_PHI: {
397 pseudo_t phi;
398 const char *s = " <-";
399 buf += sprintf(buf, "%s", show_pseudo(insn->target));
400 FOR_EACH_PTR(insn->phi_list, phi) {
401 buf += sprintf(buf, "%s %s", s, show_pseudo(phi));
402 s = ",";
403 } END_FOR_EACH_PTR(phi);
404 break;
405 }
406 case OP_LOAD: case OP_LNOP:
407 buf += sprintf(buf, "%s <- %d[%s]", show_pseudo(insn->target), insn->offset, show_pseudo(insn->src));
408 break;
409 case OP_STORE: case OP_SNOP:
410 buf += sprintf(buf, "%s -> %d[%s]", show_pseudo(insn->target), insn->offset, show_pseudo(insn->src));
411 break;
412 case OP_INLINED_CALL:
413 case OP_CALL: {
414 struct pseudo *arg;
415 if (insn->target && insn->target != VOID)
416 buf += sprintf(buf, "%s <- ", show_pseudo(insn->target));
417 buf += sprintf(buf, "%s", show_pseudo(insn->func));
418 FOR_EACH_PTR(insn->arguments, arg) {
419 buf += sprintf(buf, ", %s", show_pseudo(arg));
420 } END_FOR_EACH_PTR(arg);
421 break;
422 }
423 case OP_CAST:
424 case OP_SCAST:
425 case OP_FPCAST:
426 case OP_PTRCAST:
427 buf += sprintf(buf, "%s <- (%d) %s",
428 show_pseudo(insn->target),
429 type_size(insn->orig_type),
430 show_pseudo(insn->src));
431 break;
432 case OP_BINARY ... OP_BINARY_END:
433 case OP_BINCMP ... OP_BINCMP_END:
434 buf += sprintf(buf, "%s <- %s, %s", show_pseudo(insn->target), show_pseudo(insn->src1), show_pseudo(insn->src2));
435 break;
436
437 case OP_SEL:
438 buf += sprintf(buf, "%s <- %s, %s, %s", show_pseudo(insn->target),
439 show_pseudo(insn->src1), show_pseudo(insn->src2), show_pseudo(insn->src3));
440 break;
441
442 case OP_SLICE:
443 buf += sprintf(buf, "%s <- %s, %d, %d", show_pseudo(insn->target), show_pseudo(insn->base), insn->from, insn->len);
444 break;
445
446 case OP_NOT: case OP_NEG:
447 buf += sprintf(buf, "%s <- %s", show_pseudo(insn->target), show_pseudo(insn->src1));
448 break;
449
450 case OP_CONTEXT:
451 buf += sprintf(buf, "%s%d", insn->check ? "check: " : "", insn->increment);
452 break;
453 case OP_RANGE:
454 buf += sprintf(buf, "%s between %s..%s", show_pseudo(insn->src1), show_pseudo(insn->src2), show_pseudo(insn->src3));
455 break;
456 case OP_NOP:
457 buf += sprintf(buf, "%s <- %s", show_pseudo(insn->target), show_pseudo(insn->src1));
458 break;
459 case OP_DEATHNOTE:
460 buf += sprintf(buf, "%s", show_pseudo(insn->target));
461 break;
462 case OP_ASM:
463 buf = show_asm(buf, insn);
464 break;
465 case OP_COPY:
466 buf += sprintf(buf, "%s <- %s", show_pseudo(insn->target), show_pseudo(insn->src));
467 break;
468 default:
469 break;
470 }
471
472 if (buf >= buffer + sizeof(buffer))
473 die("instruction buffer overflowed %td\n", buf - buffer);
474 do { --buf; } while (*buf == ' ');
475 *++buf = 0;
476 return buffer;
477 }
478
479 void show_bb(struct basic_block *bb)
480 {
481 struct instruction *insn;
482
483 printf(".L%u:\n", bb->nr);
484 if (verbose) {
485 pseudo_t needs, defines;
486 printf("%s:%d\n", stream_name(bb->pos.stream), bb->pos.line);
487
488 FOR_EACH_PTR(bb->needs, needs) {
489 struct instruction *def = needs->def;
490 if (def->opcode != OP_PHI) {
491 printf(" **uses %s (from .L%u)**\n", show_pseudo(needs), def->bb->nr);
492 } else {
493 pseudo_t phi;
494 const char *sep = " ";
495 printf(" **uses %s (from", show_pseudo(needs));
496 FOR_EACH_PTR(def->phi_list, phi) {
497 if (phi == VOID)
498 continue;
499 printf("%s(%s:.L%u)", sep, show_pseudo(phi), phi->def->bb->nr);
500 sep = ", ";
501 } END_FOR_EACH_PTR(phi);
502 printf(")**\n");
503 }
504 } END_FOR_EACH_PTR(needs);
505
506 FOR_EACH_PTR(bb->defines, defines) {
507 printf(" **defines %s **\n", show_pseudo(defines));
508 } END_FOR_EACH_PTR(defines);
509
510 if (bb->parents) {
511 struct basic_block *from;
512 FOR_EACH_PTR(bb->parents, from) {
513 printf(" **from .L%u (%s:%d:%d)**\n", from->nr,
514 stream_name(from->pos.stream), from->pos.line, from->pos.pos);
515 } END_FOR_EACH_PTR(from);
516 }
517
518 if (bb->children) {
519 struct basic_block *to;
520 FOR_EACH_PTR(bb->children, to) {
521 printf(" **to .L%u (%s:%d:%d)**\n", to->nr,
522 stream_name(to->pos.stream), to->pos.line, to->pos.pos);
523 } END_FOR_EACH_PTR(to);
524 }
525 }
526
527 FOR_EACH_PTR(bb->insns, insn) {
528 if (!insn->bb && verbose < 2)
529 continue;
530 printf("\t%s\n", show_instruction(insn));
531 } END_FOR_EACH_PTR(insn);
532 if (!bb_terminated(bb))
533 printf("\tEND\n");
534 }
535
536 static void show_symbol_usage(pseudo_t pseudo)
537 {
538 struct pseudo_user *pu;
539
540 if (pseudo) {
541 FOR_EACH_PTR(pseudo->users, pu) {
542 printf("\t%s\n", show_instruction(pu->insn));
543 } END_FOR_EACH_PTR(pu);
544 }
545 }
546
547 void show_entry(struct entrypoint *ep)
548 {
549 struct symbol *sym;
550 struct basic_block *bb;
551
552 printf("%s:\n", show_ident(ep->name->ident));
553
554 if (verbose) {
555 printf("ep %p: %s\n", ep, show_ident(ep->name->ident));
556
557 FOR_EACH_PTR(ep->syms, sym) {
558 if (!sym->pseudo)
559 continue;
560 if (!sym->pseudo->users)
561 continue;
562 printf(" sym: %p %s\n", sym, show_ident(sym->ident));
563 if (sym->ctype.modifiers & (MOD_EXTERN | MOD_STATIC | MOD_ADDRESSABLE))
564 printf("\texternal visibility\n");
565 show_symbol_usage(sym->pseudo);
566 } END_FOR_EACH_PTR(sym);
567
568 printf("\n");
569 }
570
571 FOR_EACH_PTR(ep->bbs, bb) {
572 if (!bb)
573 continue;
574 if (!bb->parents && !bb->children && !bb->insns && verbose < 2)
575 continue;
576 show_bb(bb);
577 printf("\n");
578 } END_FOR_EACH_PTR(bb);
579
580 printf("\n");
581 }
582
583 static void bind_label(struct symbol *label, struct basic_block *bb, struct position pos)
584 {
585 if (label->bb_target)
586 warning(pos, "label '%s' already bound", show_ident(label->ident));
587 label->bb_target = bb;
588 }
589
590 static struct basic_block * get_bound_block(struct entrypoint *ep, struct symbol *label)
591 {
592 struct basic_block *bb = label->bb_target;
593
594 if (!bb) {
595 bb = alloc_basic_block(ep, label->pos);
596 label->bb_target = bb;
597 }
598 return bb;
599 }
600
601 static void finish_block(struct entrypoint *ep)
602 {
603 struct basic_block *src = ep->active;
604 if (bb_reachable(src))
605 ep->active = NULL;
606 }
607
608 static void add_goto(struct entrypoint *ep, struct basic_block *dst)
609 {
610 struct basic_block *src = ep->active;
611 if (bb_reachable(src)) {
