?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
remove origval from struct access_data
[smatch.git] / linearize.c
blobb22207265098db44720c74303bc1d0722ee0c8aa
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 unsigned int offset; // byte offset
847 struct position pos;
848 };
849
850 static void finish_address_gen(struct entrypoint *ep, struct access_data *ad)
851 {
852 }
853
854 static int linearize_simple_address(struct entrypoint *ep,
855 struct expression *addr,
856 struct access_data *ad)
857 {
858 if (addr->type == EXPR_SYMBOL) {
859 linearize_one_symbol(ep, addr->symbol);
860 ad->address = symbol_pseudo(ep, addr->symbol);
861 return 1;
862 }
863 if (addr->type == EXPR_BINOP) {
864 if (addr->right->type == EXPR_VALUE) {
865 if (addr->op == '+') {
866 ad->offset += get_expression_value(addr->right);
867 return linearize_simple_address(ep, addr->left, ad);
868 }
869 }
870 }
871 ad->address = linearize_expression(ep, addr);
872 return 1;
873 }
874
875 static struct symbol *base_type(struct symbol *sym)
876 {
877 struct symbol *base = sym;
878
879 if (sym) {
880 if (sym->type == SYM_NODE)
881 base = base->ctype.base_type;
882 if (base->type == SYM_BITFIELD)
883 return base->ctype.base_type;
884 }
885 return sym;
886 }
887
888 static int linearize_address_gen(struct entrypoint *ep,
889 struct expression *expr,
890 struct access_data *ad)
891 {
892 struct symbol *ctype = expr->ctype;
893
894 if (!ctype)
895 return 0;
896 ad->pos = expr->pos;
897 ad->result_type = ctype;
898 ad->source_type = base_type(ctype);
899 if (expr->type == EXPR_PREOP && expr->op == '*')
900 return linearize_simple_address(ep, expr->unop, ad);
901
902 warning(expr->pos, "generating address of non-lvalue (%d)", expr->type);
903 return 0;
904 }
905
906 static pseudo_t add_load(struct entrypoint *ep, struct access_data *ad)
907 {
908 struct instruction *insn;
909 pseudo_t new;
910
911 insn = alloc_typed_instruction(OP_LOAD, ad->source_type);
912 new = alloc_pseudo(insn);
913
914 insn->target = new;
915 insn->offset = ad->offset;
916 use_pseudo(insn, ad->address, &insn->src);
917 add_one_insn(ep, insn);
918 return new;
919 }
920
921 static void add_store(struct entrypoint *ep, struct access_data *ad, pseudo_t value)
922 {
923 struct basic_block *bb = ep->active;
924
925 if (bb_reachable(bb)) {
926 struct instruction *store = alloc_typed_instruction(OP_STORE, ad->source_type);
927 store->offset = ad->offset;
928 use_pseudo(store, value, &store->target);
929 use_pseudo(store, ad->address, &store->src);
930 add_one_insn(ep, store);
931 }
932 }
933
934 static pseudo_t linearize_store_gen(struct entrypoint *ep,
935 pseudo_t value,
936 struct access_data *ad)
937 {
938 pseudo_t store = value;
939
940 if (type_size(ad->source_type) != type_size(ad->result_type)) {
941 struct symbol *ctype = ad->result_type;
942 unsigned int shift = ctype->bit_offset;
943 unsigned int size = ctype->bit_size;
944 pseudo_t orig = add_load(ep, ad);
945 unsigned long long mask = (1ULL << size) - 1;
946
947 if (shift) {
948 store = add_binary_op(ep, ad->source_type, OP_SHL, value, value_pseudo(shift));
949 mask <<= shift;
950 }
951 orig = add_binary_op(ep, ad->source_type, OP_AND, orig, value_pseudo(~mask));
952 store = add_binary_op(ep, ad->source_type, OP_OR, orig, store);
953 }
954 add_store(ep, ad, store);
955 return value;
956 }
957
958 static pseudo_t add_binary_op(struct entrypoint *ep, struct symbol *ctype, int op, pseudo_t left, pseudo_t right)
959 {
960 struct instruction *insn = alloc_typed_instruction(op, ctype);
961 pseudo_t target = alloc_pseudo(insn);
962 insn->target = target;
963 use_pseudo(insn, left, &insn->src1);
964 use_pseudo(insn, right, &insn->src2);
965 add_one_insn(ep, insn);
966 return target;
967 }
968
969 static pseudo_t add_setval(struct entrypoint *ep, struct symbol *ctype, struct expression *val)
970 {
971 struct instruction *insn = alloc_typed_instruction(OP_SETVAL, ctype);
972 pseudo_t target = alloc_pseudo(insn);
973 insn->target = target;
974 insn->val = val;
975 add_one_insn(ep, insn);
976 return target;
977 }
978
979 static pseudo_t add_symbol_address(struct entrypoint *ep, struct symbol *sym)
980 {
981 struct instruction *insn = alloc_instruction(OP_SYMADDR, bits_in_pointer);
982 pseudo_t target = alloc_pseudo(insn);
983
984 insn->target = target;
985 use_pseudo(insn, symbol_pseudo(ep, sym), &insn->symbol);
986 add_one_insn(ep, insn);
987 return target;
988 }
989
990 static pseudo_t linearize_load_gen(struct entrypoint *ep, struct access_data *ad)
991 {
992 struct symbol *ctype = ad->result_type;
993 pseudo_t new = add_load(ep, ad);
994
995 if (ctype->bit_offset) {
996 pseudo_t shift = value_pseudo(ctype->bit_offset);
997 pseudo_t newval = add_binary_op(ep, ad->source_type, OP_LSR, new, shift);
998 new = newval;
999 }
1000 if (ctype->bit_size != type_size(ad->source_type))
1001 new = cast_pseudo(ep, new, ad->source_type, ad->result_type);
1002 return new;
1003 }
1004
1005 static pseudo_t linearize_access(struct entrypoint *ep, struct expression *expr)
1006 {
