* string.c (rb_str_splice_0): call rb_str_drop_bytes, not rb_str_drop.
[ruby-svn.git] / iseq.c
blob84c3b741e59a66dab4a25bd97108a7dbedff6c7a
1 /**********************************************************************
3 iseq.c -
5 $Author$
6 created at: 2006-07-11(Tue) 09:00:03 +0900
8 Copyright (C) 2006 Koichi Sasada
10 **********************************************************************/
12 #include "ruby/ruby.h"
13 #include "ruby/node.h"
15 /* #define MARK_FREE_DEBUG 1 */
16 #include "gc.h"
17 #include "vm_core.h"
19 #include "insns.inc"
20 #include "insns_info.inc"
22 /* compile.c */
23 void iseq_compile(VALUE self, NODE *node);
24 int iseq_translate_threaded_code(rb_iseq_t *iseq);
26 VALUE rb_cISeq;
28 static void
29 compile_data_free(struct iseq_compile_data *compile_data)
31 if (compile_data) {
32 struct iseq_compile_data_storage *cur, *next;
33 cur = compile_data->storage_head;
34 while (cur) {
35 next = cur->next;
36 ruby_xfree(cur);
37 cur = next;
39 ruby_xfree(compile_data);
43 static void
44 iseq_free(void *ptr)
46 rb_iseq_t *iseq;
47 RUBY_FREE_ENTER("iseq");
49 if (ptr) {
50 iseq = ptr;
51 if (!iseq->orig) {
52 /* It's possible that strings are freed
53 * GC_INFO("%s @ %s\n", RSTRING_PTR(iseq->name),
54 * RSTRING_PTR(iseq->filename));
56 if (iseq->iseq != iseq->iseq_encoded) {
57 RUBY_FREE_UNLESS_NULL(iseq->iseq_encoded);
60 RUBY_FREE_UNLESS_NULL(iseq->iseq);
61 RUBY_FREE_UNLESS_NULL(iseq->insn_info_table);
62 RUBY_FREE_UNLESS_NULL(iseq->local_table);
63 RUBY_FREE_UNLESS_NULL(iseq->catch_table);
64 RUBY_FREE_UNLESS_NULL(iseq->arg_opt_table);
65 compile_data_free(iseq->compile_data);
67 ruby_xfree(ptr);
69 RUBY_FREE_LEAVE("iseq");
72 static void
73 iseq_mark(void *ptr)
75 rb_iseq_t *iseq;
76 RUBY_MARK_ENTER("iseq");
78 if (ptr) {
79 iseq = ptr;
80 RUBY_GC_INFO("%s @ %s\n", RSTRING_PTR(iseq->name), RSTRING_PTR(iseq->filename));
81 RUBY_MARK_UNLESS_NULL(iseq->mark_ary);
82 RUBY_MARK_UNLESS_NULL(iseq->name);
83 RUBY_MARK_UNLESS_NULL(iseq->filename);
84 RUBY_MARK_UNLESS_NULL((VALUE)iseq->cref_stack);
85 RUBY_MARK_UNLESS_NULL(iseq->klass);
86 RUBY_MARK_UNLESS_NULL(iseq->coverage);
87 /* RUBY_MARK_UNLESS_NULL((VALUE)iseq->node); */
88 /* RUBY_MARK_UNLESS_NULL(iseq->cached_special_block); */
89 RUBY_MARK_UNLESS_NULL(iseq->orig);
91 if (iseq->compile_data != 0) {
92 RUBY_MARK_UNLESS_NULL(iseq->compile_data->mark_ary);
93 RUBY_MARK_UNLESS_NULL(iseq->compile_data->err_info);
94 RUBY_MARK_UNLESS_NULL(iseq->compile_data->catch_table_ary);
97 RUBY_MARK_LEAVE("iseq");
100 static VALUE
101 iseq_alloc(VALUE klass)
103 VALUE volatile obj;
104 rb_iseq_t *iseq;
106 obj = Data_Make_Struct(klass, rb_iseq_t, iseq_mark, iseq_free, iseq);
107 MEMZERO(iseq, rb_iseq_t, 1);
108 return obj;
111 static void
112 set_relation(rb_iseq_t *iseq, const VALUE parent)
114 const int type = iseq->type;
115 rb_thread_t *th = GET_THREAD();
117 /* set class nest stack */
118 if (type == ISEQ_TYPE_TOP) {
119 /* toplevel is private */
120 iseq->cref_stack = NEW_BLOCK(th->top_wrapper ? th->top_wrapper : rb_cObject);
121 iseq->cref_stack->nd_file = 0;
122 iseq->cref_stack->nd_visi = NOEX_PRIVATE;
124 else if (type == ISEQ_TYPE_METHOD || type == ISEQ_TYPE_CLASS) {
125 iseq->cref_stack = NEW_BLOCK(0); /* place holder */
126 iseq->cref_stack->nd_file = 0;
128 else if (RTEST(parent)) {
129 rb_iseq_t *piseq;
130 GetISeqPtr(parent, piseq);
131 iseq->cref_stack = piseq->cref_stack;
134 if (type == ISEQ_TYPE_TOP ||
135 type == ISEQ_TYPE_METHOD || type == ISEQ_TYPE_CLASS) {
136 iseq->local_iseq = iseq;
138 else if (RTEST(parent)) {
139 rb_iseq_t *piseq;
140 GetISeqPtr(parent, piseq);
141 iseq->local_iseq = piseq->local_iseq;
144 if (RTEST(parent)) {
145 rb_iseq_t *piseq;
146 GetISeqPtr(parent, piseq);
147 iseq->parent_iseq = piseq;
151 static VALUE
152 prepare_iseq_build(rb_iseq_t *iseq,
153 VALUE name, VALUE filename,
154 VALUE parent, VALUE type, VALUE block_opt,
155 const rb_compile_option_t *option)
157 OBJ_FREEZE(name);
158 OBJ_FREEZE(filename);
160 iseq->name = name;
161 iseq->filename = filename;
162 iseq->defined_method_id = 0;
163 iseq->mark_ary = rb_ary_new();
164 RBASIC(iseq->mark_ary)->klass = 0;
166 iseq->type = type;
167 iseq->arg_rest = -1;
168 iseq->arg_block = -1;
169 iseq->klass = 0;
172 * iseq->special_block_builder = GC_GUARDED_PTR_REF(block_opt);
173 * iseq->cached_special_block_builder = 0;