612 struct instruction *br = alloc_instruction(OP_BR, 0);
613 br->bb_true = dst;
614 add_bb(&dst->parents, src);
615 add_bb(&src->children, dst);
616 br->bb = src;
617 add_instruction(&src->insns, br);
618 ep->active = NULL;
619 }
620 }
621
622 static void add_one_insn(struct entrypoint *ep, struct instruction *insn)
623 {
624 struct basic_block *bb = ep->active;
625
626 if (bb_reachable(bb)) {
627 insn->bb = bb;
628 add_instruction(&bb->insns, insn);
629 }
630 }
631
632 static void set_activeblock(struct entrypoint *ep, struct basic_block *bb)
633 {
634 if (!bb_terminated(ep->active))
635 add_goto(ep, bb);
636
637 ep->active = bb;
638 if (bb_reachable(bb))
639 add_bb(&ep->bbs, bb);
640 }
641
642 static void remove_parent(struct basic_block *child, struct basic_block *parent)
643 {
644 remove_bb_from_list(&child->parents, parent, 1);
645 if (!child->parents)
646 kill_bb(child);
647 }
648
649 /* Change a "switch" into a branch */
650 void insert_branch(struct basic_block *bb, struct instruction *jmp, struct basic_block *target)
651 {
652 struct instruction *br, *old;
653 struct basic_block *child;
654
655 /* Remove the switch */
656 old = delete_last_instruction(&bb->insns);
657 assert(old == jmp);
658
659 br = alloc_instruction(OP_BR, 0);
660 br->bb = bb;
661 br->bb_true = target;
662 add_instruction(&bb->insns, br);
663
664 FOR_EACH_PTR(bb->children, child) {
665 if (child == target) {
666 target = NULL; /* Trigger just once */
667 continue;
668 }
669 DELETE_CURRENT_PTR(child);
670 remove_parent(child, bb);
671 } END_FOR_EACH_PTR(child);
672 PACK_PTR_LIST(&bb->children);
673 }
674
675
676 void insert_select(struct basic_block *bb, struct instruction *br, struct instruction *phi_node, pseudo_t if_true, pseudo_t if_false)
677 {
678 pseudo_t target;
679 struct instruction *select;
680
681 /* Remove the 'br' */
682 delete_last_instruction(&bb->insns);
683
684 select = alloc_instruction(OP_SEL, phi_node->size);
685 select->bb = bb;
686
687 assert(br->cond);
688 use_pseudo(select, br->cond, &select->src1);
689
690 target = phi_node->target;
691 assert(target->def == phi_node);
692 select->target = target;
693 target->def = select;
694
695 use_pseudo(select, if_true, &select->src2);
696 use_pseudo(select, if_false, &select->src3);
697
698 add_instruction(&bb->insns, select);
699 add_instruction(&bb->insns, br);
700 }
701
702 static inline int bb_empty(struct basic_block *bb)
703 {
704 return !bb->insns;
705 }
706
707 /* Add a label to the currently active block, return new active block */
708 static struct basic_block * add_label(struct entrypoint *ep, struct symbol *label)
709 {
710 struct basic_block *bb = label->bb_target;
711
712 if (bb) {
713 set_activeblock(ep, bb);
714 return bb;
715 }
716 bb = ep->active;
717 if (!bb_reachable(bb) || !bb_empty(bb)) {
718 bb = alloc_basic_block(ep, label->pos);
719 set_activeblock(ep, bb);
720 }
721 label->bb_target = bb;
722 return bb;
723 }
724
725 static void add_branch(struct entrypoint *ep, struct expression *expr, pseudo_t cond, struct basic_block *bb_true, struct basic_block *bb_false)
726 {
727 struct basic_block *bb = ep->active;
728 struct instruction *br;
729
730 if (bb_reachable(bb)) {
731 br = alloc_instruction(OP_CBR, 0);
732 use_pseudo(br, cond, &br->cond);
733 br->bb_true = bb_true;
734 br->bb_false = bb_false;
735 add_bb(&bb_true->parents, bb);
736 add_bb(&bb_false->parents, bb);
737 add_bb(&bb->children, bb_true);
738 add_bb(&bb->children, bb_false);
739 add_one_insn(ep, br);
740 }
741 }
742
743 /* Dummy pseudo allocator */
744 pseudo_t alloc_pseudo(struct instruction *def)
745 {
746 static int nr = 0;
747 struct pseudo * pseudo = __alloc_pseudo(0);
748 pseudo->type = PSEUDO_REG;
749 pseudo->nr = ++nr;
750 pseudo->def = def;
751 return pseudo;
752 }
753
754 static void clear_symbol_pseudos(struct entrypoint *ep)
755 {
756 pseudo_t pseudo;
757
758 FOR_EACH_PTR(ep->accesses, pseudo) {
759 pseudo->sym->pseudo = NULL;
760 } END_FOR_EACH_PTR(pseudo);
761 }
762
763 static pseudo_t symbol_pseudo(struct entrypoint *ep, struct symbol *sym)
764 {
765 pseudo_t pseudo;
766
767 if (!sym)
768 return VOID;
769
770 pseudo = sym->pseudo;
771 if (!pseudo) {
772 pseudo = __alloc_pseudo(0);
773 pseudo->nr = -1;
774 pseudo->type = PSEUDO_SYM;
775 pseudo->sym = sym;
776 pseudo->ident = sym->ident;
777 sym->pseudo = pseudo;
778 add_pseudo(&ep->accesses, pseudo);
779 }
780 /* Symbol pseudos have neither nr, usage nor def */
781 return pseudo;
782 }
783
784 pseudo_t value_pseudo(long long val)
785 {
786 #define MAX_VAL_HASH 64
787 static struct pseudo_list *prev[MAX_VAL_HASH];
788 int hash = val & (MAX_VAL_HASH-1);
789 struct pseudo_list **list = prev + hash;
790 pseudo_t pseudo;
791
792 FOR_EACH_PTR(*list, pseudo) {
793 if (pseudo->value == val)
794 return pseudo;
795 } END_FOR_EACH_PTR(pseudo);
796
797 pseudo = __alloc_pseudo(0);
798 pseudo->type = PSEUDO_VAL;
799 pseudo->value = val;
800 add_pseudo(list, pseudo);
801
802 /* Value pseudos have neither nr, usage nor def */
803 return pseudo;
804 }
805
806 static pseudo_t argument_pseudo(struct entrypoint *ep, int nr)
807 {
808 pseudo_t pseudo = __alloc_pseudo(0);
809 struct instruction *entry = ep->entry;
810
811 pseudo->type = PSEUDO_ARG;
812 pseudo->nr = nr;
813 pseudo->def = entry;
814 add_pseudo(&entry->arg_list, pseudo);
815
816 /* Argument pseudos have neither usage nor def */
817 return pseudo;
818 }
819
820 pseudo_t alloc_phi(struct basic_block *source, pseudo_t pseudo, int size)
821 {
822 struct instruction *insn = alloc_instruction(OP_PHISOURCE, size);
823 pseudo_t phi = __alloc_pseudo(0);
824 static int nr = 0;
825
826 phi->type = PSEUDO_PHI;
827 phi->nr = ++nr;
828 phi->def = insn;
829
830 use_pseudo(insn, pseudo, &insn->phi_src);
831 insn->bb = source;
832 insn->target = phi;
833 add_instruction(&source->insns, insn);
834 return phi;
835 }
836
837 /*
838 * We carry the "access_data" structure around for any accesses,
839 * which simplifies things a lot. It contains all the access
840 * information in one place.
841 */
842 struct access_data {
843 struct symbol *result_type; // result ctype
844 struct symbol *source_type; // source ctype
845 pseudo_t address; // pseudo containing address ..
846 pseudo_t origval; // pseudo for original value ..
