archival: note implicit dependencies between lzop & bbunzip
[oweals/busybox.git] / archival / lzop.c
1 /*
2    This file is part of the lzop file compressor.
3
4    Copyright (C) 1996..2003 Markus Franz Xaver Johannes Oberhumer
5    All Rights Reserved.
6
7    Markus F.X.J. Oberhumer <markus@oberhumer.com>
8    http://www.oberhumer.com/opensource/lzop/
9
10    lzop and the LZO library are free software; you can redistribute them
11    and/or modify them under the terms of the GNU General Public License as
12    published by the Free Software Foundation; either version 2 of
13    the License, or (at your option) any later version.
14
15    This program is distributed in the hope that it will be useful,
16    but WITHOUT ANY WARRANTY; without even the implied warranty of
17    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18    GNU General Public License for more details.
19
20    You should have received a copy of the GNU General Public License
21    along with this program; see the file COPYING.
22    If not, write to the Free Software Foundation, Inc.,
23    59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
24
25    "Minimalized" for busybox by Alain Knaff
26 */
27
28 //usage:#define lzop_trivial_usage
29 //usage:       "[-cfvd123456789CF] [FILE]..."
30 //usage:#define lzop_full_usage "\n\n"
31 //usage:       "        -1..9   Compression level"
32 //usage:     "\n        -d      Decompress"
33 //usage:     "\n        -c      Write to stdout"
34 //usage:     "\n        -f      Force"
35 //usage:     "\n        -v      Verbose"
36 //usage:     "\n        -F      Don't store or verify checksum"
37 //usage:     "\n        -C      Also write checksum of compressed block"
38 //usage:
39 //usage:#define lzopcat_trivial_usage
40 //usage:       "[-vCF] [FILE]..."
41 //usage:#define lzopcat_full_usage "\n\n"
42 //usage:       "        -v      Verbose"
43 //usage:     "\n        -F      Don't store or verify checksum"
44 //usage:
45 //usage:#define unlzop_trivial_usage
46 //usage:       "[-cfvCF] [FILE]..."
47 //usage:#define unlzop_full_usage "\n\n"
48 //usage:       "        -c      Write to stdout"
49 //usage:     "\n        -f      Force"
50 //usage:     "\n        -v      Verbose"
51 //usage:     "\n        -F      Don't store or verify checksum"
52
53 #include "libbb.h"
54 #include "bb_archive.h"
55 #include "liblzo_interface.h"
56
57 /* lzo-2.03/src/lzo_ptr.h */
58 #define pd(a,b)  ((unsigned)((a)-(b)))
59
60 #define lzo_version()                   LZO_VERSION
61 #define lzo_sizeof_dict_t               (sizeof(uint8_t*))
62
63 /* lzo-2.03/include/lzo/lzo1x.h */
64 #define LZO1X_1_MEM_COMPRESS    (16384 * lzo_sizeof_dict_t)
65 #define LZO1X_1_15_MEM_COMPRESS (32768 * lzo_sizeof_dict_t)
66 #define LZO1X_999_MEM_COMPRESS  (14 * 16384 * sizeof(short))
67
68 /* lzo-2.03/src/lzo1x_oo.c */
69 #define NO_LIT UINT_MAX
70
71 /**********************************************************************/
72 static void copy2(uint8_t* ip, const uint8_t* m_pos, unsigned off)
73 {
74         ip[0] = m_pos[0];
75         if (off == 1)
76                 ip[1] = m_pos[0];
77         else
78                 ip[1] = m_pos[1];
79 }
80
81 static void copy3(uint8_t* ip, const uint8_t* m_pos, unsigned off)
82 {
83         ip[0] = m_pos[0];
84         if (off == 1) {
85                 ip[2] = ip[1] = m_pos[0];
86         }
87         else if (off == 2) {
88                 ip[1] = m_pos[1];
89                 ip[2] = m_pos[0];
90         }
91         else {
92                 ip[1] = m_pos[1];
93                 ip[2] = m_pos[2];
94         }
95 }
96
97 /**********************************************************************/
98 // optimize a block of data.
