token.c

   1 /*
   2    Copyright (C) Andrew Tridgell 1996
   3    Copyright (C) Paul Mackerras 1996
   4
   5    This program is free software; you can redistribute it and/or modify
   6    it under the terms of the GNU General Public License as published by
   7    the Free Software Foundation; either version 2 of the License, or
   8    (at your option) any later version.
   9
  10    This program is distributed in the hope that it will be useful,
  11    but WITHOUT ANY WARRANTY; without even the implied warranty of
  12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13    GNU General Public License for more details.
  14
  15    You should have received a copy of the GNU General Public License
  16    along with this program; if not, write to the Free Software
  17    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  18 */
  19
  20 #include "rsync.h"
  21 #include "zlib/zlib.h"
  22
  23 extern int do_compression;
  24 extern int module_id;
  25 extern int def_compress_level;
  26
  27 static int compression_level;
  28
  29 /* determine the compression level based on a wildcard filename list */
  30 void set_compression(char *fname)
  31 {
  32         char *dont;
  33         char *tok;
  34
  35         if (!do_compression)
  36                 return;
  37
  38         compression_level = def_compress_level;
  39         dont = lp_dont_compress(module_id);
  40
  41         if (!dont || !*dont)
  42                 return;
  43
  44         if (dont[0] == '*' && !dont[1]) {
  45                 /* an optimization to skip the rest of this routine */
  46                 compression_level = 0;
  47                 return;
  48         }
  49
  50         dont = strdup(dont);
  51         fname = strdup(fname);
  52         if (!dont || !fname)
  53                 return;
  54
  55         strlower(dont);
  56         strlower(fname);
  57
  58         for (tok = strtok(dont, " "); tok; tok = strtok(NULL, " ")) {
  59                 if (wildmatch(tok, fname)) {
  60                         compression_level = 0;
  61                         break;
  62                 }
  63         }
  64         free(dont);
  65         free(fname);
  66 }
  67
  68 /* non-compressing recv token */
  69 static int32 simple_recv_token(int f, char **data)
  70 {
  71         static int32 residue;
  72         static char *buf;
  73         int32 n;
  74
  75         if (!buf) {
  76                 buf = new_array(char, CHUNK_SIZE);
  77                 if (!buf)
  78                         out_of_memory("simple_recv_token");
  79         }
  80
  81         if (residue == 0) {
  82                 int32 i = read_int(f);
  83                 if (i <= 0)
  84                         return i;
  85                 residue = i;
  86         }
  87
  88         *data = buf;
  89         n = MIN(CHUNK_SIZE,residue);
  90         residue -= n;
  91         read_buf(f,buf,n);
  92         return n;
  93 }
  94
  95
  96 /* non-compressing send token */
  97 static void simple_send_token(int f, int32 token, struct map_struct *buf,
  98                               OFF_T offset, int32 n)
  99 {
 100         if (n > 0) {
 101                 int32 len = 0;
 102                 while (len < n) {
 103                         int32 n1 = MIN(CHUNK_SIZE, n-len);
 104                         write_int(f, n1);
 105                         write_buf(f, map_ptr(buf, offset+len, n1), n1);
 106                         len += n1;
 107                 }
 108         }
 109         /* a -2 token means to send data only and no token */
 110         if (token != -2)
 111                 write_int(f, -(token+1));
 112 }
 113
 114
 115 /* Flag bytes in compressed stream are encoded as follows: */
 116 #define END_FLAG        0       /* that's all folks */
 117 #define TOKEN_LONG      0x20    /* followed by 32-bit token number */
 118 #define TOKENRUN_LONG   0x21    /* ditto with 16-bit run count */
 119 #define DEFLATED_DATA   0x40    /* + 6-bit high len, then low len byte */
 120 #define TOKEN_REL       0x80    /* + 6-bit relative token number */
 121 #define TOKENRUN_REL    0xc0    /* ditto with 16-bit run count */
 122
 123 #define MAX_DATA_COUNT  16383   /* fit 14 bit count into 2 bytes with flags */
 124
 125 /* zlib.h says that if we want to be able to compress something in a single
 126  * call, avail_out must be at least 0.1% larger than avail_in plus 12 bytes.