847 unsigned int offset, alignment; // byte offset
848 unsigned int bit_size, bit_offset; // which bits
849 struct position pos;
850 };
851
852 static void finish_address_gen(struct entrypoint *ep, struct access_data *ad)
853 {
854 }
855
856 static int linearize_simple_address(struct entrypoint *ep,
857 struct expression *addr,
858 struct access_data *ad)
859 {
860 if (addr->type == EXPR_SYMBOL) {
861 linearize_one_symbol(ep, addr->symbol);
862 ad->address = symbol_pseudo(ep, addr->symbol);
863 return 1;
864 }
865 if (addr->type == EXPR_BINOP) {
866 if (addr->right->type == EXPR_VALUE) {
867 if (addr->op == '+') {
868 ad->offset += get_expression_value(addr->right);
869 return linearize_simple_address(ep, addr->left, ad);
870 }
871 }
872 }
873 ad->address = linearize_expression(ep, addr);
874 return 1;
875 }
876
877 static struct symbol *base_type(struct symbol *sym)
878 {
879 struct symbol *base = sym;
880
881 if (sym) {
882 if (sym->type == SYM_NODE)
883 base = base->ctype.base_type;
884 if (base->type == SYM_BITFIELD)
885 return base->ctype.base_type;
886 }
887 return sym;
888 }
889
890 static int linearize_address_gen(struct entrypoint *ep,
891 struct expression *expr,
892 struct access_data *ad)
893 {
894 struct symbol *ctype = expr->ctype;
895
896 if (!ctype)
897 return 0;
898 ad->pos = expr->pos;
899 ad->result_type = ctype;
900 ad->source_type = base_type(ctype);
901 ad->bit_size = ctype->bit_size;
902 ad->alignment = ctype->ctype.alignment;
903 ad->bit_offset = ctype->bit_offset;
904 if (expr->type == EXPR_PREOP && expr->op == '*')
905 return linearize_simple_address(ep, expr->unop, ad);
906
907 warning(expr->pos, "generating address of non-lvalue (%d)", expr->type);
908 return 0;
909 }
910
911 static pseudo_t add_load(struct entrypoint *ep, struct access_data *ad)
912 {
913 struct instruction *insn;
914 pseudo_t new;
915
916 new = ad->origval;
917 if (0 && new)
918 return new;
919
920 insn = alloc_typed_instruction(OP_LOAD, ad->source_type);
921 new = alloc_pseudo(insn);
922 ad->origval = new;
923
924 insn->target = new;
925 insn->offset = ad->offset;
926 use_pseudo(insn, ad->address, &insn->src);
927 add_one_insn(ep, insn);
928 return new;
929 }
930
931 static void add_store(struct entrypoint *ep, struct access_data *ad, pseudo_t value)
932 {
933 struct basic_block *bb = ep->active;
934
935 if (bb_reachable(bb)) {
936 struct instruction *store = alloc_typed_instruction(OP_STORE, ad->source_type);
937 store->offset = ad->offset;
938 use_pseudo(store, value, &store->target);
939 use_pseudo(store, ad->address, &store->src);
940 add_one_insn(ep, store);
941 }
942 }
943
944 static pseudo_t linearize_store_gen(struct entrypoint *ep,
945 pseudo_t value,
946 struct access_data *ad)
947 {
948 pseudo_t store = value;
949
950 if (type_size(ad->source_type) != type_size(ad->result_type)) {
951 pseudo_t orig = add_load(ep, ad);
952 int shift = ad->bit_offset;
953 unsigned long long mask = (1ULL << ad->bit_size)-1;
954
955 if (shift) {
956 store = add_binary_op(ep, ad->source_type, OP_SHL, value, value_pseudo(shift));
957 mask <<= shift;
958 }
959 orig = add_binary_op(ep, ad->source_type, OP_AND, orig, value_pseudo(~mask));
960 store = add_binary_op(ep, ad->source_type, OP_OR, orig, store);
961 }
962 add_store(ep, ad, store);
963 return value;
964 }
965
966 static pseudo_t add_binary_op(struct entrypoint *ep, struct symbol *ctype, int op, pseudo_t left, pseudo_t right)
967 {
968 struct instruction *insn = alloc_typed_instruction(op, ctype);
969 pseudo_t target = alloc_pseudo(insn);
970 insn->target = target;
971 use_pseudo(insn, left, &insn->src1);
972 use_pseudo(insn, right, &insn->src2);
973 add_one_insn(ep, insn);
974 return target;
975 }
976
977 static pseudo_t add_setval(struct entrypoint *ep, struct symbol *ctype, struct expression *val)
978 {
979 struct instruction *insn = alloc_typed_instruction(OP_SETVAL, ctype);
980 pseudo_t target = alloc_pseudo(insn);
981 insn->target = target;
982 insn->val = val;
983 add_one_insn(ep, insn);
984 return target;
985 }
986
987 static pseudo_t add_symbol_address(struct entrypoint *ep, struct symbol *sym)
988 {
989 struct instruction *insn = alloc_instruction(OP_SYMADDR, bits_in_pointer);
990 pseudo_t target = alloc_pseudo(insn);
991
992 insn->target = target;
993 use_pseudo(insn, symbol_pseudo(ep, sym), &insn->symbol);
994 add_one_insn(ep, insn);
995 return target;
996 }
997
998 static pseudo_t linearize_load_gen(struct entrypoint *ep, struct access_data *ad)
999 {
1000 pseudo_t new = add_load(ep, ad);
1001
1002 if (ad->bit_offset) {
1003 pseudo_t shift = value_pseudo(ad->bit_offset);
1004 pseudo_t newval = add_binary_op(ep, ad->source_type, OP_LSR, new, shift);
1005 new = newval;
1006 }
1007 if (ad->bit_size != type_size(ad->source_type))
1008 new = cast_pseudo(ep, new, ad->source_type, ad->result_type);
1009 return new;
1010 }
1011
1012 static pseudo_t linearize_access(struct entrypoint *ep, struct expression *expr)
1013 {
1014 struct access_data ad = { NULL, };
1015 pseudo_t value;
1016
1017 if (!linearize_address_gen(ep, expr, &ad))
1018 return VOID;
1019 value = linearize_load_gen(ep, &ad);
1020 finish_address_gen(ep, &ad);
1021 return value;
1022 }
1023
1024 /* FIXME: FP */
1025 static pseudo_t linearize_inc_dec(struct entrypoint *ep, struct expression *expr, int postop)
1026 {
1027 struct access_data ad = { NULL, };
1028 pseudo_t old, new, one;
1029 int op = expr->op == SPECIAL_INCREMENT ? OP_ADD : OP_SUB;
1030
1031 if (!linearize_address_gen(ep, expr->unop, &ad))
1032 return VOID;
1033
1034 old = linearize_load_gen(ep, &ad);
1035 one = value_pseudo(expr->op_value);
1036 new = add_binary_op(ep, expr->ctype, op, old, one);
1037 linearize_store_gen(ep, new, &ad);
1038 finish_address_gen(ep, &ad);
1039 return postop ? old : new;
1040 }
1041
1042 static pseudo_t add_uniop(struct entrypoint *ep, struct expression *expr, int op, pseudo_t src)
1043 {
1044 struct instruction *insn = alloc_typed_instruction(op, expr->ctype);
1045 pseudo_t new = alloc_pseudo(insn);
1046
1047 insn->target = new;
1048 use_pseudo(insn, src, &insn->src1);
1049 add_one_insn(ep, insn);
1050 return new;
1051 }
1052
1053 static pseudo_t linearize_slice(struct entrypoint *ep, struct expression *expr)
1054 {
1055 pseudo_t pre = linearize_expression(ep, expr->base);
1056 struct instruction *insn = alloc_typed_instruction(OP_SLICE, expr->ctype);
1057 pseudo_t new = alloc_pseudo(insn);
1058
1059 insn->target = new;
1060 insn->from = expr->r_bitpos;
1061 insn->len = expr->r_nrbits;
1062 use_pseudo(insn, pre, &insn->base);
1063 add_one_insn(ep, insn);
1064 return new;
1065 }
1066
1067 static pseudo_t linearize_regular_preop(struct entrypoint *ep, struct expression *expr)
1068 {
1069 pseudo_t pre = linearize_expression(ep, expr->unop);
1070 switch (expr->op) {
1071 case '+':
1072 return pre;
1073 case '!': {
1074 pseudo_t zero = value_pseudo(0);
1075 return add_binary_op(ep, expr->ctype, OP_SET_EQ, pre, zero);
1076 }
1077 case '~':
1078 return add_uniop(ep, expr, OP_NOT, pre);
1079 case '-':
1080 return add_uniop(ep, expr, OP_NEG, pre);
1081 }
1082 return VOID;
1083 }
1084
1085 static pseudo_t linearize_preop(struct entrypoint *ep, struct expression *expr)
1086 {
1087 /*
1088 * '*' is an lvalue access, and is fundamentally different
1089 * from an arithmetic operation. Maybe it should have an
1090 * expression type of its own..
1091 */
1092 if (expr->op == '*')
1093 return linearize_access(ep, expr);
1094 if (expr->op == SPECIAL_INCREMENT || expr->op == SPECIAL_DECREMENT)
1095 return linearize_inc_dec(ep, expr, 0);
1096 return linearize_regular_preop(ep, expr);
1097 }
1098
1099 static pseudo_t linearize_postop(struct entrypoint *ep, struct expression *expr)
1100 {
1101 return linearize_inc_dec(ep, expr, 1);
1102 }
1103
1104 /*
1105 * Casts to pointers are "less safe" than other casts, since
1106 * they imply type-unsafe accesses. "void *" is a special
1107 * case, since you can't access through it anyway without another
1108 * cast.
