A dynamic tracer for Linux

ply.c 4.5KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242
  1. #include <assert.h>
  2. #include <errno.h>
  3. #include <stdio.h>
  4. #include <stdlib.h>
  5. #include <string.h>
  6. #include "ply.h"
  7. struct providers {
  8. provider_t **prov;
  9. size_t len;
  10. } providers;
  11. provider_t *provider_get(const char *name)
  12. {
  13. size_t i;
  14. for (i = 0; i < providers.len; i++) {
  15. if (strstr(providers.prov[i]->name, name)
  16. == providers.prov[i]->name)
  17. return providers.prov[i];
  18. }
  19. return NULL;
  20. }
  21. void provider_register(provider_t *prov)
  22. {
  23. assert(prov);
  24. assert(prov->probe);
  25. assert(prov->resolve);
  26. providers.prov = realloc(providers.prov,
  27. ++providers.len * sizeof(*providers.prov));
  28. providers.prov[providers.len - 1] = prov;
  29. }
  30. typedef struct pass pass_t;
  31. struct pass {
  32. int (*run)(pass_t *, prog_t *);
  33. walk_fn pre;
  34. walk_fn post;
  35. };
  36. symtab_t globals = { .sym = NULL, .len = 0 };
  37. symtab_t locals = { .sym = NULL, .len = 0 };
  38. prog_t *prog_get(void)
  39. {
  40. prog_t *prog;
  41. prog = calloc(1, sizeof(*prog));
  42. prog->locals = &locals;
  43. prog->globals = &globals;
  44. /* (@ ([ reads ((pid))) (quantize arg2)) */
  45. prog->probe = "k:SyS_read"; /* { @reads[pid()] = quantize(arg2) } */
  46. prog->ast =
  47. node_list(
  48. node_vlist(node_keyword('='),
  49. node_vlist(node_keyword('['),
  50. node_ident("@reads"),
  51. node_list(node_ident("pid")),
  52. NULL),
  53. node_vlist(node_ident("quantize"),
  54. node_ident("arg2"),
  55. NULL),
  56. NULL)
  57. );
  58. prog->provider = provider_get("k");
  59. prog->provider->probe(prog);
  60. return prog;
  61. }
  62. int pass_resolve_symbols(node_t *n, void *_prog)
  63. {
  64. prog_t *prog = _prog;
  65. provider_t *global = provider_get(":");
  66. node_t *op;
  67. int err;
  68. if (n->ntype != N_IDENT)
  69. return 0;
  70. /* .IDENT/->IDENT is a struct/union member, skip */
  71. /* op = node_prev(node_prev(n)); */
  72. /* if (op && (op->ntype == N_IDENT) && !strcmp(".", op->ident)) */
  73. /* return 0; */
  74. err = prog->provider->resolve(prog, n);
  75. if (!err || (err != -ENOENT))
  76. return err;
  77. err = global->resolve(prog, n);
  78. if (!err || (err != -ENOENT))
  79. return err;
  80. /* neither provider identifier nor global ditto => user
  81. * variable, add it as a global symbol of unknown type. */
  82. return sym_add(prog->globals, n->ident, NULL, &n->sym);
  83. }
  84. int infer_type_list(prog_t *prog, node_t *n)
  85. {
  86. type_t *t;
  87. /* list of lists (code block) => void */
  88. if (n->list->ntype == N_LIST) {
  89. n->type = &t_v;
  90. return 0;
  91. }
  92. t = n->list->type;
  93. if (!t)
  94. return 0;
  95. switch (t->ttype) {
  96. case T_FUNC:
  97. n->type = t->t.func.type;
  98. break;
  99. default:
  100. n->type = t;
  101. }
  102. return 0;
  103. }
  104. int infer_type_keyword(prog_t *prog, node_t *n)
  105. {
  106. node_t *dst, *src;
  107. switch (n->keyword) {
  108. case KW_ASSIGN:
  109. dst = node_next(n);
  110. src = node_next(dst);
  111. assert(dst && src);
  112. if (!src->type)
  113. return 0;
  114. if (!dst->type)
  115. dst->type = src->type;
  116. /* TODO: assignment is statement for now. do we need
  117. * c-style assignment expressions? e.g `a = b = 2;` */
  118. n->type = &t_v;
  119. return 0;
  120. default:
  121. return 0;
  122. }
  123. return -ENOSYS;
  124. }
  125. int infer_type_sym(prog_t *prog, node_t *n)
  126. {
  127. node_t *parent, *key;
  128. if (n->sym->type)
  129. return 0;
  130. parent = node_up(n);
  131. key = node_next(n);
  132. /* match `somemap[somekey]` where the type of the entire
  133. * expression and the type of the key is known, since that
  134. * means the type of the map itself is also known. */
  135. if (parent && parent->type
  136. && (parent->list->ntype == N_KEYWORD)
  137. && (parent->list->keyword == KW_SUBSCRIPT)
  138. && key && key->type) {
  139. n->type = type_get_map_of(key->type, parent->type);
  140. return sym_add(n->sym->st, n->ident, n->type, NULL);
  141. }
  142. return 0;
  143. }
  144. int pass_infer_types(node_t *n, void *_prog)
  145. {
  146. prog_t *prog = _prog;
  147. if (n->type)
  148. return 0;
  149. switch (n->ntype) {
  150. case N_LIST:
  151. return infer_type_list(prog, n);
  152. case N_KEYWORD:
  153. return infer_type_keyword(prog, n);
  154. case N_IDENT:
  155. return infer_type_sym(prog, n);
  156. default:
  157. break;
  158. }
  159. return 0;
  160. }
  161. int pass_walk(pass_t *pass, prog_t *prog)
  162. {
  163. return node_walk(prog->ast, pass->pre, pass->post, prog);
  164. }
  165. pass_t passes[] = {
  166. { .run = pass_walk, .post = pass_resolve_symbols },
  167. { .run = pass_walk, .post = pass_infer_types },
  168. { .run = pass_walk, .post = pass_infer_types },
  169. { NULL }
  170. };
  171. int main(void)
  172. {
  173. prog_t *prog = prog_get();
  174. pass_t *pass;
  175. int err;
  176. for (pass = passes; pass->run; pass++) {
  177. err = pass->run(pass, prog);
  178. if (err)
  179. break;
  180. }
  181. printf("AST\n===\n");
  182. node_dump(prog->ast, stdout);
  183. printf("\nLOCALS\n======\n");
  184. symtab_dump(prog->locals, stdout);
  185. printf("\nGLOBALS\n=======\n");
  186. symtab_dump(prog->globals, stdout);
  187. /* printf("\nTYPES\n=====\n"); */
  188. /* types_dump_cdecl(stdout); */
  189. return err;
  190. }