174 * iseq->cached_special_block = 0;
177 iseq->compile_data = ALLOC(struct iseq_compile_data);
178 MEMZERO(iseq->compile_data, struct iseq_compile_data, 1);
179 iseq->compile_data->mark_ary = rb_ary_new();
180 RBASIC(iseq->compile_data->mark_ary)->klass = 0;
182 iseq->compile_data->storage_head = iseq->compile_data->storage_current =
183 (struct iseq_compile_data_storage *)
184 ALLOC_N(char, INITIAL_ISEQ_COMPILE_DATA_STORAGE_BUFF_SIZE +
185 sizeof(struct iseq_compile_data_storage));
187 iseq->compile_data->catch_table_ary = rb_ary_new();
188 iseq->compile_data->storage_head->pos = 0;
189 iseq->compile_data->storage_head->next = 0;
190 iseq->compile_data->storage_head->size =
191 INITIAL_ISEQ_COMPILE_DATA_STORAGE_BUFF_SIZE;
192 iseq->compile_data->storage_head->buff =
193 (char *)(&iseq->compile_data->storage_head->buff + 1);
194 iseq->compile_data->option = option;
196 set_relation(iseq, parent);
198 iseq->coverage = Qfalse;
199 if (!GET_THREAD()->parse_in_eval) {
200 extern VALUE rb_get_coverages(void);
201 VALUE coverages = rb_get_coverages();
202 if (RTEST(coverages)) {
203 iseq->coverage = rb_hash_lookup(coverages, filename);
204 if (NIL_P(iseq->coverage)) iseq->coverage = Qfalse;
208 return Qtrue;
211 static VALUE
212 cleanup_iseq_build(rb_iseq_t *iseq)
214 struct iseq_compile_data *data = iseq->compile_data;
215 VALUE err = data->err_info;
216 iseq->compile_data = 0;
217 compile_data_free(data);
219 if (RTEST(err)) {
220 rb_funcall2(err, rb_intern("set_backtrace"), 1, &iseq->filename);
221 rb_exc_raise(err);
223 return Qtrue;
226 static rb_compile_option_t COMPILE_OPTION_DEFAULT = {
227 OPT_INLINE_CONST_CACHE, /* int inline_const_cache; */
228 OPT_PEEPHOLE_OPTIMIZATION, /* int peephole_optimization; */
229 OPT_TAILCALL_OPTIMIZATION, /* int tailcall_optimization */
230 OPT_SPECIALISED_INSTRUCTION, /* int specialized_instruction; */
231 OPT_OPERANDS_UNIFICATION, /* int operands_unification; */
232 OPT_INSTRUCTIONS_UNIFICATION, /* int instructions_unification; */
233 OPT_STACK_CACHING, /* int stack_caching; */
234 OPT_TRACE_INSTRUCTION, /* int trace_instruction */
236 static const rb_compile_option_t COMPILE_OPTION_FALSE = {0};
238 static void
239 make_compile_option(rb_compile_option_t *option, VALUE opt)
241 if (opt == Qnil) {
242 *option = COMPILE_OPTION_DEFAULT;
244 else if (opt == Qfalse) {
245 *option = COMPILE_OPTION_FALSE;
247 else if (opt == Qtrue) {
248 memset(option, 1, sizeof(rb_compile_option_t));
250 else if (CLASS_OF(opt) == rb_cHash) {
251 *option = COMPILE_OPTION_DEFAULT;
253 #define SET_COMPILE_OPTION(o, h, mem) \
254 { VALUE flag = rb_hash_aref(h, ID2SYM(rb_intern(#mem))); \
255 if (flag == Qtrue) { o->mem = 1; } \
256 else if (flag == Qfalse) { o->mem = 0; } \
258 #define SET_COMPILE_OPTION_NUM(o, h, mem) \
259 { VALUE num = rb_hash_aref(opt, ID2SYM(rb_intern(#mem))); \
260 if (!NIL_P(num)) o->mem = NUM2INT(num); \
262 SET_COMPILE_OPTION(option, opt, inline_const_cache);
263 SET_COMPILE_OPTION(option, opt, peephole_optimization);
264 SET_COMPILE_OPTION(option, opt, tailcall_optimization);
265 SET_COMPILE_OPTION(option, opt, specialized_instruction);
266 SET_COMPILE_OPTION(option, opt, operands_unification);
267 SET_COMPILE_OPTION(option, opt, instructions_unification);
268 SET_COMPILE_OPTION(option, opt, stack_caching);
269 SET_COMPILE_OPTION(option, opt, trace_instruction);
270 SET_COMPILE_OPTION_NUM(option, opt, debug_level);
271 #undef SET_COMPILE_OPTION
272 #undef SET_COMPILE_OPTION_NUM
274 else {
275 rb_raise(rb_eTypeError, "Compile option must be Hash/true/false/nil");
279 static VALUE
280 make_compile_option_value(rb_compile_option_t *option)
282 VALUE opt = rb_hash_new();
283 #define SET_COMPILE_OPTION(o, h, mem) \
284 rb_hash_aset(h, ID2SYM(rb_intern(#mem)), o->mem ? Qtrue : Qfalse)
285 #define SET_COMPILE_OPTION_NUM(o, h, mem) \
286 rb_hash_aset(h, ID2SYM(rb_intern(#mem)), INT2NUM(o->mem))
288 SET_COMPILE_OPTION(option, opt, inline_const_cache);
289 SET_COMPILE_OPTION(option, opt, peephole_optimization);
290 SET_COMPILE_OPTION(option, opt, tailcall_optimization);
291 SET_COMPILE_OPTION(option, opt, specialized_instruction);
292 SET_COMPILE_OPTION(option, opt, operands_unification);
293 SET_COMPILE_OPTION(option, opt, instructions_unification);
294 SET_COMPILE_OPTION(option, opt, stack_caching);