 127  * We'll add in 0.1%+16, just to be safe (and we'll avoid floating point,
 128  * to ensure that this is a compile-time value). */
 129 #define AVAIL_OUT_SIZE(avail_in_size) ((avail_in_size)*1001/1000+16)
 130
 131 /* For coding runs of tokens */
 132 static int32 last_token = -1;
 133 static int32 run_start;
 134 static int32 last_run_end;
 135
 136 /* Deflation state */
 137 static z_stream tx_strm;
 138
 139 /* Output buffer */
 140 static char *obuf;
 141
 142 /* We want obuf to be able to hold both MAX_DATA_COUNT+2 bytes as well as
 143  * AVAIL_OUT_SIZE(CHUNK_SIZE) bytes, so make sure that it's large enough. */
 144 #if MAX_DATA_COUNT+2 > AVAIL_OUT_SIZE(CHUNK_SIZE)
 145 #define OBUF_SIZE       (MAX_DATA_COUNT+2)
 146 #else
 147 #define OBUF_SIZE       AVAIL_OUT_SIZE(CHUNK_SIZE)
 148 #endif
 149
 150 /* Send a deflated token */
 151 static void
 152 send_deflated_token(int f, int32 token, struct map_struct *buf, OFF_T offset,
 153                     int32 nb, int32 toklen)
 154 {
 155         int32 n, r;
 156         static int init_done, flush_pending;
 157
 158         if (last_token == -1) {
 159                 /* initialization */
 160                 if (!init_done) {
 161                         tx_strm.next_in = NULL;
 162                         tx_strm.zalloc = NULL;
 163                         tx_strm.zfree = NULL;
 164                         if (deflateInit2(&tx_strm, compression_level,
 165                                          Z_DEFLATED, -15, 8,
 166                                          Z_DEFAULT_STRATEGY) != Z_OK) {
 167                                 rprintf(FERROR, "compression init failed\n");
 168                                 exit_cleanup(RERR_STREAMIO);
 169                         }
 170                         if ((obuf = new_array(char, OBUF_SIZE)) == NULL)
 171                                 out_of_memory("send_deflated_token");
 172                         init_done = 1;
 173                 } else
 174                         deflateReset(&tx_strm);
 175                 last_run_end = 0;
 176                 run_start = token;
 177                 flush_pending = 0;
 178
 179         } else if (last_token == -2) {
 180                 run_start = token;
 181
 182         } else if (nb != 0 || token != last_token + 1
 183                    || token >= run_start + 65536) {
 184                 /* output previous run */
 185                 r = run_start - last_run_end;
 186                 n = last_token - run_start;
 187                 if (r >= 0 && r <= 63) {
 188                         write_byte(f, (n==0? TOKEN_REL: TOKENRUN_REL) + r);
 189                 } else {
 190                         write_byte(f, (n==0? TOKEN_LONG: TOKENRUN_LONG));
 191                         write_int(f, run_start);
 192                 }
 193                 if (n != 0) {
 194                         write_byte(f, n);
 195                         write_byte(f, n >> 8);
 196                 }
 197                 last_run_end = last_token;
 198                 run_start = token;
 199         }
 200
 201         last_token = token;
 202
 203         if (nb != 0 || flush_pending) {
 204                 /* deflate the data starting at offset */
 205                 int flush = Z_NO_FLUSH;
 206                 tx_strm.avail_in = 0;
 207                 tx_strm.avail_out = 0;
 208                 do {
 209                         if (tx_strm.avail_in == 0 && nb != 0) {
 210                                 /* give it some more input */
 211                                 n = MIN(nb, CHUNK_SIZE);
 212                                 tx_strm.next_in = (Bytef *)
 213                                         map_ptr(buf, offset, n);
 214                                 tx_strm.avail_in = n;
 215                                 nb -= n;
 216                                 offset += n;
 217                         }
 218                         if (tx_strm.avail_out == 0) {
 219                                 tx_strm.next_out = (Bytef *)(obuf + 2);
 220                                 tx_strm.avail_out = MAX_DATA_COUNT;
 221                                 if (flush != Z_NO_FLUSH) {
 222                                         /*
 223                                          * We left the last 4 bytes in the
 224                                          * buffer, in case they are the
 225                                          * last 4.  Move them to the front.
