?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
fix crash when ep->active is NULL
[smatch.git] / linearize.c
blob12209492b4c8eb84406b7c03644392ac8bfb0a98
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 repeat_phase |= REPEAT_CFG_CLEANUP;
647 }
648
649 /* Change a "switch" or a conditional branch 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 kill_instruction(old);
659
660 br = alloc_instruction(OP_BR, 0);
661 br->bb = bb;
662 br->bb_true = target;
663 add_instruction(&bb->insns, br);
664
665 FOR_EACH_PTR(bb->children, child) {
666 if (child == target) {
667 target = NULL; /* Trigger just once */
668 continue;
669 }
670 DELETE_CURRENT_PTR(child);
671 remove_parent(child, bb);
672 } END_FOR_EACH_PTR(child);
673 PACK_PTR_LIST(&bb->children);
674 }
675
676
677 void insert_select(struct basic_block *bb, struct instruction *br, struct instruction *phi_node, pseudo_t if_true, pseudo_t if_false)
678 {
679 pseudo_t target;
680 struct instruction *select;
681
682 /* Remove the 'br' */
683 delete_last_instruction(&bb->insns);
684
685 select = alloc_instruction(OP_SEL, phi_node->size);
686 select->bb = bb;
687
688 assert(br->cond);
689 use_pseudo(select, br->cond, &select->src1);
690
691 target = phi_node->target;
692 assert(target->def == phi_node);
693 select->target = target;
694 target->def = select;
695
696 use_pseudo(select, if_true, &select->src2);
697 use_pseudo(select, if_false, &select->src3);
698
699 add_instruction(&bb->insns, select);
700 add_instruction(&bb->insns, br);
701 }
702
703 static inline int bb_empty(struct basic_block *bb)
704 {
705 return !bb->insns;
706 }
707
708 /* Add a label to the currently active block, return new active block */
709 static struct basic_block * add_label(struct entrypoint *ep, struct symbol *label)
710 {
711 struct basic_block *bb = label->bb_target;
712
713 if (bb) {
714 set_activeblock(ep, bb);
715 return bb;
716 }
717 bb = ep->active;
718 if (!bb_reachable(bb) || !bb_empty(bb)) {
719 bb = alloc_basic_block(ep, label->pos);
720 set_activeblock(ep, bb);
721 }
722 label->bb_target = bb;
723 return bb;
724 }
725
726 static void add_branch(struct entrypoint *ep, struct expression *expr, pseudo_t cond, struct basic_block *bb_true, struct basic_block *bb_false)
727 {
728 struct basic_block *bb = ep->active;
729 struct instruction *br;
730
731 if (bb_reachable(bb)) {
732 br = alloc_instruction(OP_CBR, 0);
733 use_pseudo(br, cond, &br->cond);
734 br->bb_true = bb_true;
735 br->bb_false = bb_false;
736 add_bb(&bb_true->parents, bb);
737 add_bb(&bb_false->parents, bb);
738 add_bb(&bb->children, bb_true);
739 add_bb(&bb->children, bb_false);
740 add_one_insn(ep, br);
741 }
742 }
743
744 /* Dummy pseudo allocator */
745 pseudo_t alloc_pseudo(struct instruction *def)
746 {
747 static int nr = 0;
748 struct pseudo * pseudo = __alloc_pseudo(0);
749 pseudo->type = PSEUDO_REG;
750 pseudo->nr = ++nr;
751 pseudo->def = def;
752 return pseudo;
753 }
754
755 static void clear_symbol_pseudos(struct entrypoint *ep)
756 {
757 pseudo_t pseudo;
758
759 FOR_EACH_PTR(ep->accesses, pseudo) {
760 pseudo->sym->pseudo = NULL;
761 } END_FOR_EACH_PTR(pseudo);
762 }
763
764 static pseudo_t symbol_pseudo(struct entrypoint *ep, struct symbol *sym)
765 {
766 pseudo_t pseudo;
767
768 if (!sym)
769 return VOID;
770
771 pseudo = sym->pseudo;
772 if (!pseudo) {
773 pseudo = __alloc_pseudo(0);
774 pseudo->nr = -1;
775 pseudo->type = PSEUDO_SYM;
776 pseudo->sym = sym;
777 pseudo->ident = sym->ident;
778 sym->pseudo = pseudo;
779 add_pseudo(&ep->accesses, pseudo);
780 }
781 /* Symbol pseudos have neither nr, usage nor def */
782 return pseudo;
783 }
784
785 pseudo_t value_pseudo(long long val)
786 {
787 #define MAX_VAL_HASH 64
788 static struct pseudo_list *prev[MAX_VAL_HASH];
789 int hash = val & (MAX_VAL_HASH-1);
790 struct pseudo_list **list = prev + hash;
791 pseudo_t pseudo;
792
793 FOR_EACH_PTR(*list, pseudo) {
794 if (pseudo->value == val)
795 return pseudo;
796 } END_FOR_EACH_PTR(pseudo);
797
798 pseudo = __alloc_pseudo(0);
799 pseudo->type = PSEUDO_VAL;
800 pseudo->value = val;
801 add_pseudo(list, pseudo);
802
803 /* Value pseudos have neither nr, usage nor def */
804 return pseudo;
805 }
806
807 static pseudo_t argument_pseudo(struct entrypoint *ep, int nr)
808 {
809 pseudo_t pseudo = __alloc_pseudo(0);
810 struct instruction *entry = ep->entry;
811
812 pseudo->type = PSEUDO_ARG;
813 pseudo->nr = nr;
814 pseudo->def = entry;
815 add_pseudo(&entry->arg_list, pseudo);
816
817 /* Argument pseudos have neither usage nor def */
818 return pseudo;
819 }
820
821 pseudo_t alloc_phi(struct basic_block *source, pseudo_t pseudo, int size)
822 {
823 struct instruction *insn;
824 pseudo_t phi;
825 static int nr = 0;
826
827 if (!source)
828 return VOID;
829
830 insn = alloc_instruction(OP_PHISOURCE, size);
831 phi = __alloc_pseudo(0);
832 phi->type = PSEUDO_PHI;
833 phi->nr = ++nr;
834 phi->def = insn;
835
836 use_pseudo(insn, pseudo, &insn->phi_src);
837 insn->bb = source;
838 insn->target = phi;
839 add_instruction(&source->insns, insn);
840 return phi;
841 }
842
843 /*
844 * We carry the "access_data" structure around for any accesses,
845 * which simplifies things a lot. It contains all the access
846 * information in one place.
847 */
848 struct access_data {
849 struct symbol *result_type; // result ctype
850 struct symbol *source_type; // source ctype
851 pseudo_t address; // pseudo containing address ..