295 SET_COMPILE_OPTION_NUM(option, opt, debug_level);
297 #undef SET_COMPILE_OPTION
298 #undef SET_COMPILE_OPTION_NUM
299 return opt;
302 VALUE
303 rb_iseq_new(NODE *node, VALUE name, VALUE filename,
304 VALUE parent, VALUE type)
306 return rb_iseq_new_with_opt(node, name, filename, parent, type,
307 &COMPILE_OPTION_DEFAULT);
310 static VALUE
311 rb_iseq_new_with_bopt_and_opt(NODE *node, VALUE name, VALUE filename,
312 VALUE parent, VALUE type, VALUE bopt,
313 const rb_compile_option_t *option)
315 rb_iseq_t *iseq;
316 VALUE self = iseq_alloc(rb_cISeq);
318 GetISeqPtr(self, iseq);
319 iseq->self = self;
321 prepare_iseq_build(iseq, name, filename, parent, type, bopt, option);
322 iseq_compile(self, node);
323 cleanup_iseq_build(iseq);
324 return self;
327 VALUE
328 rb_iseq_new_with_opt(NODE *node, VALUE name, VALUE filename,
329 VALUE parent, VALUE type,
330 const rb_compile_option_t *option)
332 return rb_iseq_new_with_bopt_and_opt(node, name, filename, parent, type,
333 Qfalse, option);
336 VALUE
337 rb_iseq_new_with_bopt(NODE *node, VALUE name, VALUE filename,
338 VALUE parent, VALUE type, VALUE bopt)
340 return rb_iseq_new_with_bopt_and_opt(node, name, filename, parent, type,
341 bopt, &COMPILE_OPTION_DEFAULT);
344 VALUE iseq_build_from_ary(rb_iseq_t *iseq, VALUE locals, VALUE args,
345 VALUE exception, VALUE body);
347 #define CHECK_ARRAY(v) rb_convert_type(v, T_ARRAY, "Array", "to_ary")
348 #define CHECK_STRING(v) rb_convert_type(v, T_STRING, "String", "to_str")
349 #define CHECK_SYMBOL(v) rb_convert_type(v, T_SYMBOL, "Symbol", "to_sym")
350 static inline VALUE CHECK_INTEGER(VALUE v) {NUM2LONG(v); return v;}
351 VALUE
352 iseq_load(VALUE self, VALUE data, VALUE parent, VALUE opt)
354 VALUE iseqval = iseq_alloc(rb_cISeq);
356 VALUE magic, version1, version2, format_type, misc;
357 VALUE name, filename;
358 VALUE type, body, locals, args, exception;
360 VALUE iseq_type;
361 struct st_table *type_map = 0;
362 rb_iseq_t *iseq;
363 rb_compile_option_t option;
364 int i = 0;
366 /* [magic, major_version, minor_version, format_type, misc,
367 * name, filename,
368 * type, locals, args, exception_table, body]
371 data = CHECK_ARRAY(data);
373 magic = CHECK_STRING(rb_ary_entry(data, i++));
374 version1 = CHECK_INTEGER(rb_ary_entry(data, i++));
375 version2 = CHECK_INTEGER(rb_ary_entry(data, i++));
376 format_type = CHECK_INTEGER(rb_ary_entry(data, i++));
377 misc = rb_ary_entry(data, i++); /* TODO */
379 name = CHECK_STRING(rb_ary_entry(data, i++));
380 filename = CHECK_STRING(rb_ary_entry(data, i++));
382 type = CHECK_SYMBOL(rb_ary_entry(data, i++));
383 locals = CHECK_ARRAY(rb_ary_entry(data, i++));
385 args = rb_ary_entry(data, i++);
386 if (FIXNUM_P(args) || (args = CHECK_ARRAY(args))) {
387 /* */
390 exception = CHECK_ARRAY(rb_ary_entry(data, i++));
391 body = CHECK_ARRAY(rb_ary_entry(data, i++));
393 GetISeqPtr(iseqval, iseq);
394 iseq->self = iseqval;
396 if (type_map == 0) {
397 type_map = st_init_numtable();
398 st_insert(type_map, ID2SYM(rb_intern("top")), ISEQ_TYPE_TOP);
399 st_insert(type_map, ID2SYM(rb_intern("method")), ISEQ_TYPE_METHOD);
400 st_insert(type_map, ID2SYM(rb_intern("block")), ISEQ_TYPE_BLOCK);
401 st_insert(type_map, ID2SYM(rb_intern("class")), ISEQ_TYPE_CLASS);
402 st_insert(type_map, ID2SYM(rb_intern("rescue")), ISEQ_TYPE_RESCUE);
403 st_insert(type_map, ID2SYM(rb_intern("ensure")), ISEQ_TYPE_ENSURE);
404 st_insert(type_map, ID2SYM(rb_intern("eval")), ISEQ_TYPE_EVAL);
405 st_insert(type_map, ID2SYM(rb_intern("defined_guard")), ISEQ_TYPE_DEFINED_GUARD);
408 if (st_lookup(type_map, type, &iseq_type) == 0) {
409 const char *typename = rb_id2name(type);
410 if (typename)
411 rb_raise(rb_eTypeError, "unsupport type: :%s", typename);
412 else
413 rb_raise(rb_eTypeError, "unsupport type: %p", (void *)type);
416 if (parent == Qnil) {
417 parent = 0;
420 make_compile_option(&option, opt);
421 prepare_iseq_build(iseq, name, filename,
422 parent, iseq_type, 0, &option);
424 iseq_build_from_ary(iseq, locals, args, exception, body);
426 cleanup_iseq_build(iseq);
427 return iseqval;
430 static VALUE
431 iseq_s_load(int argc, VALUE *argv, VALUE self)
433 VALUE data, opt=Qnil;
434 rb_scan_args(argc, argv, "11", &data, &opt);
436 return iseq_load(self, data, 0, opt);
439 static NODE *
440 compile_string(VALUE str, VALUE file, VALUE line)
442 VALUE parser = rb_parser_new();