 226                                          */
 227                                         memcpy(tx_strm.next_out,
 228                                                obuf+MAX_DATA_COUNT-2, 4);
 229                                         tx_strm.next_out += 4;
 230                                         tx_strm.avail_out -= 4;
 231                                 }
 232                         }
 233                         if (nb == 0 && token != -2)
 234                                 flush = Z_SYNC_FLUSH;
 235                         r = deflate(&tx_strm, flush);
 236                         if (r != Z_OK) {
 237                                 rprintf(FERROR, "deflate returned %d\n", r);
 238                                 exit_cleanup(RERR_STREAMIO);
 239                         }
 240                         if (nb == 0 || tx_strm.avail_out == 0) {
 241                                 n = MAX_DATA_COUNT - tx_strm.avail_out;
 242                                 if (flush != Z_NO_FLUSH) {
 243                                         /*
 244                                          * We have to trim off the last 4
 245                                          * bytes of output when flushing
 246                                          * (they are just 0, 0, ff, ff).
 247                                          */
 248                                         n -= 4;
 249                                 }
 250                                 if (n > 0) {
 251                                         obuf[0] = DEFLATED_DATA + (n >> 8);
 252                                         obuf[1] = n;
 253                                         write_buf(f, obuf, n+2);
 254                                 }
 255                         }
 256                 } while (nb != 0 || tx_strm.avail_out == 0);
 257                 flush_pending = token == -2;
 258         }
 259
 260         if (token == -1) {
 261                 /* end of file - clean up */
 262                 write_byte(f, END_FLAG);
 263         } else if (token != -2) {
 264                 /* Add the data in the current block to the compressor's
 265                  * history and hash table. */
 266                 do {
 267                         /* Break up long sections in the same way that
 268                          * see_deflate_token() does. */
 269                         int32 n1 = toklen > 0xffff ? 0xffff : toklen;
 270                         toklen -= n1;
 271                         tx_strm.next_in = (Bytef *)map_ptr(buf, offset, n1);
 272                         tx_strm.avail_in = n1;
 273                         tx_strm.next_out = (Bytef *) obuf;
 274                         tx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
 275                         r = deflate(&tx_strm, Z_INSERT_ONLY);
 276                         if (r != Z_OK || tx_strm.avail_in != 0) {
 277                                 rprintf(FERROR, "deflate on token returned %d (%d bytes left)\n",
 278                                         r, tx_strm.avail_in);
 279                                 exit_cleanup(RERR_STREAMIO);
 280                         }
 281                 } while (toklen > 0);
 282         }
 283 }
 284
 285
 286 /* tells us what the receiver is in the middle of doing */
 287 static enum { r_init, r_idle, r_running, r_inflating, r_inflated } recv_state;
 288
 289 /* for inflating stuff */
 290 static z_stream rx_strm;
 291 static char *cbuf;
 292 static char *dbuf;
 293
 294 /* for decoding runs of tokens */
 295 static int32 rx_token;
 296 static int32 rx_run;
 297
 298 /* Receive a deflated token and inflate it */
 299 static int32 recv_deflated_token(int f, char **data)
 300 {
 301         static int init_done;
 302         static int32 saved_flag;
 303         int32 n, flag;
 304         int r;
 305
 306         for (;;) {
 307                 switch (recv_state) {
 308                 case r_init:
 309                         if (!init_done) {
 310                                 rx_strm.next_out = NULL;
 311                                 rx_strm.zalloc = NULL;
 312                                 rx_strm.zfree = NULL;
 313                                 if (inflateInit2(&rx_strm, -15) != Z_OK) {
 314                                         rprintf(FERROR, "inflate init failed\n");
 315                                         exit_cleanup(RERR_STREAMIO);
 316                                 }
 317                                 if (!(cbuf = new_array(char, MAX_DATA_COUNT))
 318                                     || !(dbuf = new_array(char, AVAIL_OUT_SIZE(CHUNK_SIZE))))
 319                                         out_of_memory("recv_deflated_token");
 320                                 init_done = 1;
 321                         } else {
 322                                 inflateReset(&rx_strm);
 323                         }
 324                         recv_state = r_idle;
 325                         rx_token = 0;
 326                         break;
 327
 328                 case r_idle:
 329                 case r_inflated:
 330                         if (saved_flag) {
 331                                 flag = saved_flag & 0xff;
 332                                 saved_flag = 0;
 333                         } else
