A dynamic tracer for Linux

ply.c 12KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540
  1. #include <assert.h>
  2. #include <errno.h>
  3. #include <stdio.h>
  4. #include <stdlib.h>
  5. #include <string.h>
  6. #include "func.h"
  7. #include "node.h"
  8. #include "ply.h"
  9. #include "sym.h"
  10. #include "type.h"
  11. struct providers {
  12. struct provider **ps;
  13. size_t len;
  14. } providers;
  15. #define providers_foreach(_ps, _p) \
  16. for((_p) = (_ps)->ps; (_p) < &(_ps)->ps[(_ps)->len]; (_p)++)
  17. struct provider *provider_get(const char *name)
  18. {
  19. struct provider **p;
  20. providers_foreach(&providers, p) {
  21. if (strstr((*p)->name, name) == (*p)->name)
  22. return *p;
  23. }
  24. return NULL;
  25. }
  26. void provider_register(struct provider *p)
  27. {
  28. assert(p);
  29. assert(p->probe);
  30. assert(p->sym_alloc);
  31. providers.ps = realloc(providers.ps,
  32. ++providers.len * sizeof(*providers.ps));
  33. providers.ps[providers.len - 1] = p;
  34. }
  35. struct pass {
  36. int (*run)(struct pass *, struct ctx *);
  37. nwalk_fn pre;
  38. nwalk_fn post;
  39. };
  40. struct symtab syms = { .syms = NULL, .len = 0 };
  41. /* symtab_t locals = { .sym = NULL, .len = 0 }; */
  42. struct ctx *ctx_get(void)
  43. {
  44. struct ctx *ctx;
  45. struct prog *prog;
  46. ctx = calloc(1, sizeof(*ctx));
  47. ctx->globals = calloc(1, sizeof(*ctx->globals));
  48. ctx->progs = calloc(3, sizeof(*ctx->progs));
  49. /* PROBE0 */
  50. prog = calloc(1, sizeof(*prog));
  51. prog->locals = calloc(1, sizeof(*prog->locals));
  52. prog->globals = ctx->globals;
  53. prog->probe = "k:SyS_read";
  54. /* {
  55. * us = pid();
  56. * @t[0] = time();
  57. * @reads[pid()] = quantize(arg2);
  58. * }
  59. */
  60. prog->ast =
  61. node_expr(":block",
  62. node_expr(":assign",
  63. node_ident("us"),
  64. node_expr("pid", NULL),
  65. NULL),
  66. node_expr(":assign",
  67. node_expr(":map",
  68. node_ident("t"),
  69. node_num(0),
  70. node_expr("pid", NULL),
  71. NULL),
  72. node_expr("time", NULL),
  73. NULL),
  74. node_expr(":assign",
  75. node_expr(":map",
  76. node_ident("reads"),
  77. node_expr("pid", NULL),
  78. NULL),
  79. node_expr("quantize", node_ident("arg2"), NULL),
  80. NULL),
  81. NULL);
  82. prog->provider = provider_get("k");
  83. prog->provider->probe(prog);
  84. /* prog->ir = ir_new(); */
  85. ctx->progs[0] = prog;
  86. /* PROBE1 */
  87. prog = calloc(1, sizeof(*prog));
  88. prog->locals = calloc(1, sizeof(*prog->locals));
  89. prog->globals = ctx->globals;
  90. /* TODO: k -> kret */
  91. prog->probe = "k:SyS_read2";
  92. /* { @times[pid()] = quantize(time() - t0) } */
  93. prog->ast =
  94. node_expr(":assign",
  95. node_expr(":map",
  96. node_ident("times"),
  97. node_expr("pid", NULL),
  98. NULL),
  99. node_expr("quantize",
  100. node_expr("-",
  101. node_expr("time", NULL),
  102. node_ident("t0"),
  103. NULL),
  104. NULL),
  105. NULL);
  106. prog->provider = provider_get("k");
  107. prog->provider->probe(prog);
  108. /* prog->ir = ir_new(); */
  109. ctx->progs[1] = prog;
  110. return ctx;
  111. }
  112. int pass_sym_alloc(struct node *n, void *_prog)
  113. {
  114. struct prog *prog = _prog;
  115. struct provider *global = provider_get(":");
  116. int err = 0;
  117. switch (n->ntype) {
  118. case N_EXPR:
  119. case N_IDENT:
  120. err = prog->provider->sym_alloc(prog, n);
  121. if (!err || (err != -ENOENT))
  122. break;
  123. err = global->sym_alloc(prog, n);