443 NODE *node = rb_parser_compile_string(parser, StringValueCStr(file),
444 str, NUM2INT(line));
446 if (!node) {
447 rb_exc_raise(GET_THREAD()->errinfo); /* TODO: check err */
449 return node;
452 VALUE
453 rb_iseq_compile_with_option(VALUE src, VALUE file, VALUE line, VALUE opt)
455 rb_compile_option_t option;
456 NODE *node = compile_string(StringValue(src), file, line);
457 rb_thread_t *th = GET_THREAD();
458 make_compile_option(&option, opt);
460 if (th->base_block && th->base_block->iseq) {
461 return rb_iseq_new_with_opt(node, th->base_block->iseq->name,
462 file, th->base_block->iseq->self,
463 ISEQ_TYPE_EVAL, &option);
465 else {
466 return rb_iseq_new_with_opt(node, rb_str_new2("<compiled>"), file, Qfalse,
467 ISEQ_TYPE_TOP, &option);
471 VALUE
472 rb_iseq_compile(VALUE src, VALUE file, VALUE line)
474 return rb_iseq_compile_with_option(src, file, line, Qnil);
477 static VALUE
478 iseq_s_compile(int argc, VALUE *argv, VALUE self)
480 VALUE src, file = Qnil, line = INT2FIX(1), opt = Qnil;
482 rb_secure(1);
484 rb_scan_args(argc, argv, "13", &src, &file, &line, &opt);
485 file = file == Qnil ? rb_str_new2("<compiled>") : file;
486 line = line == Qnil ? INT2FIX(1) : line;
488 return rb_iseq_compile_with_option(src, file, line, opt);
491 static VALUE
492 iseq_s_compile_file(int argc, VALUE *argv, VALUE self)
494 VALUE file, line = INT2FIX(1), opt = Qnil;
495 VALUE parser;
496 VALUE f;
497 NODE *node;
498 const char *fname;
499 rb_compile_option_t option;
501 rb_secure(1);
502 rb_scan_args(argc, argv, "11", &file, &opt);
503 fname = StringValueCStr(file);
505 f = rb_file_open(fname, "r");
507 parser = rb_parser_new();
508 node = rb_parser_compile_file(parser, fname, f, NUM2INT(line));
509 make_compile_option(&option, opt);
510 return rb_iseq_new_with_opt(node, rb_str_new2("<main>"), file, Qfalse,
511 ISEQ_TYPE_TOP, &option);
514 static VALUE
515 iseq_s_compile_option_set(VALUE self, VALUE opt)
517 rb_compile_option_t option;
518 rb_secure(1);
519 make_compile_option(&option, opt);
520 COMPILE_OPTION_DEFAULT = option;
521 return opt;
524 static VALUE
525 iseq_s_compile_option_get(VALUE self)
527 return make_compile_option_value(&COMPILE_OPTION_DEFAULT);
530 static rb_iseq_t *
531 iseq_check(VALUE val)
533 rb_iseq_t *iseq;
534 GetISeqPtr(val, iseq);
535 if (!iseq->name) {
536 rb_raise(rb_eTypeError, "uninitialized InstructionSequence");
538 return iseq;
541 static VALUE
542 iseq_eval(VALUE self)
544 rb_secure(1);
545 return rb_iseq_eval(self);
548 static VALUE
549 iseq_inspect(VALUE self)
551 rb_iseq_t *iseq = iseq_check(self);
553 return rb_sprintf("<ISeq:%s@%s>",
554 RSTRING_PTR(iseq->name), RSTRING_PTR(iseq->filename));
557 VALUE iseq_data_to_ary(rb_iseq_t *iseq);
559 static VALUE
560 iseq_to_a(VALUE self)
562 rb_iseq_t *iseq = iseq_check(self);
563 rb_secure(1);
564 return iseq_data_to_ary(iseq);
567 /* TODO: search algorithm is brute force.
568 this should be binary search or so. */
570 static struct iseq_insn_info_entry *
571 get_insn_info(const rb_iseq_t *iseq, const unsigned long pos)
573 unsigned long i, size = iseq->insn_info_size;
574 struct iseq_insn_info_entry *table = iseq->insn_info_table;
576 for (i = 0; i < size; i++) {
577 if (table[i].position == pos) {
578 return &table[i];
582 return 0;
585 static unsigned short
586 find_line_no(rb_iseq_t *iseq, unsigned long pos)
588 struct iseq_insn_info_entry *entry = get_insn_info(iseq, pos);
589 if (entry) {
590 return entry->line_no;
592 else {
593 return 0;
597 static unsigned short
598 find_prev_line_no(rb_iseq_t *iseqdat, unsigned long pos)
600 unsigned long i, size = iseqdat->insn_info_size;
601 struct iseq_insn_info_entry *iiary = iseqdat->insn_info_table;
603 for (i = 0; i < size; i++) {
604 if (iiary[i].position == pos) {
605 if (i > 0) {
606 return iiary[i - 1].line_no;
608 else {
609 return 0;
614 return 0;
617 static VALUE
618 insn_operand_intern(rb_iseq_t *iseq,
619 int insn, int op_no, VALUE op,
620 int len, int pos, VALUE *pnop, VALUE child)
622 const char *types = insn_op_types(insn);
623 char type = types[op_no];
624 VALUE ret;
626 switch (type) {
627 case TS_OFFSET: /* LONG */
628 ret = rb_sprintf("%ld", pos + len + op);
629 break;
631 case TS_NUM: /* ULONG */
632 ret = rb_sprintf("%lu", op);
633 break;
635 case TS_LINDEX:
637 rb_iseq_t *ip = iseq->local_iseq;