 334                                 flag = read_byte(f);
 335                         if ((flag & 0xC0) == DEFLATED_DATA) {
 336                                 n = ((flag & 0x3f) << 8) + read_byte(f);
 337                                 read_buf(f, cbuf, n);
 338                                 rx_strm.next_in = (Bytef *)cbuf;
 339                                 rx_strm.avail_in = n;
 340                                 recv_state = r_inflating;
 341                                 break;
 342                         }
 343                         if (recv_state == r_inflated) {
 344                                 /* check previous inflated stuff ended correctly */
 345                                 rx_strm.avail_in = 0;
 346                                 rx_strm.next_out = (Bytef *)dbuf;
 347                                 rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
 348                                 r = inflate(&rx_strm, Z_SYNC_FLUSH);
 349                                 n = AVAIL_OUT_SIZE(CHUNK_SIZE) - rx_strm.avail_out;
 350                                 /*
 351                                  * Z_BUF_ERROR just means no progress was
 352                                  * made, i.e. the decompressor didn't have
 353                                  * any pending output for us.
 354                                  */
 355                                 if (r != Z_OK && r != Z_BUF_ERROR) {
 356                                         rprintf(FERROR, "inflate flush returned %d (%d bytes)\n",
 357                                                 r, n);
 358                                         exit_cleanup(RERR_STREAMIO);
 359                                 }
 360                                 if (n != 0 && r != Z_BUF_ERROR) {
 361                                         /* have to return some more data and
 362                                            save the flag for later. */
 363                                         saved_flag = flag + 0x10000;
 364                                         *data = dbuf;
 365                                         return n;
 366                                 }
 367                                 /*
 368                                  * At this point the decompressor should
 369                                  * be expecting to see the 0, 0, ff, ff bytes.
 370                                  */
 371                                 if (!inflateSyncPoint(&rx_strm)) {
 372                                         rprintf(FERROR, "decompressor lost sync!\n");
 373                                         exit_cleanup(RERR_STREAMIO);
 374                                 }
 375                                 rx_strm.avail_in = 4;
 376                                 rx_strm.next_in = (Bytef *)cbuf;
 377                                 cbuf[0] = cbuf[1] = 0;
 378                                 cbuf[2] = cbuf[3] = 0xff;
 379                                 inflate(&rx_strm, Z_SYNC_FLUSH);
 380                                 recv_state = r_idle;
 381                         }
 382                         if (flag == END_FLAG) {
 383                                 /* that's all folks */
 384                                 recv_state = r_init;
 385                                 return 0;
 386                         }
 387
 388                         /* here we have a token of some kind */
 389                         if (flag & TOKEN_REL) {
 390                                 rx_token += flag & 0x3f;
 391                                 flag >>= 6;
 392                         } else
 393                                 rx_token = read_int(f);
 394                         if (flag & 1) {
 395                                 rx_run = read_byte(f);
 396                                 rx_run += read_byte(f) << 8;
 397                                 recv_state = r_running;
 398                         }
 399                         return -1 - rx_token;
 400
 401                 case r_inflating:
 402                         rx_strm.next_out = (Bytef *)dbuf;
 403                         rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
 404                         r = inflate(&rx_strm, Z_NO_FLUSH);
 405                         n = AVAIL_OUT_SIZE(CHUNK_SIZE) - rx_strm.avail_out;
 406                         if (r != Z_OK) {
 407                                 rprintf(FERROR, "inflate returned %d (%d bytes)\n", r, n);
 408                                 exit_cleanup(RERR_STREAMIO);
 409                         }
 410                         if (rx_strm.avail_in == 0)
 411                                 recv_state = r_inflated;
 412                         if (n != 0) {
 413                                 *data = dbuf;
 414                                 return n;
 415                         }
 416                         break;
 417
 418                 case r_running:
 419                         ++rx_token;
 420                         if (--rx_run == 0)
 421                                 recv_state = r_idle;
 422                         return -1 - rx_token;
 423                 }
 424         }
 425 }
 426
 427 /*
 428  * put the data corresponding to a token that we've just returned
 429  * from recv_deflated_token into the decompressor's history buffer.
