1 /*
2  * Simple XZ decoder command line tool
3  *
4  * Author: Lasse Collin <lasse.collin@tukaani.org>
5  *
6  * This file has been put into the public domain.
7  * You can do whatever you want with this file.
8  * Modified for toybox by Isaac Dunham
9 USE_XZCAT(NEWTOY(xzcat, NULL, TOYFLAG_USR|TOYFLAG_BIN))
10 
11 config XZCAT
12   bool "xzcat"
13   default n
14   help
15     usage: xzcat [filename...]
16 
17     Decompress listed files to stdout. Use stdin if no files listed.
18 
19 */
20 #define FOR_xzcat
21 #include "toys.h"
22 
23 // BEGIN xz.h
24 
25 /**
26  * enum xz_ret - Return codes
27  * @XZ_OK:                  Everything is OK so far. More input or more
28  *                          output space is required to continue.
29  * @XZ_STREAM_END:          Operation finished successfully.
30  * @XZ_UNSUPPORTED_CHECK:   Integrity check type is not supported. Decoding
31  *                          is still possible in multi-call mode by simply
32  *                          calling xz_dec_run() again.
33  *                          Note that this return value is used only if
34  *                          XZ_DEC_ANY_CHECK was defined at build time,
35  *                          which is not used in the kernel. Unsupported
36  *                          check types return XZ_OPTIONS_ERROR if
37  *                          XZ_DEC_ANY_CHECK was not defined at build time.
38  * @XZ_MEM_ERROR:           Allocating memory failed. The amount of memory
39  *                          that was tried to be allocated was no more than the
40  *                          dict_max argument given to xz_dec_init().
41  * @XZ_MEMLIMIT_ERROR:      A bigger LZMA2 dictionary would be needed than
42  *                          allowed by the dict_max argument given to
43  *                          xz_dec_init().
44  * @XZ_FORMAT_ERROR:        File format was not recognized (wrong magic
45  *                          bytes).
46  * @XZ_OPTIONS_ERROR:       This implementation doesn't support the requested
47  *                          compression options. In the decoder this means
48  *                          that the header CRC32 matches, but the header
49  *                          itself specifies something that we don't support.
50  * @XZ_DATA_ERROR:          Compressed data is corrupt.
51  * @XZ_BUF_ERROR:           Cannot make any progress. Details are slightly
52  *                          different between multi-call and single-call
53  *                          mode; more information below.
54  *
55  * XZ_BUF_ERROR is returned when two consecutive calls to XZ code cannot
56  * consume any input and cannot produce any new output. This happens when
57  * there is no new input available, or the output buffer is full while at
58  * least one output byte is still pending. Assuming your code is not buggy,
59  * you can get this error only when decoding a compressed stream that is
60  * truncated or otherwise corrupt.
61  */
62 enum xz_ret {
63   XZ_OK,
64   XZ_STREAM_END,
65   XZ_UNSUPPORTED_CHECK,
66   XZ_MEM_ERROR,
67   XZ_MEMLIMIT_ERROR,
68   XZ_FORMAT_ERROR,
69   XZ_OPTIONS_ERROR,
70   XZ_DATA_ERROR,
71   XZ_BUF_ERROR
72 };
73 
74 /**
75  * struct xz_buf - Passing input and output buffers to XZ code
76  * @in:         Beginning of the input buffer. This may be NULL if and only
77  *              if in_pos is equal to in_size.
78  * @in_pos:     Current position in the input buffer. This must not exceed
79  *              in_size.
80  * @in_size:    Size of the input buffer
81  * @out:        Beginning of the output buffer. This may be NULL if and only
82  *              if out_pos is equal to out_size.
83  * @out_pos:    Current position in the output buffer. This must not exceed
84  *              out_size.
85  * @out_size:   Size of the output buffer
86  *
87  * Only the contents of the output buffer from out[out_pos] onward, and
88  * the variables in_pos and out_pos are modified by the XZ code.
89  */
90 struct xz_buf {
91   const uint8_t *in;
92   size_t in_pos;
93   size_t in_size;
94 
95   uint8_t *out;
96   size_t out_pos;
97   size_t out_size;
98 };
99 
100 /**
101  * struct xz_dec - Opaque type to hold the XZ decoder state
102  */
103 struct xz_dec;
104 
105 /**
106  * xz_dec_init() - Allocate and initialize a XZ decoder state
107  * @mode:       Operation mode
108  * @dict_max:   Maximum size of the LZMA2 dictionary (history buffer) for
109  *              multi-call decoding. LZMA2 dictionary is always 2^n bytes
110  *              or 2^n + 2^(n-1) bytes (the latter sizes are less common
111  *              in practice), so other values for dict_max don't make sense.
112  *              In the kernel, dictionary sizes of 64 KiB, 128 KiB, 256 KiB,
113  *              512 KiB, and 1 MiB are probably the only reasonable values,
114  *              except for kernel and initramfs images where a bigger
115  *              dictionary can be fine and useful.
116  *
117  * dict_max specifies the maximum allowed dictionary size that xz_dec_run()
118  * may allocate once it has parsed the dictionary size from the stream
119  * headers. This way excessive allocations can be avoided while still
120  * limiting the maximum memory usage to a sane value to prevent running the
121  * system out of memory when decompressing streams from untrusted sources.
122  *
123  * On success, xz_dec_init() returns a pointer to struct xz_dec, which is
124  * ready to be used with xz_dec_run(). If memory allocation fails,
125  * xz_dec_init() returns NULL.
126  */
127 struct xz_dec *xz_dec_init(uint32_t dict_max);
128 
129 /**
130  * xz_dec_run() - Run the XZ decoder
131  * @s:          Decoder state allocated using xz_dec_init()
132  * @b:          Input and output buffers
133  *
134  * The possible return values depend on build options and operation mode.
135  * See enum xz_ret for details.
136  *
137  * Note that if an error occurs in single-call mode (return value is not
138  * XZ_STREAM_END), b->in_pos and b->out_pos are not modified and the
139  * contents of the output buffer from b->out[b->out_pos] onward are
140  * undefined. This is true even after XZ_BUF_ERROR, because with some filter
141  * chains, there may be a second pass over the output buffer, and this pass
142  * cannot be properly done if the output buffer is truncated. Thus, you
143  * cannot give the single-call decoder a too small buffer and then expect to
144  * get that amount valid data from the beginning of the stream. You must use
145  * the multi-call decoder if you don't want to uncompress the whole stream.
146  */
147 enum xz_ret xz_dec_run(struct xz_dec *s, struct xz_buf *b);
148 
149 /**
150  * xz_dec_reset() - Reset an already allocated decoder state
151  * @s:          Decoder state allocated using xz_dec_init()
152  *
153  * This function can be used to reset the multi-call decoder state without
154  * freeing and reallocating memory with xz_dec_end() and xz_dec_init().
155  *
156  * In single-call mode, xz_dec_reset() is always called in the beginning of
157  * xz_dec_run(). Thus, explicit call to xz_dec_reset() is useful only in
158  * multi-call mode.
159  */
160 void xz_dec_reset(struct xz_dec *s);
161 
162 /**
163  * xz_dec_end() - Free the memory allocated for the decoder state
164  * @s:          Decoder state allocated using xz_dec_init(). If s is NULL,
165  *              this function does nothing.
166  */
167 void xz_dec_end(struct xz_dec *s);
168 
169 /*
170  * Update CRC32 value using the polynomial from IEEE-802.3. To start a new
171  * calculation, the third argument must be zero. To continue the calculation,
172  * the previously returned value is passed as the third argument.