1007 struct access_data ad = { NULL, };
1008 pseudo_t value;
1009
1010 if (!linearize_address_gen(ep, expr, &ad))
1011 return VOID;
1012 value = linearize_load_gen(ep, &ad);
1013 finish_address_gen(ep, &ad);
1014 return value;
1015 }
1016
1017 /* FIXME: FP */
1018 static pseudo_t linearize_inc_dec(struct entrypoint *ep, struct expression *expr, int postop)
1019 {
1020 struct access_data ad = { NULL, };
1021 pseudo_t old, new, one;
1022 int op = expr->op == SPECIAL_INCREMENT ? OP_ADD : OP_SUB;
1023
1024 if (!linearize_address_gen(ep, expr->unop, &ad))
1025 return VOID;
1026
1027 old = linearize_load_gen(ep, &ad);
1028 one = value_pseudo(expr->op_value);
1029 new = add_binary_op(ep, expr->ctype, op, old, one);
1030 linearize_store_gen(ep, new, &ad);
1031 finish_address_gen(ep, &ad);
1032 return postop ? old : new;
1033 }
1034
1035 static pseudo_t add_uniop(struct entrypoint *ep, struct expression *expr, int op, pseudo_t src)
1036 {
1037 struct instruction *insn = alloc_typed_instruction(op, expr->ctype);
1038 pseudo_t new = alloc_pseudo(insn);
1039
1040 insn->target = new;
1041 use_pseudo(insn, src, &insn->src1);
1042 add_one_insn(ep, insn);
1043 return new;
1044 }
1045
1046 static pseudo_t linearize_slice(struct entrypoint *ep, struct expression *expr)
1047 {
1048 pseudo_t pre = linearize_expression(ep, expr->base);
1049 struct instruction *insn = alloc_typed_instruction(OP_SLICE, expr->ctype);
1050 pseudo_t new = alloc_pseudo(insn);
1051
1052 insn->target = new;
1053 insn->from = expr->r_bitpos;
1054 insn->len = expr->r_nrbits;
1055 use_pseudo(insn, pre, &insn->base);
1056 add_one_insn(ep, insn);
1057 return new;
1058 }
1059
1060 static pseudo_t linearize_regular_preop(struct entrypoint *ep, struct expression *expr)
1061 {
1062 pseudo_t pre = linearize_expression(ep, expr->unop);
1063 switch (expr->op) {
1064 case '+':
1065 return pre;
1066 case '!': {
1067 pseudo_t zero = value_pseudo(0);
1068 return add_binary_op(ep, expr->ctype, OP_SET_EQ, pre, zero);
1069 }
1070 case '~':
1071 return add_uniop(ep, expr, OP_NOT, pre);
1072 case '-':
1073 return add_uniop(ep, expr, OP_NEG, pre);
1074 }
1075 return VOID;
1076 }
1077
1078 static pseudo_t linearize_preop(struct entrypoint *ep, struct expression *expr)
1079 {
1080 /*
1081 * '*' is an lvalue access, and is fundamentally different
1082 * from an arithmetic operation. Maybe it should have an
1083 * expression type of its own..
1084 */
1085 if (expr->op == '*')
1086 return linearize_access(ep, expr);
1087 if (expr->op == SPECIAL_INCREMENT || expr->op == SPECIAL_DECREMENT)
1088 return linearize_inc_dec(ep, expr, 0);
1089 return linearize_regular_preop(ep, expr);
1090 }
1091
1092 static pseudo_t linearize_postop(struct entrypoint *ep, struct expression *expr)
1093 {
1094 return linearize_inc_dec(ep, expr, 1);
1095 }
1096
1097 /*
1098 * Casts to pointers are "less safe" than other casts, since
1099 * they imply type-unsafe accesses. "void *" is a special
1100 * case, since you can't access through it anyway without another
1101 * cast.
1102 */
1103 static struct instruction *alloc_cast_instruction(struct symbol *src, struct symbol *ctype)
1104 {
1105 int opcode = OP_CAST;
1106 struct symbol *base = ctype;
1107
1108 if (src->ctype.modifiers & MOD_SIGNED)
1109 opcode = OP_SCAST;
1110 if (base->type == SYM_NODE)
1111 base = base->ctype.base_type;
1112 if (base->type == SYM_PTR) {
1113 base = base->ctype.base_type;
1114 if (base != &void_ctype)
1115 opcode = OP_PTRCAST;
1116 } else if (base->ctype.base_type == &fp_type)
1117 opcode = OP_FPCAST;
1118 return alloc_typed_instruction(opcode, ctype);
1119 }
1120
1121 static pseudo_t cast_pseudo(struct entrypoint *ep, pseudo_t src, struct symbol *from, struct symbol *to)
1122 {
1123 pseudo_t result;
1124 struct instruction *insn;
1125
1126 if (src == VOID)
1127 return VOID;
1128 if (!from || !to)
1129 return VOID;
1130 if (from->bit_size < 0 || to->bit_size < 0)
1131 return VOID;
1132 insn = alloc_cast_instruction(from, to);
1133 result = alloc_pseudo(insn);
1134 insn->target = result;
1135 insn->orig_type = from;
1136 use_pseudo(insn, src, &insn->src);
1137 add_one_insn(ep, insn);
1138 return result;
1139 }
1140
1141 static int opcode_sign(int opcode, struct symbol *ctype)
1142 {
1143 if (ctype && (ctype->ctype.modifiers & MOD_SIGNED)) {
1144 switch(opcode) {
1145 case OP_MULU: case OP_DIVU: case OP_MODU: case OP_LSR:
1146 opcode++;
1147 }
1148 }
1149 return opcode;
1150 }
1151
1152 static pseudo_t linearize_assignment(struct entrypoint *ep, struct expression *expr)
1153 {
1154 struct access_data ad = { NULL, };
1155 struct expression *target = expr->left;
1156 struct expression *src = expr->right;
1157 struct symbol *ctype;
1158 pseudo_t value;
1159
1160 value = linearize_expression(ep, src);
1161 if (!target || !linearize_address_gen(ep, target, &ad))
1162 return value;
1163 if (expr->op != '=') {
1164 pseudo_t oldvalue = linearize_load_gen(ep, &ad);
1165 pseudo_t dst;
1166 static const int op_trans[] = {
1167 [SPECIAL_ADD_ASSIGN - SPECIAL_BASE] = OP_ADD,
1168 [SPECIAL_SUB_ASSIGN - SPECIAL_BASE] = OP_SUB,