99 /**********************************************************************/
100 #define TEST_IP         (ip < ip_end)
101 #define TEST_OP         (op <= op_end)
102
103 static NOINLINE int lzo1x_optimize(uint8_t *in, unsigned in_len,
104                 uint8_t *out, unsigned *out_len,
105                 void* wrkmem UNUSED_PARAM)
106 {
107         uint8_t* op;
108         uint8_t* ip;
109         unsigned t;
110         uint8_t* m_pos;
111         uint8_t* const ip_end = in + in_len;
112         uint8_t* const op_end = out + *out_len;
113         uint8_t* litp = NULL;
114         unsigned lit = 0;
115         unsigned next_lit = NO_LIT;
116         unsigned nl;
117         unsigned long o_m1_a = 0, o_m1_b = 0, o_m2 = 0, o_m3_a = 0, o_m3_b = 0;
118
119 //      LZO_UNUSED(wrkmem);
120
121         *out_len = 0;
122
123         op = out;
124         ip = in;
125
126         if (*ip > 17) {
127                 t = *ip++ - 17;
128                 if (t < 4)
129                         goto match_next;
130                 goto first_literal_run;
131         }
132
133         while (TEST_IP && TEST_OP) {
134                 t = *ip++;
135                 if (t >= 16)
136                         goto match;
137                 /* a literal run */
138                 litp = ip - 1;
139                 if (t == 0) {
140                         t = 15;
141                         while (*ip == 0)
142                                 t += 255, ip++;
143                         t += *ip++;
144                 }
145                 lit = t + 3;
146                 /* copy literals */
147  copy_literal_run:
148                 *op++ = *ip++;
149                 *op++ = *ip++;
150                 *op++ = *ip++;
151  first_literal_run:
152                 do *op++ = *ip++; while (--t > 0);
153
154                 t = *ip++;
155
156                 if (t >= 16)
157                         goto match;
158 #if defined(LZO1X)
159                 m_pos = op - 1 - 0x800;
160 #elif defined(LZO1Y)
161                 m_pos = op - 1 - 0x400;
162 #endif
163                 m_pos -= t >> 2;
164                 m_pos -= *ip++ << 2;
165                 *op++ = *m_pos++;
166                 *op++ = *m_pos++;
167                 *op++ = *m_pos++;
168                 lit = 0;
169                 goto match_done;
170
171
172                 /* handle matches */
173                 do {
174                         if (t < 16) { /* a M1 match */
175                                 m_pos = op - 1;
176                                 m_pos -= t >> 2;
177                                 m_pos -= *ip++ << 2;
178
179                                 if (litp == NULL)
180                                         goto copy_m1;
181
182                                 nl = ip[-2] & 3;
183                                 /* test if a match follows */
184                                 if (nl == 0 && lit == 1 && ip[0] >= 16) {
185                                         next_lit = nl;
186                                         /* adjust length of previous short run */
187                                         lit += 2;
188                                         *litp = (unsigned char)((*litp & ~3) | lit);
189                                         /* copy over the 2 literals that replace the match */
190                                         copy2(ip-2, m_pos, pd(op, m_pos));
191                                         o_m1_a++;
192                                 }
193                                 /* test if a literal run follows */
194                                 else
195                                 if (nl == 0
196                                  && ip[0] < 16
197                                  && ip[0] != 0
198                                  && (lit + 2 + ip[0] < 16)
199                                 ) {
200                                         t = *ip++;
201                                         /* remove short run */
202                                         *litp &= ~3;
203                                         /* copy over the 2 literals that replace the match */
204                                         copy2(ip-3+1, m_pos, pd(op, m_pos));
205                                         /* move literals 1 byte ahead */
206                                         litp += 2;
207                                         if (lit > 0)
208                                                 memmove(litp+1, litp, lit);
209                                         /* insert new length of long literal run */
210                                         lit += 2 + t + 3;
211                                         *litp = (unsigned char)(lit - 3);
212
213                                         o_m1_b++;
214                                         *op++ = *m_pos++;
215                                         *op++ = *m_pos++;
216                                         goto copy_literal_run;
217                                 }
218  copy_m1:
219                                 *op++ = *m_pos++;
220                                 *op++ = *m_pos++;
221                         } else {
222  match:
223                                 if (t >= 64) {                          /* a M2 match */
224                                         m_pos = op - 1;
225 #if defined(LZO1X)
226                                         m_pos -= (t >> 2) & 7;
227                                         m_pos -= *ip++ << 3;
228                                         t = (t >> 5) - 1;
229 #elif defined(LZO1Y)
230                                         m_pos -= (t >> 2) & 3;
231                                         m_pos -= *ip++ << 2;
232                                         t = (t >> 4) - 3;
233 #endif
234                                         if (litp == NULL)