  124. break;
  125. case N_NUM:
  126. case N_STRING:
  127. err = global->sym_alloc(prog, n);
  128. }
  129. if (err) {
  130. if ((err == -ENOENT) && (n->ntype == N_EXPR))
  131. node_error(n, stderr, "unknown function '%s'",
  132. n->expr.func);
  133. }
  134. return err;
  135. }
  136. /* int infer_type_list(struct prog *prog, struct node *n) */
  137. /* { */
  138. /* type_t *t; */
  139. /* /\* list of lists (code block) => void *\/ */
  140. /* if (n->list->ntype == N_LIST) { */
  141. /* n->type = &t_void; */
  142. /* return 0; */
  143. /* } */
  144. /* t = n->list->type; */
  145. /* if (!t) */
  146. /* return 0; */
  147. /* switch (t->ttype) { */
  148. /* case T_FUNC: */
  149. /* n->type = t->t.func.type; */
  150. /* break; */
  151. /* default: */
  152. /* n->type = t; */
  153. /* } */
  154. /* return 0; */
  155. /* } */
  156. /* int infer_type_keyword(struct prog *prog, struct node *n) */
  157. /* { */
  158. /* struct node *dst, *src; */
  159. /* switch (n->keyword.class) { */
  160. /* case KW_ASSIGN: */
  161. /* dst = node_next(n); */
  162. /* src = node_next(dst); */
  163. /* assert(dst && src); */
  164. /* if (!src->type) */
  165. /* return 0; */
  166. /* /\* TODO: assignment is statement for now. do we need */
  167. /* * c-style assignment expressions? e.g `a = b = 2;` *\/ */
  168. /* n->type = &t_void; */
  169. /* if (dst->type) */
  170. /* return 0; */
  171. /* dst->type = src->type; */
  172. /* if (dst->ntype != N_IDENT) */
  173. /* return 0; */
  174. /* return sym_add(dst->sym->st, dst->ident, dst->type, NULL); */
  175. /* case KW_BINOP: */
  176. /* dst = node_next(n); */
  177. /* src = node_next(dst); */
  178. /* assert(dst && src); */
  179. /* if (!(src->type && dst->type && type_equal(src->type, dst->type))) */
  180. /* return 0; */
  181. /* n->type = dst->type; */
  182. /* return 0; */
  183. /* default: */
  184. /* n->type = &t_void; */
  185. /* return 0; */
  186. /* } */
  187. /* return -ENOSYS; */
  188. /* } */
  189. /* int infer_type_sym(struct prog *prog, struct node *n) */
  190. /* { */
  191. /* struct node *parent, *key; */
  192. /* if (n->sym->type) { */
  193. /* /\* the symbol type could have been inferred in another */
  194. /* * probe, in that case copy the type to this node. *\/ */
  195. /* if (!n->type) */
  196. /* n->type = n->sym->type; */
  197. /* return 0; */
  198. /* } */
  199. /* parent = node_up(n); */
  200. /* key = node_next(n); */
  201. /* /\* match `somemap[somekey]` where the type of the entire */
  202. /* * expression and the type of the key is known, since that */
  203. /* * means the type of the map itself is also known. *\/ */
  204. /* if (parent && parent->type */
  205. /* && (parent->list->ntype == N_KEYWORD) */
  206. /* && (parent->list->keyword.class == KW_SUBSCRIPT) */
  207. /* && key && key->type) { */
  208. /* n->type = type_map_of(key->type, parent->type); */
  209. /* return sym_add(n->sym->st, n->ident, n->type, NULL); */
  210. /* } */
  211. /* return 0; */
  212. /* } */
  213. int pass_type_infer(struct node *n, void *_prog)
  214. {
  215. struct prog *prog = _prog;
  216. if (n->sym->func->type_infer)
  217. return n->sym->func->type_infer(prog, n);
  218. return 0;
  219. }
  220. /* int validate_func(struct node *n) */
  221. /* { */
  222. /* struct node *arg; */
  223. /* field_t *f; */
  224. /* int i; */
  225. /* for (arg = node_next(n), f = n->type->t.func.args, i = 1; */
  226. /* arg && f && f->type; arg = node_next(arg), f++, i++) { */
