archival/libarchive/decompress_uncompress.c

   1 /* vi: set sw=4 ts=4: */
   2 /* uncompress for busybox -- (c) 2002 Robert Griebl
   3  *
   4  * based on the original compress42.c source
   5  * (see disclaimer below)
   6  */
   7
   8 /* (N)compress42.c - File compression ala IEEE Computer, Mar 1992.
   9  *
  10  * Authors:
  11  *   Spencer W. Thomas   (decvax!harpo!utah-cs!utah-gr!thomas)
  12  *   Jim McKie           (decvax!mcvax!jim)
  13  *   Steve Davies        (decvax!vax135!petsd!peora!srd)
  14  *   Ken Turkowski       (decvax!decwrl!turtlevax!ken)
  15  *   James A. Woods      (decvax!ihnp4!ames!jaw)
  16  *   Joe Orost           (decvax!vax135!petsd!joe)
  17  *   Dave Mack           (csu@alembic.acs.com)
  18  *   Peter Jannesen, Network Communication Systems
  19  *                       (peter@ncs.nl)
  20  *
  21  * marc@suse.de : a small security fix for a buffer overflow
  22  *
  23  * [... History snipped ...]
  24  *
  25  */
  26
  27 #include "libbb.h"
  28 #include "bb_archive.h"
  29
  30
  31 /* Default input buffer size */
  32 #define IBUFSIZ 2048
  33
  34 /* Default output buffer size */
  35 #define OBUFSIZ 2048
  36
  37 /* Defines for third byte of header */
  38 #define BIT_MASK        0x1f    /* Mask for 'number of compresssion bits'       */
  39                                 /* Masks 0x20 and 0x40 are free.                */
  40                                 /* I think 0x20 should mean that there is       */
  41                                 /* a fourth header byte (for expansion).        */
  42 #define BLOCK_MODE      0x80    /* Block compression if table is full and       */
  43                                 /* compression rate is dropping flush tables    */
  44                                 /* the next two codes should not be changed lightly, as they must not   */
  45                                 /* lie within the contiguous general code space.                        */
  46 #define FIRST   257     /* first free entry */
  47 #define CLEAR   256     /* table clear output code */
  48
  49 #define INIT_BITS 9     /* initial number of bits/code */
  50
  51
  52 /* machine variants which require cc -Dmachine:  pdp11, z8000, DOS */
  53 #define HBITS      17   /* 50% occupancy */
  54 #define HSIZE      (1<<HBITS)
  55 #define HMASK      (HSIZE-1)    /* unused */
  56 #define HPRIME     9941         /* unused */
  57 #define BITS       16
  58 #define BITS_STR   "16"
  59 #undef  MAXSEG_64K              /* unused */
  60 #define MAXCODE(n) (1L << (n))
  61
  62 #define htabof(i)               htab[i]
  63 #define codetabof(i)            codetab[i]
  64 #define tab_prefixof(i)         codetabof(i)
  65 #define tab_suffixof(i)         ((unsigned char *)(htab))[i]
  66 #define de_stack                ((unsigned char *)&(htab[HSIZE-1]))
  67 #define clear_tab_prefixof()    memset(codetab, 0, 256)
  68
  69 /*
  70  * Decompress stdin to stdout.  This routine adapts to the codes in the
  71  * file building the "string" table on-the-fly; requiring no table to
  72  * be stored in the compressed file.