235                                                 goto copy_m;
236
237                                         nl = ip[-2] & 3;
238                                         /* test if in beetween two long literal runs */
239                                         if (t == 1 && lit > 3 && nl == 0
240                                          && ip[0] < 16 && ip[0] != 0 && (lit + 3 + ip[0] < 16)
241                                         ) {
242                                                 t = *ip++;
243                                                 /* copy over the 3 literals that replace the match */
244                                                 copy3(ip-1-2, m_pos, pd(op, m_pos));
245                                                 /* set new length of previous literal run */
246                                                 lit += 3 + t + 3;
247                                                 *litp = (unsigned char)(lit - 3);
248                                                 o_m2++;
249                                                 *op++ = *m_pos++;
250                                                 *op++ = *m_pos++;
251                                                 *op++ = *m_pos++;
252                                                 goto copy_literal_run;
253                                         }
254                                 } else {
255                                         if (t >= 32) {                  /* a M3 match */
256                                                 t &= 31;
257                                                 if (t == 0) {
258                                                         t = 31;
259                                                         while (*ip == 0)
260                                                                 t += 255, ip++;
261                                                         t += *ip++;
262                                                 }
263                                                 m_pos = op - 1;
264                                                 m_pos -= *ip++ >> 2;
265                                                 m_pos -= *ip++ << 6;
266                                         } else {                                        /* a M4 match */
267                                                 m_pos = op;
268                                                 m_pos -= (t & 8) << 11;
269                                                 t &= 7;
270                                                 if (t == 0) {
271                                                         t = 7;
272                                                         while (*ip == 0)
273                                                                 t += 255, ip++;
274                                                         t += *ip++;
275                                                 }
276                                                 m_pos -= *ip++ >> 2;
277                                                 m_pos -= *ip++ << 6;
278                                                 if (m_pos == op)
279                                                         goto eof_found;
280                                                 m_pos -= 0x4000;
281                                         }
282                                         if (litp == NULL)
283                                                 goto copy_m;
284
285                                         nl = ip[-2] & 3;
286                                         /* test if in beetween two matches */
287                                         if (t == 1 && lit == 0 && nl == 0 && ip[0] >= 16) {
288                                                 next_lit = nl;
289                                                 /* make a previous short run */
290                                                 lit += 3;
291                                                 *litp = (unsigned char)((*litp & ~3) | lit);
292                                                 /* copy over the 3 literals that replace the match */
293                                                 copy3(ip-3, m_pos, pd(op, m_pos));
294                                                 o_m3_a++;
295                                         }
296                                         /* test if a literal run follows */
297                                         else if (t == 1 && lit <= 3 && nl == 0
298                                          && ip[0] < 16 && ip[0] != 0 && (lit + 3 + ip[0] < 16)
299                                         ) {
300                                                 t = *ip++;
301                                                 /* remove short run */
302                                                 *litp &= ~3;
303                                                 /* copy over the 3 literals that replace the match */
304                                                 copy3(ip-4+1, m_pos, pd(op, m_pos));
305                                                 /* move literals 1 byte ahead */
306                                                 litp += 2;
307                                                 if (lit > 0)
308                                                         memmove(litp+1,litp,lit);
309                                                 /* insert new length of long literal run */
310                                                 lit += 3 + t + 3;
311                                                 *litp = (unsigned char)(lit - 3);
312
313                                                 o_m3_b++;
314                                                 *op++ = *m_pos++;
315                                                 *op++ = *m_pos++;
316                                                 *op++ = *m_pos++;
