lj_buf.c 7.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305
  1. /*
  2. ** Buffer handling.
  3. ** Copyright (C) 2005-2023 Mike Pall. See Copyright Notice in luajit.h
  4. */
  5. #define lj_buf_c
  6. #define LUA_CORE
  7. #include "lj_obj.h"
  8. #include "lj_gc.h"
  9. #include "lj_err.h"
  10. #include "lj_buf.h"
  11. #include "lj_str.h"
  12. #include "lj_tab.h"
  13. #include "lj_strfmt.h"
  14. /* -- Buffer management --------------------------------------------------- */
  15. static void buf_grow(SBuf *sb, MSize sz)
  16. {
  17. MSize osz = sbufsz(sb), len = sbuflen(sb), nsz = osz;
  18. char *b;
  19. GCSize flag;
  20. if (nsz < LJ_MIN_SBUF) nsz = LJ_MIN_SBUF;
  21. while (nsz < sz) nsz += nsz;
  22. flag = sbufflag(sb);
  23. if ((flag & SBUF_FLAG_COW)) { /* Copy-on-write semantics. */
  24. lj_assertG_(G(sbufL(sb)), sb->w == sb->e, "bad SBuf COW");
  25. b = (char *)lj_mem_new(sbufL(sb), nsz);
  26. setsbufflag(sb, flag & ~(GCSize)SBUF_FLAG_COW);
  27. setgcrefnull(sbufX(sb)->cowref);
  28. memcpy(b, sb->b, osz);
  29. } else {
  30. b = (char *)lj_mem_realloc(sbufL(sb), sb->b, osz, nsz);
  31. }
  32. if ((flag & SBUF_FLAG_EXT)) {
  33. sbufX(sb)->r = sbufX(sb)->r - sb->b + b; /* Adjust read pointer, too. */
  34. }
  35. /* Adjust buffer pointers. */
  36. sb->b = b;
  37. sb->w = b + len;
  38. sb->e = b + nsz;
  39. if ((flag & SBUF_FLAG_BORROW)) { /* Adjust borrowed buffer pointers. */
  40. SBuf *bsb = mref(sbufX(sb)->bsb, SBuf);
  41. bsb->b = b;
  42. bsb->w = b + len;
  43. bsb->e = b + nsz;
  44. }
  45. }
  46. LJ_NOINLINE char *LJ_FASTCALL lj_buf_need2(SBuf *sb, MSize sz)
  47. {
  48. lj_assertG_(G(sbufL(sb)), sz > sbufsz(sb), "SBuf overflow");
  49. if (LJ_UNLIKELY(sz > LJ_MAX_BUF))
  50. lj_err_mem(sbufL(sb));
  51. buf_grow(sb, sz);
  52. return sb->b;
  53. }
  54. LJ_NOINLINE char *LJ_FASTCALL lj_buf_more2(SBuf *sb, MSize sz)
  55. {
  56. if (sbufisext(sb)) {
  57. SBufExt *sbx = (SBufExt *)sb;
  58. MSize len = sbufxlen(sbx);
  59. if (LJ_UNLIKELY(sz > LJ_MAX_BUF || len + sz > LJ_MAX_BUF))
  60. lj_err_mem(sbufL(sbx));
  61. if (len + sz > sbufsz(sbx)) { /* Must grow. */
  62. buf_grow((SBuf *)sbx, len + sz);
  63. } else if (sbufiscow(sb) || sbufxslack(sbx) < (sbufsz(sbx) >> 3)) {
  64. /* Also grow to avoid excessive compactions, if slack < size/8. */
  65. buf_grow((SBuf *)sbx, sbuflen(sbx) + sz); /* Not sbufxlen! */
  66. return sbx->w;
  67. }
  68. if (sbx->r != sbx->b) { /* Compact by moving down. */
  69. memmove(sbx->b, sbx->r, len);
  70. sbx->r = sbx->b;
  71. sbx->w = sbx->b + len;
  72. lj_assertG_(G(sbufL(sbx)), len + sz <= sbufsz(sbx), "bad SBuf compact");
  73. }
  74. } else {
  75. MSize len = sbuflen(sb);
  76. lj_assertG_(G(sbufL(sb)), sz > sbufleft(sb), "SBuf overflow");
  77. if (LJ_UNLIKELY(sz > LJ_MAX_BUF || len + sz > LJ_MAX_BUF))
  78. lj_err_mem(sbufL(sb));
  79. buf_grow(sb, len + sz);
  80. }
  81. return sb->w;
  82. }