  73  */
  74
  75 IF_DESKTOP(long long) int FAST_FUNC
  76 unpack_Z_stream(int fd_in, int fd_out)
  77 {
  78         IF_DESKTOP(long long total_written = 0;)
  79         IF_DESKTOP(long long) int retval = -1;
  80         unsigned char *stackp;
  81         long code;
  82         int finchar;
  83         long oldcode;
  84         long incode;
  85         int inbits;
  86         int posbits;
  87         int outpos;
  88         int insize;
  89         int bitmask;
  90         long free_ent;
  91         long maxcode;
  92         long maxmaxcode;
  93         int n_bits;
  94         int rsize = 0;
  95         unsigned char *inbuf; /* were eating insane amounts of stack - */
  96         unsigned char *outbuf; /* bad for some embedded targets */
  97         unsigned char *htab;
  98         unsigned short *codetab;
  99
 100         /* Hmm, these were statics - why?! */
 101         /* user settable max # bits/code */
 102         int maxbits; /* = BITS; */
 103         /* block compress mode -C compatible with 2.0 */
 104         int block_mode; /* = BLOCK_MODE; */
 105
 106         inbuf = xzalloc(IBUFSIZ + 64);
 107         outbuf = xzalloc(OBUFSIZ + 2048);
 108         htab = xzalloc(HSIZE);  /* wsn't zeroed out before, maybe can xmalloc? */
 109         codetab = xzalloc(HSIZE * sizeof(codetab[0]));
 110
 111         insize = 0;
 112
 113         /* xread isn't good here, we have to return - caller may want
 114          * to do some cleanup (e.g. delete incomplete unpacked file etc) */
 115         if (full_read(fd_in, inbuf, 1) != 1) {
 116                 bb_error_msg("short read");
 117                 goto err;
 118         }
 119
 120         maxbits = inbuf[0] & BIT_MASK;
 121         block_mode = inbuf[0] & BLOCK_MODE;
 122         maxmaxcode = MAXCODE(maxbits);
 123
 124         if (maxbits > BITS) {
 125                 bb_error_msg("compressed with %d bits, can only handle "
 126                                 BITS_STR" bits", maxbits);
 127                 goto err;
 128         }
 129
 130         n_bits = INIT_BITS;
 131         maxcode = MAXCODE(INIT_BITS) - 1;
 132         bitmask = (1 << INIT_BITS) - 1;
 133         oldcode = -1;
 134         finchar = 0;
 135         outpos = 0;
 136         posbits = 0 << 3;
 137
 138         free_ent = ((block_mode) ? FIRST : 256);
 139
 140         /* As above, initialize the first 256 entries in the table. */
 141         /*clear_tab_prefixof(); - done by xzalloc */
 142
 143         for (code = 255; code >= 0; --code) {
 144                 tab_suffixof(code) = (unsigned char) code;
 145         }
 146
 147         do {
 148  resetbuf:
 149                 {
 150                         int i;
 151                         int e;
 152                         int o;
 153
 154                         o = posbits >> 3;
 155                         e = insize - o;
 156
 157                         for (i = 0; i < e; ++i)
 158                                 inbuf[i] = inbuf[i + o];
 159
 160                         insize = e;
 161                         posbits = 0;
 162                 }
 163
 164                 if (insize < (int) (IBUFSIZ + 64) - IBUFSIZ) {
 165                         rsize = safe_read(fd_in, inbuf + insize, IBUFSIZ);
 166                         if (rsize < 0)
 167                                 bb_error_msg_and_die(bb_msg_read_error);
 168                         insize += rsize;
 169                 }
 170
 171                 inbits = ((rsize > 0) ? (insize - insize % n_bits) << 3 :
 172                                   (insize << 3) - (n_bits - 1));
 173
 174                 while (inbits > posbits) {
 175                         if (free_ent > maxcode) {
 176                                 posbits =
 177                                         ((posbits - 1) +
 178                                          ((n_bits << 3) -
 179                                           (posbits - 1 + (n_bits << 3)) % (n_bits << 3)));
 180                                 ++n_bits;
 181                                 if (n_bits == maxbits) {
 182                                         maxcode = maxmaxcode;
 183                                 } else {
 184                                         maxcode = MAXCODE(n_bits) - 1;
 185                                 }
 186                                 bitmask = (1 << n_bits) - 1;
 187                                 goto resetbuf;
 188                         }
 189                         {
 190                                 unsigned char *p = &inbuf[posbits >> 3];
 191
 192                                 code = ((((long) (p[0])) | ((long) (p[1]) << 8) |
 193                                          ((long) (p[2]) << 16)) >> (posbits & 0x7)) & bitmask;
 194                         }
 195                         posbits += n_bits;
 196
 197
 198                         if (oldcode == -1) {
 199                                 if (code >= 256)
 200                                         bb_error_msg_and_die("corrupted data"); /* %ld", code); */
 201                                 oldcode = code;
 202                                 finchar = (int) oldcode;
 203                                 outbuf[outpos++] = (unsigned char) finchar;