1169 [SPECIAL_MUL_ASSIGN - SPECIAL_BASE] = OP_MULU,
1170 [SPECIAL_DIV_ASSIGN - SPECIAL_BASE] = OP_DIVU,
1171 [SPECIAL_MOD_ASSIGN - SPECIAL_BASE] = OP_MODU,
1172 [SPECIAL_SHL_ASSIGN - SPECIAL_BASE] = OP_SHL,
1173 [SPECIAL_SHR_ASSIGN - SPECIAL_BASE] = OP_LSR,
1174 [SPECIAL_AND_ASSIGN - SPECIAL_BASE] = OP_AND,
1175 [SPECIAL_OR_ASSIGN - SPECIAL_BASE] = OP_OR,
1176 [SPECIAL_XOR_ASSIGN - SPECIAL_BASE] = OP_XOR
1177 };
1178 int opcode;
1179
1180 if (!src)
1181 return VOID;
1182
1183 ctype = src->ctype;
1184 oldvalue = cast_pseudo(ep, oldvalue, target->ctype, ctype);
1185 opcode = opcode_sign(op_trans[expr->op - SPECIAL_BASE], ctype);
1186 dst = add_binary_op(ep, ctype, opcode, oldvalue, value);
1187 value = cast_pseudo(ep, dst, ctype, expr->ctype);
1188 }
1189 value = linearize_store_gen(ep, value, &ad);
1190 finish_address_gen(ep, &ad);
1191 return value;
1192 }
1193
1194 static pseudo_t linearize_call_expression(struct entrypoint *ep, struct expression *expr)
1195 {
1196 struct expression *arg, *fn;
1197 struct instruction *insn = alloc_typed_instruction(OP_CALL, expr->ctype);
1198 pseudo_t retval, call;
1199 struct ctype *ctype = NULL;
1200 struct symbol *fntype;
1201 struct context *context;
1202
1203 if (!expr->ctype) {
1204 warning(expr->pos, "call with no type!");
1205 return VOID;
1206 }
1207
1208 FOR_EACH_PTR(expr->args, arg) {
1209 pseudo_t new = linearize_expression(ep, arg);
1210 use_pseudo(insn, new, add_pseudo(&insn->arguments, new));
1211 } END_FOR_EACH_PTR(arg);
1212
1213 fn = expr->fn;
1214
1215 if (fn->ctype)
1216 ctype = &fn->ctype->ctype;
1217
1218 fntype = fn->ctype;
1219 if (fntype) {
1220 if (fntype->type == SYM_NODE)
1221 fntype = fntype->ctype.base_type;
1222 }
1223 insn->fntype = fntype;
1224
1225 if (fn->type == EXPR_PREOP) {
1226 if (fn->unop->type == EXPR_SYMBOL) {
1227 struct symbol *sym = fn->unop->symbol;
1228 if (sym->ctype.base_type->type == SYM_FN)
1229 fn = fn->unop;
1230 }
1231 }
1232 if (fn->type == EXPR_SYMBOL) {
1233 call = symbol_pseudo(ep, fn->symbol);
1234 } else {
1235 call = linearize_expression(ep, fn);
1236 }
1237 use_pseudo(insn, call, &insn->func);
1238 retval = VOID;
1239 if (expr->ctype != &void_ctype)
1240 retval = alloc_pseudo(insn);
1241 insn->target = retval;
1242 add_one_insn(ep, insn);
1243
1244 if (ctype) {
1245 FOR_EACH_PTR(ctype->contexts, context) {
1246 int in = context->in;
1247 int out = context->out;
1248 int check = 0;
1249 int context_diff;
1250 if (in < 0) {
1251 check = 1;
1252 in = 0;
1253 }
1254 if (out < 0) {
1255 check = 0;
1256 out = 0;
1257 }
1258 context_diff = out - in;
1259 if (check || context_diff) {
1260 insn = alloc_instruction(OP_CONTEXT, 0);
1261 insn->increment = context_diff;
1262 insn->check = check;
1263 insn->context_expr = context->context;
1264 add_one_insn(ep, insn);
1265 }
1266 } END_FOR_EACH_PTR(context);
1267 }
1268
1269 return retval;
1270 }
1271
1272 static pseudo_t linearize_binop(struct entrypoint *ep, struct expression *expr)
1273 {
1274 pseudo_t src1, src2, dst;
1275 static const int opcode[] = {
1276 ['+'] = OP_ADD, ['-'] = OP_SUB,
1277 ['*'] = OP_MULU, ['/'] = OP_DIVU,
1278 ['%'] = OP_MODU, ['&'] = OP_AND,
1279 ['|'] = OP_OR, ['^'] = OP_XOR,
1280 [SPECIAL_LEFTSHIFT] = OP_SHL,
1281 [SPECIAL_RIGHTSHIFT] = OP_LSR,
1282 [SPECIAL_LOGICAL_AND] = OP_AND_BOOL,
1283 [SPECIAL_LOGICAL_OR] = OP_OR_BOOL,
1284 };
1285 int op;
1286
1287 src1 = linearize_expression(ep, expr->left);
1288 src2 = linearize_expression(ep, expr->right);
1289 op = opcode_sign(opcode[expr->op], expr->ctype);
1290 dst = add_binary_op(ep, expr->ctype, op, src1, src2);
1291 return dst;
1292 }
1293
1294 static pseudo_t linearize_logical_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false);
1295
1296 pseudo_t linearize_cond_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false);
1297
1298 static pseudo_t linearize_select(struct entrypoint *ep, struct expression *expr)
1299 {
1300 pseudo_t cond, true, false, res;
1301 struct instruction *insn;
1302
1303 true = linearize_expression(ep, expr->cond_true);
1304 false = linearize_expression(ep, expr->cond_false);
1305 cond = linearize_expression(ep, expr->conditional);
1306
1307 insn = alloc_typed_instruction(OP_SEL, expr->ctype);
1308 if (!expr->cond_true)
1309 true = cond;
1310 use_pseudo(insn, cond, &insn->src1);
1311 use_pseudo(insn, true, &insn->src2);
1312 use_pseudo(insn, false, &insn->src3);
1313
1314 res = alloc_pseudo(insn);
1315 insn->target = res;
1316 add_one_insn(ep, insn);
1317 return res;
1318 }
1319
1320 static pseudo_t add_join_conditional(struct entrypoint *ep, struct expression *expr,
1321 pseudo_t phi1, pseudo_t phi2)
1322 {
1323 pseudo_t target;
1324 struct instruction *phi_node;
1325
1326 if (phi1 == VOID)
1327 return phi2;
1328 if (phi2 == VOID)
1329 return phi1;
1330
1331 phi_node = alloc_typed_instruction(OP_PHI, expr->ctype);
1332 use_pseudo(phi_node, phi1, add_pseudo(&phi_node->phi_list, phi1));
1333 use_pseudo(phi_node, phi2, add_pseudo(&phi_node->phi_list, phi2));
1334 phi_node->target = target = alloc_pseudo(phi_node);