852 unsigned int offset; // byte offset
853 struct position pos;
854 };
855
856 static void finish_address_gen(struct entrypoint *ep, struct access_data *ad)
857 {
858 }
859
860 static int linearize_simple_address(struct entrypoint *ep,
861 struct expression *addr,
862 struct access_data *ad)
863 {
864 if (addr->type == EXPR_SYMBOL) {
865 linearize_one_symbol(ep, addr->symbol);
866 ad->address = symbol_pseudo(ep, addr->symbol);
867 return 1;
868 }
869 if (addr->type == EXPR_BINOP) {
870 if (addr->right->type == EXPR_VALUE) {
871 if (addr->op == '+') {
872 ad->offset += get_expression_value(addr->right);
873 return linearize_simple_address(ep, addr->left, ad);
874 }
875 }
876 }
877 ad->address = linearize_expression(ep, addr);
878 return 1;
879 }
880
881 static struct symbol *base_type(struct symbol *sym)
882 {
883 struct symbol *base = sym;
884
885 if (sym) {
886 if (sym->type == SYM_NODE)
887 base = base->ctype.base_type;
888 if (base->type == SYM_BITFIELD)
889 return base->ctype.base_type;
890 }
891 return sym;
892 }
893
894 static int linearize_address_gen(struct entrypoint *ep,
895 struct expression *expr,
896 struct access_data *ad)
897 {
898 struct symbol *ctype = expr->ctype;
899
900 if (!ctype)
901 return 0;
902 ad->pos = expr->pos;
903 ad->result_type = ctype;
904 ad->source_type = base_type(ctype);
905 if (expr->type == EXPR_PREOP && expr->op == '*')
906 return linearize_simple_address(ep, expr->unop, ad);
907
908 warning(expr->pos, "generating address of non-lvalue (%d)", expr->type);
909 return 0;
910 }
911
912 static pseudo_t add_load(struct entrypoint *ep, struct access_data *ad)
913 {
914 struct instruction *insn;
915 pseudo_t new;
916
917 insn = alloc_typed_instruction(OP_LOAD, ad->source_type);
918 new = alloc_pseudo(insn);
919
920 insn->target = new;
921 insn->offset = ad->offset;
922 use_pseudo(insn, ad->address, &insn->src);
923 add_one_insn(ep, insn);
924 return new;
925 }
926
927 static void add_store(struct entrypoint *ep, struct access_data *ad, pseudo_t value)
928 {
929 struct basic_block *bb = ep->active;
930
931 if (bb_reachable(bb)) {
932 struct instruction *store = alloc_typed_instruction(OP_STORE, ad->source_type);
933 store->offset = ad->offset;
934 use_pseudo(store, value, &store->target);
935 use_pseudo(store, ad->address, &store->src);
936 add_one_insn(ep, store);
937 }
938 }
939
940 static pseudo_t linearize_store_gen(struct entrypoint *ep,
941 pseudo_t value,
942 struct access_data *ad)
943 {
944 pseudo_t store = value;
945
946 if (type_size(ad->source_type) != type_size(ad->result_type)) {
947 struct symbol *ctype = ad->result_type;
948 unsigned int shift = ctype->bit_offset;
949 unsigned int size = ctype->bit_size;
950 pseudo_t orig = add_load(ep, ad);
951 unsigned long long mask = (1ULL << size) - 1;
952
953 if (shift) {
954 store = add_binary_op(ep, ad->source_type, OP_SHL, value, value_pseudo(shift));
955 mask <<= shift;
956 }
957 orig = add_binary_op(ep, ad->source_type, OP_AND, orig, value_pseudo(~mask));
958 store = add_binary_op(ep, ad->source_type, OP_OR, orig, store);
959 }
960 add_store(ep, ad, store);
961 return value;
962 }
963
964 static pseudo_t add_binary_op(struct entrypoint *ep, struct symbol *ctype, int op, pseudo_t left, pseudo_t right)
965 {
966 struct instruction *insn = alloc_typed_instruction(op, ctype);
967 pseudo_t target = alloc_pseudo(insn);
968 insn->target = target;
969 use_pseudo(insn, left, &insn->src1);
970 use_pseudo(insn, right, &insn->src2);
971 add_one_insn(ep, insn);
972 return target;
973 }
974
975 static pseudo_t add_setval(struct entrypoint *ep, struct symbol *ctype, struct expression *val)
976 {
977 struct instruction *insn = alloc_typed_instruction(OP_SETVAL, ctype);
978 pseudo_t target = alloc_pseudo(insn);
979 insn->target = target;
980 insn->val = val;
981 add_one_insn(ep, insn);
982 return target;
983 }
984
985 static pseudo_t add_symbol_address(struct entrypoint *ep, struct symbol *sym)
986 {
987 struct instruction *insn = alloc_instruction(OP_SYMADDR, bits_in_pointer);
988 pseudo_t target = alloc_pseudo(insn);
989
990 insn->target = target;
991 use_pseudo(insn, symbol_pseudo(ep, sym), &insn->symbol);
992 add_one_insn(ep, insn);
993 return target;
994 }
995
996 static pseudo_t linearize_load_gen(struct entrypoint *ep, struct access_data *ad)
997 {
998 struct symbol *ctype = ad->result_type;
999 pseudo_t new = add_load(ep, ad);
1000
1001 if (ctype->bit_offset) {
1002 pseudo_t shift = value_pseudo(ctype->bit_offset);
1003 pseudo_t newval = add_binary_op(ep, ad->source_type, OP_LSR, new, shift);
1004 new = newval;
1005 }
1006 if (ctype->bit_size != type_size(ad->source_type))
1007 new = cast_pseudo(ep, new, ad->source_type, ad->result_type);
1008 return new;
1009 }
1010
1011 static pseudo_t linearize_access(struct entrypoint *ep, struct expression *expr)
1012 {
1013 struct access_data ad = { NULL, };
1014 pseudo_t value;
1015
1016 if (!linearize_address_gen(ep, expr, &ad))
1017 return VOID;
1018 value = linearize_load_gen(ep, &ad);
1019 finish_address_gen(ep, &ad);
1020 return value;
1021 }
1022
1023 /* FIXME: FP */
1024 static pseudo_t linearize_inc_dec(struct entrypoint *ep, struct expression *expr, int postop)
1025 {
1026 struct access_data ad = { NULL, };
1027 pseudo_t old, new, one;
1028 int op = expr->op == SPECIAL_INCREMENT ? OP_ADD : OP_SUB;
1029
1030 if (!linearize_address_gen(ep, expr->unop, &ad))
1031 return VOID;
1032
1033 old = linearize_load_gen(ep, &ad);
1034 one = value_pseudo(expr->op_value);
1035 new = add_binary_op(ep, expr->ctype, op, old, one);
1036 linearize_store_gen(ep, new, &ad);
1037 finish_address_gen(ep, &ad);
1038 return postop ? old : new;
1039 }
1040
1041 static pseudo_t add_uniop(struct entrypoint *ep, struct expression *expr, int op, pseudo_t src)
1042 {
1043 struct instruction *insn = alloc_typed_instruction(op, expr->ctype);
1044 pseudo_t new = alloc_pseudo(insn);
1045
1046 insn->target = new;
1047 use_pseudo(insn, src, &insn->src1);
1048 add_one_insn(ep, insn);
1049 return new;
1050 }
1051
1052 static pseudo_t linearize_slice(struct entrypoint *ep, struct expression *expr)
1053 {
1054 pseudo_t pre = linearize_expression(ep, expr->base);
1055 struct instruction *insn = alloc_typed_instruction(OP_SLICE, expr->ctype);
1056 pseudo_t new = alloc_pseudo(insn);
1057
1058 insn->target = new;
1059 insn->from = expr->r_bitpos;
1060 insn->len = expr->r_nrbits;
1061 use_pseudo(insn, pre, &insn->base);
1062 add_one_insn(ep, insn);
1063 return new;
1064 }
1065
1066 static pseudo_t linearize_regular_preop(struct entrypoint *ep, struct expression *expr)
1067 {
1068 pseudo_t pre = linearize_expression(ep, expr->unop);
1069 switch (expr->op) {
1070 case '+':
1071 return pre;
1072 case '!': {
1073 pseudo_t zero = value_pseudo(0);
1074 return add_binary_op(ep, expr->ctype, OP_SET_EQ, pre, zero);
1075 }
1076 case '~':
1077 return add_uniop(ep, expr, OP_NOT, pre);
1078 case '-':
1079 return add_uniop(ep, expr, OP_NEG, pre);
1080 }
1081 return VOID;
1082 }
1083
1084 static pseudo_t linearize_preop(struct entrypoint *ep, struct expression *expr)
1085 {
1086 /*
1087 * '*' is an lvalue access, and is fundamentally different
1088 * from an arithmetic operation. Maybe it should have an
1089 * expression type of its own..