 430  */
 431 static void see_deflate_token(char *buf, int32 len)
 432 {
 433         int r;
 434         int32 blklen;
 435         unsigned char hdr[5];
 436
 437         rx_strm.avail_in = 0;
 438         blklen = 0;
 439         hdr[0] = 0;
 440         do {
 441                 if (rx_strm.avail_in == 0 && len != 0) {
 442                         if (blklen == 0) {
 443                                 /* Give it a fake stored-block header. */
 444                                 rx_strm.next_in = (Bytef *)hdr;
 445                                 rx_strm.avail_in = 5;
 446                                 blklen = len;
 447                                 if (blklen > 0xffff)
 448                                         blklen = 0xffff;
 449                                 hdr[1] = blklen;
 450                                 hdr[2] = blklen >> 8;
 451                                 hdr[3] = ~hdr[1];
 452                                 hdr[4] = ~hdr[2];
 453                         } else {
 454                                 rx_strm.next_in = (Bytef *)buf;
 455                                 rx_strm.avail_in = blklen;
 456                                 len -= blklen;
 457                                 blklen = 0;
 458                         }
 459                 }
 460                 rx_strm.next_out = (Bytef *)dbuf;
 461                 rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
 462                 r = inflate(&rx_strm, Z_SYNC_FLUSH);
 463                 if (r != Z_OK) {
 464                         rprintf(FERROR, "inflate (token) returned %d\n", r);
 465                         exit_cleanup(RERR_STREAMIO);
 466                 }
 467         } while (len || rx_strm.avail_out == 0);
 468 }
 469
 470 /**
 471  * Transmit a verbatim buffer of length @p n followed by a token.
 472  * If token == -1 then we have reached EOF
 473  * If n == 0 then don't send a buffer
 474  */
 475 void send_token(int f, int32 token, struct map_struct *buf, OFF_T offset,
 476                 int32 n, int32 toklen)
 477 {
 478         if (!do_compression)
 479                 simple_send_token(f, token, buf, offset, n);
 480         else
 481                 send_deflated_token(f, token, buf, offset, n, toklen);
 482 }
 483
 484
 485 /*
 486  * receive a token or buffer from the other end. If the reurn value is >0 then
 487  * it is a data buffer of that length, and *data will point at the data.
 488  * if the return value is -i then it represents token i-1
 489  * if the return value is 0 then the end has been reached
 490  */
 491 int32 recv_token(int f, char **data)
 492 {
 493         int tok;
 494
 495         if (!do_compression) {
 496                 tok = simple_recv_token(f,data);
 497         } else {
 498                 tok = recv_deflated_token(f, data);
 499         }
 500         return tok;
 501 }
 502
 503 /*
 504  * look at the data corresponding to a token, if necessary
 505  */
 506 void see_token(char *data, int32 toklen)
 507 {
 508         if (do_compression)
 509                 see_deflate_token(data, toklen);
 510 }