173  */
174 static uint32_t xz_crc32_table[256];
175 
xz_crc32(const uint8_t * buf,size_t size,uint32_t crc)176 uint32_t xz_crc32(const uint8_t *buf, size_t size, uint32_t crc)
177 {
178   crc = ~crc;
179 
180   while (size != 0) {
181     crc = xz_crc32_table[*buf++ ^ (crc & 0xFF)] ^ (crc >> 8);
182     --size;
183   }
184 
185   return ~crc;
186 }
187 
188 static uint64_t xz_crc64_table[256];
189 
190 
191 // END xz.h
192 
193 static uint8_t in[BUFSIZ];
194 static uint8_t out[BUFSIZ];
195 
do_xzcat(int fd,char * name)196 void do_xzcat(int fd, char *name)
197 {
198   struct xz_buf b;
199   struct xz_dec *s;
200   enum xz_ret ret;
201   const char *msg;
202 
203   crc_init(xz_crc32_table, 1);
204   const uint64_t poly = 0xC96C5795D7870F42ULL;
205   uint32_t i;
206   uint32_t j;
207   uint64_t r;
208 
209   /* initialize CRC64 table*/
210   for (i = 0; i < 256; ++i) {
211     r = i;
212     for (j = 0; j < 8; ++j)
213       r = (r >> 1) ^ (poly & ~((r & 1) - 1));
214 
215     xz_crc64_table[i] = r;
216   }
217 
218   /*
219    * Support up to 64 MiB dictionary. The actually needed memory
220    * is allocated once the headers have been parsed.
221    */
222   s = xz_dec_init(1 << 26);
223   if (s == NULL) {
224     msg = "Memory allocation failed\n";
225     goto error;
226   }
227 
228   b.in = in;
229   b.in_pos = 0;
230   b.in_size = 0;
231   b.out = out;
232   b.out_pos = 0;
233   b.out_size = BUFSIZ;
234 
235   for (;;) {
236     if (b.in_pos == b.in_size) {
237       b.in_size = read(fd, in, sizeof(in));
238       b.in_pos = 0;
239     }
240 
241     ret = xz_dec_run(s, &b);
242 
243     if (b.out_pos == sizeof(out)) {
244       if (fwrite(out, 1, b.out_pos, stdout) != b.out_pos) {
245         msg = "Write error\n";
246         goto error;
247       }
248 
249       b.out_pos = 0;
250     }
251 
252     if (ret == XZ_OK)
253       continue;
254 
255     if (ret == XZ_UNSUPPORTED_CHECK)
256       continue;
257 
258     if (fwrite(out, 1, b.out_pos, stdout) != b.out_pos) {
259       msg = "Write error\n";
260       goto error;
261     }
262 
263     switch (ret) {
264     case XZ_STREAM_END:
265       xz_dec_end(s);
266       return;
267 
268     case XZ_MEM_ERROR:
269       msg = "Memory allocation failed\n";
270       goto error;
271 
272     case XZ_MEMLIMIT_ERROR:
273       msg = "Memory usage limit reached\n";
274       goto error;
275 
276     case XZ_FORMAT_ERROR:
277       msg = "Not a .xz file\n";
278       goto error;
279 
280     case XZ_OPTIONS_ERROR:
281       msg = "Unsupported options in the .xz headers\n";
282       goto error;
283 
284     case XZ_DATA_ERROR:
285     case XZ_BUF_ERROR:
286       msg = "File is corrupt\n";
287       goto error;
288 
289     default:
290       msg = "Bug!\n";
291       goto error;
292     }
293   }
294 
295 error:
296   xz_dec_end(s);
297   error_exit("%s", msg);
298 }
299 
xzcat_main(void)300 void xzcat_main(void)
301 {
302   loopfiles(toys.optargs, do_xzcat);
303 }
304 
305 // BEGIN xz_private.h
306 
307 
308 /* Uncomment as needed to enable BCJ filter decoders.
309  * These cost about 2.5 k when all are enabled; SPARC and IA64 make 0.7 k
310  * */
311 
312 #define XZ_DEC_X86
313 #define XZ_DEC_POWERPC
314 #define XZ_DEC_IA64
315 #define XZ_DEC_ARM
316 #define XZ_DEC_ARMTHUMB
317 #define XZ_DEC_SPARC
318 
319 
320 #define memeq(a, b, size) (memcmp(a, b, size) == 0)
321 
322 #ifndef min
323 #	define min(x, y) ((x) < (y) ? (x) : (y))
324 #endif
325 #define min_t(type, x, y) min(x, y)
326 
327 
328 /* Inline functions to access unaligned unsigned 32-bit integers */
329 #ifndef get_unaligned_le32
get_unaligned_le32(const uint8_t * buf)330 static inline uint32_t get_unaligned_le32(const uint8_t *buf)
331 {
332   return (uint32_t)buf[0]
333       | ((uint32_t)buf[1] << 8)
334       | ((uint32_t)buf[2] << 16)
335       | ((uint32_t)buf[3] << 24);
336 }
337 #endif
338 
339 #ifndef get_unaligned_be32
get_unaligned_be32(const uint8_t * buf)340 static inline uint32_t get_unaligned_be32(const uint8_t *buf)
341 {
342   return (uint32_t)(buf[0] << 24)
343       | ((uint32_t)buf[1] << 16)
344       | ((uint32_t)buf[2] << 8)
345       | (uint32_t)buf[3];
346 }
347 #endif
348 
349 #ifndef put_unaligned_le32
put_unaligned_le32(uint32_t val,uint8_t * buf)350 static inline void put_unaligned_le32(uint32_t val, uint8_t *buf)
351 {
352   buf[0] = (uint8_t)val;
353   buf[1] = (uint8_t)(val >> 8);
354   buf[2] = (uint8_t)(val >> 16);
355   buf[3] = (uint8_t)(val >> 24);
356 }
357 #endif
358 
359 #ifndef put_unaligned_be32
put_unaligned_be32(uint32_t val,uint8_t * buf)360 static inline void put_unaligned_be32(uint32_t val, uint8_t *buf)
361 {
362   buf[0] = (uint8_t)(val >> 24);
363   buf[1] = (uint8_t)(val >> 16);
364   buf[2] = (uint8_t)(val >> 8);
365   buf[3] = (uint8_t)val;
366 }
367 #endif
368 
369 /*
370  * Use get_unaligned_le32() also for aligned access for simplicity. On
371  * little endian systems, #define get_le32(ptr) (*(const uint32_t *)(ptr))
372  * could save a few bytes in code size.
373  */
374 #ifndef get_le32
375 #	define get_le32 get_unaligned_le32
376 #endif
377 
378 /*
379  * If any of the BCJ filter decoders are wanted, define XZ_DEC_BCJ.
380  * XZ_DEC_BCJ is used to enable generic support for BCJ decoders.
381  */
382 #ifndef XZ_DEC_BCJ
383 #	if defined(XZ_DEC_X86) || defined(XZ_DEC_POWERPC) \
384       || defined(XZ_DEC_IA64) || defined(XZ_DEC_ARM) \
385       || defined(XZ_DEC_ARM) || defined(XZ_DEC_ARMTHUMB) \
386       || defined(XZ_DEC_SPARC)
387 #		define XZ_DEC_BCJ
388 #	endif
389 #endif
390 
391 /*
392  * Allocate memory for LZMA2 decoder. xz_dec_lzma2_reset() must be used
393  * before calling xz_dec_lzma2_run().
