A dynamic tracer for Linux

ir.c 8.7KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466
  1. #include <assert.h>
  2. #include <inttypes.h>
  3. #include <stdio.h>
  4. #include <string.h>
  5. #include <linux/bpf.h>
  6. #include "ir.h"
  7. #include "sym.h"
  8. #include "type.h"
  9. const uint16_t vreg_base = 0x8000;
  10. static const char *bpf_func_name(enum bpf_func_id id)
  11. {
  12. switch (id) {
  13. case BPF_FUNC_get_current_comm:
  14. return "get_current_comm";
  15. case BPF_FUNC_get_current_pid_tgid:
  16. return "get_current_pid_tgid";
  17. case BPF_FUNC_get_current_uid_gid:
  18. return "get_current_uid_gid";
  19. case BPF_FUNC_get_stackid:
  20. return "get_stackid";
  21. case BPF_FUNC_ktime_get_ns:
  22. return "ktime_get_ns";
  23. case BPF_FUNC_map_delete_elem:
  24. return "map_delete_elem";
  25. case BPF_FUNC_map_lookup_elem:
  26. return "map_lookup_elem";
  27. case BPF_FUNC_map_update_elem:
  28. return "map_update_elem";
  29. case BPF_FUNC_perf_event_output:
  30. return "perf_event_output";
  31. case BPF_FUNC_probe_read:
  32. return "probe_read";
  33. case BPF_FUNC_trace_printk:
  34. return "trace_printk";
  35. default:
  36. return NULL;
  37. }
  38. }
  39. static void reg_name(uint16_t reg, char *name)
  40. {
  41. if (reg & vreg_base) {
  42. sprintf(name, "v%u", reg & ~vreg_base);
  43. } else if (reg == BPF_REG_10) {
  44. strcpy(name, "bp");
  45. } else {
  46. sprintf(name, "r%u", reg);
  47. }
  48. }
  49. static void reg_dump(uint16_t reg, int16_t off, FILE *fp)
  50. {
  51. char name[8];
  52. reg_name(reg, name);
  53. if (off < 0)
  54. fprintf(fp, "[%s - 0x%x]", name, -off);
  55. else if (off > 0)
  56. fprintf(fp, "[%s + 0x%x]", name, off);
  57. else
  58. fprintf(fp, "%s", name);
  59. }
  60. static char size_name(uint8_t code)
  61. {
  62. switch (BPF_SIZE(code)) {
  63. case BPF_B: return 'b';
  64. case BPF_H: return 'h';
  65. case BPF_W: return 'w';
  66. case BPF_DW: return 'q';
  67. }
  68. return '?';
  69. }
  70. static void alu_dump(uint8_t code, FILE *fp)
  71. {
  72. switch (BPF_OP(code)) {
  73. case BPF_MOV: fputs("mov", fp); break;
  74. case BPF_ADD: fputs("add", fp); break;
  75. case BPF_SUB: fputs("sub", fp); break;
  76. case BPF_MUL: fputs("mul", fp); break;
  77. case BPF_DIV: fputs("div", fp); break;
  78. case BPF_OR : fputs("or", fp); break;
  79. case BPF_AND: fputs("and", fp); break;
  80. case BPF_LSH: fputs("lsh", fp); break;
  81. case BPF_RSH: fputs("rsh", fp); break;
  82. case BPF_NEG: fputs("neg", fp); break;
  83. case BPF_MOD: fputs("mod", fp); break;
  84. case BPF_XOR: fputs("xor", fp); break;
  85. }
  86. switch (BPF_CLASS(code)) {
  87. case BPF_ALU: fputc(size_name(BPF_W), fp);
  88. case BPF_ALU64: fputc(size_name(BPF_DW), fp);
  89. }
  90. }
  91. static void offset_dump(int16_t off, FILE *fp)
  92. {
  93. if (off < 0)
  94. fprintf(fp, "L%d", -off);
  95. else
  96. fprintf(fp, "+%d", off);
  97. }
  98. static void __insn_dump(const struct bpf_insn insn, uint16_t dst, uint16_t src,
  99. FILE *fp)
  100. {
  101. const char *name;
  102. enum {
  103. OFF_NONE,
  104. OFF_DST,
  105. OFF_SRC,
  106. OFF_EXP,
  107. } off = OFF_NONE;
  108. switch (BPF_CLASS(insn.code)) {
  109. case BPF_LD:
  110. case BPF_LDX:
  111. off = OFF_SRC;
  112. fprintf(fp, "ld%c", size_name(insn.code));
  113. break;
  114. case BPF_ST:
  115. case BPF_STX:
  116. off = OFF_DST;
  117. fprintf(fp, "st%c", size_name(insn.code));
  118. break;
  119. case BPF_ALU:
  120. case BPF_ALU64:
  121. alu_dump(insn.code, fp);
  122. break;
  123. case BPF_JMP:
  124. off = OFF_EXP;
  125. switch (BPF_OP(insn.code)) {