1090 */
1091 if (expr->op == '*')
1092 return linearize_access(ep, expr);
1093 if (expr->op == SPECIAL_INCREMENT || expr->op == SPECIAL_DECREMENT)
1094 return linearize_inc_dec(ep, expr, 0);
1095 return linearize_regular_preop(ep, expr);
1096 }
1097
1098 static pseudo_t linearize_postop(struct entrypoint *ep, struct expression *expr)
1099 {
1100 return linearize_inc_dec(ep, expr, 1);
1101 }
1102
1103 /*
1104 * Casts to pointers are "less safe" than other casts, since
1105 * they imply type-unsafe accesses. "void *" is a special
1106 * case, since you can't access through it anyway without another
1107 * cast.
1108 */
1109 static struct instruction *alloc_cast_instruction(struct symbol *src, struct symbol *ctype)
1110 {
1111 int opcode = OP_CAST;
1112 struct symbol *base = ctype;
1113
1114 if (src->ctype.modifiers & MOD_SIGNED)
1115 opcode = OP_SCAST;
1116 if (base->type == SYM_NODE)
1117 base = base->ctype.base_type;
1118 if (base->type == SYM_PTR) {
1119 base = base->ctype.base_type;
1120 if (base != &void_ctype)
1121 opcode = OP_PTRCAST;
1122 } else if (base->ctype.base_type == &fp_type)
1123 opcode = OP_FPCAST;
1124 return alloc_typed_instruction(opcode, ctype);
1125 }
1126
1127 static pseudo_t cast_pseudo(struct entrypoint *ep, pseudo_t src, struct symbol *from, struct symbol *to)
1128 {
1129 pseudo_t result;
1130 struct instruction *insn;
1131
1132 if (src == VOID)
1133 return VOID;
1134 if (!from || !to)
1135 return VOID;
1136 if (from->bit_size < 0 || to->bit_size < 0)
1137 return VOID;
1138 insn = alloc_cast_instruction(from, to);
1139 result = alloc_pseudo(insn);
1140 insn->target = result;
1141 insn->orig_type = from;
1142 use_pseudo(insn, src, &insn->src);
1143 add_one_insn(ep, insn);
1144 return result;
1145 }
1146
1147 static int opcode_sign(int opcode, struct symbol *ctype)
1148 {
1149 if (ctype && (ctype->ctype.modifiers & MOD_SIGNED)) {
1150 switch(opcode) {
1151 case OP_MULU: case OP_DIVU: case OP_MODU: case OP_LSR:
1152 opcode++;
1153 }
1154 }
1155 return opcode;
1156 }
1157
1158 static inline pseudo_t add_convert_to_bool(struct entrypoint *ep, pseudo_t src, struct symbol *type)
1159 {
1160 pseudo_t zero;
1161 int op;
1162
1163 if (is_bool_type(type))
1164 return src;
1165 zero = value_pseudo(0);
1166 op = OP_SET_NE;
1167 return add_binary_op(ep, &bool_ctype, op, src, zero);
1168 }
1169
1170 static pseudo_t linearize_expression_to_bool(struct entrypoint *ep, struct expression *expr)
1171 {
1172 pseudo_t dst;
1173 dst = linearize_expression(ep, expr);
1174 dst = add_convert_to_bool(ep, dst, expr->ctype);
1175 return dst;
1176 }
1177
1178 static pseudo_t linearize_assignment(struct entrypoint *ep, struct expression *expr)
1179 {
1180 struct access_data ad = { NULL, };
1181 struct expression *target = expr->left;
1182 struct expression *src = expr->right;
1183 struct symbol *ctype;
1184 pseudo_t value;
1185
1186 value = linearize_expression(ep, src);
1187 if (!target || !linearize_address_gen(ep, target, &ad))
1188 return value;
1189 if (expr->op != '=') {
1190 pseudo_t oldvalue = linearize_load_gen(ep, &ad);
1191 pseudo_t dst;
1192 static const int op_trans[] = {
1193 [SPECIAL_ADD_ASSIGN - SPECIAL_BASE] = OP_ADD,
1194 [SPECIAL_SUB_ASSIGN - SPECIAL_BASE] = OP_SUB,
1195 [SPECIAL_MUL_ASSIGN - SPECIAL_BASE] = OP_MULU,
1196 [SPECIAL_DIV_ASSIGN - SPECIAL_BASE] = OP_DIVU,
1197 [SPECIAL_MOD_ASSIGN - SPECIAL_BASE] = OP_MODU,
1198 [SPECIAL_SHL_ASSIGN - SPECIAL_BASE] = OP_SHL,
1199 [SPECIAL_SHR_ASSIGN - SPECIAL_BASE] = OP_LSR,
1200 [SPECIAL_AND_ASSIGN - SPECIAL_BASE] = OP_AND,
1201 [SPECIAL_OR_ASSIGN - SPECIAL_BASE] = OP_OR,
1202 [SPECIAL_XOR_ASSIGN - SPECIAL_BASE] = OP_XOR
1203 };
1204 int opcode;
1205
1206 if (!src)
1207 return VOID;
1208
1209 ctype = src->ctype;
1210 oldvalue = cast_pseudo(ep, oldvalue, target->ctype, ctype);
1211 opcode = opcode_sign(op_trans[expr->op - SPECIAL_BASE], ctype);
1212 dst = add_binary_op(ep, ctype, opcode, oldvalue, value);
1213 value = cast_pseudo(ep, dst, ctype, expr->ctype);
1214 }
1215 value = linearize_store_gen(ep, value, &ad);
1216 finish_address_gen(ep, &ad);
1217 return value;
1218 }
1219
1220 static pseudo_t linearize_call_expression(struct entrypoint *ep, struct expression *expr)
1221 {
1222 struct expression *arg, *fn;
1223 struct instruction *insn = alloc_typed_instruction(OP_CALL, expr->ctype);
1224 pseudo_t retval, call;
1225 struct ctype *ctype = NULL;
1226 struct symbol *fntype;
1227 struct context *context;
1228
1229 if (!expr->ctype) {
1230 warning(expr->pos, "call with no type!");
1231 return VOID;
1232 }
1233
1234 FOR_EACH_PTR(expr->args, arg) {
1235 pseudo_t new = linearize_expression(ep, arg);
1236 use_pseudo(insn, new, add_pseudo(&insn->arguments, new));
1237 } END_FOR_EACH_PTR(arg);
1238
1239 fn = expr->fn;
1240
1241 if (fn->ctype)
1242 ctype = &fn->ctype->ctype;
1243
1244 fntype = fn->ctype;
1245 if (fntype) {
1246 if (fntype->type == SYM_NODE)
1247 fntype = fntype->ctype.base_type;
1248 }
1249 insn->fntype = fntype;
1250
1251 if (fn->type == EXPR_PREOP) {
1252 if (fn->unop->type == EXPR_SYMBOL) {
1253 struct symbol *sym = fn->unop->symbol;
1254 if (sym->ctype.base_type->type == SYM_FN)
1255 fn = fn->unop;
1256 }
1257 }
1258 if (fn->type == EXPR_SYMBOL) {
1259 call = symbol_pseudo(ep, fn->symbol);
1260 } else {
1261 call = linearize_expression(ep, fn);
1262 }
1263 use_pseudo(insn, call, &insn->func);
1264 retval = VOID;
1265 if (expr->ctype != &void_ctype)
1266 retval = alloc_pseudo(insn);
1267 insn->target = retval;
1268 add_one_insn(ep, insn);
1269
1270 if (ctype) {
1271 FOR_EACH_PTR(ctype->contexts, context) {
1272 int in = context->in;
1273 int out = context->out;
1274 int check = 0;
1275 int context_diff;
1276 if (in < 0) {
1277 check = 1;
1278 in = 0;
1279 }
1280 if (out < 0) {
1281 check = 0;
1282 out = 0;
1283 }
1284 context_diff = out - in;
1285 if (check || context_diff) {
1286 insn = alloc_instruction(OP_CONTEXT, 0);
1287 insn->increment = context_diff;
1288 insn->check = check;
1289 insn->context_expr = context->context;
1290 add_one_insn(ep, insn);
1291 }
1292 } END_FOR_EACH_PTR(context);
1293 }
1294
1295 return retval;
1296 }
1297
1298 static pseudo_t linearize_binop_bool(struct entrypoint *ep, struct expression *expr)
1299 {
1300 pseudo_t src1, src2, dst;
1301 int op = (expr->op == SPECIAL_LOGICAL_OR) ? OP_OR_BOOL : OP_AND_BOOL;