394  */
395 struct xz_dec_lzma2 *xz_dec_lzma2_create(uint32_t dict_max);
396 
397 /*
398  * Decode the LZMA2 properties (one byte) and reset the decoder. Return
399  * XZ_OK on success, XZ_MEMLIMIT_ERROR if the preallocated dictionary is not
400  * big enough, and XZ_OPTIONS_ERROR if props indicates something that this
401  * decoder doesn't support.
402  */
403 enum xz_ret xz_dec_lzma2_reset(struct xz_dec_lzma2 *s,
404            uint8_t props);
405 
406 /* Decode raw LZMA2 stream from b->in to b->out. */
407 enum xz_ret xz_dec_lzma2_run(struct xz_dec_lzma2 *s,
408                struct xz_buf *b);
409 
410 // END "xz_private.h"
411 
412 
413 
414 
415 /*
416  * Branch/Call/Jump (BCJ) filter decoders
417  * The rest of the code is inside this ifdef. It makes things a little more
418  * convenient when building without support for any BCJ filters.
419  */
420 #ifdef XZ_DEC_BCJ
421 
422 struct xz_dec_bcj {
423   /* Type of the BCJ filter being used */
424   enum {
425     BCJ_X86 = 4,        /* x86 or x86-64 */
426     BCJ_POWERPC = 5,    /* Big endian only */
427     BCJ_IA64 = 6,       /* Big or little endian */
428     BCJ_ARM = 7,        /* Little endian only */
429     BCJ_ARMTHUMB = 8,   /* Little endian only */
430     BCJ_SPARC = 9       /* Big or little endian */
431   } type;
432 
433   /*
434    * Return value of the next filter in the chain. We need to preserve
435    * this information across calls, because we must not call the next
436    * filter anymore once it has returned XZ_STREAM_END.
437    */
438   enum xz_ret ret;
439 
440   /*
441    * Absolute position relative to the beginning of the uncompressed
442    * data (in a single .xz Block). We care only about the lowest 32
443    * bits so this doesn't need to be uint64_t even with big files.
444    */
445   uint32_t pos;
446 
447   /* x86 filter state */
448   uint32_t x86_prev_mask;
449 
450   /* Temporary space to hold the variables from struct xz_buf */
451   uint8_t *out;
452   size_t out_pos;
453   size_t out_size;
454 
455   struct {
456     /* Amount of already filtered data in the beginning of buf */
457     size_t filtered;
458 
459     /* Total amount of data currently stored in buf  */
460     size_t size;
461 
462     /*
463      * Buffer to hold a mix of filtered and unfiltered data. This
464      * needs to be big enough to hold Alignment + 2 * Look-ahead:
465      *
466      * Type         Alignment   Look-ahead
467      * x86              1           4
468      * PowerPC          4           0
469      * IA-64           16           0
470      * ARM              4           0
471      * ARM-Thumb        2           2
472      * SPARC            4           0
473      */
474     uint8_t buf[16];
475   } temp;
476 };
477 
478 /*
479  * Decode the Filter ID of a BCJ filter. This implementation doesn't
480  * support custom start offsets, so no decoding of Filter Properties
481  * is needed. Returns XZ_OK if the given Filter ID is supported.
482  * Otherwise XZ_OPTIONS_ERROR is returned.
483  */
484 enum xz_ret xz_dec_bcj_reset(struct xz_dec_bcj *s, uint8_t id);
485 
486 /*
487  * Decode raw BCJ + LZMA2 stream. This must be used only if there actually is
488  * a BCJ filter in the chain. If the chain has only LZMA2, xz_dec_lzma2_run()
489  * must be called directly.
490  */
491 enum xz_ret xz_dec_bcj_run(struct xz_dec_bcj *s,
492              struct xz_dec_lzma2 *lzma2,
493              struct xz_buf *b);
494 
495 #ifdef XZ_DEC_X86
496 /*
497  * This is used to test the most significant byte of a memory address
498  * in an x86 instruction.
499  */
bcj_x86_test_msbyte(uint8_t b)500 static inline int bcj_x86_test_msbyte(uint8_t b)
501 {
502   return b == 0x00 || b == 0xFF;
503 }
504 
bcj_x86(struct xz_dec_bcj * s,uint8_t * buf,size_t size)505 static size_t bcj_x86(struct xz_dec_bcj *s, uint8_t *buf, size_t size)
506 {
507   static const int mask_to_allowed_status[8]
508     = { 1,1,1,0,1,0,0,0 };
509 
510   static const uint8_t mask_to_bit_num[8] = { 0, 1, 2, 2, 3, 3, 3, 3 };
511 
512   size_t i;
513   size_t prev_pos = (size_t)-1;
514   uint32_t prev_mask = s->x86_prev_mask;
515   uint32_t src;
516   uint32_t dest;
517   uint32_t j;
518   uint8_t b;
519 
520   if (size <= 4)
521     return 0;
522 
523   size -= 4;
524   for (i = 0; i < size; ++i) {
525     if ((buf[i] & 0xFE) != 0xE8)
526       continue;
527 
528     prev_pos = i - prev_pos;
529     if (prev_pos > 3) {
530       prev_mask = 0;
531     } else {
532       prev_mask = (prev_mask << (prev_pos - 1)) & 7;
533       if (prev_mask != 0) {
534         b = buf[i + 4 - mask_to_bit_num[prev_mask]];
535         if (!mask_to_allowed_status[prev_mask]
536             || bcj_x86_test_msbyte(b)) {
537           prev_pos = i;
538           prev_mask = (prev_mask << 1) | 1;
539           continue;
540         }
541       }
542     }
543 
544     prev_pos = i;
545 
546     if (bcj_x86_test_msbyte(buf[i + 4])) {
547       src = get_unaligned_le32(buf + i + 1);
548       for (;;) {
549         dest = src - (s->pos + (uint32_t)i + 5);
550         if (prev_mask == 0)
551           break;
552 
553         j = mask_to_bit_num[prev_mask] * 8;
554         b = (uint8_t)(dest >> (24 - j));
555         if (!bcj_x86_test_msbyte(b))
556           break;
557 
558         src = dest ^ (((uint32_t)1 << (32 - j)) - 1);
559       }
560 
561       dest &= 0x01FFFFFF;
562       dest |= (uint32_t)0 - (dest & 0x01000000);
563       put_unaligned_le32(dest, buf + i + 1);
564       i += 4;
565     } else {
566       prev_mask = (prev_mask << 1) | 1;
567     }
568   }
569 
570   prev_pos = i - prev_pos;
571   s->x86_prev_mask = prev_pos > 3 ? 0 : prev_mask << (prev_pos - 1);
572   return i;
573 }
574 #endif
575 
576 #ifdef XZ_DEC_POWERPC
bcj_powerpc(struct xz_dec_bcj * s,uint8_t * buf,size_t size)577 static size_t bcj_powerpc(struct xz_dec_bcj *s, uint8_t *buf, size_t size)
578 {
579   size_t i;
580   uint32_t instr;
581 
582   for (i = 0; i + 4 <= size; i += 4) {
583     instr = get_unaligned_be32(buf + i);
584     if ((instr & 0xFC000003) == 0x48000001) {
585       instr &= 0x03FFFFFC;
586       instr -= s->pos + (uint32_t)i;
587       instr &= 0x03FFFFFC;
588       instr |= 0x48000001;
589       put_unaligned_be32(instr, buf + i);
590     }
591   }
592 
593   return i;
594 }
595 #endif
596 
597 #ifdef XZ_DEC_IA64
bcj_ia64(struct xz_dec_bcj * s,uint8_t * buf,size_t size)598 static size_t bcj_ia64(struct xz_dec_bcj *s, uint8_t *buf, size_t size)
599 {
600   static const uint8_t branch_table[32] = {
601     0, 0, 0, 0, 0, 0, 0, 0,
602     0, 0, 0, 0, 0, 0, 0, 0,
603     4, 4, 6, 6, 0, 0, 7, 7,
604     4, 4, 0, 0, 4, 4, 0, 0
605   };
606 
607   /*
608    * The local variables take a little bit stack space, but it's less
609    * than what LZMA2 decoder takes, so it doesn't make sense to reduce
610    * stack usage here without doing that for the LZMA2 decoder too.