1109 */
1110 static struct instruction *alloc_cast_instruction(struct symbol *src, struct symbol *ctype)
1111 {
1112 int opcode = OP_CAST;
1113 struct symbol *base = ctype;
1114
1115 if (src->ctype.modifiers & MOD_SIGNED)
1116 opcode = OP_SCAST;
1117 if (base->type == SYM_NODE)
1118 base = base->ctype.base_type;
1119 if (base->type == SYM_PTR) {
1120 base = base->ctype.base_type;
1121 if (base != &void_ctype)
1122 opcode = OP_PTRCAST;
1123 } else if (base->ctype.base_type == &fp_type)
1124 opcode = OP_FPCAST;
1125 return alloc_typed_instruction(opcode, ctype);
1126 }
1127
1128 static pseudo_t cast_pseudo(struct entrypoint *ep, pseudo_t src, struct symbol *from, struct symbol *to)
1129 {
1130 pseudo_t result;
1131 struct instruction *insn;
1132
1133 if (src == VOID)
1134 return VOID;
1135 if (!from || !to)
1136 return VOID;
1137 if (from->bit_size < 0 || to->bit_size < 0)
1138 return VOID;
1139 insn = alloc_cast_instruction(from, to);
1140 result = alloc_pseudo(insn);
1141 insn->target = result;
1142 insn->orig_type = from;
1143 use_pseudo(insn, src, &insn->src);
1144 add_one_insn(ep, insn);
1145 return result;
1146 }
1147
1148 static int opcode_sign(int opcode, struct symbol *ctype)
1149 {
1150 if (ctype && (ctype->ctype.modifiers & MOD_SIGNED)) {
1151 switch(opcode) {
1152 case OP_MULU: case OP_DIVU: case OP_MODU: case OP_LSR:
1153 opcode++;
1154 }
1155 }
1156 return opcode;
1157 }
1158
1159 static inline pseudo_t add_convert_to_bool(struct entrypoint *ep, pseudo_t src, struct symbol *type)
1160 {
1161 pseudo_t zero;
1162 int op;
1163
1164 if (is_bool_type(type))
1165 return src;
1166 zero = value_pseudo(0);
1167 op = OP_SET_NE;
1168 return add_binary_op(ep, &bool_ctype, op, src, zero);
1169 }
1170
1171 static pseudo_t linearize_expression_to_bool(struct entrypoint *ep, struct expression *expr)
1172 {
1173 pseudo_t dst;
1174 dst = linearize_expression(ep, expr);
1175 dst = add_convert_to_bool(ep, dst, expr->ctype);
1176 return dst;
1177 }
1178
1179 static pseudo_t linearize_assignment(struct entrypoint *ep, struct expression *expr)
1180 {
1181 struct access_data ad = { NULL, };
1182 struct expression *target = expr->left;
1183 struct expression *src = expr->right;
1184 struct symbol *ctype;
1185 pseudo_t value;
1186
1187 value = linearize_expression(ep, src);
1188 if (!target || !linearize_address_gen(ep, target, &ad))
1189 return value;
1190 if (expr->op != '=') {
1191 pseudo_t oldvalue = linearize_load_gen(ep, &ad);
1192 pseudo_t dst;
1193 static const int op_trans[] = {
1194 [SPECIAL_ADD_ASSIGN - SPECIAL_BASE] = OP_ADD,
1195 [SPECIAL_SUB_ASSIGN - SPECIAL_BASE] = OP_SUB,
1196 [SPECIAL_MUL_ASSIGN - SPECIAL_BASE] = OP_MULU,
1197 [SPECIAL_DIV_ASSIGN - SPECIAL_BASE] = OP_DIVU,
1198 [SPECIAL_MOD_ASSIGN - SPECIAL_BASE] = OP_MODU,
1199 [SPECIAL_SHL_ASSIGN - SPECIAL_BASE] = OP_SHL,
1200 [SPECIAL_SHR_ASSIGN - SPECIAL_BASE] = OP_LSR,
1201 [SPECIAL_AND_ASSIGN - SPECIAL_BASE] = OP_AND,
1202 [SPECIAL_OR_ASSIGN - SPECIAL_BASE] = OP_OR,
1203 [SPECIAL_XOR_ASSIGN - SPECIAL_BASE] = OP_XOR
1204 };
1205 int opcode;
1206
1207 if (!src)
1208 return VOID;
1209
1210 ctype = src->ctype;
1211 oldvalue = cast_pseudo(ep, oldvalue, target->ctype, ctype);
1212 opcode = opcode_sign(op_trans[expr->op - SPECIAL_BASE], ctype);
1213 dst = add_binary_op(ep, ctype, opcode, oldvalue, value);
1214 value = cast_pseudo(ep, dst, ctype, expr->ctype);
1215 }
1216 value = linearize_store_gen(ep, value, &ad);
1217 finish_address_gen(ep, &ad);
1218 return value;
1219 }
1220
1221 static pseudo_t linearize_call_expression(struct entrypoint *ep, struct expression *expr)
1222 {
1223 struct expression *arg, *fn;
1224 struct instruction *insn = alloc_typed_instruction(OP_CALL, expr->ctype);
1225 pseudo_t retval, call;
1226 struct ctype *ctype = NULL;
1227 struct symbol *fntype;
1228 struct context *context;
1229
1230 if (!expr->ctype) {
1231 warning(expr->pos, "call with no type!");
1232 return VOID;
1233 }
1234
1235 FOR_EACH_PTR(expr->args, arg) {
1236 pseudo_t new = linearize_expression(ep, arg);
1237 use_pseudo(insn, new, add_pseudo(&insn->arguments, new));
1238 } END_FOR_EACH_PTR(arg);
1239
1240 fn = expr->fn;
1241
1242 if (fn->ctype)
1243 ctype = &fn->ctype->ctype;
1244
1245 fntype = fn->ctype;
1246 if (fntype) {
1247 if (fntype->type == SYM_NODE)
1248 fntype = fntype->ctype.base_type;
1249 }
1250 insn->fntype = fntype;
1251
1252 if (fn->type == EXPR_PREOP) {
1253 if (fn->unop->type == EXPR_SYMBOL) {
1254 struct symbol *sym = fn->unop->symbol;
1255 if (sym->ctype.base_type->type == SYM_FN)
1256 fn = fn->unop;
1257 }
1258 }
1259 if (fn->type == EXPR_SYMBOL) {
1260 call = symbol_pseudo(ep, fn->symbol);
1261 } else {
1262 call = linearize_expression(ep, fn);
1263 }
1264 use_pseudo(insn, call, &insn->func);
1265 retval = VOID;
1266 if (expr->ctype != &void_ctype)
1267 retval = alloc_pseudo(insn);
1268 insn->target = retval;
1269 add_one_insn(ep, insn);
1270
1271 if (ctype) {
1272 FOR_EACH_PTR(ctype->contexts, context) {
1273 int in = context->in;
1274 int out = context->out;
1275 int check = 0;
1276 int context_diff;
1277 if (in < 0) {
1278 check = 1;
1279 in = 0;
1280 }
1281 if (out < 0) {
1282 check = 0;
1283 out = 0;
1284 }
1285 context_diff = out - in;
1286 if (check || context_diff) {
1287 insn = alloc_instruction(OP_CONTEXT, 0);
1288 insn->increment = context_diff;
1289 insn->check = check;
1290 insn->context_expr = context->context;
1291 add_one_insn(ep, insn);
1292 }
1293 } END_FOR_EACH_PTR(context);
1294 }
1295
1296 return retval;
1297 }
1298
1299 static pseudo_t linearize_binop_bool(struct entrypoint *ep, struct expression *expr)
1300 {
1301 pseudo_t src1, src2, dst;
1302 int op = (expr->op == SPECIAL_LOGICAL_OR) ? OP_OR_BOOL : OP_AND_BOOL;
1303
1304 src1 = linearize_expression_to_bool(ep, expr->left);
1305 src2 = linearize_expression_to_bool(ep, expr->right);
1306 dst = add_binary_op(ep, &bool_ctype, op, src1, src2);
1307 if (expr->ctype != &bool_ctype)
1308 dst = cast_pseudo(ep, dst, &bool_ctype, expr->ctype);
1309 return dst;