  126. case BPF_EXIT:
  127. fputs("exit", fp);
  128. return;
  129. case BPF_CALL:
  130. fputs("call\t", fp);
  131. name = bpf_func_name(insn.imm);
  132. if (name)
  133. fputs(name, fp);
  134. else
  135. fprintf(fp, "%d", insn.imm);
  136. return;
  137. case BPF_JA:
  138. fputs("ja\t", fp);
  139. offset_dump(insn.off, fp);
  140. return;
  141. case BPF_JEQ: fputs("jeq", fp); break;
  142. case BPF_JNE: fputs("jne", fp); break;
  143. case BPF_JGT: fputs("jgt", fp); break;
  144. case BPF_JGE: fputs("jge", fp); break;
  145. case BPF_JSGE: fputs("jsge", fp); break;
  146. case BPF_JSGT: fputs("jsgt", fp); break;
  147. default:
  148. goto unknown;
  149. }
  150. break;
  151. default:
  152. goto unknown;
  153. }
  154. fputc('\t', fp);
  155. reg_dump(dst, off == OFF_DST ? insn.off : 0, fp);
  156. fputs(", ", fp);
  157. if (BPF_CLASS(insn.code) == BPF_LDX || BPF_CLASS(insn.code) == BPF_STX)
  158. goto reg_src;
  159. switch (BPF_SRC(insn.code)) {
  160. case BPF_K:
  161. fprintf(fp, "#%s0x%x", insn.imm < 0 ? "-" : "",
  162. insn.imm < 0 ? -insn.imm : insn.imm);
  163. break;
  164. case BPF_X:
  165. reg_src:
  166. reg_dump(src, off == OFF_SRC ? insn.off : 0, fp);
  167. break;
  168. }
  169. if (off == OFF_EXP) {
  170. fputs(", ", fp);
  171. offset_dump(insn.off, fp);
  172. }
  173. return;
  174. unknown:
  175. fprintf(fp, "data\t0x%16.16" PRIx64 "\n", *((uint64_t *)&insn));
  176. }
  177. void insn_dump(struct bpf_insn insn, FILE *fp)
  178. {
  179. __insn_dump(insn, insn.dst_reg, insn.src_reg, fp);
  180. }
  181. void vinsn_dump(struct vinsn *vi, FILE *fp)
  182. {
  183. switch (vi->vitype) {
  184. case VI_INSN:
  185. __insn_dump(vi->insn.bpf, vi->insn.dst, vi->insn.src, fp);
  186. return;
  187. case VI_LDMAP:
  188. fputs("ldmap\t", fp); reg_dump(vi->map.reg, 0, fp);
  189. fprintf(fp, ", %s", vi->map.sym->name);
  190. return;
  191. case VI_LABEL:
  192. offset_dump(vi->label, fp);
  193. fputc(':', fp);
  194. return;
  195. case VI_REG_GET:
  196. case VI_REG_PUT:
  197. fputs((vi->vitype == VI_REG_GET) ? "+ " : "- ", fp);
  198. reg_dump(vi->reg, 0, fp);
  199. return;
  200. }
  201. }
  202. void ir_dump(struct ir *ir, FILE *fp)
  203. {
  204. size_t i;
  205. for (i = 0; i < ir->len; i++) {
  206. struct vinsn *vi = &ir->vi[i];
  207. switch (vi->vitype) {
  208. case VI_INSN:
  209. case VI_LDMAP:
  210. fputc('\t', fp);
  211. break;
  212. case VI_REG_GET:
  213. case VI_REG_PUT:
  214. fputs("\e[2m", fp);
  215. case VI_LABEL:
  216. default:
  217. break;
  218. }
  219. vinsn_dump(vi, fp);
  220. /* print multiple gets/puts on one line */
  221. switch (vi->vitype) {
  222. case VI_REG_GET:
  223. for (; (vi + 1)->vitype == VI_REG_GET; vi++, i++) {
  224. fputs(", ", fp);
  225. reg_dump((vi + 1)->reg, 0, fp);
  226. }
  227. fputs("\e[0m", fp);
  228. break;
  229. case VI_REG_PUT:
  230. for (; (vi + 1)->vitype == VI_REG_PUT; vi++, i++) {
  231. fputs(", ", fp);
  232. reg_dump((vi + 1)->reg, 0, fp);
  233. }
  234. fputs("\e[0m", fp);
  235. break;
  236. default:
  237. break;
  238. }
  239. fputc('\n', fp);
  240. }
  241. }
  242. static void ir_emit(struct ir *ir, struct vinsn *vi)
  243. {
  244. ir->vi = realloc(ir->vi, (++ir->len)*sizeof(*vi));
  245. assert(ir->vi);
  246. ir->vi[ir->len - 1] = *vi;
  247. }
  248. void ir_emit_insn(struct ir *ir, struct bpf_insn bpf, uint16_t dst, uint16_t src)
  249. {
  250. struct vinsn vi;
  251. vi.vitype = VI_INSN;
  252. vi.insn.bpf = bpf;
  253. vi.insn.dst = dst;
  254. vi.insn.src = src;
  255. ir_emit(ir, &vi);
  256. }
  257. void ir_emit_ldmap(struct ir *ir, uint16_t dst, struct sym *map)