317                                                 goto copy_literal_run;
318                                         }
319                                 }
320  copy_m:
321                                 *op++ = *m_pos++;
322                                 *op++ = *m_pos++;
323                                 do *op++ = *m_pos++; while (--t > 0);
324                         }
325
326  match_done:
327                         if (next_lit == NO_LIT) {
328                                 t = ip[-2] & 3;
329                                 lit = t;
330                                 litp = ip - 2;
331                         }
332                         else
333                                 t = next_lit;
334                         next_lit = NO_LIT;
335                         if (t == 0)
336                                 break;
337                         /* copy literals */
338  match_next:
339                         do *op++ = *ip++; while (--t > 0);
340                         t = *ip++;
341                 } while (TEST_IP && TEST_OP);
342         }
343
344         /* no EOF code was found */
345         *out_len = pd(op, out);
346         return LZO_E_EOF_NOT_FOUND;
347
348  eof_found:
349 //      LZO_UNUSED(o_m1_a); LZO_UNUSED(o_m1_b); LZO_UNUSED(o_m2);
350 //      LZO_UNUSED(o_m3_a); LZO_UNUSED(o_m3_b);
351         *out_len = pd(op, out);
352         return (ip == ip_end ? LZO_E_OK :
353                 (ip < ip_end ? LZO_E_INPUT_NOT_CONSUMED : LZO_E_INPUT_OVERRUN));
354 }
355
356 /**********************************************************************/
357 #define F_OS F_OS_UNIX
358 #define F_CS F_CS_NATIVE
359
360 /**********************************************************************/
361 #define ADLER32_INIT_VALUE 1
362 #define CRC32_INIT_VALUE   0
363
364 /**********************************************************************/
365 enum {
366         M_LZO1X_1    = 1,
367         M_LZO1X_1_15 = 2,
368         M_LZO1X_999  = 3,
369 };
370
371 /**********************************************************************/
372 /* header flags */
373 #define F_ADLER32_D     0x00000001L
374 #define F_ADLER32_C     0x00000002L
375 #define F_H_EXTRA_FIELD 0x00000040L
376 #define F_H_GMTDIFF     0x00000080L
377 #define F_CRC32_D       0x00000100L
378 #define F_CRC32_C       0x00000200L
379 #define F_H_FILTER      0x00000800L
380 #define F_H_CRC32       0x00001000L
381 #define F_MASK          0x00003FFFL
382
383 /* operating system & file system that created the file [mostly unused] */
384 #define F_OS_UNIX       0x03000000L
385 #define F_OS_SHIFT      24
386 #define F_OS_MASK       0xff000000L
387
388 /* character set for file name encoding [mostly unused] */
389 #define F_CS_NATIVE     0x00000000L
390 #define F_CS_SHIFT      20
391 #define F_CS_MASK       0x00f00000L
392
393 /* these bits must be zero */
394 #define F_RESERVED      ((F_MASK | F_OS_MASK | F_CS_MASK) ^ 0xffffffffL)
395
396 typedef struct chksum_t {
397         uint32_t f_adler32;
398         uint32_t f_crc32;
399 } chksum_t;
400
401 typedef struct header_t {
402         unsigned version;
403         unsigned lib_version;
404         unsigned version_needed_to_extract;
405         uint32_t flags;
406         uint32_t mode;
407         uint32_t mtime;
408         uint32_t gmtdiff;
409         uint32_t header_checksum;
410
411         uint32_t extra_field_len;
412         uint32_t extra_field_checksum;
413
414         unsigned char method;
415         unsigned char level;
416
417         /* info */
418         char name[255+1];
419 } header_t;
420
421 struct globals {
422         /*const uint32_t *lzo_crc32_table;*/
423         chksum_t chksum_in;
424         chksum_t chksum_out;
425 } FIX_ALIASING;
426 #define G (*(struct globals*)&bb_common_bufsiz1)
427 #define INIT_G() do { } while (0)
428 //#define G (*ptr_to_globals)
429 //#define INIT_G() do {
430 //      SET_PTR_TO_GLOBALS(xzalloc(sizeof(G)));
431 //} while (0)
432
433
434 /**********************************************************************/
435 #define LZOP_VERSION            0x1010
436 //#define LZOP_VERSION_STRING     "1.01"
437 //#define LZOP_VERSION_DATE       "Apr 27th 2003"
438
439 #define OPTION_STRING "cfvqdt123456789CF"
440
441 /* Note: must be kept in sync with archival/bbunzip.c */
442 enum {
443         OPT_STDOUT      = (1 << 0),
444         OPT_FORCE       = (1 << 1),
445         OPT_VERBOSE     = (1 << 2),
446         OPT_QUIET       = (1 << 3),
447         OPT_DECOMPRESS  = (1 << 4),
448         OPT_TEST        = (1 << 5),
449         OPT_1           = (1 << 6),
450         OPT_2           = (1 << 7),
451         OPT_3           = (1 << 8),
452         OPT_4           = (1 << 9),
453         OPT_5           = (1 << 10),
454         OPT_6           = (1 << 11),
455         OPT_789         = (7 << 12),
456         OPT_7           = (1 << 13),
457         OPT_8           = (1 << 14),
458         OPT_C           = (1 << 15),
459         OPT_F           = (1 << 16),
460 };
461
462 /**********************************************************************/
463 // adler32 checksum
464 // adapted from free code by Mark Adler <madler@alumni.caltech.edu>
465 // see http://www.zlib.org/
466 /**********************************************************************/
467 static FAST_FUNC uint32_t
468 lzo_adler32(uint32_t adler, const uint8_t* buf, unsigned len)
469 {
470         enum {
471                 LZO_BASE = 65521, /* largest prime smaller than 65536 */
472                 /* NMAX is the largest n such that
473                  * 255n(n+1)/2 + (n+1)(BASE-1) <= 2^32-1 */
474                 LZO_NMAX = 5552,
475         };
476         uint32_t s1 = adler & 0xffff;
477         uint32_t s2 = (adler >> 16) & 0xffff;
478         unsigned k;
479