1335 add_one_insn(ep, phi_node);
1336 return target;
1337 }
1338
1339 static pseudo_t linearize_short_conditional(struct entrypoint *ep, struct expression *expr,
1340 struct expression *cond,
1341 struct expression *expr_false)
1342 {
1343 pseudo_t src1, src2;
1344 struct basic_block *bb_false;
1345 struct basic_block *merge = alloc_basic_block(ep, expr->pos);
1346 pseudo_t phi1, phi2;
1347 int size = type_size(expr->ctype);
1348
1349 if (!expr_false || !ep->active)
1350 return VOID;
1351
1352 bb_false = alloc_basic_block(ep, expr_false->pos);
1353 src1 = linearize_expression(ep, cond);
1354 phi1 = alloc_phi(ep->active, src1, size);
1355 add_branch(ep, expr, src1, merge, bb_false);
1356
1357 set_activeblock(ep, bb_false);
1358 src2 = linearize_expression(ep, expr_false);
1359 phi2 = alloc_phi(ep->active, src2, size);
1360 set_activeblock(ep, merge);
1361
1362 return add_join_conditional(ep, expr, phi1, phi2);
1363 }
1364
1365 static pseudo_t linearize_conditional(struct entrypoint *ep, struct expression *expr,
1366 struct expression *cond,
1367 struct expression *expr_true,
1368 struct expression *expr_false)
1369 {
1370 pseudo_t src1, src2;
1371 pseudo_t phi1, phi2;
1372 struct basic_block *bb_true, *bb_false, *merge;
1373 int size = type_size(expr->ctype);
1374
1375 if (!cond || !expr_true || !expr_false || !ep->active)
1376 return VOID;
1377 bb_true = alloc_basic_block(ep, expr_true->pos);
1378 bb_false = alloc_basic_block(ep, expr_false->pos);
1379 merge = alloc_basic_block(ep, expr->pos);
1380
1381 linearize_cond_branch(ep, cond, bb_true, bb_false);
1382
1383 set_activeblock(ep, bb_true);
1384 src1 = linearize_expression(ep, expr_true);
1385 phi1 = alloc_phi(ep->active, src1, size);
1386 add_goto(ep, merge);
1387
1388 set_activeblock(ep, bb_false);
1389 src2 = linearize_expression(ep, expr_false);
1390 phi2 = alloc_phi(ep->active, src2, size);
1391 set_activeblock(ep, merge);
1392
1393 return add_join_conditional(ep, expr, phi1, phi2);
1394 }
1395
1396 static pseudo_t linearize_logical(struct entrypoint *ep, struct expression *expr)
1397 {
1398 struct expression *shortcut;
1399
1400 shortcut = alloc_const_expression(expr->pos, expr->op == SPECIAL_LOGICAL_OR);
1401 shortcut->ctype = expr->ctype;
1402 if (expr->op == SPECIAL_LOGICAL_OR)
1403 return linearize_conditional(ep, expr, expr->left, shortcut, expr->right);
1404 return linearize_conditional(ep, expr, expr->left, expr->right, shortcut);
1405 }
1406
1407 static pseudo_t linearize_compare(struct entrypoint *ep, struct expression *expr)
1408 {
1409 static const int cmpop[] = {
1410 ['>'] = OP_SET_GT, ['<'] = OP_SET_LT,
1411 [SPECIAL_EQUAL] = OP_SET_EQ,
1412 [SPECIAL_NOTEQUAL] = OP_SET_NE,
1413 [SPECIAL_GTE] = OP_SET_GE,
1414 [SPECIAL_LTE] = OP_SET_LE,
1415 [SPECIAL_UNSIGNED_LT] = OP_SET_B,
1416 [SPECIAL_UNSIGNED_GT] = OP_SET_A,
1417 [SPECIAL_UNSIGNED_LTE] = OP_SET_BE,
1418 [SPECIAL_UNSIGNED_GTE] = OP_SET_AE,
1419 };
1420
1421 pseudo_t src1 = linearize_expression(ep, expr->left);
1422 pseudo_t src2 = linearize_expression(ep, expr->right);
1423 pseudo_t dst = add_binary_op(ep, expr->ctype, cmpop[expr->op], src1, src2);
1424 return dst;
1425 }
1426
1427
1428 pseudo_t linearize_cond_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false)
1429 {
1430 pseudo_t cond;
1431
1432 if (!expr || !bb_reachable(ep->active))
1433 return VOID;
1434
1435 switch (expr->type) {
1436
1437 case EXPR_STRING:
1438 case EXPR_VALUE:
1439 add_goto(ep, expr->value ? bb_true : bb_false);
1440 return VOID;
1441
1442 case EXPR_FVALUE:
1443 add_goto(ep, expr->fvalue ? bb_true : bb_false);
1444 return VOID;
1445
1446 case EXPR_LOGICAL:
1447 linearize_logical_branch(ep, expr, bb_true, bb_false);
1448 return VOID;
1449
1450 case EXPR_COMPARE:
1451 cond = linearize_compare(ep, expr);
1452 add_branch(ep, expr, cond, bb_true, bb_false);
1453 break;
1454
1455 case EXPR_PREOP:
1456 if (expr->op == '!')
1457 return linearize_cond_branch(ep, expr->unop, bb_false, bb_true);
1458 /* fall through */
1459 default: {
1460 cond = linearize_expression(ep, expr);
1461 add_branch(ep, expr, cond, bb_true, bb_false);
1462
1463 return VOID;
1464 }
1465 }
1466 return VOID;
1467 }
1468
1469
1470
1471 static pseudo_t linearize_logical_branch(struct entrypoint *ep, struct expression *expr, struct basic_block *bb_true, struct basic_block *bb_false)
1472 {
1473 struct basic_block *next = alloc_basic_block(ep, expr->pos);
1474
1475 if (expr->op == SPECIAL_LOGICAL_OR)
1476 linearize_cond_branch(ep, expr->left, bb_true, next);
1477 else
1478 linearize_cond_branch(ep, expr->left, next, bb_false);
1479 set_activeblock(ep, next);
1480 linearize_cond_branch(ep, expr->right, bb_true, bb_false);
1481 return VOID;
1482 }
1483
1484 static pseudo_t linearize_cast(struct entrypoint *ep, struct expression *expr)
1485 {
1486 pseudo_t src;
1487 struct expression *orig = expr->cast_expression;
1488
1489 if (!orig)
1490 return VOID;
1491
1492 src = linearize_expression(ep, orig);
1493 return cast_pseudo(ep, src, orig->ctype, expr->ctype);
1494 }
1495