638 int lidx = ip->local_size - op;
639 const char *name = rb_id2name(ip->local_table[lidx]);
641 if (name) {
642 ret = rb_str_new2(name);
644 else {
645 ret = rb_str_new2("*");
647 break;
649 case TS_DINDEX:{
650 if (insn == BIN(getdynamic) || insn == BIN(setdynamic)) {
651 rb_iseq_t *ip = iseq;
652 int level = *pnop, i;
653 const char *name;
654 for (i = 0; i < level; i++) {
655 ip = ip->parent_iseq;
657 name = rb_id2name(ip->local_table[ip->local_size - op]);
659 if (!name) {
660 name = "*";
662 ret = rb_str_new2(name);
664 else {
665 ret = rb_inspect(INT2FIX(op));
667 break;
669 case TS_ID: /* ID (symbol) */
670 op = ID2SYM(op);
672 case TS_VALUE: /* VALUE */
673 ret = rb_inspect(op);
674 if (CLASS_OF(op) == rb_cISeq) {
675 rb_ary_push(child, op);
677 break;
679 case TS_ISEQ: /* iseq */
681 rb_iseq_t *iseq = (rb_iseq_t *)op;
682 if (iseq) {
683 ret = iseq->name;
684 if (child) {
685 rb_ary_push(child, iseq->self);
688 else {
689 ret = rb_str_new2("nil");
691 break;
693 case TS_GENTRY:
695 struct global_entry *entry = (struct global_entry *)op;
696 ret = rb_str_dup(rb_id2str(entry->id));
698 break;
700 case TS_IC:
701 ret = rb_str_new2("<ic>");
702 break;
704 case TS_CDHASH:
705 ret = rb_str_new2("<cdhash>");
706 break;
708 case TS_FUNCPTR:
709 ret = rb_str_new2("<funcptr>");
710 break;
712 default:
713 rb_bug("ruby_iseq_disasm: unknown operand type: %c", type);
715 return ret;
719 * Disassemble a instruction
720 * Iseq -> Iseq inspect object
722 VALUE
723 ruby_iseq_disasm_insn(VALUE ret, VALUE *iseq, int pos,
724 rb_iseq_t *iseqdat, VALUE child)
726 int insn = iseq[pos];
727 int len = insn_len(insn);
728 int j;
729 const char *types = insn_op_types(insn);
730 VALUE str = rb_str_new(0, 0);
731 const char *insn_name_buff;
733 insn_name_buff = insn_name(insn);
734 if (1) {
735 rb_str_catf(str, "%04d %-16s ", pos, insn_name_buff);
737 else {
738 rb_str_catf(str, "%04d %-16.*s ", pos,
739 (int)strcspn(insn_name_buff, "_"), insn_name_buff);
742 for (j = 0; types[j]; j++) {
743 const char *types = insn_op_types(insn);
744 VALUE opstr = insn_operand_intern(iseqdat, insn, j, iseq[pos + j + 1],
745 len, pos, &iseq[pos + j + 2],
746 child);
747 rb_str_concat(str, opstr);
749 if (types[j + 1]) {
750 rb_str_cat2(str, ", ");
754 if (1) {
755 int line_no = find_line_no(iseqdat, pos);
756 int prev = find_prev_line_no(iseqdat, pos);
757 if (line_no && line_no != prev) {
758 long slen = RSTRING_LEN(str);
759 slen = (slen > 70) ? 0 : (70 - slen);
760 str = rb_str_catf(str, "%*s(%4d)", (int)slen, "", line_no);
763 else {
764 /* for debug */
765 struct iseq_insn_info_entry *entry = get_insn_info(iseqdat, pos);
766 long slen = RSTRING_LEN(str);
767 slen = (slen > 60) ? 0 : (60 - slen);
768 str = rb_str_catf(str, "%*s(line: %d, sp: %d)",
769 (int)slen, "", entry->line_no, entry->sp);
772 if (ret) {
773 rb_str_cat2(str, "\n");
774 rb_str_concat(ret, str);
776 else {
777 printf("%s\n", RSTRING_PTR(str));
779 return len;
782 static const char *
783 catch_type(int type)
785 switch (type) {
786 case CATCH_TYPE_RESCUE:
787 return "rescue";
788 case CATCH_TYPE_ENSURE:
789 return "ensure";
790 case CATCH_TYPE_RETRY:
791 return "retry";
792 case CATCH_TYPE_BREAK:
793 return "break";
794 case CATCH_TYPE_REDO:
795 return "redo";
796 case CATCH_TYPE_NEXT:
797 return "next";
798 default:
799 rb_bug("unknown catch type (%d)", type);
800 return 0;
804 VALUE
805 ruby_iseq_disasm(VALUE self)
807 rb_iseq_t *iseqdat = iseq_check(self);
808 VALUE *iseq;
809 VALUE str = rb_str_new(0, 0);
810 VALUE child = rb_ary_new();
811 unsigned long size;
812 int i;
813 ID *tbl;
814 enum {header_minlen = 72};
816 rb_secure(1);
818 iseq = iseqdat->iseq;
819 size = iseqdat->iseq_size;
821 rb_str_cat2(str, "== disasm: ");
823 rb_str_concat(str, iseq_inspect(iseqdat->self));
824 if ((i = RSTRING_LEN(str)) < header_minlen) {
825 rb_str_resize(str, header_minlen);
826 memset(RSTRING_PTR(str) + i, '=', header_minlen - i);
828 rb_str_cat2(str, "\n");
830 /* show catch table information */
831 if (iseqdat->catch_table_size != 0) {
832 rb_str_cat2(str, "== catch table\n");
834 for (i = 0; i < iseqdat->catch_table_size; i++) {
835 struct iseq_catch_table_entry *entry = &iseqdat->catch_table[i];
836 rb_str_catf(str,
837 "| catch type: %-6s st: %04d ed: %04d sp: %04d cont: %04d\n",
838 catch_type((int)entry->type), (int)entry->start,