480         if (buf == NULL)
481                 return 1;
482
483         while (len > 0) {
484                 k = len < LZO_NMAX ? (unsigned) len : LZO_NMAX;
485                 len -= k;
486                 if (k != 0) do {
487                         s1 += *buf++;
488                         s2 += s1;
489                 } while (--k > 0);
490                 s1 %= LZO_BASE;
491                 s2 %= LZO_BASE;
492         }
493         return (s2 << 16) | s1;
494 }
495
496 static FAST_FUNC uint32_t
497 lzo_crc32(uint32_t c, const uint8_t* buf, unsigned len)
498 {
499         //if (buf == NULL) - impossible
500         //      return 0;
501
502         return ~crc32_block_endian0(~c, buf, len, global_crc32_table);
503 }
504
505 /**********************************************************************/
506 static void init_chksum(chksum_t *ct)
507 {
508         ct->f_adler32 = ADLER32_INIT_VALUE;
509         ct->f_crc32 = CRC32_INIT_VALUE;
510 }
511
512 static void add_bytes_to_chksum(chksum_t *ct, const void* buf, int cnt)
513 {
514         /* We need to handle the two checksums at once, because at the
515          * beginning of the header, we don't know yet which one we'll
516          * eventually need */
517         ct->f_adler32 = lzo_adler32(ct->f_adler32, (const uint8_t*)buf, cnt);
518         ct->f_crc32 = lzo_crc32(ct->f_crc32, (const uint8_t*)buf, cnt);
519 }
520
521 static uint32_t chksum_getresult(chksum_t *ct, const header_t *h)
522 {
523         return (h->flags & F_H_CRC32) ? ct->f_crc32 : ct->f_adler32;
524 }
525
526 /**********************************************************************/
527 static uint32_t read32(void)
528 {
529         uint32_t v;
530         xread(0, &v, 4);
531         return ntohl(v);
532 }
533
534 static void write32(uint32_t v)
535 {
536         v = htonl(v);
537         xwrite(1, &v, 4);
538 }
539
540 static void f_write(const void* buf, int cnt)
541 {
542         xwrite(1, buf, cnt);
543         add_bytes_to_chksum(&G.chksum_out, buf, cnt);
544 }
545
546 static void f_read(void* buf, int cnt)
547 {
548         xread(0, buf, cnt);
549         add_bytes_to_chksum(&G.chksum_in, buf, cnt);
550 }
551
552 static int f_read8(void)
553 {
554         uint8_t v;
555         f_read(&v, 1);
556         return v;
557 }
558
559 static void f_write8(uint8_t v)
560 {
561         f_write(&v, 1);
562 }
563
564 static unsigned f_read16(void)
565 {
566         uint16_t v;
567         f_read(&v, 2);
568         return ntohs(v);
569 }
570
571 static void f_write16(uint16_t v)
572 {
573         v = htons(v);
574         f_write(&v, 2);
575 }
576
577 static uint32_t f_read32(void)
578 {
579         uint32_t v;
580         f_read(&v, 4);
581         return ntohl(v);
582 }
583
584 static void f_write32(uint32_t v)
585 {
586         v = htonl(v);
587         f_write(&v, 4);
588 }
589
590 /**********************************************************************/
591 static int lzo_get_method(header_t *h)
592 {
593         /* check method */
594         if (h->method == M_LZO1X_1) {
595                 if (h->level == 0)
596                         h->level = 3;
597         } else if (h->method == M_LZO1X_1_15) {
598                 if (h->level == 0)
599                         h->level = 1;
600         } else if (h->method == M_LZO1X_999) {
601                 if (h->level == 0)
602                         h->level = 9;
603         } else
604                 return -1;              /* not a LZO method */
605
606         /* check compression level */
607         if (h->level < 1 || h->level > 9)
608                 return 15;
609
610         return 0;
611 }
612
613 /**********************************************************************/
614 #define LZO_BLOCK_SIZE  (256 * 1024l)
615 #define MAX_BLOCK_SIZE  (64 * 1024l * 1024l)    /* DO NOT CHANGE */
616
617 /* LZO may expand uncompressible data by a small amount */
618 #define MAX_COMPRESSED_SIZE(x)  ((x) + (x) / 16 + 64 + 3)
619
620 /**********************************************************************/
621 // compress a file
622 /**********************************************************************/
623 static NOINLINE smallint lzo_compress(const header_t *h)
624 {
625         unsigned block_size = LZO_BLOCK_SIZE;
626         int r = 0; /* LZO_E_OK */
627         uint8_t *const b1 = xzalloc(block_size);
628         uint8_t *const b2 = xzalloc(MAX_COMPRESSED_SIZE(block_size));
629         unsigned src_len = 0, dst_len = 0;
630         uint32_t d_adler32 = ADLER32_INIT_VALUE;
631         uint32_t d_crc32 = CRC32_INIT_VALUE;
632         int l;
633         smallint ok = 1;
634         uint8_t *wrk_mem = NULL;
635
636         if (h->method == M_LZO1X_1)
637                 wrk_mem = xzalloc(LZO1X_1_MEM_COMPRESS);
638         else if (h->method == M_LZO1X_1_15)
639                 wrk_mem = xzalloc(LZO1X_1_15_MEM_COMPRESS);
640         else if (h->method == M_LZO1X_999)
641                 wrk_mem = xzalloc(LZO1X_999_MEM_COMPRESS);
642
643         for (;;) {
644                 /* read a block */
645                 l = full_read(0, b1, block_size);
646                 src_len = (l > 0 ? l : 0);
647
648                 /* write uncompressed block size */
649                 write32(src_len);
650
651                 /* exit if last block */
652                 if (src_len == 0)
653                         break;
654
655                 /* compute checksum of uncompressed block */
656                 if (h->flags & F_ADLER32_D)
657                         d_adler32 = lzo_adler32(ADLER32_INIT_VALUE, b1, src_len);
658                 if (h->flags & F_CRC32_D)
659                         d_crc32 = lzo_crc32(CRC32_INIT_VALUE, b1, src_len);
660
661                 /* compress */
662                 if (h->method == M_LZO1X_1)