1496 static pseudo_t linearize_position(struct entrypoint *ep, struct expression *pos, struct access_data *ad)
1497 {
1498 struct expression *init_expr = pos->init_expr;
1499
1500 ad->offset = pos->init_offset;
1501 ad->source_type = base_type(init_expr->ctype);
1502 ad->result_type = init_expr->ctype;
1503 return linearize_initializer(ep, init_expr, ad);
1504 }
1505
1506 static pseudo_t linearize_initializer(struct entrypoint *ep, struct expression *initializer, struct access_data *ad)
1507 {
1508 switch (initializer->type) {
1509 case EXPR_INITIALIZER: {
1510 struct expression *expr;
1511 FOR_EACH_PTR(initializer->expr_list, expr) {
1512 linearize_initializer(ep, expr, ad);
1513 } END_FOR_EACH_PTR(expr);
1514 break;
1515 }
1516 case EXPR_POS:
1517 linearize_position(ep, initializer, ad);
1518 break;
1519 default: {
1520 pseudo_t value = linearize_expression(ep, initializer);
1521 ad->source_type = base_type(initializer->ctype);
1522 ad->result_type = initializer->ctype;
1523 linearize_store_gen(ep, value, ad);
1524 return value;
1525 }
1526 }
1527
1528 return VOID;
1529 }
1530
1531 static void linearize_argument(struct entrypoint *ep, struct symbol *arg, int nr)
1532 {
1533 struct access_data ad = { NULL, };
1534
1535 ad.source_type = arg;
1536 ad.result_type = arg;
1537 ad.address = symbol_pseudo(ep, arg);
1538 linearize_store_gen(ep, argument_pseudo(ep, nr), &ad);
1539 finish_address_gen(ep, &ad);
1540 }
1541
1542 pseudo_t linearize_expression(struct entrypoint *ep, struct expression *expr)
1543 {
1544 if (!expr)
1545 return VOID;
1546
1547 current_pos = expr->pos;
1548 switch (expr->type) {
1549 case EXPR_SYMBOL:
1550 linearize_one_symbol(ep, expr->symbol);
1551 return add_symbol_address(ep, expr->symbol);
1552
1553 case EXPR_VALUE:
1554 return value_pseudo(expr->value);
1555
1556 case EXPR_STRING: case EXPR_FVALUE: case EXPR_LABEL:
1557 return add_setval(ep, expr->ctype, expr);
1558
1559 case EXPR_STATEMENT:
1560 return linearize_statement(ep, expr->statement);
1561
1562 case EXPR_CALL:
1563 return linearize_call_expression(ep, expr);
1564
1565 case EXPR_BINOP:
1566 return linearize_binop(ep, expr);
1567
1568 case EXPR_LOGICAL:
1569 return linearize_logical(ep, expr);
1570
1571 case EXPR_COMPARE:
1572 return linearize_compare(ep, expr);
1573
1574 case EXPR_SELECT:
1575 return linearize_select(ep, expr);
1576
1577 case EXPR_CONDITIONAL:
1578 if (!expr->cond_true)
1579 return linearize_short_conditional(ep, expr, expr->conditional, expr->cond_false);
1580
1581 return linearize_conditional(ep, expr, expr->conditional,
1582 expr->cond_true, expr->cond_false);
1583
1584 case EXPR_COMMA:
1585 linearize_expression(ep, expr->left);
1586 return linearize_expression(ep, expr->right);
1587
1588 case EXPR_ASSIGNMENT:
1589 return linearize_assignment(ep, expr);
1590
1591 case EXPR_PREOP:
1592 return linearize_preop(ep, expr);
1593
1594 case EXPR_POSTOP:
1595 return linearize_postop(ep, expr);
1596
1597 case EXPR_CAST:
1598 case EXPR_FORCE_CAST:
1599 case EXPR_IMPLIED_CAST:
1600 return linearize_cast(ep, expr);
1601
1602 case EXPR_SLICE:
1603 return linearize_slice(ep, expr);
1604
1605 case EXPR_INITIALIZER:
1606 case EXPR_POS:
1607 warning(expr->pos, "unexpected initializer expression (%d %d)", expr->type, expr->op);
1608 return VOID;
1609 default:
1610 warning(expr->pos, "unknown expression (%d %d)", expr->type, expr->op);
1611 return VOID;
1612 }
1613 return VOID;
1614 }
1615
1616 static pseudo_t linearize_one_symbol(struct entrypoint *ep, struct symbol *sym)
1617 {
1618 struct access_data ad = { NULL, };
1619 pseudo_t value;
1620
1621 if (!sym || !sym->initializer || sym->initialized)
1622 return VOID;
1623
1624 /* We need to output these puppies some day too.. */
1625 if (sym->ctype.modifiers & (MOD_STATIC | MOD_TOPLEVEL))
1626 return VOID;
1627
1628 sym->initialized = 1;
1629 ad.address = symbol_pseudo(ep, sym);
1630
1631 if (sym->initializer && !is_scalar_type(sym)) {
1632 // default zero initialization [6.7.9.21]
1633 // FIXME: this init the whole aggregate while
1634 // only the existing fields need to be initialized.
1635 // FIXME: this init the whole aggregate even if
1636 // all fields arelater explicitely initialized.
1637 struct expression *expr = sym->initializer;
1638 ad.pos = expr->pos;
1639 ad.result_type = sym;
1640 ad.source_type = base_type(sym);
1641 ad.address = symbol_pseudo(ep, sym);
1642 linearize_store_gen(ep, value_pseudo(0), &ad);
1643 }
1644
1645 value = linearize_initializer(ep, sym->initializer, &ad);
1646 finish_address_gen(ep, &ad);
1647 return value;
1648 }
1649
1650 static pseudo_t linearize_compound_statement(struct entrypoint *ep, struct statement *stmt)
1651 {
1652 pseudo_t pseudo;
1653 struct statement *s;
1654 struct symbol *ret = stmt->ret;
1655
1656 pseudo = VOID;
1657 FOR_EACH_PTR(stmt->stmts, s) {
1658 pseudo = linearize_statement(ep, s);
1659 } END_FOR_EACH_PTR(s);
1660
1661 if (ret) {
1662 struct basic_block *bb = add_label(ep, ret);
1663 struct instruction *phi_node = first_instruction(bb->insns);
1664
1665 if (!phi_node)