611    */
612 
613   /* Loop counters */
614   size_t i;
615   size_t j;
616 
617   /* Instruction slot (0, 1, or 2) in the 128-bit instruction word */
618   uint32_t slot;
619 
620   /* Bitwise offset of the instruction indicated by slot */
621   uint32_t bit_pos;
622 
623   /* bit_pos split into byte and bit parts */
624   uint32_t byte_pos;
625   uint32_t bit_res;
626 
627   /* Address part of an instruction */
628   uint32_t addr;
629 
630   /* Mask used to detect which instructions to convert */
631   uint32_t mask;
632 
633   /* 41-bit instruction stored somewhere in the lowest 48 bits */
634   uint64_t instr;
635 
636   /* Instruction normalized with bit_res for easier manipulation */
637   uint64_t norm;
638 
639   for (i = 0; i + 16 <= size; i += 16) {
640     mask = branch_table[buf[i] & 0x1F];
641     for (slot = 0, bit_pos = 5; slot < 3; ++slot, bit_pos += 41) {
642       if (((mask >> slot) & 1) == 0)
643         continue;
644 
645       byte_pos = bit_pos >> 3;
646       bit_res = bit_pos & 7;
647       instr = 0;
648       for (j = 0; j < 6; ++j)
649         instr |= (uint64_t)(buf[i + j + byte_pos])
650             << (8 * j);
651 
652       norm = instr >> bit_res;
653 
654       if (((norm >> 37) & 0x0F) == 0x05
655           && ((norm >> 9) & 0x07) == 0) {
656         addr = (norm >> 13) & 0x0FFFFF;
657         addr |= ((uint32_t)(norm >> 36) & 1) << 20;
658         addr <<= 4;
659         addr -= s->pos + (uint32_t)i;
660         addr >>= 4;
661 
662         norm &= ~((uint64_t)0x8FFFFF << 13);
663         norm |= (uint64_t)(addr & 0x0FFFFF) << 13;
664         norm |= (uint64_t)(addr & 0x100000)
665             << (36 - 20);
666 
667         instr &= (1 << bit_res) - 1;
668         instr |= norm << bit_res;
669 
670         for (j = 0; j < 6; j++)
671           buf[i + j + byte_pos]
672             = (uint8_t)(instr >> (8 * j));
673       }
674     }
675   }
676 
677   return i;
678 }
679 #endif
680 
681 #ifdef XZ_DEC_ARM
bcj_arm(struct xz_dec_bcj * s,uint8_t * buf,size_t size)682 static size_t bcj_arm(struct xz_dec_bcj *s, uint8_t *buf, size_t size)
683 {
684   size_t i;
685   uint32_t addr;
686 
687   for (i = 0; i + 4 <= size; i += 4) {
688     if (buf[i + 3] == 0xEB) {
689       addr = (uint32_t)buf[i] | ((uint32_t)buf[i + 1] << 8)
690           | ((uint32_t)buf[i + 2] << 16);
691       addr <<= 2;
692       addr -= s->pos + (uint32_t)i + 8;
693       addr >>= 2;
694       buf[i] = (uint8_t)addr;
695       buf[i + 1] = (uint8_t)(addr >> 8);
696       buf[i + 2] = (uint8_t)(addr >> 16);
697     }
698   }
699 
700   return i;
701 }
702 #endif
703 
704 #ifdef XZ_DEC_ARMTHUMB
bcj_armthumb(struct xz_dec_bcj * s,uint8_t * buf,size_t size)705 static size_t bcj_armthumb(struct xz_dec_bcj *s, uint8_t *buf, size_t size)
706 {
707   size_t i;
708   uint32_t addr;
709 
710   for (i = 0; i + 4 <= size; i += 2) {
711     if ((buf[i + 1] & 0xF8) == 0xF0
712         && (buf[i + 3] & 0xF8) == 0xF8) {
713       addr = (((uint32_t)buf[i + 1] & 0x07) << 19)
714           | ((uint32_t)buf[i] << 11)
715           | (((uint32_t)buf[i + 3] & 0x07) << 8)
716           | (uint32_t)buf[i + 2];
717       addr <<= 1;
718       addr -= s->pos + (uint32_t)i + 4;
719       addr >>= 1;
720       buf[i + 1] = (uint8_t)(0xF0 | ((addr >> 19) & 0x07));
721       buf[i] = (uint8_t)(addr >> 11);
722       buf[i + 3] = (uint8_t)(0xF8 | ((addr >> 8) & 0x07));
723       buf[i + 2] = (uint8_t)addr;
724       i += 2;
725     }
726   }
727 
728   return i;
729 }
730 #endif
731 
732 #ifdef XZ_DEC_SPARC
bcj_sparc(struct xz_dec_bcj * s,uint8_t * buf,size_t size)733 static size_t bcj_sparc(struct xz_dec_bcj *s, uint8_t *buf, size_t size)
734 {
735   size_t i;
736   uint32_t instr;
737 
738   for (i = 0; i + 4 <= size; i += 4) {
739     instr = get_unaligned_be32(buf + i);
740     if ((instr >> 22) == 0x100 || (instr >> 22) == 0x1FF) {
741       instr <<= 2;
742       instr -= s->pos + (uint32_t)i;
743       instr >>= 2;
744       instr = ((uint32_t)0x40000000 - (instr & 0x400000))
745           | 0x40000000 | (instr & 0x3FFFFF);
746       put_unaligned_be32(instr, buf + i);
747     }
748   }
749 
750   return i;
751 }
752 #endif
753 
754 /*
755  * Apply the selected BCJ filter. Update *pos and s->pos to match the amount
756  * of data that got filtered.
757  *
758  * NOTE: This is implemented as a switch statement to avoid using function
759  * pointers, which could be problematic in the kernel boot code, which must
760  * avoid pointers to static data (at least on x86).