1310 }
1311
1312 static pseudo_t linearize_binop(struct entrypoint *ep, struct expression *expr)
1313 {
1314 pseudo_t src1, src2, dst;
1315 static const int opcode[] = {
1316 ['+'] = OP_ADD, ['-'] = OP_SUB,
1317 ['*'] = OP_MULU, ['/'] = OP_DIVU,
1318 ['%'] = OP_MODU, ['&'] = OP_AND,
1319 ['|'] = OP_OR, ['^'] = OP_XOR,
1320 [SPECIAL_LEFTSHIFT] = OP_SHL,
1321 [SPECIAL_RIGHTSHIFT] = OP_LSR,
1322 };
1323 int op;
1324
1325 src1 = linearize_expression(ep, expr->left);
1326 src2 = linearize_expression(ep, expr->right);
1327 op = opcode_sign(opcode[expr->op], expr->ctype);
1328 dst = add_binary_op(ep, expr->ctype, op, src1, src2);
1329 return dst;
1330 }
1331
1332 static pseudo_t linearize_logical_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false);
1333
1334 pseudo_t linearize_cond_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false);
1335
1336 static pseudo_t linearize_select(struct entrypoint *ep, struct expression *expr)
1337 {
1338 pseudo_t cond, true, false, res;
1339 struct instruction *insn;
1340
1341 true = linearize_expression(ep, expr->cond_true);
1342 false = linearize_expression(ep, expr->cond_false);
1343 cond = linearize_expression(ep, expr->conditional);
1344
1345 insn = alloc_typed_instruction(OP_SEL, expr->ctype);
1346 if (!expr->cond_true)
1347 true = cond;
1348 use_pseudo(insn, cond, &insn->src1);
1349 use_pseudo(insn, true, &insn->src2);
1350 use_pseudo(insn, false, &insn->src3);
1351
1352 res = alloc_pseudo(insn);
1353 insn->target = res;
1354 add_one_insn(ep, insn);
1355 return res;
1356 }
1357
1358 static pseudo_t add_join_conditional(struct entrypoint *ep, struct expression *expr,
1359 pseudo_t phi1, pseudo_t phi2)
1360 {
1361 pseudo_t target;
1362 struct instruction *phi_node;
1363
1364 if (phi1 == VOID)
1365 return phi2;
1366 if (phi2 == VOID)
1367 return phi1;
1368
1369 phi_node = alloc_typed_instruction(OP_PHI, expr->ctype);
1370 use_pseudo(phi_node, phi1, add_pseudo(&phi_node->phi_list, phi1));
1371 use_pseudo(phi_node, phi2, add_pseudo(&phi_node->phi_list, phi2));
1372 phi_node->target = target = alloc_pseudo(phi_node);
1373 add_one_insn(ep, phi_node);
1374 return target;
1375 }
1376
1377 static pseudo_t linearize_short_conditional(struct entrypoint *ep, struct expression *expr,
1378 struct expression *cond,
1379 struct expression *expr_false)
1380 {
1381 pseudo_t src1, src2;
1382 struct basic_block *bb_false;
1383 struct basic_block *merge = alloc_basic_block(ep, expr->pos);
1384 pseudo_t phi1, phi2;
1385 int size = type_size(expr->ctype);
1386
1387 if (!expr_false || !ep->active)
1388 return VOID;
1389
1390 bb_false = alloc_basic_block(ep, expr_false->pos);
1391 src1 = linearize_expression(ep, cond);
1392 phi1 = alloc_phi(ep->active, src1, size);
1393 add_branch(ep, expr, src1, merge, bb_false);
1394
1395 set_activeblock(ep, bb_false);
1396 src2 = linearize_expression(ep, expr_false);
1397 phi2 = alloc_phi(ep->active, src2, size);
1398 set_activeblock(ep, merge);
1399
1400 return add_join_conditional(ep, expr, phi1, phi2);
1401 }
1402
1403 static pseudo_t linearize_conditional(struct entrypoint *ep, struct expression *expr,
1404 struct expression *cond,
1405 struct expression *expr_true,
1406 struct expression *expr_false)
1407 {
1408 pseudo_t src1, src2;
1409 pseudo_t phi1, phi2;
1410 struct basic_block *bb_true, *bb_false, *merge;
1411 int size = type_size(expr->ctype);
1412
1413 if (!cond || !expr_true || !expr_false || !ep->active)
1414 return VOID;
1415 bb_true = alloc_basic_block(ep, expr_true->pos);
1416 bb_false = alloc_basic_block(ep, expr_false->pos);
1417 merge = alloc_basic_block(ep, expr->pos);
1418
1419 linearize_cond_branch(ep, cond, bb_true, bb_false);
1420
1421 set_activeblock(ep, bb_true);
1422 src1 = linearize_expression(ep, expr_true);
1423 phi1 = alloc_phi(ep->active, src1, size);
1424 add_goto(ep, merge);
1425
1426 set_activeblock(ep, bb_false);
1427 src2 = linearize_expression(ep, expr_false);
1428 phi2 = alloc_phi(ep->active, src2, size);
1429 set_activeblock(ep, merge);
1430
1431 return add_join_conditional(ep, expr, phi1, phi2);
1432 }
1433
1434 static pseudo_t linearize_logical(struct entrypoint *ep, struct expression *expr)
1435 {
1436 struct expression *shortcut;
1437
1438 shortcut = alloc_const_expression(expr->pos, expr->op == SPECIAL_LOGICAL_OR);
1439 shortcut->ctype = expr->ctype;
1440 if (expr->op == SPECIAL_LOGICAL_OR)
1441 return linearize_conditional(ep, expr, expr->left, shortcut, expr->right);
1442 return linearize_conditional(ep, expr, expr->left, expr->right, shortcut);
1443 }
1444
1445 static pseudo_t linearize_compare(struct entrypoint *ep, struct expression *expr)
1446 {
1447 static const int cmpop[] = {
1448 ['>'] = OP_SET_GT, ['<'] = OP_SET_LT,
1449 [SPECIAL_EQUAL] = OP_SET_EQ,
1450 [SPECIAL_NOTEQUAL] = OP_SET_NE,
1451 [SPECIAL_GTE] = OP_SET_GE,
1452 [SPECIAL_LTE] = OP_SET_LE,
1453 [SPECIAL_UNSIGNED_LT] = OP_SET_B,
1454 [SPECIAL_UNSIGNED_GT] = OP_SET_A,
1455 [SPECIAL_UNSIGNED_LTE] = OP_SET_BE,
1456 [SPECIAL_UNSIGNED_GTE] = OP_SET_AE,
1457 };
1458
1459 pseudo_t src1 = linearize_expression(ep, expr->left);
1460 pseudo_t src2 = linearize_expression(ep, expr->right);
1461 pseudo_t dst = add_binary_op(ep, expr->ctype, cmpop[expr->op], src1, src2);
1462 return dst;
1463 }
1464
1465
1466 pseudo_t linearize_cond_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false)
1467 {
1468 pseudo_t cond;
1469
1470 if (!expr || !bb_reachable(ep->active))
1471 return VOID;
1472
1473 switch (expr->type) {
1474
1475 case EXPR_STRING:
1476 case EXPR_VALUE:
1477 add_goto(ep, expr->value ? bb_true : bb_false);
1478 return VOID;
1479
1480 case EXPR_FVALUE:
1481 add_goto(ep, expr->fvalue ? bb_true : bb_false);
1482 return VOID;
1483
1484 case EXPR_LOGICAL:
1485 linearize_logical_branch(ep, expr, bb_true, bb_false);
1486 return VOID;
1487
1488 case EXPR_COMPARE:
1489 cond = linearize_compare(ep, expr);
1490 add_branch(ep, expr, cond, bb_true, bb_false);
1491 break;
1492
1493 case EXPR_PREOP:
1494 if (expr->op == '!')