  227. /* if (type_compatible(arg->type, f->type)) */
  228. /* continue; */
  229. /* node_print(n, stderr); */
  230. /* fprintf(stderr, ": incompatible type of argument %d (", i); */
  231. /* type_dump(arg->type, stderr); */
  232. /* fputs("), expected ", stderr); */
  233. /* type_dump(f->type, stderr); */
  234. /* fputs("\n", stderr); */
  235. /* return -EINVAL; */
  236. /* } */
  237. /* if (!arg && (!f || !f->type)) */
  238. /* return 0; */
  239. /* if (arg) { */
  240. /* node_print(n, stderr); */
  241. /* fprintf(stderr, ": too many arguments, expected %d", i); */
  242. /* return -EINVAL; */
  243. /* } */
  244. /* if (f->optional) */
  245. /* return 0; */
  246. /* node_print(n, stderr); */
  247. /* fputs(": too few arguments", stderr); */
  248. /* return -EINVAL; */
  249. /* } */
  250. /* int pass_validate_types(struct node *n, void *_prog) */
  251. /* { */
  252. /* struct prog *prog = _prog; */
  253. /* node_print(n, stdout); putchar('\n'); */
  254. /* if (!n->type) { */
  255. /* node_print(n, stderr); */
  256. /* fputs(": type unknown\n", stderr); */
  257. /* return -EINVAL; */
  258. /* } */
  259. /* if (n->ntype != N_LIST) */
  260. /* return 0; */
  261. /* if (n->list->ntype != N_IDENT) */
  262. /* return 0; */
  263. /* assert(n->list->type->ttype == T_FUNC); */
  264. /* return validate_func(n->list); */
  265. /* } */
  266. /* int validate_syms(struct prog *prog) */
  267. /* { */
  268. /* return 0; */
  269. /* } */
  270. /* int run_validate_types(struct pass *pass, struct ctx *ctx) */
  271. /* { */
  272. /* struct prog **prog; */
  273. /* int err; */
  274. /* for (prog = ctx->progs; *prog; prog++) { */
  275. /* /\* check syms first to give better error messages. */
  276. /* * e.g. "i: type unknown", not "b-: type unknown" *\/ */
  277. /* err = validate_syms(*prog); */
  278. /* if (err) */
  279. /* return err; */
  280. /* err = node_walk((*prog)->ast, pass->pre, pass->post, *prog); */
  281. /* if (err) */
  282. /* return err; */
  283. /* } */
  284. /* return 0; */
  285. /* } */
  286. /* int rewrite_const_math(struct node *n) */
  287. /* { */
  288. /* int64_t result; */
  289. /* struct node *a, *b, *new; */
  290. /* int op; */
  291. /* /\* TODO: handle L/UL/ULL correctly *\/ */
  292. /* op = n->list->keyword.op; */
  293. /* a = node_next(n->list); */
  294. /* b = node_next(a); */
  295. /* switch (op) { */
  296. /* case '*': result = a->num * b->num; break; */
  297. /* case '/': result = a->num / b->num; break; */
  298. /* case '%': result = a->num % b->num; break; */
  299. /* case '+': result = a->num + b->num; break; */
  300. /* case '-': result = a->num - b->num; break; */
  301. /* case '<': result = a->num << b->num; break; */
  302. /* case '>': result = a->num >> b->num; break; */
  303. /* default: return 0; */
  304. /* } */
  305. /* new = node_num(result); */
  306. /* new->type = n->type; */
  307. /* return node_replace(n, new); */
  308. /* } */
  309. /* int pass_rewrite_ast(struct node *n, void *_prog) */
  310. /* { */
  311. /* struct prog *prog = _prog; */
  312. /* provider_t *global = provider_get(":"); */
  313. /* int err; */
  314. /* if (prog->provider->rewrite_node) { */
  315. /* err = prog->provider->rewrite_node(prog, n); */
  316. /* if (err) */
  317. /* return err; */
  318. /* } */
  319. /* if (global->rewrite_node) { */
  320. /* err = global->rewrite_node(prog, n); */
  321. /* if (err) */
  322. /* return err; */
  323. /* } */
  324. /* /\* pre-compute binops where both sides are constants *\/ */