761  */
bcj_apply(struct xz_dec_bcj * s,uint8_t * buf,size_t * pos,size_t size)762 static void bcj_apply(struct xz_dec_bcj *s,
763           uint8_t *buf, size_t *pos, size_t size)
764 {
765   size_t filtered;
766 
767   buf += *pos;
768   size -= *pos;
769 
770   switch (s->type) {
771 #ifdef XZ_DEC_X86
772   case BCJ_X86:
773     filtered = bcj_x86(s, buf, size);
774     break;
775 #endif
776 #ifdef XZ_DEC_POWERPC
777   case BCJ_POWERPC:
778     filtered = bcj_powerpc(s, buf, size);
779     break;
780 #endif
781 #ifdef XZ_DEC_IA64
782   case BCJ_IA64:
783     filtered = bcj_ia64(s, buf, size);
784     break;
785 #endif
786 #ifdef XZ_DEC_ARM
787   case BCJ_ARM:
788     filtered = bcj_arm(s, buf, size);
789     break;
790 #endif
791 #ifdef XZ_DEC_ARMTHUMB
792   case BCJ_ARMTHUMB:
793     filtered = bcj_armthumb(s, buf, size);
794     break;
795 #endif
796 #ifdef XZ_DEC_SPARC
797   case BCJ_SPARC:
798     filtered = bcj_sparc(s, buf, size);
799     break;
800 #endif
801   default:
802     /* Never reached but silence compiler warnings. */
803     filtered = 0;
804     break;
805   }
806 
807   *pos += filtered;
808   s->pos += filtered;
809 }
810 
811 /*
812  * Flush pending filtered data from temp to the output buffer.
813  * Move the remaining mixture of possibly filtered and unfiltered
814  * data to the beginning of temp.
815  */
bcj_flush(struct xz_dec_bcj * s,struct xz_buf * b)816 static void bcj_flush(struct xz_dec_bcj *s, struct xz_buf *b)
817 {
818   size_t copy_size;
819 
820   copy_size = min_t(size_t, s->temp.filtered, b->out_size - b->out_pos);
821   memcpy(b->out + b->out_pos, s->temp.buf, copy_size);
822   b->out_pos += copy_size;
823 
824   s->temp.filtered -= copy_size;
825   s->temp.size -= copy_size;
826   memmove(s->temp.buf, s->temp.buf + copy_size, s->temp.size);
827 }
828 
829 /*
830  * The BCJ filter functions are primitive in sense that they process the
831  * data in chunks of 1-16 bytes. To hide this issue, this function does
832  * some buffering.
833  */
xz_dec_bcj_run(struct xz_dec_bcj * s,struct xz_dec_lzma2 * lzma2,struct xz_buf * b)834 enum xz_ret xz_dec_bcj_run(struct xz_dec_bcj *s,
835              struct xz_dec_lzma2 *lzma2,
836              struct xz_buf *b)
837 {
838   size_t out_start;
839 
840   /*
841    * Flush pending already filtered data to the output buffer. Return
842    * immediatelly if we couldn't flush everything, or if the next
843    * filter in the chain had already returned XZ_STREAM_END.
844    */
845   if (s->temp.filtered > 0) {
846     bcj_flush(s, b);
847     if (s->temp.filtered > 0)
848       return XZ_OK;
849 
850     if (s->ret == XZ_STREAM_END)
851       return XZ_STREAM_END;
852   }
853 
854   /*
855    * If we have more output space than what is currently pending in
856    * temp, copy the unfiltered data from temp to the output buffer
857    * and try to fill the output buffer by decoding more data from the
858    * next filter in the chain. Apply the BCJ filter on the new data
859    * in the output buffer. If everything cannot be filtered, copy it
860    * to temp and rewind the output buffer position accordingly.
861    *
862    * This needs to be always run when temp.size == 0 to handle a special
863    * case where the output buffer is full and the next filter has no
864    * more output coming but hasn't returned XZ_STREAM_END yet.
865    */
866   if (s->temp.size < b->out_size - b->out_pos || s->temp.size == 0) {
867     out_start = b->out_pos;
868     memcpy(b->out + b->out_pos, s->temp.buf, s->temp.size);
869     b->out_pos += s->temp.size;
870 
871     s->ret = xz_dec_lzma2_run(lzma2, b);
872     if (s->ret != XZ_STREAM_END
873         && (s->ret != XZ_OK ))
874       return s->ret;
875 
876     bcj_apply(s, b->out, &out_start, b->out_pos);
877 
878     /*
879      * As an exception, if the next filter returned XZ_STREAM_END,
880      * we can do that too, since the last few bytes that remain
881      * unfiltered are meant to remain unfiltered.
882      */
883     if (s->ret == XZ_STREAM_END)
884       return XZ_STREAM_END;
885 
886     s->temp.size = b->out_pos - out_start;
887     b->out_pos -= s->temp.size;
888     memcpy(s->temp.buf, b->out + b->out_pos, s->temp.size);
889 
890     /*
891      * If there wasn't enough input to the next filter to fill
892      * the output buffer with unfiltered data, there's no point
893      * to try decoding more data to temp.
894      */
895     if (b->out_pos + s->temp.size < b->out_size)
896       return XZ_OK;
897   }
898 
899   /*
900    * We have unfiltered data in temp. If the output buffer isn't full
901    * yet, try to fill the temp buffer by decoding more data from the
902    * next filter. Apply the BCJ filter on temp. Then we hopefully can
903    * fill the actual output buffer by copying filtered data from temp.
904    * A mix of filtered and unfiltered data may be left in temp; it will
905    * be taken care on the next call to this function.
906    */
907   if (b->out_pos < b->out_size) {
908     /* Make b->out{,_pos,_size} temporarily point to s->temp. */
909     s->out = b->out;
910     s->out_pos = b->out_pos;
911     s->out_size = b->out_size;
912     b->out = s->temp.buf;
913     b->out_pos = s->temp.size;
914     b->out_size = sizeof(s->temp.buf);
915 
916     s->ret = xz_dec_lzma2_run(lzma2, b);
917 
918     s->temp.size = b->out_pos;
919     b->out = s->out;
920     b->out_pos = s->out_pos;
921     b->out_size = s->out_size;
922 
923     if (s->ret != XZ_OK && s->ret != XZ_STREAM_END)
924       return s->ret;
925 
926     bcj_apply(s, s->temp.buf, &s->temp.filtered, s->temp.size);
927 
928     /*
929      * If the next filter returned XZ_STREAM_END, we mark that
930      * everything is filtered, since the last unfiltered bytes
931      * of the stream are meant to be left as is.
932      */
933     if (s->ret == XZ_STREAM_END)
934       s->temp.filtered = s->temp.size;
935 
936     bcj_flush(s, b);
937     if (s->temp.filtered > 0)
938       return XZ_OK;
939   }
940 
941   return s->ret;
942 }
943 
xz_dec_bcj_reset(struct xz_dec_bcj * s,uint8_t id)944 enum xz_ret xz_dec_bcj_reset(struct xz_dec_bcj *s, uint8_t id)
945 {
946   switch (id) {
947 #ifdef XZ_DEC_X86
948   case BCJ_X86:
949 #endif
950 #ifdef XZ_DEC_POWERPC
951   case BCJ_POWERPC:
952 #endif
953 #ifdef XZ_DEC_IA64
954   case BCJ_IA64:
955 #endif
956 #ifdef XZ_DEC_ARM
957   case BCJ_ARM:
958 #endif
959 #ifdef XZ_DEC_ARMTHUMB
960   case BCJ_ARMTHUMB:
961 #endif
962 #ifdef XZ_DEC_SPARC
963   case BCJ_SPARC:
964 #endif
965     break;
966 
967   default:
968     /* Unsupported Filter ID */
969     return XZ_OPTIONS_ERROR;
970   }
971 
972   s->type = id;
973   s->ret = XZ_OK;
974   s->pos = 0;
975   s->x86_prev_mask = 0;
976   s->temp.filtered = 0;
977   s->temp.size = 0;
978 
979   return XZ_OK;
980 }
981 
982 #endif
983 /*
984  * LZMA2 decoder
985  */
986 
987 
988 // BEGIN xz_lzma2.h
989 /*
990  * LZMA2 definitions
991  *
992  */
993 
994 
995 /* Range coder constants */
996 #define RC_SHIFT_BITS 8
997 #define RC_TOP_BITS 24
998 #define RC_TOP_VALUE (1 << RC_TOP_BITS)
999 #define RC_BIT_MODEL_TOTAL_BITS 11
1000 #define RC_BIT_MODEL_TOTAL (1 << RC_BIT_MODEL_TOTAL_BITS)
1001 #define RC_MOVE_BITS 5
1002 
1003 /*
1004  * Maximum number of position states. A position state is the lowest pb
1005  * number of bits of the current uncompressed offset. In some places there
1006  * are different sets of probabilities for different position states.