 204                                 continue;
 205                         }
 206
 207                         if (code == CLEAR && block_mode) {
 208                                 clear_tab_prefixof();
 209                                 free_ent = FIRST - 1;
 210                                 posbits =
 211                                         ((posbits - 1) +
 212                                          ((n_bits << 3) -
 213                                           (posbits - 1 + (n_bits << 3)) % (n_bits << 3)));
 214                                 n_bits = INIT_BITS;
 215                                 maxcode = MAXCODE(INIT_BITS) - 1;
 216                                 bitmask = (1 << INIT_BITS) - 1;
 217                                 goto resetbuf;
 218                         }
 219
 220                         incode = code;
 221                         stackp = de_stack;
 222
 223                         /* Special case for KwKwK string. */
 224                         if (code >= free_ent) {
 225                                 if (code > free_ent) {
 226                                         unsigned char *p;
 227
 228                                         posbits -= n_bits;
 229                                         p = &inbuf[posbits >> 3];
 230
 231                                         bb_error_msg
 232                                                 ("insize:%d posbits:%d inbuf:%02X %02X %02X %02X %02X (%d)",
 233                                                  insize, posbits, p[-1], p[0], p[1], p[2], p[3],
 234                                                  (posbits & 07));
 235                                         bb_error_msg("corrupted data");
 236                                         goto err;
 237                                 }
 238
 239                                 *--stackp = (unsigned char) finchar;
 240                                 code = oldcode;
 241                         }
 242
 243                         /* Generate output characters in reverse order */
 244                         while ((long) code >= (long) 256) {
 245                                 if (stackp <= &htabof(0))
 246                                         bb_error_msg_and_die("corrupted data");
 247                                 *--stackp = tab_suffixof(code);
 248                                 code = tab_prefixof(code);
 249                         }
 250
 251                         finchar = tab_suffixof(code);
 252                         *--stackp = (unsigned char) finchar;
 253
 254                         /* And put them out in forward order */
 255                         {
 256                                 int i;
 257
 258                                 i = de_stack - stackp;
 259                                 if (outpos + i >= OBUFSIZ) {
 260                                         do {
 261                                                 if (i > OBUFSIZ - outpos) {
 262                                                         i = OBUFSIZ - outpos;
 263                                                 }
 264
 265                                                 if (i > 0) {
 266                                                         memcpy(outbuf + outpos, stackp, i);
 267                                                         outpos += i;
 268                                                 }
 269
 270                                                 if (outpos >= OBUFSIZ) {
 271                                                         xwrite(fd_out, outbuf, outpos);
 272                                                         IF_DESKTOP(total_written += outpos;)
 273                                                         outpos = 0;
 274                                                 }
 275                                                 stackp += i;
 276                                                 i = de_stack - stackp;
 277                                         } while (i > 0);
 278                                 } else {
 279                                         memcpy(outbuf + outpos, stackp, i);
 280                                         outpos += i;
 281                                 }
 282                         }
 283
 284                         /* Generate the new entry. */
 285                         code = free_ent;
 286                         if (code < maxmaxcode) {
 287                                 tab_prefixof(code) = (unsigned short) oldcode;
 288                                 tab_suffixof(code) = (unsigned char) finchar;
 289                                 free_ent = code + 1;
 290                         }
 291
 292                         /* Remember previous code.  */
 293                         oldcode = incode;
 294                 }
 295
 296         } while (rsize > 0);
 297
 298         if (outpos > 0) {
 299                 xwrite(fd_out, outbuf, outpos);
 300                 IF_DESKTOP(total_written += outpos;)
 301         }
 302
 303         retval = IF_DESKTOP(total_written) + 0;
 304  err:
 305         free(inbuf);
 306         free(outbuf);
 307         free(htab);
 308         free(codetab);
 309         return retval;
 310 }