1495 return linearize_cond_branch(ep, expr->unop, bb_false, bb_true);
1496 /* fall through */
1497 default: {
1498 cond = linearize_expression(ep, expr);
1499 add_branch(ep, expr, cond, bb_true, bb_false);
1500
1501 return VOID;
1502 }
1503 }
1504 return VOID;
1505 }
1506
1507
1508
1509 static pseudo_t linearize_logical_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false)
1510 {
1511 struct basic_block *next = alloc_basic_block(ep, expr->pos);
1512
1513 if (expr->op == SPECIAL_LOGICAL_OR)
1514 linearize_cond_branch(ep, expr->left, bb_true, next);
1515 else
1516 linearize_cond_branch(ep, expr->left, next, bb_false);
1517 set_activeblock(ep, next);
1518 linearize_cond_branch(ep, expr->right, bb_true, bb_false);
1519 return VOID;
1520 }
1521
1522 static pseudo_t linearize_cast(struct entrypoint *ep, struct expression *expr)
1523 {
1524 pseudo_t src;
1525 struct expression *orig = expr->cast_expression;
1526
1527 if (!orig)
1528 return VOID;
1529
1530 src = linearize_expression(ep, orig);
1531 return cast_pseudo(ep, src, orig->ctype, expr->ctype);
1532 }
1533
1534 static pseudo_t linearize_position(struct entrypoint *ep, struct expression *pos, struct access_data *ad)
1535 {
1536 struct expression *init_expr = pos->init_expr;
1537
1538 ad->offset = pos->init_offset;
1539 ad->source_type = base_type(init_expr->ctype);
1540 ad->result_type = init_expr->ctype;
1541 return linearize_initializer(ep, init_expr, ad);
1542 }
1543
1544 static pseudo_t linearize_initializer(struct entrypoint *ep, struct expression *initializer, struct access_data *ad)
1545 {
1546 switch (initializer->type) {
1547 case EXPR_INITIALIZER: {
1548 struct expression *expr;
1549 FOR_EACH_PTR(initializer->expr_list, expr) {
1550 linearize_initializer(ep, expr, ad);
1551 } END_FOR_EACH_PTR(expr);
1552 break;
1553 }
1554 case EXPR_POS:
1555 linearize_position(ep, initializer, ad);
1556 break;
1557 default: {
1558 pseudo_t value = linearize_expression(ep, initializer);
1559 ad->source_type = base_type(initializer->ctype);
1560 ad->result_type = initializer->ctype;
1561 linearize_store_gen(ep, value, ad);
1562 return value;
1563 }
1564 }
1565
1566 return VOID;
1567 }
1568
1569 static void linearize_argument(struct entrypoint *ep, struct symbol *arg, int nr)
1570 {
1571 struct access_data ad = { NULL, };
1572
1573 ad.source_type = arg;
1574 ad.result_type = arg;
1575 ad.address = symbol_pseudo(ep, arg);
1576 linearize_store_gen(ep, argument_pseudo(ep, nr), &ad);
1577 finish_address_gen(ep, &ad);
1578 }
1579
1580 pseudo_t linearize_expression(struct entrypoint *ep, struct expression *expr)
1581 {
1582 if (!expr)
1583 return VOID;
1584
1585 current_pos = expr->pos;
1586 switch (expr->type) {
1587 case EXPR_SYMBOL:
1588 linearize_one_symbol(ep, expr->symbol);
1589 return add_symbol_address(ep, expr->symbol);
1590
1591 case EXPR_VALUE:
1592 return value_pseudo(expr->value);
1593
1594 case EXPR_STRING: case EXPR_FVALUE: case EXPR_LABEL:
1595 return add_setval(ep, expr->ctype, expr);
1596
1597 case EXPR_STATEMENT:
1598 return linearize_statement(ep, expr->statement);
1599
1600 case EXPR_CALL:
1601 return linearize_call_expression(ep, expr);
1602
1603 case EXPR_BINOP:
1604 if (expr->op == SPECIAL_LOGICAL_AND || expr->op == SPECIAL_LOGICAL_OR)
1605 return linearize_binop_bool(ep, expr);
1606 return linearize_binop(ep, expr);
1607
1608 case EXPR_LOGICAL:
1609 return linearize_logical(ep, expr);
1610
1611 case EXPR_COMPARE:
1612 return linearize_compare(ep, expr);
1613
1614 case EXPR_SELECT:
1615 return linearize_select(ep, expr);
1616
1617 case EXPR_CONDITIONAL:
1618 if (!expr->cond_true)
1619 return linearize_short_conditional(ep, expr, expr->conditional, expr->cond_false);
1620
1621 return linearize_conditional(ep, expr, expr->conditional,
1622 expr->cond_true, expr->cond_false);
1623
1624 case EXPR_COMMA:
1625 linearize_expression(ep, expr->left);
1626 return linearize_expression(ep, expr->right);
1627
1628 case EXPR_ASSIGNMENT:
1629 return linearize_assignment(ep, expr);
1630
1631 case EXPR_PREOP:
1632 return linearize_preop(ep, expr);
1633
1634 case EXPR_POSTOP:
1635 return linearize_postop(ep, expr);
1636
1637 case EXPR_CAST:
1638 case EXPR_FORCE_CAST:
1639 case EXPR_IMPLIED_CAST:
1640 return linearize_cast(ep, expr);
1641
1642 case EXPR_SLICE:
1643 return linearize_slice(ep, expr);
1644
1645 case EXPR_INITIALIZER:
1646 case EXPR_POS:
1647 warning(expr->pos, "unexpected initializer expression (%d %d)", expr->type, expr->op);
1648 return VOID;
1649 default:
1650 warning(expr->pos, "unknown expression (%d %d)", expr->type, expr->op);
1651 return VOID;
1652 }
1653 return VOID;
1654 }
1655
1656 static pseudo_t linearize_one_symbol(struct entrypoint *ep, struct symbol *sym)
1657 {
1658 struct access_data ad = { NULL, };
1659 pseudo_t value;
1660
1661 if (!sym || !sym->initializer || sym->initialized)
1662 return VOID;
1663
1664 /* We need to output these puppies some day too.. */
1665 if (sym->ctype.modifiers & (MOD_STATIC | MOD_TOPLEVEL))
1666 return VOID;
1667
1668 sym->initialized = 1;
1669 ad.address = symbol_pseudo(ep, sym);
1670 value = linearize_initializer(ep, sym->initializer, &ad);
1671 finish_address_gen(ep, &ad);
1672 return value;
1673 }
1674
1675 static pseudo_t linearize_compound_statement(struct entrypoint *ep, struct statement *stmt)
1676 {
1677 pseudo_t pseudo;
1678 struct statement *s;
1679 struct symbol *ret = stmt->ret;
1680
1681 pseudo = VOID;
1682 FOR_EACH_PTR(stmt->stmts, s) {
1683 pseudo = linearize_statement(ep, s);
1684 } END_FOR_EACH_PTR(s);
1685
1686 if (ret) {
1687 struct basic_block *bb = add_label(ep, ret);
1688 struct instruction *phi_node = first_instruction(bb->insns);
1689
1690 if (!phi_node)
1691 return pseudo;
1692
1693 if (pseudo_list_size(phi_node->phi_list)==1) {
1694 pseudo = first_pseudo(phi_node->phi_list);
1695 assert(pseudo->type == PSEUDO_PHI);
1696 return pseudo->def->src1;
1697 }
1698 return phi_node->target;
1699 }
1700
1701 return pseudo;
1702 }
1703
1704 static pseudo_t linearize_inlined_call(struct entrypoint *ep, struct statement *stmt)
1705 {
1706 struct instruction *insn = alloc_instruction(OP_INLINED_CALL, 0);
1707 struct statement *args = stmt->args;
1708 struct basic_block *bb;
1709 pseudo_t pseudo;
1710
1711 if (args) {
1712 struct symbol *sym;
1713
1714 concat_symbol_list(args->declaration, &ep->syms);
1715 FOR_EACH_PTR(args->declaration, sym) {
1716 pseudo_t value = linearize_one_symbol(ep, sym);
1717 use_pseudo(insn, value, add_pseudo(&insn->arguments, value));
1718 } END_FOR_EACH_PTR(sym);
1719 }
1720
1721 insn->target = pseudo = linearize_compound_statement(ep, stmt);
1722 use_pseudo(insn, symbol_pseudo(ep, stmt->inline_fn), &insn->func);
1723 bb = ep->active;
1724 if (bb && !bb->insns)
1725 bb->pos = stmt->pos;
1726 add_one_insn(ep, insn);
1727 return pseudo;
1728 }
1729
1730 static pseudo_t linearize_context(struct entrypoint *ep, struct statement *stmt)
1731 {
1732 struct instruction *insn = alloc_instruction(OP_CONTEXT, 0);
1733 struct expression *expr = stmt->expression;
1734 int value = 0;
1735
1736 if (expr->type == EXPR_VALUE)
1737 value = expr->value;
1738
1739 insn->increment = value;
1740 insn->context_expr = stmt->context;
1741 add_one_insn(ep, insn);
1742 return VOID;
1743 }
1744
1745 static pseudo_t linearize_range(struct entrypoint *ep, struct statement *stmt)
1746 {
1747 struct instruction *insn = alloc_instruction(OP_RANGE, 0);