1007  */
1008 #define POS_STATES_MAX (1 << 4)
1009 
1010 /*
1011  * This enum is used to track which LZMA symbols have occurred most recently
1012  * and in which order. This information is used to predict the next symbol.
1013  *
1014  * Symbols:
1015  *  - Literal: One 8-bit byte
1016  *  - Match: Repeat a chunk of data at some distance
1017  *  - Long repeat: Multi-byte match at a recently seen distance
1018  *  - Short repeat: One-byte repeat at a recently seen distance
1019  *
1020  * The symbol names are in from STATE_oldest_older_previous. REP means
1021  * either short or long repeated match, and NONLIT means any non-literal.
1022  */
1023 enum lzma_state {
1024   STATE_LIT_LIT,
1025   STATE_MATCH_LIT_LIT,
1026   STATE_REP_LIT_LIT,
1027   STATE_SHORTREP_LIT_LIT,
1028   STATE_MATCH_LIT,
1029   STATE_REP_LIT,
1030   STATE_SHORTREP_LIT,
1031   STATE_LIT_MATCH,
1032   STATE_LIT_LONGREP,
1033   STATE_LIT_SHORTREP,
1034   STATE_NONLIT_MATCH,
1035   STATE_NONLIT_REP
1036 };
1037 
1038 /* Total number of states */
1039 #define STATES 12
1040 
1041 /* The lowest 7 states indicate that the previous state was a literal. */
1042 #define LIT_STATES 7
1043 
1044 /* Indicate that the latest symbol was a literal. */
lzma_state_literal(enum lzma_state * state)1045 static inline void lzma_state_literal(enum lzma_state *state)
1046 {
1047   if (*state <= STATE_SHORTREP_LIT_LIT)
1048     *state = STATE_LIT_LIT;
1049   else if (*state <= STATE_LIT_SHORTREP)
1050     *state -= 3;
1051   else
1052     *state -= 6;
1053 }
1054 
1055 /* Indicate that the latest symbol was a match. */
lzma_state_match(enum lzma_state * state)1056 static inline void lzma_state_match(enum lzma_state *state)
1057 {
1058   *state = *state < LIT_STATES ? STATE_LIT_MATCH : STATE_NONLIT_MATCH;
1059 }
1060 
1061 /* Indicate that the latest state was a long repeated match. */
lzma_state_long_rep(enum lzma_state * state)1062 static inline void lzma_state_long_rep(enum lzma_state *state)
1063 {
1064   *state = *state < LIT_STATES ? STATE_LIT_LONGREP : STATE_NONLIT_REP;
1065 }
1066 
1067 /* Indicate that the latest symbol was a short match. */
lzma_state_short_rep(enum lzma_state * state)1068 static inline void lzma_state_short_rep(enum lzma_state *state)
1069 {
1070   *state = *state < LIT_STATES ? STATE_LIT_SHORTREP : STATE_NONLIT_REP;
1071 }
1072 
1073 /* Test if the previous symbol was a literal. */
lzma_state_is_literal(enum lzma_state state)1074 static inline int lzma_state_is_literal(enum lzma_state state)
1075 {
1076   return state < LIT_STATES;
1077 }
1078 
1079 /* Each literal coder is divided in three sections:
1080  *   - 0x001-0x0FF: Without match byte
1081  *   - 0x101-0x1FF: With match byte; match bit is 0
1082  *   - 0x201-0x2FF: With match byte; match bit is 1
1083  *
1084  * Match byte is used when the previous LZMA symbol was something else than
1085  * a literal (that is, it was some kind of match).
1086  */
1087 #define LITERAL_CODER_SIZE 0x300
1088 
1089 /* Maximum number of literal coders */
1090 #define LITERAL_CODERS_MAX (1 << 4)
1091 
1092 /* Minimum length of a match is two bytes. */
1093 #define MATCH_LEN_MIN 2
1094 
1095 /* Match length is encoded with 4, 5, or 10 bits.
1096  *
1097  * Length   Bits
1098  *  2-9      4 = Choice=0 + 3 bits
1099  * 10-17     5 = Choice=1 + Choice2=0 + 3 bits
1100  * 18-273   10 = Choice=1 + Choice2=1 + 8 bits
1101  */
1102 #define LEN_LOW_BITS 3
1103 #define LEN_LOW_SYMBOLS (1 << LEN_LOW_BITS)
1104 #define LEN_MID_BITS 3
1105 #define LEN_MID_SYMBOLS (1 << LEN_MID_BITS)
1106 #define LEN_HIGH_BITS 8
1107 #define LEN_HIGH_SYMBOLS (1 << LEN_HIGH_BITS)
1108 #define LEN_SYMBOLS (LEN_LOW_SYMBOLS + LEN_MID_SYMBOLS + LEN_HIGH_SYMBOLS)
1109 
1110 /*
1111  * Maximum length of a match is 273 which is a result of the encoding
1112  * described above.
1113  */
1114 #define MATCH_LEN_MAX (MATCH_LEN_MIN + LEN_SYMBOLS - 1)
1115 
1116 /*
1117  * Different sets of probabilities are used for match distances that have
1118  * very short match length: Lengths of 2, 3, and 4 bytes have a separate
1119  * set of probabilities for each length. The matches with longer length
1120  * use a shared set of probabilities.
1121  */
1122 #define DIST_STATES 4
1123 
1124 /*
1125  * Get the index of the appropriate probability array for decoding
1126  * the distance slot.
1127  */
lzma_get_dist_state(uint32_t len)1128 static inline uint32_t lzma_get_dist_state(uint32_t len)
1129 {
1130   return len < DIST_STATES + MATCH_LEN_MIN
1131       ? len - MATCH_LEN_MIN : DIST_STATES - 1;
1132 }
1133 
1134 /*
1135  * The highest two bits of a 32-bit match distance are encoded using six bits.
1136  * This six-bit value is called a distance slot. This way encoding a 32-bit
1137  * value takes 6-36 bits, larger values taking more bits.
1138  */
1139 #define DIST_SLOT_BITS 6
1140 #define DIST_SLOTS (1 << DIST_SLOT_BITS)
1141 
1142 /* Match distances up to 127 are fully encoded using probabilities. Since
1143  * the highest two bits (distance slot) are always encoded using six bits,
1144  * the distances 0-3 don't need any additional bits to encode, since the
1145  * distance slot itself is the same as the actual distance. DIST_MODEL_START
1146  * indicates the first distance slot where at least one additional bit is
1147  * needed.
1148  */
1149 #define DIST_MODEL_START 4
1150 
1151 /*
1152  * Match distances greater than 127 are encoded in three pieces:
1153  *   - distance slot: the highest two bits
1154  *   - direct bits: 2-26 bits below the highest two bits
1155  *   - alignment bits: four lowest bits
1156  *
1157  * Direct bits don't use any probabilities.