839 (int)entry->end, (int)entry->sp, (int)entry->cont);
840 if (entry->iseq) {
841 rb_str_concat(str, ruby_iseq_disasm(entry->iseq));
844 if (iseqdat->catch_table_size != 0) {
845 rb_str_cat2(str, "|-------------------------------------"
846 "-----------------------------------\n");
849 /* show local table information */
850 tbl = iseqdat->local_table;
852 if (tbl) {
853 rb_str_catf(str,
854 "local table (size: %d, argc: %d "
855 "[opts: %d, rest: %d, post: %d, block: %d] s%d)\n",
856 iseqdat->local_size, iseqdat->argc,
857 iseqdat->arg_opts, iseqdat->arg_rest,
858 iseqdat->arg_post_len, iseqdat->arg_block,
859 iseqdat->arg_simple);
861 for (i = 0; i < iseqdat->local_table_size; i++) {
862 const char *name = rb_id2name(tbl[i]);
863 char info[0x100];
864 char argi[0x100] = "";
865 char opti[0x100] = "";
867 if (iseqdat->arg_opts) {
868 int argc = iseqdat->argc;
869 int opts = iseqdat->arg_opts;
870 if (i >= argc && i < argc + opts - 1) {
871 snprintf(opti, sizeof(opti), "Opt=%ld",
872 iseqdat->arg_opt_table[i - argc]);
876 snprintf(argi, sizeof(argi), "%s%s%s%s%s", /* arg, opts, rest, post block */
877 iseqdat->argc > i ? "Arg" : "",
878 opti,
879 iseqdat->arg_rest == i ? "Rest" : "",
880 (iseqdat->arg_post_start <= i &&
881 i < iseqdat->arg_post_start + iseqdat->arg_post_len) ? "Post" : "",
882 iseqdat->arg_block == i ? "Block" : "");
884 snprintf(info, sizeof(info), "%s%s%s%s", name ? name : "?",
885 *argi ? "<" : "", argi, *argi ? ">" : "");
887 rb_str_catf(str, "[%2d] %-11s", iseqdat->local_size - i, info);
889 rb_str_cat2(str, "\n");
892 /* show each line */
893 for (i = 0; i < size;) {
894 i += ruby_iseq_disasm_insn(str, iseq, i, iseqdat, child);
897 for (i = 0; i < RARRAY_LEN(child); i++) {
898 VALUE isv = rb_ary_entry(child, i);
899 rb_str_concat(str, ruby_iseq_disasm(isv));
902 return str;
905 static VALUE
906 iseq_s_disasm(VALUE klass, VALUE body)
908 extern NODE *rb_method_body(VALUE body);
909 NODE *node;
910 VALUE ret = Qnil;
912 rb_secure(1);
914 if ((node = rb_method_body(body)) != 0) {
915 if (nd_type(node) == RUBY_VM_METHOD_NODE) {
916 VALUE iseqval = (VALUE)node->nd_body;
917 ret = ruby_iseq_disasm(iseqval);
921 return ret;
924 const char *
925 ruby_node_name(int node)
927 switch (node) {
928 #include "node_name.inc"
929 default:
930 rb_bug("unknown node (%d)", node);
931 return 0;
935 #define DECL_SYMBOL(name) \
936 static VALUE sym_##name
938 #define INIT_SYMBOL(name) \
939 sym_##name = ID2SYM(rb_intern(#name))
941 static VALUE
942 register_label(struct st_table *table, int idx)
944 VALUE sym;
945 char buff[8 + (sizeof(idx) * CHAR_BIT * 32 / 100)];
947 snprintf(buff, sizeof(buff), "label_%u", idx);
948 sym = ID2SYM(rb_intern(buff));
949 st_insert(table, idx, sym);
950 return sym;
953 static VALUE
954 exception_type2symbol(VALUE type)
956 ID id;
957 switch(type) {
958 case CATCH_TYPE_RESCUE: CONST_ID(id, "rescue"); break;
959 case CATCH_TYPE_ENSURE: CONST_ID(id, "ensure"); break;
960 case CATCH_TYPE_RETRY: CONST_ID(id, "retry"); break;
961 case CATCH_TYPE_BREAK: CONST_ID(id, "break"); break;
962 case CATCH_TYPE_REDO: CONST_ID(id, "redo"); break;
963 case CATCH_TYPE_NEXT: CONST_ID(id, "next"); break;
964 default:
965 rb_bug("...");
967 return ID2SYM(id);
970 static int
971 cdhash_each(VALUE key, VALUE value, VALUE ary)
973 rb_ary_push(ary, key);
974 rb_ary_push(ary, value);
975 return ST_CONTINUE;
978 VALUE
979 iseq_data_to_ary(rb_iseq_t *iseq)
981 int i, pos, line = 0;
982 VALUE *seq;
984 VALUE val = rb_ary_new();
985 VALUE type; /* Symbol */
986 VALUE locals = rb_ary_new();
987 VALUE args = rb_ary_new();
988 VALUE body = rb_ary_new(); /* [[:insn1, ...], ...] */
989 VALUE nbody;
990 VALUE exception = rb_ary_new(); /* [[....]] */
991 VALUE misc = rb_hash_new();
993 static VALUE insn_syms[VM_INSTRUCTION_SIZE];
994 struct st_table *labels_table = st_init_numtable();
996 DECL_SYMBOL(top);
997 DECL_SYMBOL(method);
998 DECL_SYMBOL(block);
999 DECL_SYMBOL(class);
1000 DECL_SYMBOL(rescue);
1001 DECL_SYMBOL(ensure);
1002 DECL_SYMBOL(eval);
1003 DECL_SYMBOL(defined_guard);
1005 if (sym_top == 0) {
1006 int i;
1007 for (i=0; i<VM_INSTRUCTION_SIZE; i++) {
1008 insn_syms[i] = ID2SYM(rb_intern(insn_name(i)));
1010 INIT_SYMBOL(top);
1011 INIT_SYMBOL(method);
1012 INIT_SYMBOL(block);
1013 INIT_SYMBOL(class);
1014 INIT_SYMBOL(rescue);