1748
1749 use_pseudo(insn, linearize_expression(ep, stmt->range_expression), &insn->src1);
1750 use_pseudo(insn, linearize_expression(ep, stmt->range_low), &insn->src2);
1751 use_pseudo(insn, linearize_expression(ep, stmt->range_high), &insn->src3);
1752 add_one_insn(ep, insn);
1753 return VOID;
1754 }
1755
1756 ALLOCATOR(asm_rules, "asm rules");
1757 ALLOCATOR(asm_constraint, "asm constraints");
1758
1759 static void add_asm_input(struct entrypoint *ep, struct instruction *insn, struct expression *expr,
1760 const char *constraint, const struct ident *ident)
1761 {
1762 pseudo_t pseudo = linearize_expression(ep, expr);
1763 struct asm_constraint *rule = __alloc_asm_constraint(0);
1764
1765 rule->ident = ident;
1766 rule->constraint = constraint;
1767 use_pseudo(insn, pseudo, &rule->pseudo);
1768 add_ptr_list(&insn->asm_rules->inputs, rule);
1769 }
1770
1771 static void add_asm_output(struct entrypoint *ep, struct instruction *insn, struct expression *expr,
1772 const char *constraint, const struct ident *ident)
1773 {
1774 struct access_data ad = { NULL, };
1775 pseudo_t pseudo = alloc_pseudo(insn);
1776 struct asm_constraint *rule;
1777
1778 if (!expr || !linearize_address_gen(ep, expr, &ad))
1779 return;
1780 linearize_store_gen(ep, pseudo, &ad);
1781 finish_address_gen(ep, &ad);
1782 rule = __alloc_asm_constraint(0);
1783 rule->ident = ident;
1784 rule->constraint = constraint;
1785 use_pseudo(insn, pseudo, &rule->pseudo);
1786 add_ptr_list(&insn->asm_rules->outputs, rule);
1787 }
1788
1789 static pseudo_t linearize_asm_statement(struct entrypoint *ep, struct statement *stmt)
1790 {
1791 int state;
1792 struct expression *expr;
1793 struct instruction *insn;
1794 struct asm_rules *rules;
1795 const char *constraint;
1796 struct ident *ident;
1797
1798 insn = alloc_instruction(OP_ASM, 0);
1799 expr = stmt->asm_string;
1800 if (!expr || expr->type != EXPR_STRING) {
1801 warning(stmt->pos, "expected string in inline asm");
1802 return VOID;
1803 }
1804 insn->string = expr->string->data;
1805
1806 rules = __alloc_asm_rules(0);
1807 insn->asm_rules = rules;
1808
1809 /* Gather the inputs.. */
1810 state = 0;
1811 ident = NULL;
1812 constraint = NULL;
1813 FOR_EACH_PTR(stmt->asm_inputs, expr) {
1814 switch (state) {
1815 case 0: /* Identifier */
1816 state = 1;
1817 ident = (struct ident *)expr;
1818 continue;
1819
1820 case 1: /* Constraint */
1821 state = 2;
1822 constraint = expr ? expr->string->data : "";
1823 continue;
1824
1825 case 2: /* Expression */
1826 state = 0;
1827 add_asm_input(ep, insn, expr, constraint, ident);
1828 }
1829 } END_FOR_EACH_PTR(expr);
1830
1831 add_one_insn(ep, insn);
1832
1833 /* Assign the outputs */
1834 state = 0;
1835 ident = NULL;
1836 constraint = NULL;
1837 FOR_EACH_PTR(stmt->asm_outputs, expr) {
1838 switch (state) {
1839 case 0: /* Identifier */
1840 state = 1;
1841 ident = (struct ident *)expr;
1842 continue;
1843
1844 case 1: /* Constraint */
1845 state = 2;
1846 constraint = expr ? expr->string->data : "";
1847 continue;
1848
1849 case 2:
1850 state = 0;
1851 add_asm_output(ep, insn, expr, constraint, ident);
1852 }
1853 } END_FOR_EACH_PTR(expr);
1854
1855 return VOID;
1856 }
1857
1858 static int multijmp_cmp(const void *_a, const void *_b)
1859 {
1860 const struct multijmp *a = _a;
1861 const struct multijmp *b = _b;
1862
1863 // "default" case?
1864 if (a->begin > a->end) {
1865 if (b->begin > b->end)
1866 return 0;
1867 return 1;
1868 }
1869 if (b->begin > b->end)
1870 return -1;
1871 if (a->begin == b->begin) {
1872 if (a->end == b->end)
1873 return 0;
1874 return (a->end < b->end) ? -1 : 1;
1875 }
1876 return a->begin < b->begin ? -1 : 1;
1877 }
1878
1879 static void sort_switch_cases(struct instruction *insn)
1880 {
1881 sort_list((struct ptr_list **)&insn->multijmp_list, multijmp_cmp);
1882 }
1883
1884 static pseudo_t linearize_declaration(struct entrypoint *ep, struct statement *stmt)
1885 {
1886 struct symbol *sym;
1887
1888 concat_symbol_list(stmt->declaration, &ep->syms);
1889
1890 FOR_EACH_PTR(stmt->declaration, sym) {
1891 linearize_one_symbol(ep, sym);
1892 } END_FOR_EACH_PTR(sym);
1893 return VOID;
1894 }
1895
1896 static pseudo_t linearize_return(struct entrypoint *ep, struct statement *stmt)
1897 {
1898 struct expression *expr = stmt->expression;
1899 struct basic_block *bb_return = get_bound_block(ep, stmt->ret_target);
1900 struct basic_block *active;
1901 pseudo_t src = linearize_expression(ep, expr);
1902 active = ep->active;
1903 if (active && src != VOID) {
1904 struct instruction *phi_node = first_instruction(bb_return->insns);
1905 pseudo_t phi;
1906 if (!phi_node) {
1907 phi_node = alloc_typed_instruction(OP_PHI, expr->ctype);
1908 phi_node->target = alloc_pseudo(phi_node);
1909 phi_node->bb = bb_return;
1910 add_instruction(&bb_return->insns, phi_node);
1911 }
1912 phi = alloc_phi(active, src, type_size(expr->ctype));
1913 phi->ident = &return_ident;
1914 use_pseudo(phi_node, phi, add_pseudo(&phi_node->phi_list, phi));
1915 }
1916 add_goto(ep, bb_return);
1917 return VOID;
1918 }
1919
1920 static pseudo_t linearize_switch(struct entrypoint *ep, struct statement *stmt)
1921 {
1922 struct symbol *sym;
1923 struct instruction *switch_ins;
1924 struct basic_block *switch_end = alloc_basic_block(ep, stmt->pos);
1925 struct basic_block *active, *default_case;
1926 struct multijmp *jmp;
1927 pseudo_t pseudo;
1928
1929 pseudo = linearize_expression(ep, stmt->switch_expression);
1930
1931 active = ep->active;
1932 if (!bb_reachable(active))
1933 return VOID;
1934
1935 switch_ins = alloc_instruction(OP_SWITCH, 0);
1936 use_pseudo(switch_ins, pseudo, &switch_ins->cond);
1937 add_one_insn(ep, switch_ins);
1938 finish_block(ep);
1939
1940 default_case = NULL;
1941 FOR_EACH_PTR(stmt->switch_case->symbol_list, sym) {
1942 struct statement *case_stmt = sym->stmt;
1943 struct basic_block *bb_case = get_bound_block(ep, sym);
1944
1945 if (!case_stmt->case_expression) {
1946 default_case = bb_case;
1947 continue;
1948 } else {
1949 int begin, end;
1950
1951 begin = end = case_stmt->case_expression->value;
1952 if (case_stmt->case_to)
1953 end = case_stmt->case_to->value;
1954 if (begin > end)
1955 jmp = alloc_multijmp(bb_case, end, begin);
1956 else
1957 jmp = alloc_multijmp(bb_case, begin, end);
1958
1959 }
1960 add_multijmp(&switch_ins->multijmp_list, jmp);
1961 add_bb(&bb_case->parents, active);
1962 add_bb(&active->children, bb_case);
1963 } END_FOR_EACH_PTR(sym);
1964
1965 bind_label(stmt->switch_break, switch_end, stmt->pos);
1966
1967 /* And linearize the actual statement */
1968 linearize_statement(ep, stmt->switch_statement);
1969 set_activeblock(ep, switch_end);
1970
1971 if (!default_case)
1972 default_case = switch_end;
1973
1974 jmp = alloc_multijmp(default_case, 1, 0);
1975 add_multijmp(&switch_ins->multijmp_list, jmp);
1976 add_bb(&default_case->parents, active);
1977 add_bb(&active->children, default_case);
1978 sort_switch_cases(switch_ins);
1979
1980 return VOID;
1981 }
1982
1983 static pseudo_t linearize_iterator(struct entrypoint *ep, struct statement *stmt)
1984 {
1985 struct statement *pre_statement = stmt->iterator_pre_statement;
1986 struct expression *pre_condition = stmt->iterator_pre_condition;
1987 struct statement *statement = stmt->iterator_statement;
1988 struct statement *post_statement = stmt->iterator_post_statement;
1989 struct expression *post_condition = stmt->iterator_post_condition;