  258. {
  259. struct vinsn vi;
  260. vi.vitype = VI_LDMAP;
  261. vi.map.reg = dst;
  262. vi.map.sym = map;
  263. ir_emit(ir, &vi);
  264. }
  265. void ir_emit_label (struct ir *ir, int16_t label)
  266. {
  267. struct vinsn vi;
  268. vi.vitype = VI_LABEL;
  269. vi.label = label;
  270. ir_emit(ir, &vi);
  271. }
  272. /* void ir_emit_reg_get(struct ir *ir, uint16_t reg) */
  273. /* { */
  274. /* struct vinsn vi; */
  275. /* vi.vitype = VI_REG_GET; */
  276. /* vi.reg = reg; */
  277. /* ir_emit(ir, &vi); */
  278. /* } */
  279. /* void ir_emit_reg_put(struct ir *ir, uint16_t reg) */
  280. /* { */
  281. /* struct vinsn vi; */
  282. /* vi.vitype = VI_REG_PUT; */
  283. /* vi.reg = reg; */
  284. /* ir_emit(ir, &vi); */
  285. /* } */
  286. void ir_emit_sym_to_reg(struct ir *ir, uint16_t dst, struct sym *src)
  287. {
  288. struct irstate *irs = &src->irs;
  289. switch (irs->loc) {
  290. case LOC_IMM:
  291. ir_emit_insn(ir, MOV_IMM(0, irs->imm), dst, 0);
  292. break;
  293. case LOC_REG:
  294. ir_emit_insn(ir, MOV(0, 0), dst, irs->reg);
  295. break;
  296. case LOC_STACK:
  297. switch (irs->size) {
  298. case 1:
  299. ir_emit_insn(ir, LDXB(0, irs->stack, 0),
  300. dst, BPF_REG_BP);
  301. break;
  302. case 2:
  303. ir_emit_insn(ir, LDXH(0, irs->stack, 0),
  304. dst, BPF_REG_BP);
  305. break;
  306. case 4:
  307. ir_emit_insn(ir, LDXW(0, irs->stack, 0),
  308. dst, BPF_REG_BP);
  309. break;
  310. case 8:
  311. ir_emit_insn(ir, LDXDW(0, irs->stack, 0),
  312. dst, BPF_REG_BP);
  313. break;
  314. default:
  315. assert(0);
  316. }
  317. break;
  318. default:
  319. assert(0);
  320. }
  321. }
  322. void ir_emit_reg_to_sym(struct ir *ir, struct sym *dst, uint16_t src)
  323. {
  324. struct irstate *irs = &dst->irs;
  325. switch (irs->loc) {
  326. case LOC_REG:
  327. ir_emit_insn(ir, MOV(0, 0), irs->reg, src);
  328. break;
  329. case LOC_STACK:
  330. switch (irs->size) {
  331. case 1:
  332. ir_emit_insn(ir, STXB(0, irs->stack, 0),
  333. BPF_REG_BP, src);
  334. break;
  335. case 2:
  336. ir_emit_insn(ir, STXH(0, irs->stack, 0),
  337. BPF_REG_BP, src);
  338. break;
  339. case 4:
  340. ir_emit_insn(ir, STXW(0, irs->stack, 0),
  341. BPF_REG_BP, src);
  342. break;
  343. case 8:
  344. ir_emit_insn(ir, STXDW(0, irs->stack, 0),
  345. BPF_REG_BP, src);
  346. break;
  347. default:
  348. assert(0);
  349. }
  350. break;
  351. default:
  352. assert(0);
  353. }
  354. }
  355. /* void ir_emit_xfer(struct ir *ir, */
  356. int16_t ir_alloc_label (struct ir *ir)
  357. {
  358. return ir->next_label--;
  359. }
  360. uint16_t ir_alloc_register(struct ir *ir)
  361. {
  362. return ir->next_reg++;
  363. }
  364. ssize_t ir_alloc_stack(struct ir *ir, size_t size)
  365. {
  366. ir->sp -= size;
  367. assert(ir->sp > INT32_MIN);
  368. return ir->sp;
  369. }
  370. void ir_init_sym(struct ir *ir, struct sym *sym)
  371. {
  372. struct irstate *irs = &sym->irs;
  373. struct type *t = type_base(sym->type);
  374. if (irs->loc)
  375. return;
  376. irs->size = type_sizeof(t);
  377. if (irs->hint.stack)
  378. goto alloc_stack;
  379. switch (t->ttype) {
  380. case T_SCALAR:
  381. case T_POINTER:
  382. irs->loc = LOC_REG;
  383. irs->reg = ir_alloc_register(ir);
  384. return;
  385. default:
  386. alloc_stack:
  387. irs->loc = LOC_STACK;
  388. irs->stack = ir_alloc_stack(ir, irs->size);
  389. }
  390. }
  391. struct ir *ir_new(void)
  392. {
  393. struct ir *ir;
  394. ir = calloc(1, sizeof(*ir));
  395. assert(ir);
  396. ir->next_reg = vreg_base;
  397. ir->next_label = -1;
  398. return ir;
  399. }