1666 return pseudo;
1667
1668 if (pseudo_list_size(phi_node->phi_list)==1) {
1669 pseudo = first_pseudo(phi_node->phi_list);
1670 assert(pseudo->type == PSEUDO_PHI);
1671 return pseudo->def->src1;
1672 }
1673 return phi_node->target;
1674 }
1675
1676 return pseudo;
1677 }
1678
1679 static pseudo_t linearize_inlined_call(struct entrypoint *ep, struct statement *stmt)
1680 {
1681 struct instruction *insn = alloc_instruction(OP_INLINED_CALL, 0);
1682 struct statement *args = stmt->args;
1683 struct basic_block *bb;
1684 pseudo_t pseudo;
1685
1686 if (args) {
1687 struct symbol *sym;
1688
1689 concat_symbol_list(args->declaration, &ep->syms);
1690 FOR_EACH_PTR(args->declaration, sym) {
1691 pseudo_t value = linearize_one_symbol(ep, sym);
1692 use_pseudo(insn, value, add_pseudo(&insn->arguments, value));
1693 } END_FOR_EACH_PTR(sym);
1694 }
1695
1696 insn->target = pseudo = linearize_compound_statement(ep, stmt);
1697 use_pseudo(insn, symbol_pseudo(ep, stmt->inline_fn), &insn->func);
1698 bb = ep->active;
1699 if (bb && !bb->insns)
1700 bb->pos = stmt->pos;
1701 add_one_insn(ep, insn);
1702 return pseudo;
1703 }
1704
1705 static pseudo_t linearize_context(struct entrypoint *ep, struct statement *stmt)
1706 {
1707 struct instruction *insn = alloc_instruction(OP_CONTEXT, 0);
1708 struct expression *expr = stmt->expression;
1709 int value = 0;
1710
1711 if (expr->type == EXPR_VALUE)
1712 value = expr->value;
1713
1714 insn->increment = value;
1715 insn->context_expr = stmt->context;
1716 add_one_insn(ep, insn);
1717 return VOID;
1718 }
1719
1720 static pseudo_t linearize_range(struct entrypoint *ep, struct statement *stmt)
1721 {
1722 struct instruction *insn = alloc_instruction(OP_RANGE, 0);
1723
1724 use_pseudo(insn, linearize_expression(ep, stmt->range_expression), &insn->src1);
1725 use_pseudo(insn, linearize_expression(ep, stmt->range_low), &insn->src2);
1726 use_pseudo(insn, linearize_expression(ep, stmt->range_high), &insn->src3);
1727 add_one_insn(ep, insn);
1728 return VOID;
1729 }
1730
1731 ALLOCATOR(asm_rules, "asm rules");
1732 ALLOCATOR(asm_constraint, "asm constraints");
1733
1734 static void add_asm_input(struct entrypoint *ep, struct instruction *insn, struct expression *expr,
1735 const char *constraint, const struct ident *ident)
1736 {
1737 pseudo_t pseudo = linearize_expression(ep, expr);
1738 struct asm_constraint *rule = __alloc_asm_constraint(0);
1739
1740 rule->ident = ident;
1741 rule->constraint = constraint;
1742 use_pseudo(insn, pseudo, &rule->pseudo);
1743 add_ptr_list(&insn->asm_rules->inputs, rule);
1744 }
1745
1746 static void add_asm_output(struct entrypoint *ep, struct instruction *insn, struct expression *expr,
1747 const char *constraint, const struct ident *ident)
1748 {
1749 struct access_data ad = { NULL, };
1750 pseudo_t pseudo = alloc_pseudo(insn);
1751 struct asm_constraint *rule;
1752
1753 if (!expr || !linearize_address_gen(ep, expr, &ad))
1754 return;
1755 linearize_store_gen(ep, pseudo, &ad);
1756 finish_address_gen(ep, &ad);
1757 rule = __alloc_asm_constraint(0);
1758 rule->ident = ident;
1759 rule->constraint = constraint;
1760 use_pseudo(insn, pseudo, &rule->pseudo);
1761 add_ptr_list(&insn->asm_rules->outputs, rule);
1762 }
1763
1764 static pseudo_t linearize_asm_statement(struct entrypoint *ep, struct statement *stmt)
1765 {
1766 int state;
1767 struct expression *expr;
1768 struct instruction *insn;
1769 struct asm_rules *rules;
1770 const char *constraint;
1771 struct ident *ident;
1772
1773 insn = alloc_instruction(OP_ASM, 0);
1774 expr = stmt->asm_string;
1775 if (!expr || expr->type != EXPR_STRING) {
1776 warning(stmt->pos, "expected string in inline asm");
1777 return VOID;
1778 }
1779 insn->string = expr->string->data;
1780
1781 rules = __alloc_asm_rules(0);
1782 insn->asm_rules = rules;
1783
1784 /* Gather the inputs.. */
1785 state = 0;
1786 ident = NULL;
1787 constraint = NULL;
1788 FOR_EACH_PTR(stmt->asm_inputs, expr) {
1789 switch (state) {
1790 case 0: /* Identifier */
1791 state = 1;
1792 ident = (struct ident *)expr;
1793 continue;
1794
1795 case 1: /* Constraint */
1796 state = 2;
1797 constraint = expr ? expr->string->data : "";
1798 continue;
1799
1800 case 2: /* Expression */
1801 state = 0;
1802 add_asm_input(ep, insn, expr, constraint, ident);
1803 }
1804 } END_FOR_EACH_PTR(expr);
1805
1806 add_one_insn(ep, insn);
1807
1808 /* Assign the outputs */
1809 state = 0;
1810 ident = NULL;
1811 constraint = NULL;
1812 FOR_EACH_PTR(stmt->asm_outputs, expr) {
1813 switch (state) {
1814 case 0: /* Identifier */
1815 state = 1;
1816 ident = (struct ident *)expr;
1817 continue;
1818
1819 case 1: /* Constraint */
1820 state = 2;
1821 constraint = expr ? expr->string->data : "";
1822 continue;
1823
1824 case 2:
1825 state = 0;
1826 add_asm_output(ep, insn, expr, constraint, ident);
1827 }
1828 } END_FOR_EACH_PTR(expr);
1829
1830 return VOID;
1831 }
1832
1833 static int multijmp_cmp(const void *_a, const void *_b)
1834 {
1835 const struct multijmp *a = _a;
1836 const struct multijmp *b = _b;
1837
1838 // "default" case?