663                         r = lzo1x_1_compress(b1, src_len, b2, &dst_len, wrk_mem);
664                 else if (h->method == M_LZO1X_1_15)
665                         r = lzo1x_1_15_compress(b1, src_len, b2, &dst_len, wrk_mem);
666 #if ENABLE_LZOP_COMPR_HIGH
667                 else if (h->method == M_LZO1X_999)
668                         r = lzo1x_999_compress_level(b1, src_len, b2, &dst_len,
669                                                 wrk_mem, h->level);
670 #endif
671                 else
672                         bb_error_msg_and_die("internal error");
673
674                 if (r != 0) /* not LZO_E_OK */
675                         bb_error_msg_and_die("internal error - compression failed");
676
677                 /* write compressed block size */
678                 if (dst_len < src_len) {
679                         /* optimize */
680                         if (h->method == M_LZO1X_999) {
681                                 unsigned new_len = src_len;
682                                 r = lzo1x_optimize(b2, dst_len, b1, &new_len, NULL);
683                                 if (r != 0 /*LZO_E_OK*/ || new_len != src_len)
684                                         bb_error_msg_and_die("internal error - optimization failed");
685                         }
686                         write32(dst_len);
687                 } else {
688                         /* data actually expanded => store data uncompressed */
689                         write32(src_len);
690                 }
691
692                 /* write checksum of uncompressed block */
693                 if (h->flags & F_ADLER32_D)
694                         write32(d_adler32);
695                 if (h->flags & F_CRC32_D)
696                         write32(d_crc32);
697
698                 if (dst_len < src_len) {
699                         /* write checksum of compressed block */
700                         if (h->flags & F_ADLER32_C)
701                                 write32(lzo_adler32(ADLER32_INIT_VALUE, b2, dst_len));
702                         if (h->flags & F_CRC32_C)
703                                 write32(lzo_crc32(CRC32_INIT_VALUE, b2, dst_len));
704                         /* write compressed block data */
705                         xwrite(1, b2, dst_len);
706                 } else {
707                         /* write uncompressed block data */
708                         xwrite(1, b1, src_len);
709                 }
710         }
711
712         free(wrk_mem);
713         free(b1);
714         free(b2);
715         return ok;
716 }
717
718 static FAST_FUNC void lzo_check(
719                 uint32_t init,
720                 uint8_t* buf, unsigned len,
721                 uint32_t FAST_FUNC (*fn)(uint32_t, const uint8_t*, unsigned),
722                 uint32_t ref)
723 {
724         /* This function, by having the same order of parameters
725          * as fn, and by being marked FAST_FUNC (same as fn),
726          * saves a dozen bytes of code.
727          */
728         uint32_t c = fn(init, buf, len);
729         if (c != ref)
730                 bb_error_msg_and_die("checksum error");
731 }
732
733 /**********************************************************************/
734 // decompress a file
735 /**********************************************************************/
736 static NOINLINE smallint lzo_decompress(const header_t *h)
737 {
738         unsigned block_size = LZO_BLOCK_SIZE;
739         int r;
740         uint32_t src_len, dst_len;
741         uint32_t c_adler32 = ADLER32_INIT_VALUE;
742         uint32_t d_adler32 = ADLER32_INIT_VALUE;
743         uint32_t c_crc32 = CRC32_INIT_VALUE, d_crc32 = CRC32_INIT_VALUE;
744         smallint ok = 1;
745         uint8_t *b1;
746         uint32_t mcs_block_size = MAX_COMPRESSED_SIZE(block_size);
747         uint8_t *b2 = NULL;
748
749         for (;;) {
750                 uint8_t *dst;
751
752                 /* read uncompressed block size */
753                 dst_len = read32();
754
755                 /* exit if last block */
756                 if (dst_len == 0)
757                         break;
758
759                 /* error if split file */
760                 if (dst_len == 0xffffffffL)
761                         /* should not happen - not yet implemented */
762                         bb_error_msg_and_die("this file is a split lzop file");
763
764                 if (dst_len > MAX_BLOCK_SIZE)
765                         bb_error_msg_and_die("corrupted data");
766
767                 /* read compressed block size */
768                 src_len = read32();
769                 if (src_len <= 0 || src_len > dst_len)
770                         bb_error_msg_and_die("corrupted data");
771
772                 if (dst_len > block_size) {
773                         if (b2) {
774                                 free(b2);
775                                 b2 = NULL;
776                         }
777                         block_size = dst_len;
778                         mcs_block_size = MAX_COMPRESSED_SIZE(block_size);
779                 }
780
781                 /* read checksum of uncompressed block */
782                 if (h->flags & F_ADLER32_D)
783                         d_adler32 = read32();
784                 if (h->flags & F_CRC32_D)
785                         d_crc32 = read32();
786
787                 /* read checksum of compressed block */
788                 if (src_len < dst_len) {
789                         if (h->flags & F_ADLER32_C)
790                                 c_adler32 = read32();
791                         if (h->flags & F_CRC32_C)
792                                 c_crc32 = read32();
793                 }
794
795                 if (b2 == NULL)
796                         b2 = xzalloc(mcs_block_size);