1302
1303 src1 = linearize_expression_to_bool(ep, expr->left);
1304 src2 = linearize_expression_to_bool(ep, expr->right);
1305 dst = add_binary_op(ep, &bool_ctype, op, src1, src2);
1306 if (expr->ctype != &bool_ctype)
1307 dst = cast_pseudo(ep, dst, &bool_ctype, expr->ctype);
1308 return dst;
1309 }
1310
1311 static pseudo_t linearize_binop(struct entrypoint *ep, struct expression *expr)
1312 {
1313 pseudo_t src1, src2, dst;
1314 static const int opcode[] = {
1315 ['+'] = OP_ADD, ['-'] = OP_SUB,
1316 ['*'] = OP_MULU, ['/'] = OP_DIVU,
1317 ['%'] = OP_MODU, ['&'] = OP_AND,
1318 ['|'] = OP_OR, ['^'] = OP_XOR,
1319 [SPECIAL_LEFTSHIFT] = OP_SHL,
1320 [SPECIAL_RIGHTSHIFT] = OP_LSR,
1321 };
1322 int op;
1323
1324 src1 = linearize_expression(ep, expr->left);
1325 src2 = linearize_expression(ep, expr->right);
1326 op = opcode_sign(opcode[expr->op], expr->ctype);
1327 dst = add_binary_op(ep, expr->ctype, op, src1, src2);
1328 return dst;
1329 }
1330
1331 static pseudo_t linearize_logical_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false);
1332
1333 pseudo_t linearize_cond_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false);
1334
1335 static pseudo_t linearize_select(struct entrypoint *ep, struct expression *expr)
1336 {
1337 pseudo_t cond, true, false, res;
1338 struct instruction *insn;
1339
1340 true = linearize_expression(ep, expr->cond_true);
1341 false = linearize_expression(ep, expr->cond_false);
1342 cond = linearize_expression(ep, expr->conditional);
1343
1344 insn = alloc_typed_instruction(OP_SEL, expr->ctype);
1345 if (!expr->cond_true)
1346 true = cond;
1347 use_pseudo(insn, cond, &insn->src1);
1348 use_pseudo(insn, true, &insn->src2);
1349 use_pseudo(insn, false, &insn->src3);
1350
1351 res = alloc_pseudo(insn);
1352 insn->target = res;
1353 add_one_insn(ep, insn);
1354 return res;
1355 }
1356
1357 static pseudo_t add_join_conditional(struct entrypoint *ep, struct expression *expr,
1358 pseudo_t phi1, pseudo_t phi2)
1359 {
1360 pseudo_t target;
1361 struct instruction *phi_node;
1362
1363 if (phi1 == VOID)
1364 return phi2;
1365 if (phi2 == VOID)
1366 return phi1;
1367
1368 phi_node = alloc_typed_instruction(OP_PHI, expr->ctype);
1369 use_pseudo(phi_node, phi1, add_pseudo(&phi_node->phi_list, phi1));
1370 use_pseudo(phi_node, phi2, add_pseudo(&phi_node->phi_list, phi2));
1371 phi_node->target = target = alloc_pseudo(phi_node);
1372 add_one_insn(ep, phi_node);
1373 return target;
1374 }
1375
1376 static pseudo_t linearize_short_conditional(struct entrypoint *ep, struct expression *expr,
1377 struct expression *cond,
1378 struct expression *expr_false)
1379 {
1380 pseudo_t src1, src2;
1381 struct basic_block *bb_false;
1382 struct basic_block *merge = alloc_basic_block(ep, expr->pos);
1383 pseudo_t phi1, phi2;
1384 int size = type_size(expr->ctype);
1385
1386 if (!expr_false || !ep->active)
1387 return VOID;
1388
1389 bb_false = alloc_basic_block(ep, expr_false->pos);
1390 src1 = linearize_expression(ep, cond);
1391 phi1 = alloc_phi(ep->active, src1, size);
1392 add_branch(ep, expr, src1, merge, bb_false);
1393
1394 set_activeblock(ep, bb_false);
1395 src2 = linearize_expression(ep, expr_false);
1396 phi2 = alloc_phi(ep->active, src2, size);
1397 set_activeblock(ep, merge);
1398
1399 return add_join_conditional(ep, expr, phi1, phi2);
1400 }
1401
1402 static pseudo_t linearize_conditional(struct entrypoint *ep, struct expression *expr,
1403 struct expression *cond,
1404 struct expression *expr_true,
1405 struct expression *expr_false)
1406 {
1407 pseudo_t src1, src2;
1408 pseudo_t phi1, phi2;
1409 struct basic_block *bb_true, *bb_false, *merge;
1410 int size = type_size(expr->ctype);
1411
1412 if (!cond || !expr_true || !expr_false || !ep->active)
1413 return VOID;
1414 bb_true = alloc_basic_block(ep, expr_true->pos);
1415 bb_false = alloc_basic_block(ep, expr_false->pos);
1416 merge = alloc_basic_block(ep, expr->pos);
1417
1418 linearize_cond_branch(ep, cond, bb_true, bb_false);
1419
1420 set_activeblock(ep, bb_true);
1421 src1 = linearize_expression(ep, expr_true);
1422 phi1 = alloc_phi(ep->active, src1, size);
1423 add_goto(ep, merge);
1424
1425 set_activeblock(ep, bb_false);
1426 src2 = linearize_expression(ep, expr_false);
1427 phi2 = alloc_phi(ep->active, src2, size);
1428 set_activeblock(ep, merge);
1429
1430 return add_join_conditional(ep, expr, phi1, phi2);
1431 }
1432
1433 static pseudo_t linearize_logical(struct entrypoint *ep, struct expression *expr)
1434 {
1435 struct expression *shortcut;
1436
1437 shortcut = alloc_const_expression(expr->pos, expr->op == SPECIAL_LOGICAL_OR);
1438 shortcut->ctype = expr->ctype;
1439 if (expr->op == SPECIAL_LOGICAL_OR)
1440 return linearize_conditional(ep, expr, expr->left, shortcut, expr->right);
1441 return linearize_conditional(ep, expr, expr->left, expr->right, shortcut);
1442 }
1443
1444 static pseudo_t linearize_compare(struct entrypoint *ep, struct expression *expr)
1445 {
1446 static const int cmpop[] = {
1447 ['>'] = OP_SET_GT, ['<'] = OP_SET_LT,
1448 [SPECIAL_EQUAL] = OP_SET_EQ,
1449 [SPECIAL_NOTEQUAL] = OP_SET_NE,
1450 [SPECIAL_GTE] = OP_SET_GE,
1451 [SPECIAL_LTE] = OP_SET_LE,
1452 [SPECIAL_UNSIGNED_LT] = OP_SET_B,
1453 [SPECIAL_UNSIGNED_GT] = OP_SET_A,
1454 [SPECIAL_UNSIGNED_LTE] = OP_SET_BE,
1455 [SPECIAL_UNSIGNED_GTE] = OP_SET_AE,
1456 };
1457
1458 pseudo_t src1 = linearize_expression(ep, expr->left);
1459 pseudo_t src2 = linearize_expression(ep, expr->right);
1460 pseudo_t dst = add_binary_op(ep, expr->ctype, cmpop[expr->op], src1, src2);
1461 return dst;
1462 }
1463
1464
1465 pseudo_t linearize_cond_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false)
1466 {
1467 pseudo_t cond;
1468
1469 if (!expr || !bb_reachable(ep->active))
1470 return VOID;
1471
1472 switch (expr->type) {
1473
1474 case EXPR_STRING:
1475 case EXPR_VALUE:
1476 add_goto(ep, expr->value ? bb_true : bb_false);
1477 return VOID;
1478
1479 case EXPR_FVALUE:
1480 add_goto(ep, expr->fvalue ? bb_true : bb_false);
1481 return VOID;
1482
1483 case EXPR_LOGICAL:
1484 linearize_logical_branch(ep, expr, bb_true, bb_false);
1485 return VOID;
1486
1487 case EXPR_COMPARE:
1488 cond = linearize_compare(ep, expr);
1489 add_branch(ep, expr, cond, bb_true, bb_false);
1490 break;
1491
1492 case EXPR_PREOP:
1493 if (expr->op == '!')