  325. /* if ((n->ntype == N_LIST) */
  326. /* && (n->list->ntype == N_KEYWORD) */
  327. /* && (n->list->keyword.class == KW_BINOP) */
  328. /* && (node_next(n->list)->ntype == N_NUM) */
  329. /* && (node_next(node_next(n->list))->ntype == N_NUM)) */
  330. /* return rewrite_const_math(n); */
  331. /* return 0; */
  332. /* } */
  333. /* int generate_ir_ident(struct prog *prog, struct node *n) */
  334. /* { */
  335. /* switch (n->sym->type->ttype) { */
  336. /* case T_FUNC: */
  337. /* return n->sym->type->t.func.generate_ir(prog, n); */
  338. /* case T_MAP: */
  339. /* ir_emit_ldmap(prog->ir, BPF_REG_0, n->sym); */
  340. /* return 0; */
  341. /* default: */
  342. /* break; */
  343. /* } */
  344. /* return 0; */
  345. /* } */
  346. /* int pass_generate_ir(struct node *n, void *_prog) */
  347. /* { */
  348. /* struct prog *prog = _prog; */
  349. /* switch (n->ntype) { */
  350. /* case N_LIST: */
  351. /* return 0; */
  352. /* case N_IDENT: */
  353. /* return generate_ir_ident(prog, n); */
  354. /* default: */
  355. /* break; */
  356. /* } */
  357. /* return 0; */
  358. /* } */
  359. /* int run_generate_ir(struct pass *pass, struct ctx *ctx) */
  360. /* { */
  361. /* struct prog **progp; */
  362. /* int err; */
  363. /* for (progp = ctx->progs; *progp; progp++) { */
  364. /* struct prog *prog = *progp; */
  365. /* int return_label = ir_alloc_label(prog->ir); */
  366. /* err = prog->provider->ir_prologue ? */
  367. /* prog->provider->ir_prologue(prog) : 0; */
  368. /* if (err) */
  369. /* return err; */
  370. /* err = node_walk(prog->ast, NULL, pass_generate_ir, prog); */
  371. /* if (err) */
  372. /* return err; */
  373. /* err = prog->provider->ir_epilogue ? */
  374. /* prog->provider->ir_epilogue(prog) : 0; */
  375. /* if (err) */
  376. /* return err; */
  377. /* ir_emit_label(prog->ir, return_label); */
  378. /* ir_emit_insn(prog->ir, EXIT, 0, 0); */
  379. /* } */
  380. /* return 0; */
  381. /* } */
  382. int run_walk(struct pass *pass, struct ctx *ctx)
  383. {
  384. struct prog **prog;
  385. int err;
  386. for (prog = ctx->progs; *prog; prog++) {
  387. err = node_walk((*prog)->ast, pass->pre, pass->post, *prog);
  388. if (err)
  389. return err;
  390. }
  391. return 0;
  392. }
  393. struct pass passes[] = {
  394. { .run = run_walk, .post = pass_sym_alloc },
  395. { .run = run_walk, .post = pass_type_infer },
  396. /* { .run = run_walk, .post = pass_infer_types }, */
  397. /* { .run = run_walk, .post = pass_infer_types }, */
  398. /* { .run = run_validate_types, .post = pass_validate_types }, */
  399. /* { .run = run_walk, .post = pass_rewrite_ast }, */
  400. /* { .run = run_generate_ir }, */
  401. { NULL }
  402. };
  403. int main(void)
  404. {
  405. struct ctx *ctx = ctx_get();
  406. struct prog **prog;
  407. struct pass *pass;
  408. int err = 0;
  409. for (pass = passes; pass->run; pass++) {
  410. err = pass->run(pass, ctx);
  411. if (err)
  412. break;
  413. }
  414. for (prog = ctx->progs; *prog; prog++) {
  415. printf("\n\e[34m%s\e[0m\n", (*prog)->probe);
  416. node_dump((*prog)->ast, stdout);
  417. printf("\n-- locals\n");
  418. symtab_dump((*prog)->locals, stdout);
  419. /* printf("-- ir\n"); */
  420. /* ir_dump((*prog)->ir, stdout); */
  421. }
  422. printf("\n\n-- globals\n");
  423. symtab_dump(ctx->globals, stdout);
  424. printf("\n\n-- decls\n");
  425. type_dump_decls(stdout);
  426. if (err)
  427. printf("ERR: %d\n", err);
  428. return err;
  429. }