1015 INIT_SYMBOL(ensure);
1016 INIT_SYMBOL(eval);
1017 INIT_SYMBOL(defined_guard);
1020 /* type */
1021 switch(iseq->type) {
1022 case ISEQ_TYPE_TOP: type = sym_top; break;
1023 case ISEQ_TYPE_METHOD: type = sym_method; break;
1024 case ISEQ_TYPE_BLOCK: type = sym_block; break;
1025 case ISEQ_TYPE_CLASS: type = sym_class; break;
1026 case ISEQ_TYPE_RESCUE: type = sym_rescue; break;
1027 case ISEQ_TYPE_ENSURE: type = sym_ensure; break;
1028 case ISEQ_TYPE_EVAL: type = sym_eval; break;
1029 case ISEQ_TYPE_DEFINED_GUARD: type = sym_defined_guard; break;
1030 default: rb_bug("unsupported iseq type");
1033 /* locals */
1034 for (i=0; i<iseq->local_table_size; i++) {
1035 ID lid = iseq->local_table[i];
1036 if (lid) {
1037 if (rb_id2str(lid)) rb_ary_push(locals, ID2SYM(lid));
1039 else {
1040 rb_ary_push(locals, ID2SYM(rb_intern("#arg_rest")));
1044 /* args */
1047 * [argc, # argc
1048 * [label1, label2, ...] # opts
1049 * rest index,
1050 * post_len
1051 * post_start
1052 * block index,
1053 * simple,
1056 VALUE arg_opt_labels = rb_ary_new();
1057 int j;
1059 for (j=0; j<iseq->arg_opts; j++) {
1060 rb_ary_push(arg_opt_labels,
1061 register_label(labels_table, iseq->arg_opt_table[j]));
1064 /* commit */
1065 if (iseq->arg_simple == 1) {
1066 args = INT2FIX(iseq->argc);
1068 else {
1069 rb_ary_push(args, INT2FIX(iseq->argc));
1070 rb_ary_push(args, arg_opt_labels);
1071 rb_ary_push(args, INT2FIX(iseq->arg_post_len));
1072 rb_ary_push(args, INT2FIX(iseq->arg_post_start));
1073 rb_ary_push(args, INT2FIX(iseq->arg_rest));
1074 rb_ary_push(args, INT2FIX(iseq->arg_block));
1075 rb_ary_push(args, INT2FIX(iseq->arg_simple));
1079 /* body */
1080 for (seq = iseq->iseq; seq < iseq->iseq + iseq->iseq_size; ) {
1081 VALUE insn = *seq++;
1082 int j, len = insn_len(insn);
1083 VALUE *nseq = seq + len - 1;
1084 VALUE ary = rb_ary_new2(len);
1086 rb_ary_push(ary, insn_syms[insn]);
1087 for (j=0; j<len-1; j++, seq++) {
1088 switch (insn_op_type(insn, j)) {
1089 case TS_OFFSET: {
1090 unsigned int idx = nseq - iseq->iseq + *seq;
1091 rb_ary_push(ary, register_label(labels_table, idx));
1092 break;
1094 case TS_LINDEX:
1095 case TS_DINDEX:
1096 case TS_NUM:
1097 rb_ary_push(ary, INT2FIX(*seq));
1098 break;
1099 case TS_VALUE:
1100 rb_ary_push(ary, *seq);
1101 break;
1102 case TS_ISEQ:
1104 rb_iseq_t *iseq = (rb_iseq_t *)*seq;
1105 if (iseq) {
1106 VALUE val = iseq_data_to_ary(iseq);
1107 rb_ary_push(ary, val);
1109 else {
1110 rb_ary_push(ary, Qnil);
1113 break;
1114 case TS_GENTRY:
1116 struct global_entry *entry = (struct global_entry *)*seq;
1117 rb_ary_push(ary, ID2SYM(entry->id));
1119 break;
1120 case TS_IC:
1121 rb_ary_push(ary, Qnil);
1122 break;
1123 case TS_ID:
1124 rb_ary_push(ary, ID2SYM(*seq));
1125 break;
1126 case TS_CDHASH:
1128 VALUE hash = *seq;
1129 VALUE val = rb_ary_new();
1130 int i;
1132 rb_hash_foreach(hash, cdhash_each, val);
1134 for (i=0; i<RARRAY_LEN(val); i+=2) {
1135 VALUE pos = FIX2INT(rb_ary_entry(val, i+1));
1136 unsigned int idx = nseq - iseq->iseq + pos;
1138 rb_ary_store(val, i+1,
1139 register_label(labels_table, idx));
1141 rb_ary_push(ary, val);
1143 break;
1144 default:
1145 rb_bug("unknown operand: %c", insn_op_type(insn, j));
1148 rb_ary_push(body, ary);
1151 nbody = body;
1153 /* exception */
1154 for (i=0; i<iseq->catch_table_size; i++) {
1155 VALUE ary = rb_ary_new();
1156 struct iseq_catch_table_entry *entry = &iseq->catch_table[i];
1157 rb_ary_push(ary, exception_type2symbol(entry->type));
1158 if (entry->iseq) {
1159 rb_iseq_t *eiseq;
1160 GetISeqPtr(entry->iseq, eiseq);
1161 rb_ary_push(ary, iseq_data_to_ary(eiseq));
1163 else {
1164 rb_ary_push(ary, Qnil);
1166 rb_ary_push(ary, register_label(labels_table, entry->start));
1167 rb_ary_push(ary, register_label(labels_table, entry->end));
1168 rb_ary_push(ary, register_label(labels_table, entry->cont));
1169 rb_ary_push(ary, INT2FIX(entry->sp));
1170 rb_ary_push(exception, ary);
1173 /* make body with labels and insert line number */
1174 body = rb_ary_new();
1176 for (i=0, pos=0; i<RARRAY_LEN(nbody); i++) {
1177 VALUE ary = RARRAY_PTR(nbody)[i];
1178 VALUE label;
1180 if (st_lookup(labels_table, pos, &label)) {
1181 rb_ary_push(body, label);
1184 if (iseq->insn_info_table[i].line_no != line) {
1185 line = iseq->insn_info_table[i].line_no;
1186 rb_ary_push(body, INT2FIX(line));