1158  *
1159  * The distance slot value of 14 is for distances 128-191.
1160  */
1161 #define DIST_MODEL_END 14
1162 
1163 /* Distance slots that indicate a distance <= 127. */
1164 #define FULL_DISTANCES_BITS (DIST_MODEL_END / 2)
1165 #define FULL_DISTANCES (1 << FULL_DISTANCES_BITS)
1166 
1167 /*
1168  * For match distances greater than 127, only the highest two bits and the
1169  * lowest four bits (alignment) is encoded using probabilities.
1170  */
1171 #define ALIGN_BITS 4
1172 #define ALIGN_SIZE (1 << ALIGN_BITS)
1173 #define ALIGN_MASK (ALIGN_SIZE - 1)
1174 
1175 /* Total number of all probability variables */
1176 #define PROBS_TOTAL (1846 + LITERAL_CODERS_MAX * LITERAL_CODER_SIZE)
1177 
1178 /*
1179  * LZMA remembers the four most recent match distances. Reusing these
1180  * distances tends to take less space than re-encoding the actual
1181  * distance value.
1182  */
1183 #define REPS 4
1184 
1185 
1186 // END xz_lzma2.h
1187 
1188 /*
1189  * Range decoder initialization eats the first five bytes of each LZMA chunk.
1190  */
1191 #define RC_INIT_BYTES 5
1192 
1193 /*
1194  * Minimum number of usable input buffer to safely decode one LZMA symbol.
1195  * The worst case is that we decode 22 bits using probabilities and 26
1196  * direct bits. This may decode at maximum of 20 bytes of input. However,
1197  * lzma_main() does an extra normalization before returning, thus we
1198  * need to put 21 here.
1199  */
1200 #define LZMA_IN_REQUIRED 21
1201 
1202 /*
1203  * Dictionary (history buffer)
1204  *
1205  * These are always true:
1206  *    start <= pos <= full <= end
1207  *    pos <= limit <= end
1208  *    end == size
1209  *    size <= size_max
1210  *    allocated <= size
1211  *
1212  * Most of these variables are size_t as a relic of single-call mode,
1213  * in which the dictionary variables address the actual output
1214  * buffer directly.
1215  */
1216 struct dictionary {
1217   /* Beginning of the history buffer */
1218   uint8_t *buf;
1219 
1220   /* Old position in buf (before decoding more data) */
1221   size_t start;
1222 
1223   /* Position in buf */
1224   size_t pos;
1225 
1226   /*
1227    * How full dictionary is. This is used to detect corrupt input that
1228    * would read beyond the beginning of the uncompressed stream.
1229    */
1230   size_t full;
1231 
1232   /* Write limit; we don't write to buf[limit] or later bytes. */
1233   size_t limit;
1234 
1235   /* End of the dictionary buffer. This is the same as the dictionary size. */
1236   size_t end;
1237 
1238   /*
1239    * Size of the dictionary as specified in Block Header. This is used
1240    * together with "full" to detect corrupt input that would make us
1241    * read beyond the beginning of the uncompressed stream.
1242    */
1243   uint32_t size;
1244 
1245   /*
1246    * Maximum allowed dictionary size.
1247    */
1248   uint32_t size_max;
1249 
1250   /*
1251    * Amount of memory currently allocated for the dictionary.
1252    */
1253   uint32_t allocated;
1254 };
1255 
1256 /* Range decoder */
1257 struct rc_dec {
1258   uint32_t range;
1259   uint32_t code;
1260 
1261   /*
1262    * Number of initializing bytes remaining to be read
1263    * by rc_read_init().
1264    */
1265   uint32_t init_bytes_left;
1266 
1267   /*
1268    * Buffer from which we read our input. It can be either
1269    * temp.buf or the caller-provided input buffer.
1270    */
1271   const uint8_t *in;
1272   size_t in_pos;
1273   size_t in_limit;
1274 };
1275 
1276 /* Probabilities for a length decoder. */
1277 struct lzma_len_dec {
1278   /* Probability of match length being at least 10 */
1279   uint16_t choice;
1280 
1281   /* Probability of match length being at least 18 */
1282   uint16_t choice2;
1283 
1284   /* Probabilities for match lengths 2-9 */
1285   uint16_t low[POS_STATES_MAX][LEN_LOW_SYMBOLS];
1286 
1287   /* Probabilities for match lengths 10-17 */
1288   uint16_t mid[POS_STATES_MAX][LEN_MID_SYMBOLS];
1289 
1290   /* Probabilities for match lengths 18-273 */
1291   uint16_t high[LEN_HIGH_SYMBOLS];
1292 };
1293 
1294 struct lzma_dec {
1295   /* Distances of latest four matches */
1296   uint32_t rep0;
1297   uint32_t rep1;
1298   uint32_t rep2;
1299   uint32_t rep3;
1300 
1301   /* Types of the most recently seen LZMA symbols */
1302   enum lzma_state state;
1303 
1304   /*
1305    * Length of a match. This is updated so that dict_repeat can
1306    * be called again to finish repeating the whole match.
1307    */
1308   uint32_t len;
1309 
1310   /*
1311    * LZMA properties or related bit masks (number of literal
1312    * context bits, a mask dervied from the number of literal
1313    * position bits, and a mask dervied from the number
1314    * position bits)
1315    */
1316   uint32_t lc;
1317   uint32_t literal_pos_mask; /* (1 << lp) - 1 */
1318   uint32_t pos_mask;         /* (1 << pb) - 1 */
1319 
1320   /* If 1, it's a match. Otherwise it's a single 8-bit literal. */
1321   uint16_t is_match[STATES][POS_STATES_MAX];
1322 
1323   /* If 1, it's a repeated match. The distance is one of rep0 .. rep3. */
1324   uint16_t is_rep[STATES];
1325 
1326   /*
1327    * If 0, distance of a repeated match is rep0.
1328    * Otherwise check is_rep1.
1329    */
1330   uint16_t is_rep0[STATES];
1331 
1332   /*
1333    * If 0, distance of a repeated match is rep1.
1334    * Otherwise check is_rep2.
1335    */
1336   uint16_t is_rep1[STATES];
1337 
1338   /* If 0, distance of a repeated match is rep2. Otherwise it is rep3. */
1339   uint16_t is_rep2[STATES];
1340 
1341   /*
1342    * If 1, the repeated match has length of one byte. Otherwise
1343    * the length is decoded from rep_len_decoder.
1344    */
1345   uint16_t is_rep0_long[STATES][POS_STATES_MAX];
1346 
1347   /*
1348    * Probability tree for the highest two bits of the match
1349    * distance. There is a separate probability tree for match
1350    * lengths of 2 (i.e. MATCH_LEN_MIN), 3, 4, and [5, 273].
1351    */
1352   uint16_t dist_slot[DIST_STATES][DIST_SLOTS];
1353 
1354   /*
1355    * Probility trees for additional bits for match distance
1356    * when the distance is in the range [4, 127].
1357    */
1358   uint16_t dist_special[FULL_DISTANCES - DIST_MODEL_END];
1359 
1360   /*
1361    * Probability tree for the lowest four bits of a match
1362    * distance that is equal to or greater than 128.