797                 /* read the block into the end of our buffer */
798                 b1 = b2 + mcs_block_size - src_len;
799                 xread(0, b1, src_len);
800
801                 if (src_len < dst_len) {
802                         unsigned d = dst_len;
803
804                         if (!(option_mask32 & OPT_F)) {
805                                 /* verify checksum of compressed block */
806                                 if (h->flags & F_ADLER32_C)
807                                         lzo_check(ADLER32_INIT_VALUE,
808                                                         b1, src_len,
809                                                         lzo_adler32, c_adler32);
810                                 if (h->flags & F_CRC32_C)
811                                         lzo_check(CRC32_INIT_VALUE,
812                                                         b1, src_len,
813                                                         lzo_crc32, c_crc32);
814                         }
815
816                         /* decompress */
817 //                      if (option_mask32 & OPT_F)
818 //                              r = lzo1x_decompress(b1, src_len, b2, &d, NULL);
819 //                      else
820                                 r = lzo1x_decompress_safe(b1, src_len, b2, &d, NULL);
821
822                         if (r != 0 /*LZO_E_OK*/ || dst_len != d) {
823                                 bb_error_msg_and_die("corrupted data");
824                         }
825                         dst = b2;
826                 } else {
827                         /* "stored" block => no decompression */
828                         dst = b1;
829                 }
830
831                 if (!(option_mask32 & OPT_F)) {
832                         /* verify checksum of uncompressed block */
833                         if (h->flags & F_ADLER32_D)
834                                 lzo_check(ADLER32_INIT_VALUE,
835                                         dst, dst_len,
836                                         lzo_adler32, d_adler32);
837                         if (h->flags & F_CRC32_D)
838                                 lzo_check(CRC32_INIT_VALUE,
839                                         dst, dst_len,
840                                         lzo_crc32, d_crc32);
841                 }
842
843                 /* write uncompressed block data */
844                 xwrite(1, dst, dst_len);
845         }
846
847         free(b2);
848         return ok;
849 }
850
851 /**********************************************************************/
852 // lzop file signature (shamelessly borrowed from PNG)
853 /**********************************************************************/
854 /*
855  * The first nine bytes of a lzop file always contain the following values:
856  *
857  *                                 0   1   2   3   4   5   6   7   8
858  *                               --- --- --- --- --- --- --- --- ---
859  * (hex)                          89  4c  5a  4f  00  0d  0a  1a  0a
860  * (decimal)                     137  76  90  79   0  13  10  26  10
861  * (C notation - ASCII)         \211   L   Z   O  \0  \r  \n \032 \n
862  */
863
864 /* (vda) comparison with lzop v1.02rc1 ("lzop -1 <FILE" cmd):
865  * Only slight differences in header:
866  * -00000000  89 4c 5a 4f 00 0d 0a 1a 0a 10 20 20 20 09 40 02
867  * +00000000  89 4c 5a 4f 00 0d 0a 1a 0a 10 10 20 30 09 40 02
868  *                                       ^^^^^ ^^^^^
869  *                                     version lib_version
870  * -00000010  01 03 00 00 0d 00 00 81 a4 49 f7 a6 3f 00 00 00
871  * +00000010  01 03 00 00 01 00 00 00 00 00 00 00 00 00 00 00
872  *               ^^^^^^^^^^^ ^^^^^^^^^^^ ^^^^^^^^^^^
873  *               flags       mode        mtime
874  * -00000020  00 00 2d 67 04 17 00 04 00 00 00 03 ed ec 9d 6d
875  * +00000020  00 00 10 5f 00 c1 00 04 00 00 00 03 ed ec 9d 6d
876  *                  ^^^^^^^^^^^
877  *                  chksum_out
878  * The rest is identical.
879 */
880 static const unsigned char lzop_magic[9] = {
881         0x89, 0x4c, 0x5a, 0x4f, 0x00, 0x0d, 0x0a, 0x1a, 0x0a
882 };
883
884 /* This coding is derived from Alexander Lehmann's pngcheck code. */
885 static void check_magic(void)
886 {
887         unsigned char magic[sizeof(lzop_magic)];
888         xread(0, magic, sizeof(magic));
889         if (memcmp(magic, lzop_magic, sizeof(lzop_magic)) != 0)
890                 bb_error_msg_and_die("bad magic number");
891 }
892
893 /**********************************************************************/
894 // lzop file header
895 /**********************************************************************/
896 static void write_header(const header_t *h)
897 {
898         int l;
899
900         xwrite(1, lzop_magic, sizeof(lzop_magic));
901
902         init_chksum(&G.chksum_out);
903
904         f_write16(h->version);
905         f_write16(h->lib_version);
906         f_write16(h->version_needed_to_extract);
907         f_write8(h->method);
908         f_write8(h->level);
909         f_write32(h->flags);
910         f_write32(h->mode);
911         f_write32(h->mtime);
912         f_write32(h->gmtdiff);
913
914         l = (int) strlen(h->name);
915         f_write8(l);
916         if (l)
917                 f_write(h->name, l);
918
919         f_write32(chksum_getresult(&G.chksum_out, h));
920 }
921
922 static int read_header(header_t *h)
923 {
924         int r;
925         int l;
926         uint32_t checksum;
927
928         memset(h, 0, sizeof(*h));
929         h->version_needed_to_extract = 0x0900;  /* first lzop version */
930         h->level = 0;
931
932         init_chksum(&G.chksum_in);
933
934         h->version = f_read16();
935         if (h->version < 0x0900)
936                 return 3;
937         h->lib_version = f_read16();