1494 return linearize_cond_branch(ep, expr->unop, bb_false, bb_true);
1495 /* fall through */
1496 default: {
1497 cond = linearize_expression(ep, expr);
1498 add_branch(ep, expr, cond, bb_true, bb_false);
1499
1500 return VOID;
1501 }
1502 }
1503 return VOID;
1504 }
1505
1506
1507
1508 static pseudo_t linearize_logical_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false)
1509 {
1510 struct basic_block *next = alloc_basic_block(ep, expr->pos);
1511
1512 if (expr->op == SPECIAL_LOGICAL_OR)
1513 linearize_cond_branch(ep, expr->left, bb_true, next);
1514 else
1515 linearize_cond_branch(ep, expr->left, next, bb_false);
1516 set_activeblock(ep, next);
1517 linearize_cond_branch(ep, expr->right, bb_true, bb_false);
1518 return VOID;
1519 }
1520
1521 static pseudo_t linearize_cast(struct entrypoint *ep, struct expression *expr)
1522 {
1523 pseudo_t src;
1524 struct expression *orig = expr->cast_expression;
1525
1526 if (!orig)
1527 return VOID;
1528
1529 src = linearize_expression(ep, orig);
1530 return cast_pseudo(ep, src, orig->ctype, expr->ctype);
1531 }
1532
1533 static pseudo_t linearize_position(struct entrypoint *ep, struct expression *pos, struct access_data *ad)
1534 {
1535 struct expression *init_expr = pos->init_expr;
1536
1537 ad->offset = pos->init_offset;
1538 ad->source_type = base_type(init_expr->ctype);
1539 ad->result_type = init_expr->ctype;
1540 return linearize_initializer(ep, init_expr, ad);
1541 }
1542
1543 static pseudo_t linearize_initializer(struct entrypoint *ep, struct expression *initializer, struct access_data *ad)
1544 {
1545 switch (initializer->type) {
1546 case EXPR_INITIALIZER: {
1547 struct expression *expr;
1548 FOR_EACH_PTR(initializer->expr_list, expr) {
1549 linearize_initializer(ep, expr, ad);
1550 } END_FOR_EACH_PTR(expr);
1551 break;
1552 }
1553 case EXPR_POS:
1554 linearize_position(ep, initializer, ad);
1555 break;
1556 default: {
1557 pseudo_t value = linearize_expression(ep, initializer);
1558 ad->source_type = base_type(initializer->ctype);
1559 ad->result_type = initializer->ctype;
1560 linearize_store_gen(ep, value, ad);
1561 return value;
1562 }
1563 }
1564
1565 return VOID;
1566 }
1567
1568 static void linearize_argument(struct entrypoint *ep, struct symbol *arg, int nr)
1569 {
1570 struct access_data ad = { NULL, };
1571
1572 ad.source_type = arg;
1573 ad.result_type = arg;
1574 ad.address = symbol_pseudo(ep, arg);
1575 linearize_store_gen(ep, argument_pseudo(ep, nr), &ad);
1576 finish_address_gen(ep, &ad);
1577 }
1578
1579 pseudo_t linearize_expression(struct entrypoint *ep, struct expression *expr)
1580 {
1581 if (!expr)
1582 return VOID;
1583
1584 current_pos = expr->pos;
1585 switch (expr->type) {
1586 case EXPR_SYMBOL:
1587 linearize_one_symbol(ep, expr->symbol);
1588 return add_symbol_address(ep, expr->symbol);
1589
1590 case EXPR_VALUE:
1591 return value_pseudo(expr->value);
1592
1593 case EXPR_STRING: case EXPR_FVALUE: case EXPR_LABEL:
1594 return add_setval(ep, expr->ctype, expr);
1595
1596 case EXPR_STATEMENT:
1597 return linearize_statement(ep, expr->statement);
1598
1599 case EXPR_CALL:
1600 return linearize_call_expression(ep, expr);
1601
1602 case EXPR_BINOP:
1603 if (expr->op == SPECIAL_LOGICAL_AND || expr->op == SPECIAL_LOGICAL_OR)
1604 return linearize_binop_bool(ep, expr);
1605 return linearize_binop(ep, expr);
1606
1607 case EXPR_LOGICAL:
1608 return linearize_logical(ep, expr);
1609
1610 case EXPR_COMPARE:
1611 return linearize_compare(ep, expr);
1612
1613 case EXPR_SELECT:
1614 return linearize_select(ep, expr);
1615
1616 case EXPR_CONDITIONAL:
1617 if (!expr->cond_true)
1618 return linearize_short_conditional(ep, expr, expr->conditional, expr->cond_false);
1619
1620 return linearize_conditional(ep, expr, expr->conditional,
1621 expr->cond_true, expr->cond_false);
1622
1623 case EXPR_COMMA:
1624 linearize_expression(ep, expr->left);
1625 return linearize_expression(ep, expr->right);
1626
1627 case EXPR_ASSIGNMENT:
1628 return linearize_assignment(ep, expr);
1629
1630 case EXPR_PREOP:
1631 return linearize_preop(ep, expr);
1632
1633 case EXPR_POSTOP:
1634 return linearize_postop(ep, expr);
1635
1636 case EXPR_CAST:
1637 case EXPR_FORCE_CAST:
1638 case EXPR_IMPLIED_CAST:
1639 return linearize_cast(ep, expr);
1640
1641 case EXPR_SLICE:
1642 return linearize_slice(ep, expr);
1643
1644 case EXPR_INITIALIZER:
1645 case EXPR_POS:
1646 warning(expr->pos, "unexpected initializer expression (%d %d)", expr->type, expr->op);
1647 return VOID;
1648 default:
1649 warning(expr->pos, "unknown expression (%d %d)", expr->type, expr->op);
1650 return VOID;
1651 }
1652 return VOID;
1653 }
1654
1655 static pseudo_t linearize_one_symbol(struct entrypoint *ep, struct symbol *sym)
1656 {
1657 struct access_data ad = { NULL, };
1658 pseudo_t value;
1659
1660 if (!sym || !sym->initializer || sym->initialized)
1661 return VOID;
1662
1663 /* We need to output these puppies some day too.. */
1664 if (sym->ctype.modifiers & (MOD_STATIC | MOD_TOPLEVEL))
1665 return VOID;
1666
1667 sym->initialized = 1;
1668 ad.address = symbol_pseudo(ep, sym);
1669
1670 if (sym->initializer && !is_scalar_type(sym)) {
1671 // default zero initialization [6.7.9.21]
1672 // FIXME: this init the whole aggregate while
1673 // only the existing fields need to be initialized.