1990 struct basic_block *loop_top, *loop_body, *loop_continue, *loop_end;
1991 struct symbol *sym;
1992
1993 FOR_EACH_PTR(stmt->iterator_syms, sym) {
1994 linearize_one_symbol(ep, sym);
1995 } END_FOR_EACH_PTR(sym);
1996 concat_symbol_list(stmt->iterator_syms, &ep->syms);
1997 linearize_statement(ep, pre_statement);
1998
1999 loop_body = loop_top = alloc_basic_block(ep, stmt->pos);
2000 loop_continue = alloc_basic_block(ep, stmt->pos);
2001 loop_end = alloc_basic_block(ep, stmt->pos);
2002
2003 /* An empty post-condition means that it's the same as the pre-condition */
2004 if (!post_condition) {
2005 loop_top = alloc_basic_block(ep, stmt->pos);
2006 set_activeblock(ep, loop_top);
2007 }
2008
2009 if (pre_condition)
2010 linearize_cond_branch(ep, pre_condition, loop_body, loop_end);
2011
2012 bind_label(stmt->iterator_continue, loop_continue, stmt->pos);
2013 bind_label(stmt->iterator_break, loop_end, stmt->pos);
2014
2015 set_activeblock(ep, loop_body);
2016 linearize_statement(ep, statement);
2017 add_goto(ep, loop_continue);
2018
2019 set_activeblock(ep, loop_continue);
2020 linearize_statement(ep, post_statement);
2021 if (!post_condition)
2022 add_goto(ep, loop_top);
2023 else
2024 linearize_cond_branch(ep, post_condition, loop_top, loop_end);
2025 set_activeblock(ep, loop_end);
2026
2027 return VOID;
2028 }
2029
2030 pseudo_t linearize_statement(struct entrypoint *ep, struct statement *stmt)
2031 {
2032 struct basic_block *bb;
2033
2034 if (!stmt)
2035 return VOID;
2036
2037 bb = ep->active;
2038 if (bb && !bb->insns)
2039 bb->pos = stmt->pos;
2040 current_pos = stmt->pos;
2041
2042 switch (stmt->type) {
2043 case STMT_NONE:
2044 break;
2045
2046 case STMT_DECLARATION:
2047 return linearize_declaration(ep, stmt);
2048
2049 case STMT_CONTEXT:
2050 return linearize_context(ep, stmt);
2051
2052 case STMT_RANGE:
2053 return linearize_range(ep, stmt);
2054
2055 case STMT_EXPRESSION:
2056 return linearize_expression(ep, stmt->expression);
2057
2058 case STMT_ASM:
2059 return linearize_asm_statement(ep, stmt);
2060
2061 case STMT_RETURN:
2062 return linearize_return(ep, stmt);
2063
2064 case STMT_CASE: {
2065 add_label(ep, stmt->case_label);
2066 linearize_statement(ep, stmt->case_statement);
2067 break;
2068 }
2069
2070 case STMT_LABEL: {
2071 struct symbol *label = stmt->label_identifier;
2072
2073 if (label->used) {
2074 add_label(ep, label);
2075 }
2076 return linearize_statement(ep, stmt->label_statement);
2077 }
2078
2079 case STMT_GOTO: {
2080 struct symbol *sym;
2081 struct expression *expr;
2082 struct instruction *goto_ins;
2083 struct basic_block *active;
2084 pseudo_t pseudo;
2085
2086 active = ep->active;
2087 if (!bb_reachable(active))
2088 break;
2089
2090 if (stmt->goto_label) {
2091 add_goto(ep, get_bound_block(ep, stmt->goto_label));
2092 break;
2093 }
2094
2095 expr = stmt->goto_expression;
2096 if (!expr)
2097 break;
2098
2099 /* This can happen as part of simplification */
2100 if (expr->type == EXPR_LABEL) {
2101 add_goto(ep, get_bound_block(ep, expr->label_symbol));
2102 break;
2103 }
2104
2105 pseudo = linearize_expression(ep, expr);
2106 goto_ins = alloc_instruction(OP_COMPUTEDGOTO, 0);
2107 use_pseudo(goto_ins, pseudo, &goto_ins->target);
2108 add_one_insn(ep, goto_ins);
2109
2110 FOR_EACH_PTR(stmt->target_list, sym) {
2111 struct basic_block *bb_computed = get_bound_block(ep, sym);
2112 struct multijmp *jmp = alloc_multijmp(bb_computed, 1, 0);
2113 add_multijmp(&goto_ins->multijmp_list, jmp);
2114 add_bb(&bb_computed->parents, ep->active);
2115 add_bb(&active->children, bb_computed);
2116 } END_FOR_EACH_PTR(sym);
2117
2118 finish_block(ep);
2119 break;
2120 }
2121
2122 case STMT_COMPOUND:
2123 if (stmt->inline_fn)
2124 return linearize_inlined_call(ep, stmt);
2125 return linearize_compound_statement(ep, stmt);
2126
2127 /*
2128 * This could take 'likely/unlikely' into account, and
2129 * switch the arms around appropriately..
2130 */
2131 case STMT_IF: {
2132 struct basic_block *bb_true, *bb_false, *endif;
2133 struct expression *cond = stmt->if_conditional;
2134
2135 bb_true = alloc_basic_block(ep, stmt->pos);
2136 bb_false = endif = alloc_basic_block(ep, stmt->pos);
2137
2138 linearize_cond_branch(ep, cond, bb_true, bb_false);
2139
2140 set_activeblock(ep, bb_true);
2141 linearize_statement(ep, stmt->if_true);
2142
2143 if (stmt->if_false) {
2144 endif = alloc_basic_block(ep, stmt->pos);
2145 add_goto(ep, endif);
2146 set_activeblock(ep, bb_false);
2147 linearize_statement(ep, stmt->if_false);
2148 }
2149 set_activeblock(ep, endif);
2150 break;
2151 }
2152
2153 case STMT_SWITCH:
2154 return linearize_switch(ep, stmt);
2155
2156 case STMT_ITERATOR:
2157 return linearize_iterator(ep, stmt);
2158
2159 default:
2160 break;
2161 }
2162 return VOID;
2163 }
2164
2165 static struct entrypoint *linearize_fn(struct symbol *sym, struct symbol *base_type)
2166 {
2167 struct entrypoint *ep;
2168 struct basic_block *bb;
2169 struct symbol *arg;
2170 struct instruction *entry;
2171 pseudo_t result;
2172 int i;
2173
2174 if (!base_type->stmt)
2175 return NULL;
2176
2177 ep = alloc_entrypoint();
2178 bb = alloc_basic_block(ep, sym->pos);
2179
2180 ep->name = sym;
2181 sym->ep = ep;
2182 set_activeblock(ep, bb);
2183
2184 entry = alloc_instruction(OP_ENTRY, 0);
2185 add_one_insn(ep, entry);
2186 ep->entry = entry;
2187
2188 concat_symbol_list(base_type->arguments, &ep->syms);
2189
2190 /* FIXME!! We should do something else about varargs.. */
2191 i = 0;
2192 FOR_EACH_PTR(base_type->arguments, arg) {
2193 linearize_argument(ep, arg, ++i);
2194 } END_FOR_EACH_PTR(arg);
2195
2196 result = linearize_statement(ep, base_type->stmt);
2197 if (bb_reachable(ep->active) && !bb_terminated(ep->active)) {
2198 struct symbol *ret_type = base_type->ctype.base_type;
2199 struct instruction *insn = alloc_typed_instruction(OP_RET, ret_type);
2200
2201 if (type_size(ret_type) > 0)
2202 use_pseudo(insn, result, &insn->src);
2203 add_one_insn(ep, insn);
2204 }
2205
2206 /*
2207 * Do trivial flow simplification - branches to
2208 * branches, kill dead basicblocks etc
2209 */
2210 kill_unreachable_bbs(ep);
2211
2212 /*
2213 * Turn symbols into pseudos
2214 */
2215 simplify_symbol_usage(ep);
2216
2217 repeat:
2218 /*
2219 * Remove trivial instructions, and try to CSE
2220 * the rest.
2221 */
2222 do {
2223 cleanup_and_cse(ep);
2224 pack_basic_blocks(ep);
2225 } while (repeat_phase & REPEAT_CSE);
2226
2227 kill_unreachable_bbs(ep);
2228 vrfy_flow(ep);
2229
2230 /* Cleanup */
2231 clear_symbol_pseudos(ep);
2232
2233 /* And track pseudo register usage */
2234 track_pseudo_liveness(ep);
2235
2236 /*
2237 * Some flow optimizations can only effectively
2238 * be done when we've done liveness analysis. But
2239 * if they trigger, we need to start all over
2240 * again
2241 */
2242 if (simplify_flow(ep)) {
2243 clear_liveness(ep);
2244 goto repeat;
2245 }
2246
2247 /* Finally, add deathnotes to pseudos now that we have them */
2248 if (dbg_dead)
2249 track_pseudo_death(ep);
2250
2251 return ep;
2252 }
2253
2254 struct entrypoint *linearize_symbol(struct symbol *sym)
2255 {
2256 struct symbol *base_type;
2257
2258 if (!sym)
2259 return NULL;
2260 current_pos = sym->pos;
2261 base_type = sym->ctype.base_type;
2262 if (!base_type)
2263 return NULL;
2264 if (base_type->type == SYM_FN)
2265 return linearize_fn(sym, base_type);
2266 return NULL;
2267 }
Static analysis for C