938         if (h->version >= 0x0940) {
939                 h->version_needed_to_extract = f_read16();
940                 if (h->version_needed_to_extract > LZOP_VERSION)
941                         return 16;
942                 if (h->version_needed_to_extract < 0x0900)
943                         return 3;
944         }
945         h->method = f_read8();
946         if (h->version >= 0x0940)
947                 h->level = f_read8();
948         h->flags = f_read32();
949         if (h->flags & F_H_FILTER)
950                 return 16; /* filter not supported */
951         h->mode = f_read32();
952         h->mtime = f_read32();
953         if (h->version >= 0x0940)
954                 h->gmtdiff = f_read32();
955
956         l = f_read8();
957         if (l > 0)
958                 f_read(h->name, l);
959         h->name[l] = 0;
960
961         checksum = chksum_getresult(&G.chksum_in, h);
962         h->header_checksum = f_read32();
963         if (h->header_checksum != checksum)
964                 return 2;
965
966         if (h->method <= 0)
967                 return 14;
968         r = lzo_get_method(h);
969         if (r != 0)
970                 return r;
971
972         /* check reserved flags */
973         if (h->flags & F_RESERVED)
974                 return -13;
975
976         /* skip extra field [not used yet] */
977         if (h->flags & F_H_EXTRA_FIELD) {
978                 uint32_t k;
979
980                 /* note: the checksum also covers the length */
981                 init_chksum(&G.chksum_in);
982                 h->extra_field_len = f_read32();
983                 for (k = 0; k < h->extra_field_len; k++)
984                         f_read8();
985                 checksum = chksum_getresult(&G.chksum_in, h);
986                 h->extra_field_checksum = f_read32();
987                 if (h->extra_field_checksum != checksum)
988                         return 3;
989         }
990
991         return 0;
992 }
993
994 static void p_header(header_t *h)
995 {
996         int r;
997
998         r = read_header(h);
999         if (r == 0)
1000                 return;
1001         bb_error_msg_and_die("header_error %d", r);
1002 }
1003
1004 /**********************************************************************/
1005 // compress
1006 /**********************************************************************/
1007 static void lzo_set_method(header_t *h)
1008 {
1009         int level = 1;
1010
1011         if (option_mask32 & OPT_1) {
1012                 h->method = M_LZO1X_1_15;
1013         } else if (option_mask32 & OPT_789) {
1014 #if ENABLE_LZOP_COMPR_HIGH
1015                 h->method = M_LZO1X_999;
1016                 if (option_mask32 & OPT_7)
1017                         level = 7;
1018                 else if (option_mask32 & OPT_8)
1019                         level = 8;
1020                 else
1021                         level = 9;
1022 #else
1023                 bb_error_msg_and_die("high compression not compiled in");
1024 #endif
1025         } else { /* levels 2..6 or none (defaults to level 3) */
1026                 h->method = M_LZO1X_1;
1027                 level = 5; /* levels 2-6 are actually the same */
1028         }
1029
1030         h->level = level;
1031 }
1032
1033 static smallint do_lzo_compress(void)
1034 {
1035         header_t header;
1036
1037 #define h (&header)
1038         memset(h, 0, sizeof(*h));
1039
1040         lzo_set_method(h);
1041
1042         h->version = (LZOP_VERSION & 0xffff);
1043         h->version_needed_to_extract = 0x0940;
1044         h->lib_version = lzo_version() & 0xffff;
1045
1046         h->flags = (F_OS & F_OS_MASK) | (F_CS & F_CS_MASK);
1047
1048         if (!(option_mask32 & OPT_F) || h->method == M_LZO1X_999) {
1049                 h->flags |= F_ADLER32_D;
1050                 if (option_mask32 & OPT_C)
1051                         h->flags |= F_ADLER32_C;
1052         }
1053         write_header(h);
1054         return lzo_compress(h);
1055 #undef h
1056 }
1057
1058 /**********************************************************************/
1059 // decompress
1060 /**********************************************************************/
1061 static smallint do_lzo_decompress(void)
1062 {
1063         header_t header;
1064
1065         check_magic();
1066         p_header(&header);
1067         return lzo_decompress(&header);
1068 }
1069
1070 static char* FAST_FUNC make_new_name_lzop(char *filename, const char *expected_ext UNUSED_PARAM)
1071 {
1072         if (option_mask32 & OPT_DECOMPRESS) {
1073                 char *extension = strrchr(filename, '.');
1074                 if (!extension || strcmp(extension + 1, "lzo") != 0)
1075                         return xasprintf("%s.out", filename);
1076                 *extension = '\0';
1077                 return filename;
1078         }
1079         return xasprintf("%s.lzo", filename);
1080 }
1081
1082 static IF_DESKTOP(long long) int FAST_FUNC pack_lzop(transformer_aux_data_t *aux UNUSED_PARAM)
1083 {
1084         if (option_mask32 & OPT_DECOMPRESS)
1085                 return do_lzo_decompress();
1086         return do_lzo_compress();
1087 }
1088
1089 int lzop_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
1090 int lzop_main(int argc UNUSED_PARAM, char **argv)
1091 {
1092         getopt32(argv, OPTION_STRING);
1093         argv += optind;
1094         /* lzopcat? */
1095         if (applet_name[4] == 'c')
1096                 option_mask32 |= (OPT_STDOUT | OPT_DECOMPRESS);
1097         /* unlzop? */
1098         if (applet_name[4] == 'o')
1099                 option_mask32 |= OPT_DECOMPRESS;
1100
1101         global_crc32_table = crc32_filltable(NULL, 0);
1102         return bbunpack(argv, pack_lzop, make_new_name_lzop, /*unused:*/ NULL);
1103 }