1674 // FIXME: this init the whole aggregate even if
1675 // all fields arelater explicitely initialized.
1676 struct expression *expr = sym->initializer;
1677 ad.pos = expr->pos;
1678 ad.result_type = sym;
1679 ad.source_type = base_type(sym);
1680 ad.address = symbol_pseudo(ep, sym);
1681 linearize_store_gen(ep, value_pseudo(0), &ad);
1682 }
1683
1684 value = linearize_initializer(ep, sym->initializer, &ad);
1685 finish_address_gen(ep, &ad);
1686 return value;
1687 }
1688
1689 static pseudo_t linearize_compound_statement(struct entrypoint *ep, struct statement *stmt)
1690 {
1691 pseudo_t pseudo;
1692 struct statement *s;
1693 struct symbol *ret = stmt->ret;
1694
1695 pseudo = VOID;
1696 FOR_EACH_PTR(stmt->stmts, s) {
1697 pseudo = linearize_statement(ep, s);
1698 } END_FOR_EACH_PTR(s);
1699
1700 if (ret) {
1701 struct basic_block *bb = add_label(ep, ret);
1702 struct instruction *phi_node = first_instruction(bb->insns);
1703
1704 if (!phi_node)
1705 return pseudo;
1706
1707 if (pseudo_list_size(phi_node->phi_list)==1) {
1708 pseudo = first_pseudo(phi_node->phi_list);
1709 assert(pseudo->type == PSEUDO_PHI);
1710 return pseudo->def->src1;
1711 }
1712 return phi_node->target;
1713 }
1714
1715 return pseudo;
1716 }
1717
1718 static pseudo_t linearize_inlined_call(struct entrypoint *ep, struct statement *stmt)
1719 {
1720 struct instruction *insn = alloc_instruction(OP_INLINED_CALL, 0);
1721 struct statement *args = stmt->args;
1722 struct basic_block *bb;
1723 pseudo_t pseudo;
1724
1725 if (args) {
1726 struct symbol *sym;
1727
1728 concat_symbol_list(args->declaration, &ep->syms);
1729 FOR_EACH_PTR(args->declaration, sym) {
1730 pseudo_t value = linearize_one_symbol(ep, sym);
1731 use_pseudo(insn, value, add_pseudo(&insn->arguments, value));
1732 } END_FOR_EACH_PTR(sym);
1733 }
1734
1735 insn->target = pseudo = linearize_compound_statement(ep, stmt);
1736 use_pseudo(insn, symbol_pseudo(ep, stmt->inline_fn), &insn->func);
1737 bb = ep->active;
1738 if (bb && !bb->insns)
1739 bb->pos = stmt->pos;
1740 add_one_insn(ep, insn);
1741 return pseudo;
1742 }
1743
1744 static pseudo_t linearize_context(struct entrypoint *ep, struct statement *stmt)
1745 {
1746 struct instruction *insn = alloc_instruction(OP_CONTEXT, 0);
1747 struct expression *expr = stmt->expression;
1748 int value = 0;
1749
1750 if (expr->type == EXPR_VALUE)
1751 value = expr->value;
1752
1753 insn->increment = value;
1754 insn->context_expr = stmt->context;
1755 add_one_insn(ep, insn);
1756 return VOID;
1757 }
1758
1759 static pseudo_t linearize_range(struct entrypoint *ep, struct statement *stmt)
1760 {
1761 struct instruction *insn = alloc_instruction(OP_RANGE, 0);
1762
1763 use_pseudo(insn, linearize_expression(ep, stmt->range_expression), &insn->src1);
1764 use_pseudo(insn, linearize_expression(ep, stmt->range_low), &insn->src2);
1765 use_pseudo(insn, linearize_expression(ep, stmt->range_high), &insn->src3);
1766 add_one_insn(ep, insn);
1767 return VOID;
1768 }
1769
1770 ALLOCATOR(asm_rules, "asm rules");
1771 ALLOCATOR(asm_constraint, "asm constraints");
1772
1773 static void add_asm_input(struct entrypoint *ep, struct instruction *insn, struct expression *expr,
1774 const char *constraint, const struct ident *ident)
1775 {
1776 pseudo_t pseudo = linearize_expression(ep, expr);
1777 struct asm_constraint *rule = __alloc_asm_constraint(0);
1778
1779 rule->ident = ident;
1780 rule->constraint = constraint;
1781 use_pseudo(insn, pseudo, &rule->pseudo);
1782 add_ptr_list(&insn->asm_rules->inputs, rule);
1783 }
1784
1785 static void add_asm_output(struct entrypoint *ep, struct instruction *insn, struct expression *expr,
1786 const char *constraint, const struct ident *ident)
1787 {
1788 struct access_data ad = { NULL, };
1789 pseudo_t pseudo = alloc_pseudo(insn);
1790 struct asm_constraint *rule;
1791
1792 if (!expr || !linearize_address_gen(ep, expr, &ad))
1793 return;
1794 linearize_store_gen(ep, pseudo, &ad);
1795 finish_address_gen(ep, &ad);
1796 rule = __alloc_asm_constraint(0);
1797 rule->ident = ident;
1798 rule->constraint = constraint;
1799 use_pseudo(insn, pseudo, &rule->pseudo);
1800 add_ptr_list(&insn->asm_rules->outputs, rule);
1801 }
1802
1803 static pseudo_t linearize_asm_statement(struct entrypoint *ep, struct statement *stmt)
1804 {
1805 int state;
1806 struct expression *expr;
1807 struct instruction *insn;
1808 struct asm_rules *rules;
1809 const char *constraint;
1810 struct ident *ident;
1811
1812 insn = alloc_instruction(OP_ASM, 0);
1813 expr = stmt->asm_string;
1814 if (!expr || expr->type != EXPR_STRING) {
1815 warning(stmt->pos, "expected string in inline asm");
1816 return VOID;
1817 }
1818 insn->string = expr->string->data;
1819
1820 rules = __alloc_asm_rules(0);
1821 insn->asm_rules = rules;
1822
1823 /* Gather the inputs.. */
1824 state = 0;
1825 ident = NULL;
1826 constraint = NULL;
1827 FOR_EACH_PTR(stmt->asm_inputs, expr) {
1828 switch (state) {
1829 case 0: /* Identifier */
1830 state = 1;
1831 ident = (struct ident *)expr;
1832 continue;
1833
1834 case 1: /* Constraint */
1835 state = 2;
1836 constraint = expr ? expr->string->data : "";
1837 continue;
1838
1839 case 2: /* Expression */
1840 state = 0;
1841 add_asm_input(ep, insn, expr, constraint, ident);
1842 }
1843 } END_FOR_EACH_PTR(expr);
1844
1845 add_one_insn(ep, insn);
1846
1847 /* Assign the outputs */
1848 state = 0;
1849 ident = NULL;
1850 constraint = NULL;
1851 FOR_EACH_PTR(stmt->asm_outputs, expr) {
1852 switch (state) {
1853 case 0: /* Identifier */
1854 state = 1;
1855 ident = (struct ident *)expr;
1856 continue;
1857
1858 case 1: /* Constraint */
1859 state = 2;
1860 constraint = expr ? expr->string->data : "";
1861 continue;
1862
1863 case 2:
1864 state = 0;
1865 add_asm_output(ep, insn, expr, constraint, ident);
1866 }
1867 } END_FOR_EACH_PTR(expr);
1868
1869 return VOID;
1870 }
1871
1872 static int multijmp_cmp(const void *_a, const void *_b)
1873 {
1874 const struct multijmp *a = _a;
1875 const struct multijmp *b = _b;
1876
1877 // "default" case?