1839 if (a->begin > a->end) {
1840 if (b->begin > b->end)
1841 return 0;
1842 return 1;
1843 }
1844 if (b->begin > b->end)
1845 return -1;
1846 if (a->begin == b->begin) {
1847 if (a->end == b->end)
1848 return 0;
1849 return (a->end < b->end) ? -1 : 1;
1850 }
1851 return a->begin < b->begin ? -1 : 1;
1852 }
1853
1854 static void sort_switch_cases(struct instruction *insn)
1855 {
1856 sort_list((struct ptr_list **)&insn->multijmp_list, multijmp_cmp);
1857 }
1858
1859 static pseudo_t linearize_declaration(struct entrypoint *ep, struct statement *stmt)
1860 {
1861 struct symbol *sym;
1862
1863 concat_symbol_list(stmt->declaration, &ep->syms);
1864
1865 FOR_EACH_PTR(stmt->declaration, sym) {
1866 linearize_one_symbol(ep, sym);
1867 } END_FOR_EACH_PTR(sym);
1868 return VOID;
1869 }
1870
1871 static pseudo_t linearize_return(struct entrypoint *ep, struct statement *stmt)
1872 {
1873 struct expression *expr = stmt->expression;
1874 struct basic_block *bb_return = get_bound_block(ep, stmt->ret_target);
1875 struct basic_block *active;
1876 pseudo_t src = linearize_expression(ep, expr);
1877 active = ep->active;
1878 if (active && src != VOID) {
1879 struct instruction *phi_node = first_instruction(bb_return->insns);
1880 pseudo_t phi;
1881 if (!phi_node) {
1882 phi_node = alloc_typed_instruction(OP_PHI, expr->ctype);
1883 phi_node->target = alloc_pseudo(phi_node);
1884 phi_node->bb = bb_return;
1885 add_instruction(&bb_return->insns, phi_node);
1886 }
1887 phi = alloc_phi(active, src, type_size(expr->ctype));
1888 phi->ident = &return_ident;
1889 use_pseudo(phi_node, phi, add_pseudo(&phi_node->phi_list, phi));
1890 }
1891 add_goto(ep, bb_return);
1892 return VOID;
1893 }
1894
1895 static pseudo_t linearize_switch(struct entrypoint *ep, struct statement *stmt)
1896 {
1897 struct symbol *sym;
1898 struct instruction *switch_ins;
1899 struct basic_block *switch_end = alloc_basic_block(ep, stmt->pos);
1900 struct basic_block *active, *default_case;
1901 struct multijmp *jmp;
1902 pseudo_t pseudo;
1903
1904 pseudo = linearize_expression(ep, stmt->switch_expression);
1905
1906 active = ep->active;
1907 if (!bb_reachable(active))
1908 return VOID;
1909
1910 switch_ins = alloc_instruction(OP_SWITCH, 0);
1911 use_pseudo(switch_ins, pseudo, &switch_ins->cond);
1912 add_one_insn(ep, switch_ins);
1913 finish_block(ep);
1914
1915 default_case = NULL;
1916 FOR_EACH_PTR(stmt->switch_case->symbol_list, sym) {
1917 struct statement *case_stmt = sym->stmt;
1918 struct basic_block *bb_case = get_bound_block(ep, sym);
1919
1920 if (!case_stmt->case_expression) {
1921 default_case = bb_case;
1922 continue;
1923 } else {
1924 int begin, end;
1925
1926 begin = end = case_stmt->case_expression->value;
1927 if (case_stmt->case_to)
1928 end = case_stmt->case_to->value;
1929 if (begin > end)
1930 jmp = alloc_multijmp(bb_case, end, begin);
1931 else
1932 jmp = alloc_multijmp(bb_case, begin, end);
1933
1934 }
1935 add_multijmp(&switch_ins->multijmp_list, jmp);
1936 add_bb(&bb_case->parents, active);
1937 add_bb(&active->children, bb_case);
1938 } END_FOR_EACH_PTR(sym);
1939
1940 bind_label(stmt->switch_break, switch_end, stmt->pos);
1941
1942 /* And linearize the actual statement */
1943 linearize_statement(ep, stmt->switch_statement);
1944 set_activeblock(ep, switch_end);
1945
1946 if (!default_case)
1947 default_case = switch_end;
1948
1949 jmp = alloc_multijmp(default_case, 1, 0);
1950 add_multijmp(&switch_ins->multijmp_list, jmp);
1951 add_bb(&default_case->parents, active);
1952 add_bb(&active->children, default_case);
1953 sort_switch_cases(switch_ins);
1954
1955 return VOID;
1956 }
1957
1958 static pseudo_t linearize_iterator(struct entrypoint *ep, struct statement *stmt)
1959 {
1960 struct statement *pre_statement = stmt->iterator_pre_statement;
1961 struct expression *pre_condition = stmt->iterator_pre_condition;
1962 struct statement *statement = stmt->iterator_statement;
1963 struct statement *post_statement = stmt->iterator_post_statement;
1964 struct expression *post_condition = stmt->iterator_post_condition;
1965 struct basic_block *loop_top, *loop_body, *loop_continue, *loop_end;
1966 struct symbol *sym;
1967
1968 FOR_EACH_PTR(stmt->iterator_syms, sym) {
1969 linearize_one_symbol(ep, sym);
1970 } END_FOR_EACH_PTR(sym);
1971 concat_symbol_list(stmt->iterator_syms, &ep->syms);
1972 linearize_statement(ep, pre_statement);
1973
1974 loop_body = loop_top = alloc_basic_block(ep, stmt->pos);
1975 loop_continue = alloc_basic_block(ep, stmt->pos);
1976 loop_end = alloc_basic_block(ep, stmt->pos);
1977
1978 /* An empty post-condition means that it's the same as the pre-condition */
1979 if (!post_condition) {
1980 loop_top = alloc_basic_block(ep, stmt->pos);
1981 set_activeblock(ep, loop_top);
1982 }
1983
1984 if (pre_condition)
1985 linearize_cond_branch(ep, pre_condition, loop_body, loop_end);
1986
1987 bind_label(stmt->iterator_continue, loop_continue, stmt->pos);
1988 bind_label(stmt->iterator_break, loop_end, stmt->pos);
1989
1990 set_activeblock(ep, loop_body);
1991 linearize_statement(ep, statement);
1992 add_goto(ep, loop_continue);
1993
1994 set_activeblock(ep, loop_continue);
1995 linearize_statement(ep, post_statement);
1996 if (!post_condition)
1997 add_goto(ep, loop_top);
1998 else
1999 linearize_cond_branch(ep, post_condition, loop_top, loop_end);
2000 set_activeblock(ep, loop_end);
2001
2002 return VOID;
2003 }
2004
2005 pseudo_t linearize_statement(struct entrypoint *ep, struct statement *stmt)
2006 {
2007 struct basic_block *bb;
2008
2009 if (!stmt)
2010 return VOID;
2011
2012 bb = ep->active;
2013 if (bb && !bb->insns)
2014 bb->pos = stmt->pos;
2015 current_pos = stmt->pos;
2016
2017 switch (stmt->type) {
2018 case STMT_NONE:
2019 break;
2020
2021 case STMT_DECLARATION:
2022 return linearize_declaration(ep, stmt);
2023
2024 case STMT_CONTEXT:
2025 return linearize_context(ep, stmt);
2026
2027 case STMT_RANGE:
2028 return linearize_range(ep, stmt);
2029
2030 case STMT_EXPRESSION:
2031 return linearize_expression(ep, stmt->expression);
2032
2033 case STMT_ASM:
2034 return linearize_asm_statement(ep, stmt);
2035
2036 case STMT_RETURN:
2037 return linearize_return(ep, stmt);
2038
2039 case STMT_CASE: {
2040 add_label(ep, stmt->case_label);
2041 linearize_statement(ep, stmt->case_statement);
2042 break;
2043 }
2044
2045 case STMT_LABEL: {
2046 struct symbol *label = stmt->label_identifier;
2047
2048 if (label->used) {
2049 add_label(ep, label);
2050 }
2051 return linearize_statement(ep, stmt->label_statement);
2052 }
2053
2054 case STMT_GOTO: {
2055 struct symbol *sym;
2056 struct expression *expr;
2057 struct instruction *goto_ins;
2058 struct basic_block *active;
2059 pseudo_t pseudo;
2060
2061 active = ep->active;
2062 if (!bb_reachable(active))
2063 break;
2064
2065 if (stmt->goto_label) {
2066 add_goto(ep, get_bound_block(ep, stmt->goto_label));
2067 break;
2068 }
2069
2070 expr = stmt->goto_expression;
2071 if (!expr)
2072 break;
2073
2074 /* This can happen as part of simplification */
2075 if (expr->type == EXPR_LABEL) {
2076 add_goto(ep, get_bound_block(ep, expr->label_symbol));
2077 break;
2078 }
2079
2080 pseudo = linearize_expression(ep, expr);
2081 goto_ins = alloc_instruction(OP_COMPUTEDGOTO, 0);
2082 use_pseudo(goto_ins, pseudo, &goto_ins->target);
2083 add_one_insn(ep, goto_ins);
2084
2085 FOR_EACH_PTR(stmt->target_list, sym) {
2086 struct basic_block *bb_computed = get_bound_block(ep, sym);
2087 struct multijmp *jmp = alloc_multijmp(bb_computed, 1, 0);
2088 add_multijmp(&goto_ins->multijmp_list, jmp);
2089 add_bb(&bb_computed->parents, ep->active);
2090 add_bb(&active->children, bb_computed);
2091 } END_FOR_EACH_PTR(sym);
2092
2093 finish_block(ep);
2094 break;
2095 }
2096
2097 case STMT_COMPOUND:
2098 if (stmt->inline_fn)
2099 return linearize_inlined_call(ep, stmt);
2100 return linearize_compound_statement(ep, stmt);
2101
2102 /*
2103 * This could take 'likely/unlikely' into account, and
2104 * switch the arms around appropriately..
2105 */
2106 case STMT_IF: {
2107 struct basic_block *bb_true, *bb_false, *endif;
2108 struct expression *cond = stmt->if_conditional;
2109
2110 bb_true = alloc_basic_block(ep, stmt->pos);
2111 bb_false = endif = alloc_basic_block(ep, stmt->pos);
2112
2113 linearize_cond_branch(ep, cond, bb_true, bb_false);
2114
2115 set_activeblock(ep, bb_true);
2116 linearize_statement(ep, stmt->if_true);
2117
2118 if (stmt->if_false) {
2119 endif = alloc_basic_block(ep, stmt->pos);
2120 add_goto(ep, endif);
2121 set_activeblock(ep, bb_false);
2122 linearize_statement(ep, stmt->if_false);
2123 }
2124 set_activeblock(ep, endif);
2125 break;
2126 }
2127
2128 case STMT_SWITCH:
2129 return linearize_switch(ep, stmt);
2130
2131 case STMT_ITERATOR:
2132 return linearize_iterator(ep, stmt);
2133
2134 default:
2135 break;
2136 }
2137 return VOID;
2138 }
2139
2140 static struct entrypoint *linearize_fn(struct symbol *sym, struct symbol *base_type)
2141 {
2142 struct entrypoint *ep;
2143 struct basic_block *bb;
2144 struct symbol *arg;
2145 struct instruction *entry;
2146 pseudo_t result;
2147 int i;
2148
2149 if (!base_type->stmt)
2150 return NULL;
2151
2152 ep = alloc_entrypoint();
2153 bb = alloc_basic_block(ep, sym->pos);
2154
2155 ep->name = sym;
2156 sym->ep = ep;
2157 set_activeblock(ep, bb);
2158
2159 entry = alloc_instruction(OP_ENTRY, 0);
2160 add_one_insn(ep, entry);
2161 ep->entry = entry;
2162
2163 concat_symbol_list(base_type->arguments, &ep->syms);
2164
2165 /* FIXME!! We should do something else about varargs.. */
2166 i = 0;
2167 FOR_EACH_PTR(base_type->arguments, arg) {
2168 linearize_argument(ep, arg, ++i);
2169 } END_FOR_EACH_PTR(arg);
2170
2171 result = linearize_statement(ep, base_type->stmt);
2172 if (bb_reachable(ep->active) && !bb_terminated(ep->active)) {
2173 struct symbol *ret_type = base_type->ctype.base_type;
2174 struct instruction *insn = alloc_typed_instruction(OP_RET, ret_type);
2175
2176 if (type_size(ret_type) > 0)
2177 use_pseudo(insn, result, &insn->src);
2178 add_one_insn(ep, insn);
2179 }
2180
2181 /*
2182 * Do trivial flow simplification - branches to
2183 * branches, kill dead basicblocks etc
2184 */
2185 kill_unreachable_bbs(ep);
2186
2187 /*
2188 * Turn symbols into pseudos
2189 */
2190 simplify_symbol_usage(ep);
2191
2192 repeat:
2193 /*
2194 * Remove trivial instructions, and try to CSE
2195 * the rest.
2196 */
2197 do {
2198 cleanup_and_cse(ep);
2199 pack_basic_blocks(ep);
2200 } while (repeat_phase & REPEAT_CSE);
2201
2202 kill_unreachable_bbs(ep);
2203 vrfy_flow(ep);
2204
2205 /* Cleanup */
2206 clear_symbol_pseudos(ep);
2207
2208 /* And track pseudo register usage */
2209 track_pseudo_liveness(ep);
2210
2211 /*
2212 * Some flow optimizations can only effectively
2213 * be done when we've done liveness analysis. But
2214 * if they trigger, we need to start all over
2215 * again
2216 */
2217 if (simplify_flow(ep)) {
2218 clear_liveness(ep);
2219 goto repeat;
2220 }
2221
2222 /* Finally, add deathnotes to pseudos now that we have them */
2223 if (dbg_dead)
2224 track_pseudo_death(ep);
2225
2226 return ep;
2227 }
2228
2229 struct entrypoint *linearize_symbol(struct symbol *sym)
2230 {
2231 struct symbol *base_type;
2232
2233 if (!sym)
2234 return NULL;
2235 current_pos = sym->pos;
2236 base_type = sym->ctype.base_type;
2237 if (!base_type)
2238 return NULL;
2239 if (base_type->type == SYM_FN)
2240 return linearize_fn(sym, base_type);
2241 return NULL;
2242 }
Static analysis for C