1189 rb_ary_push(body, ary);
1190 pos += RARRAY_LEN(ary);
1193 st_free_table(labels_table);
1195 rb_hash_aset(misc, ID2SYM(rb_intern("arg_size")), INT2FIX(iseq->arg_size));
1196 rb_hash_aset(misc, ID2SYM(rb_intern("local_size")), INT2FIX(iseq->local_size));
1197 rb_hash_aset(misc, ID2SYM(rb_intern("stack_max")), INT2FIX(iseq->stack_max));
1200 * [:magic, :major_version, :minor_version, :format_type, :misc,
1201 * :name, :filename, :type, :locals, :args,
1202 * :catch_table, :bytecode]
1204 rb_ary_push(val, rb_str_new2("YARVInstructionSequence/SimpleDataFormat"));
1205 rb_ary_push(val, INT2FIX(1));
1206 rb_ary_push(val, INT2FIX(1));
1207 rb_ary_push(val, INT2FIX(1));
1208 rb_ary_push(val, misc);
1209 rb_ary_push(val, iseq->name);
1210 rb_ary_push(val, iseq->filename);
1211 rb_ary_push(val, type);
1212 rb_ary_push(val, locals);
1213 rb_ary_push(val, args);
1214 rb_ary_push(val, exception);
1215 rb_ary_push(val, body);
1216 return val;
1219 struct st_table *
1220 insn_make_insn_table(void)
1222 struct st_table *table;
1223 int i;
1224 table = st_init_numtable();
1226 for (i=0; i<VM_INSTRUCTION_SIZE; i++) {
1227 st_insert(table, ID2SYM(rb_intern(insn_name(i))), i);
1230 return table;
1233 VALUE
1234 rb_iseq_clone(VALUE iseqval, VALUE newcbase)
1236 VALUE newiseq = iseq_alloc(rb_cISeq);
1237 rb_iseq_t *iseq0, *iseq1;
1239 GetISeqPtr(iseqval, iseq0);
1240 GetISeqPtr(newiseq, iseq1);
1242 *iseq1 = *iseq0;
1243 iseq1->self = newiseq;
1244 if (!iseq1->orig) {
1245 iseq1->orig = iseqval;
1247 if (newcbase) {
1248 iseq1->cref_stack = NEW_BLOCK(newcbase);
1249 if (iseq0->cref_stack->nd_next) {
1250 iseq1->cref_stack->nd_next = iseq0->cref_stack->nd_next;
1254 return newiseq;
1257 /* ruby2cext */
1259 VALUE
1260 rb_iseq_build_for_ruby2cext(
1261 const rb_iseq_t *iseq_template,
1262 const rb_insn_func_t *func,
1263 const struct iseq_insn_info_entry *insn_info_table,
1264 const char **local_table,
1265 const VALUE *arg_opt_table,
1266 const struct iseq_catch_table_entry *catch_table,
1267 const char *name,
1268 const char *filename)
1270 int i;
1271 VALUE iseqval = iseq_alloc(rb_cISeq);
1272 rb_iseq_t *iseq;
1273 GetISeqPtr(iseqval, iseq);
1275 /* copy iseq */
1276 *iseq = *iseq_template;
1277 iseq->name = rb_str_new2(name);
1278 iseq->filename = rb_str_new2(filename);
1279 iseq->mark_ary = rb_ary_new();
1280 iseq->self = iseqval;
1282 iseq->iseq = ALLOC_N(VALUE, iseq->iseq_size);
1284 for (i=0; i<iseq->iseq_size; i+=2) {
1285 iseq->iseq[i] = BIN(opt_call_c_function);
1286 iseq->iseq[i+1] = (VALUE)func;
1289 iseq_translate_threaded_code(iseq);
1291 #define ALLOC_AND_COPY(dst, src, type, size) do { \
1292 if (size) { \
1293 (dst) = ALLOC_N(type, (size)); \
1294 MEMCPY((dst), (src), type, (size)); \
1296 } while (0)
1298 ALLOC_AND_COPY(iseq->insn_info_table, insn_info_table,
1299 struct iseq_insn_info_entry, iseq->insn_info_size);
1301 ALLOC_AND_COPY(iseq->catch_table, catch_table,
1302 struct iseq_catch_table_entry, iseq->catch_table_size);
1304 ALLOC_AND_COPY(iseq->arg_opt_table, arg_opt_table,
1305 VALUE, iseq->arg_opts);
1307 set_relation(iseq, 0);
1309 return iseqval;
1312 void
1313 Init_ISeq(void)
1315 /* declare ::VM::InstructionSequence */
1316 rb_cISeq = rb_define_class_under(rb_cRubyVM, "InstructionSequence", rb_cObject);
1317 rb_define_alloc_func(rb_cISeq, iseq_alloc);
1318 rb_define_method(rb_cISeq, "inspect", iseq_inspect, 0);
1319 rb_define_method(rb_cISeq, "disasm", ruby_iseq_disasm, 0);
1320 rb_define_method(rb_cISeq, "disassemble", ruby_iseq_disasm, 0);
1321 rb_define_method(rb_cISeq, "to_a", iseq_to_a, 0);
1322 rb_define_method(rb_cISeq, "eval", iseq_eval, 0);
1324 /* disable this feature because there is no verifier. */
1325 /* rb_define_singleton_method(rb_cISeq, "load", iseq_s_load, -1); */
1326 (void)iseq_s_load;
1328 rb_define_singleton_method(rb_cISeq, "compile", iseq_s_compile, -1);
1329 rb_define_singleton_method(rb_cISeq, "new", iseq_s_compile, -1);
1330 rb_define_singleton_method(rb_cISeq, "compile_file", iseq_s_compile_file, -1);
1331 rb_define_singleton_method(rb_cISeq, "compile_option", iseq_s_compile_option_get, 0);
1332 rb_define_singleton_method(rb_cISeq, "compile_option=", iseq_s_compile_option_set, 1);
1333 rb_define_singleton_method(rb_cISeq, "disasm", iseq_s_disasm, 1);
1334 rb_define_singleton_method(rb_cISeq, "disassemble", iseq_s_disasm, 1);