  83. void LJ_FASTCALL lj_buf_shrink(lua_State *L, SBuf *sb)
  84. {
  85. char *b = sb->b;
  86. MSize osz = (MSize)(sb->e - b);
  87. if (osz > 2*LJ_MIN_SBUF) {
  88. MSize n = (MSize)(sb->w - b);
  89. b = lj_mem_realloc(L, b, osz, (osz >> 1));
  90. sb->b = b;
  91. sb->w = b + n;
  92. sb->e = b + (osz >> 1);
  93. }
  94. lj_assertG_(G(sbufL(sb)), !sbufisext(sb), "YAGNI shrink SBufExt");
  95. }
  96. char * LJ_FASTCALL lj_buf_tmp(lua_State *L, MSize sz)
  97. {
  98. SBuf *sb = &G(L)->tmpbuf;
  99. setsbufL(sb, L);
  100. return lj_buf_need(sb, sz);
  101. }
  102. #if LJ_HASBUFFER && LJ_HASJIT
  103. void lj_bufx_set(SBufExt *sbx, const char *p, MSize len, GCobj *ref)
  104. {
  105. lua_State *L = sbufL(sbx);
  106. lj_bufx_free(L, sbx);
  107. lj_bufx_set_cow(L, sbx, p, len);
  108. setgcref(sbx->cowref, ref);
  109. lj_gc_objbarrier(L, (GCudata *)sbx - 1, ref);
  110. }
  111. #if LJ_HASFFI
  112. MSize LJ_FASTCALL lj_bufx_more(SBufExt *sbx, MSize sz)
  113. {
  114. lj_buf_more((SBuf *)sbx, sz);
  115. return sbufleft(sbx);
  116. }
  117. #endif
  118. #endif
  119. /* -- Low-level buffer put operations ------------------------------------- */
  120. SBuf *lj_buf_putmem(SBuf *sb, const void *q, MSize len)
  121. {
  122. char *w = lj_buf_more(sb, len);
  123. w = lj_buf_wmem(w, q, len);
  124. sb->w = w;
  125. return sb;
  126. }
  127. #if LJ_HASJIT || LJ_HASFFI
  128. static LJ_NOINLINE SBuf * LJ_FASTCALL lj_buf_putchar2(SBuf *sb, int c)
  129. {
  130. char *w = lj_buf_more2(sb, 1);
  131. *w++ = (char)c;
  132. sb->w = w;
  133. return sb;
  134. }
  135. SBuf * LJ_FASTCALL lj_buf_putchar(SBuf *sb, int c)
  136. {
  137. char *w = sb->w;
  138. if (LJ_LIKELY(w < sb->e)) {
  139. *w++ = (char)c;
  140. sb->w = w;
  141. return sb;
  142. }
  143. return lj_buf_putchar2(sb, c);
  144. }
  145. #endif
  146. SBuf * LJ_FASTCALL lj_buf_putstr(SBuf *sb, GCstr *s)
  147. {
  148. MSize len = s->len;
  149. char *w = lj_buf_more(sb, len);
  150. w = lj_buf_wmem(w, strdata(s), len);
  151. sb->w = w;
  152. return sb;
  153. }
  154. /* -- High-level buffer put operations ------------------------------------ */
  155. SBuf * LJ_FASTCALL lj_buf_putstr_reverse(SBuf *sb, GCstr *s)
  156. {
  157. MSize len = s->len;
  158. char *w = lj_buf_more(sb, len), *e = w+len;
  159. const char *q = strdata(s)+len-1;
  160. while (w < e)
  161. *w++ = *q--;
  162. sb->w = w;
  163. return sb;
  164. }
  165. SBuf * LJ_FASTCALL lj_buf_putstr_lower(SBuf *sb, GCstr *s)
  166. {
  167. MSize len = s->len;
  168. char *w = lj_buf_more(sb, len), *e = w+len;
  169. const char *q = strdata(s);
  170. for (; w < e; w++, q++) {
  171. uint32_t c = *(unsigned char *)q;
  172. #if LJ_TARGET_PPC
  173. *w = c + ((c >= 'A' && c <= 'Z') << 5);
  174. #else
  175. if (c >= 'A' && c <= 'Z') c += 0x20;
  176. *w = c;
  177. #endif
  178. }
  179. sb->w = w;
  180. return sb;
  181. }
  182. SBuf * LJ_FASTCALL lj_buf_putstr_upper(SBuf *sb, GCstr *s)
  183. {