1878 if (a->begin > a->end) {
1879 if (b->begin > b->end)
1880 return 0;
1881 return 1;
1882 }
1883 if (b->begin > b->end)
1884 return -1;
1885 if (a->begin == b->begin) {
1886 if (a->end == b->end)
1887 return 0;
1888 return (a->end < b->end) ? -1 : 1;
1889 }
1890 return a->begin < b->begin ? -1 : 1;
1891 }
1892
1893 static void sort_switch_cases(struct instruction *insn)
1894 {
1895 sort_list((struct ptr_list **)&insn->multijmp_list, multijmp_cmp);
1896 }
1897
1898 static pseudo_t linearize_declaration(struct entrypoint *ep, struct statement *stmt)
1899 {
1900 struct symbol *sym;
1901
1902 concat_symbol_list(stmt->declaration, &ep->syms);
1903
1904 FOR_EACH_PTR(stmt->declaration, sym) {
1905 linearize_one_symbol(ep, sym);
1906 } END_FOR_EACH_PTR(sym);
1907 return VOID;
1908 }
1909
1910 static pseudo_t linearize_return(struct entrypoint *ep, struct statement *stmt)
1911 {
1912 struct expression *expr = stmt->expression;
1913 struct basic_block *bb_return = get_bound_block(ep, stmt->ret_target);
1914 struct basic_block *active;
1915 pseudo_t src = linearize_expression(ep, expr);
1916 active = ep->active;
1917 if (active && src != VOID) {
1918 struct instruction *phi_node = first_instruction(bb_return->insns);
1919 pseudo_t phi;
1920 if (!phi_node) {
1921 phi_node = alloc_typed_instruction(OP_PHI, expr->ctype);
1922 phi_node->target = alloc_pseudo(phi_node);
1923 phi_node->bb = bb_return;
1924 add_instruction(&bb_return->insns, phi_node);
1925 }
1926 phi = alloc_phi(active, src, type_size(expr->ctype));
1927 phi->ident = &return_ident;
1928 use_pseudo(phi_node, phi, add_pseudo(&phi_node->phi_list, phi));
1929 }
1930 add_goto(ep, bb_return);
1931 return VOID;
1932 }
1933
1934 static pseudo_t linearize_switch(struct entrypoint *ep, struct statement *stmt)
1935 {
1936 struct symbol *sym;
1937 struct instruction *switch_ins;
1938 struct basic_block *switch_end = alloc_basic_block(ep, stmt->pos);
1939 struct basic_block *active, *default_case;
1940 struct multijmp *jmp;
1941 pseudo_t pseudo;
1942
1943 pseudo = linearize_expression(ep, stmt->switch_expression);
1944
1945 active = ep->active;
1946 if (!bb_reachable(active))
1947 return VOID;
1948
1949 switch_ins = alloc_instruction(OP_SWITCH, 0);
1950 use_pseudo(switch_ins, pseudo, &switch_ins->cond);
1951 add_one_insn(ep, switch_ins);
1952 finish_block(ep);
1953
1954 default_case = NULL;
1955 FOR_EACH_PTR(stmt->switch_case->symbol_list, sym) {
1956 struct statement *case_stmt = sym->stmt;
1957 struct basic_block *bb_case = get_bound_block(ep, sym);
1958
1959 if (!case_stmt->case_expression) {
1960 default_case = bb_case;
1961 continue;
1962 } else {
1963 int begin, end;
1964
1965 begin = end = case_stmt->case_expression->value;
1966 if (case_stmt->case_to)
1967 end = case_stmt->case_to->value;
1968 if (begin > end)
1969 jmp = alloc_multijmp(bb_case, end, begin);
1970 else
1971 jmp = alloc_multijmp(bb_case, begin, end);
1972
1973 }
1974 add_multijmp(&switch_ins->multijmp_list, jmp);
1975 add_bb(&bb_case->parents, active);
1976 add_bb(&active->children, bb_case);
1977 } END_FOR_EACH_PTR(sym);
1978
1979 bind_label(stmt->switch_break, switch_end, stmt->pos);
1980
1981 /* And linearize the actual statement */
1982 linearize_statement(ep, stmt->switch_statement);
1983 set_activeblock(ep, switch_end);
1984
1985 if (!default_case)
1986 default_case = switch_end;
1987
1988 jmp = alloc_multijmp(default_case, 1, 0);
1989 add_multijmp(&switch_ins->multijmp_list, jmp);
1990 add_bb(&default_case->parents, active);
1991 add_bb(&active->children, default_case);
1992 sort_switch_cases(switch_ins);
1993
1994 return VOID;
1995 }
1996
1997 static pseudo_t linearize_iterator(struct entrypoint *ep, struct statement *stmt)
1998 {
1999 struct statement *pre_statement = stmt->iterator_pre_statement;
2000 struct expression *pre_condition = stmt->iterator_pre_condition;
2001 struct statement *statement = stmt->iterator_statement;
2002 struct statement *post_statement = stmt->iterator_post_statement;
2003 struct expression *post_condition = stmt->iterator_post_condition;
2004 struct basic_block *loop_top, *loop_body, *loop_continue, *loop_end;
2005 struct symbol *sym;
2006
2007 FOR_EACH_PTR(stmt->iterator_syms, sym) {
2008 linearize_one_symbol(ep, sym);
2009 } END_FOR_EACH_PTR(sym);
2010 concat_symbol_list(stmt->iterator_syms, &ep->syms);
2011 linearize_statement(ep, pre_statement);
2012
2013 loop_body = loop_top = alloc_basic_block(ep, stmt->pos);
2014 loop_continue = alloc_basic_block(ep, stmt->pos);
2015 loop_end = alloc_basic_block(ep, stmt->pos);
2016
2017 /* An empty post-condition means that it's the same as the pre-condition */
2018 if (!post_condition) {
2019 loop_top = alloc_basic_block(ep, stmt->pos);
2020 set_activeblock(ep, loop_top);
2021 }
2022
2023 if (pre_condition)
2024 linearize_cond_branch(ep, pre_condition, loop_body, loop_end);
2025
2026 bind_label(stmt->iterator_continue, loop_continue, stmt->pos);
2027 bind_label(stmt->iterator_break, loop_end, stmt->pos);
2028
2029 set_activeblock(ep, loop_body);
2030 linearize_statement(ep, statement);
2031 add_goto(ep, loop_continue);
2032
2033 set_activeblock(ep, loop_continue);
2034 linearize_statement(ep, post_statement);
2035 if (!post_condition)
2036 add_goto(ep, loop_top);
2037 else
2038 linearize_cond_branch(ep, post_condition, loop_top, loop_end);
2039 set_activeblock(ep, loop_end);
2040
2041 return VOID;
2042 }
2043
2044 pseudo_t linearize_statement(struct entrypoint *ep, struct statement *stmt)
2045 {
2046 struct basic_block *bb;
2047
2048 if (!stmt)
2049 return VOID;
2050
2051 bb = ep->active;
2052 if (bb && !bb->insns)
2053 bb->pos = stmt->pos;
2054 current_pos = stmt->pos;
2055
2056 switch (stmt->type) {
2057 case STMT_NONE:
2058 break;
2059
2060 case STMT_DECLARATION:
2061 return linearize_declaration(ep, stmt);
2062
2063 case STMT_CONTEXT:
2064 return linearize_context(ep, stmt);
2065
2066 case STMT_RANGE:
2067 return linearize_range(ep, stmt);
2068
2069 case STMT_EXPRESSION:
2070 return linearize_expression(ep, stmt->expression);
2071
2072 case STMT_ASM:
2073 return linearize_asm_statement(ep, stmt);
2074
2075 case STMT_RETURN:
2076 return linearize_return(ep, stmt);
2077
2078 case STMT_CASE: {
2079 add_label(ep, stmt->case_label);
2080 linearize_statement(ep, stmt->case_statement);
2081 break;
2082 }
2083
2084 case STMT_LABEL: {
2085 struct symbol *label = stmt->label_identifier;
2086
2087 if (label->used) {
2088 add_label(ep, label);
2089 }
2090 return linearize_statement(ep, stmt->label_statement);
2091 }
2092
2093 case STMT_GOTO: {
2094 struct symbol *sym;
2095 struct expression *expr;
2096 struct instruction *goto_ins;
2097 struct basic_block *active;
2098 pseudo_t pseudo;
2099
2100 active = ep->active;
2101 if (!bb_reachable(active))
2102 break;
2103
2104 if (stmt->goto_label) {
2105 add_goto(ep, get_bound_block(ep, stmt->goto_label));
2106 break;
2107 }
2108
2109 expr = stmt->goto_expression;
2110 if (!expr)
2111 break;
2112
2113 /* This can happen as part of simplification */
2114 if (expr->type == EXPR_LABEL) {
2115 add_goto(ep, get_bound_block(ep, expr->label_symbol));
2116 break;
2117 }
2118
2119 pseudo = linearize_expression(ep, expr);
2120 goto_ins = alloc_instruction(OP_COMPUTEDGOTO, 0);
2121 use_pseudo(goto_ins, pseudo, &goto_ins->target);
2122 add_one_insn(ep, goto_ins);
2123
2124 FOR_EACH_PTR(stmt->target_list, sym) {
2125 struct basic_block *bb_computed = get_bound_block(ep, sym);
2126 struct multijmp *jmp = alloc_multijmp(bb_computed, 1, 0);
2127 add_multijmp(&goto_ins->multijmp_list, jmp);
2128 add_bb(&bb_computed->parents, ep->active);
2129 add_bb(&active->children, bb_computed);
2130 } END_FOR_EACH_PTR(sym);
2131
2132 finish_block(ep);
2133 break;
2134 }
2135
2136 case STMT_COMPOUND:
2137 if (stmt->inline_fn)
2138 return linearize_inlined_call(ep, stmt);
2139 return linearize_compound_statement(ep, stmt);
2140
2141 /*
2142 * This could take 'likely/unlikely' into account, and
2143 * switch the arms around appropriately..
2144 */
2145 case STMT_IF: {
2146 struct basic_block *bb_true, *bb_false, *endif;
2147 struct expression *cond = stmt->if_conditional;
2148
2149 bb_true = alloc_basic_block(ep, stmt->pos);
2150 bb_false = endif = alloc_basic_block(ep, stmt->pos);
2151
2152 linearize_cond_branch(ep, cond, bb_true, bb_false);
2153
2154 set_activeblock(ep, bb_true);
2155 linearize_statement(ep, stmt->if_true);
2156
2157 if (stmt->if_false) {
2158 endif = alloc_basic_block(ep, stmt->pos);
2159 add_goto(ep, endif);
2160 set_activeblock(ep, bb_false);
2161 linearize_statement(ep, stmt->if_false);
2162 }
2163 set_activeblock(ep, endif);
2164 break;
2165 }
2166
2167 case STMT_SWITCH:
2168 return linearize_switch(ep, stmt);
2169
2170 case STMT_ITERATOR:
2171 return linearize_iterator(ep, stmt);
2172
2173 default:
2174 break;
2175 }
2176 return VOID;
2177 }
2178
2179 static struct entrypoint *linearize_fn(struct symbol *sym, struct symbol *base_type)
2180 {
2181 struct entrypoint *ep;
2182 struct basic_block *bb;
2183 struct symbol *arg;
2184 struct instruction *entry;
2185 pseudo_t result;
2186 int i;
2187
2188 if (!base_type->stmt)
2189 return NULL;
2190
2191 ep = alloc_entrypoint();
2192 bb = alloc_basic_block(ep, sym->pos);
2193
2194 ep->name = sym;
2195 sym->ep = ep;
2196 set_activeblock(ep, bb);
2197
2198 entry = alloc_instruction(OP_ENTRY, 0);
2199 add_one_insn(ep, entry);
2200 ep->entry = entry;
2201
2202 concat_symbol_list(base_type->arguments, &ep->syms);
2203
2204 /* FIXME!! We should do something else about varargs.. */
2205 i = 0;
2206 FOR_EACH_PTR(base_type->arguments, arg) {
2207 linearize_argument(ep, arg, ++i);
2208 } END_FOR_EACH_PTR(arg);
2209
2210 result = linearize_statement(ep, base_type->stmt);
2211 if (bb_reachable(ep->active) && !bb_terminated(ep->active)) {
2212 struct symbol *ret_type = base_type->ctype.base_type;
2213 struct instruction *insn = alloc_typed_instruction(OP_RET, ret_type);
2214
2215 if (type_size(ret_type) > 0)
2216 use_pseudo(insn, result, &insn->src);
2217 add_one_insn(ep, insn);
2218 }
2219
2220 if (fdump_linearize) {
2221 if (fdump_linearize == 2)
2222 return ep;
2223 show_entry(ep);
2224 }
2225
2226 /*
2227 * Do trivial flow simplification - branches to
2228 * branches, kill dead basicblocks etc
2229 */
2230 kill_unreachable_bbs(ep);
2231
2232 /*
2233 * Turn symbols into pseudos
2234 */
2235 simplify_symbol_usage(ep);
2236
2237 repeat:
2238 /*
2239 * Remove trivial instructions, and try to CSE
2240 * the rest.
2241 */
2242 do {
2243 cleanup_and_cse(ep);
2244 pack_basic_blocks(ep);
2245 } while (repeat_phase & REPEAT_CSE);
2246
2247 kill_unreachable_bbs(ep);
2248 vrfy_flow(ep);
2249
2250 /* Cleanup */
2251 clear_symbol_pseudos(ep);
2252
2253 /* And track pseudo register usage */
2254 track_pseudo_liveness(ep);
2255
2256 /*
2257 * Some flow optimizations can only effectively
2258 * be done when we've done liveness analysis. But
2259 * if they trigger, we need to start all over
2260 * again
2261 */
2262 if (simplify_flow(ep)) {
2263 clear_liveness(ep);
2264 goto repeat;
2265 }
2266
2267 /* Finally, add deathnotes to pseudos now that we have them */
2268 if (dbg_dead)
2269 track_pseudo_death(ep);
2270
2271 return ep;
2272 }
2273
2274 struct entrypoint *linearize_symbol(struct symbol *sym)
2275 {
2276 struct symbol *base_type;
2277
2278 if (!sym)
2279 return NULL;
2280 current_pos = sym->pos;
2281 base_type = sym->ctype.base_type;
2282 if (!base_type)
2283 return NULL;
2284 if (base_type->type == SYM_FN)
2285 return linearize_fn(sym, base_type);
2286 return NULL;
2287 }
Static analysis for C