  184. MSize len = s->len;
  185. char *w = lj_buf_more(sb, len), *e = w+len;
  186. const char *q = strdata(s);
  187. for (; w < e; w++, q++) {
  188. uint32_t c = *(unsigned char *)q;
  189. #if LJ_TARGET_PPC
  190. *w = c - ((c >= 'a' && c <= 'z') << 5);
  191. #else
  192. if (c >= 'a' && c <= 'z') c -= 0x20;
  193. *w = c;
  194. #endif
  195. }
  196. sb->w = w;
  197. return sb;
  198. }
  199. SBuf *lj_buf_putstr_rep(SBuf *sb, GCstr *s, int32_t rep)
  200. {
  201. MSize len = s->len;
  202. if (rep > 0 && len) {
  203. uint64_t tlen = (uint64_t)rep * len;
  204. char *w;
  205. if (LJ_UNLIKELY(tlen > LJ_MAX_STR))
  206. lj_err_mem(sbufL(sb));
  207. w = lj_buf_more(sb, (MSize)tlen);
  208. if (len == 1) { /* Optimize a common case. */
  209. uint32_t c = strdata(s)[0];
  210. do { *w++ = c; } while (--rep > 0);
  211. } else {
  212. const char *e = strdata(s) + len;
  213. do {
  214. const char *q = strdata(s);
  215. do { *w++ = *q++; } while (q < e);
  216. } while (--rep > 0);
  217. }
  218. sb->w = w;
  219. }
  220. return sb;
  221. }
  222. SBuf *lj_buf_puttab(SBuf *sb, GCtab *t, GCstr *sep, int32_t i, int32_t e)
  223. {
  224. MSize seplen = sep ? sep->len : 0;
  225. if (i <= e) {
  226. for (;;) {
  227. cTValue *o = lj_tab_getint(t, i);
  228. char *w;
  229. if (!o) {
  230. badtype: /* Error: bad element type. */
  231. sb->w = (char *)(intptr_t)i; /* Store failing index. */
  232. return NULL;
  233. } else if (tvisstr(o)) {
  234. MSize len = strV(o)->len;
  235. w = lj_buf_wmem(lj_buf_more(sb, len + seplen), strVdata(o), len);
  236. } else if (tvisint(o)) {
  237. w = lj_strfmt_wint(lj_buf_more(sb, STRFMT_MAXBUF_INT+seplen), intV(o));
  238. } else if (tvisnum(o)) {
  239. w = lj_buf_more(lj_strfmt_putfnum(sb, STRFMT_G14, numV(o)), seplen);
  240. } else {
  241. goto badtype;
  242. }
  243. if (i++ == e) {
  244. sb->w = w;
  245. break;
  246. }
  247. if (seplen) w = lj_buf_wmem(w, strdata(sep), seplen);
  248. sb->w = w;
  249. }
  250. }
  251. return sb;
  252. }
  253. /* -- Miscellaneous buffer operations ------------------------------------- */
  254. GCstr * LJ_FASTCALL lj_buf_tostr(SBuf *sb)
  255. {
  256. return lj_str_new(sbufL(sb), sb->b, sbuflen(sb));
  257. }
  258. /* Concatenate two strings. */
  259. GCstr *lj_buf_cat2str(lua_State *L, GCstr *s1, GCstr *s2)
  260. {
  261. MSize len1 = s1->len, len2 = s2->len;
  262. char *buf = lj_buf_tmp(L, len1 + len2);
  263. memcpy(buf, strdata(s1), len1);
  264. memcpy(buf+len1, strdata(s2), len2);
  265. return lj_str_new(L, buf, len1 + len2);
  266. }
  267. /* Read ULEB128 from buffer. */
  268. uint32_t LJ_FASTCALL lj_buf_ruleb128(const char **pp)
  269. {
  270. const uint8_t *w = (const uint8_t *)*pp;
  271. uint32_t v = *w++;
  272. if (LJ_UNLIKELY(v >= 0x80)) {
  273. int sh = 0;
  274. v &= 0x7f;
  275. do { v |= ((*w & 0x7f) << (sh += 7)); } while (*w++ >= 0x80);
  276. }
  277. *pp = (const char *)w;
  278. return v;
  279. }