1363    */
1364   uint16_t dist_align[ALIGN_SIZE];
1365 
1366   /* Length of a normal match */
1367   struct lzma_len_dec match_len_dec;
1368 
1369   /* Length of a repeated match */
1370   struct lzma_len_dec rep_len_dec;
1371 
1372   /* Probabilities of literals */
1373   uint16_t literal[LITERAL_CODERS_MAX][LITERAL_CODER_SIZE];
1374 };
1375 
1376 struct lzma2_dec {
1377   /* Position in xz_dec_lzma2_run(). */
1378   enum lzma2_seq {
1379     SEQ_CONTROL,
1380     SEQ_UNCOMPRESSED_1,
1381     SEQ_UNCOMPRESSED_2,
1382     SEQ_COMPRESSED_0,
1383     SEQ_COMPRESSED_1,
1384     SEQ_PROPERTIES,
1385     SEQ_LZMA_PREPARE,
1386     SEQ_LZMA_RUN,
1387     SEQ_COPY
1388   } sequence;
1389 
1390   /* Next position after decoding the compressed size of the chunk. */
1391   enum lzma2_seq next_sequence;
1392 
1393   /* Uncompressed size of LZMA chunk (2 MiB at maximum) */
1394   uint32_t uncompressed;
1395 
1396   /*
1397    * Compressed size of LZMA chunk or compressed/uncompressed
1398    * size of uncompressed chunk (64 KiB at maximum)
1399    */
1400   uint32_t compressed;
1401 
1402   /*
1403    * True if dictionary reset is needed. This is false before
1404    * the first chunk (LZMA or uncompressed).
1405    */
1406   int need_dict_reset;
1407 
1408   /*
1409    * True if new LZMA properties are needed. This is false
1410    * before the first LZMA chunk.
1411    */
1412   int need_props;
1413 };
1414 
1415 struct xz_dec_lzma2 {
1416   /*
1417    * The order below is important on x86 to reduce code size and
1418    * it shouldn't hurt on other platforms. Everything up to and
1419    * including lzma.pos_mask are in the first 128 bytes on x86-32,
1420    * which allows using smaller instructions to access those
1421    * variables. On x86-64, fewer variables fit into the first 128
1422    * bytes, but this is still the best order without sacrificing
1423    * the readability by splitting the structures.
1424    */
1425   struct rc_dec rc;
1426   struct dictionary dict;
1427   struct lzma2_dec lzma2;
1428   struct lzma_dec lzma;
1429 
1430   /*
1431    * Temporary buffer which holds small number of input bytes between
1432    * decoder calls. See lzma2_lzma() for details.
1433    */
1434   struct {
1435     uint32_t size;
1436     uint8_t buf[3 * LZMA_IN_REQUIRED];
1437   } temp;
1438 };
1439 
1440 /**************
1441  * Dictionary *
1442  **************/
1443 
1444 /* Reset the dictionary state. */
dict_reset(struct dictionary * dict)1445 static void dict_reset(struct dictionary *dict)
1446 {
1447   dict->start = 0;
1448   dict->pos = 0;
1449   dict->limit = 0;
1450   dict->full = 0;
1451 }
1452 
1453 /* Set dictionary write limit */
dict_limit(struct dictionary * dict,size_t out_max)1454 static void dict_limit(struct dictionary *dict, size_t out_max)
1455 {
1456   if (dict->end - dict->pos <= out_max)
1457     dict->limit = dict->end;
1458   else
1459     dict->limit = dict->pos + out_max;
1460 }
1461 
1462 /* Return true if at least one byte can be written into the dictionary. */
dict_has_space(const struct dictionary * dict)1463 static inline int dict_has_space(const struct dictionary *dict)
1464 {
1465   return dict->pos < dict->limit;
1466 }
1467 
1468 /*
1469  * Get a byte from the dictionary at the given distance. The distance is
1470  * assumed to valid, or as a special case, zero when the dictionary is
1471  * still empty. This special case is needed for single-call decoding to
1472  * avoid writing a '\0' to the end of the destination buffer.
1473  */
dict_get(const struct dictionary * dict,uint32_t dist)1474 static inline uint32_t dict_get(const struct dictionary *dict, uint32_t dist)
1475 {
1476   size_t offset = dict->pos - dist - 1;
1477 
1478   if (dist >= dict->pos)
1479     offset += dict->end;
1480 
1481   return dict->full > 0 ? dict->buf[offset] : 0;
1482 }
1483 
1484 /*
1485  * Put one byte into the dictionary. It is assumed that there is space for it.
1486  */
dict_put(struct dictionary * dict,uint8_t byte)1487 static inline void dict_put(struct dictionary *dict, uint8_t byte)
1488 {
1489   dict->buf[dict->pos++] = byte;
1490 
1491   if (dict->full < dict->pos)
1492     dict->full = dict->pos;
1493 }
1494 
1495 /*
1496  * Repeat given number of bytes from the given distance. If the distance is
1497  * invalid, false is returned. On success, true is returned and *len is
1498  * updated to indicate how many bytes were left to be repeated.
1499  */
dict_repeat(struct dictionary * dict,uint32_t * len,uint32_t dist)1500 static int dict_repeat(struct dictionary *dict, uint32_t *len, uint32_t dist)
1501 {
1502   size_t back;
1503   uint32_t left;
1504 
1505   if (dist >= dict->full || dist >= dict->size) return 0;
1506 
1507   left = min_t(size_t, dict->limit - dict->pos, *len);
1508   *len -= left;
1509 
1510   back = dict->pos - dist - 1;
1511   if (dist >= dict->pos)
1512     back += dict->end;
1513 
1514   do {
1515     dict->buf[dict->pos++] = dict->buf[back++];
1516     if (back == dict->end)
1517       back = 0;
1518   } while (--left > 0);
1519 
1520   if (dict->full < dict->pos)
1521     dict->full = dict->pos;
1522 
1523   return 1;
1524 }
1525 
1526 /* Copy uncompressed data as is from input to dictionary and output buffers. */
dict_uncompressed(struct dictionary * dict,struct xz_buf * b,uint32_t * left)1527 static void dict_uncompressed(struct dictionary *dict, struct xz_buf *b,
1528             uint32_t *left)
1529 {
1530   size_t copy_size;
1531 
1532   while (*left > 0 && b->in_pos < b->in_size
1533       && b->out_pos < b->out_size) {
1534     copy_size = min(b->in_size - b->in_pos,
1535         b->out_size - b->out_pos);
1536     if (copy_size > dict->end - dict->pos)
1537       copy_size = dict->end - dict->pos;
1538     if (copy_size > *left)
1539       copy_size = *left;
1540 
1541     *left -= copy_size;
1542 
1543     memcpy(dict->buf + dict->pos, b->in + b->in_pos, copy_size);
1544     dict->pos += copy_size;
1545 
1546     if (dict->full < dict->pos)
1547       dict->full = dict->pos;
1548 
1549     if (dict->pos == dict->end)
1550       dict->pos = 0;
1551 
1552     memcpy(b->out + b->out_pos, b->in + b->in_pos,
1553         copy_size);
1554 
1555     dict->start = dict->pos;
1556 
1557     b->out_pos += copy_size;
1558     b->in_pos += copy_size;
1559   }
1560 }
1561 
1562 /*
1563  * Flush pending data from dictionary to b->out. It is assumed that there is
1564  * enough space in b->out. This is guaranteed because caller uses dict_limit()
1565  * before decoding data into the dictionary.
1566  */
dict_flush(struct dictionary * dict,struct xz_buf * b)1567 static uint32_t dict_flush(struct dictionary *dict, struct xz_buf *b)
1568 {
1569   size_t copy_size = dict->pos - dict->start;
1570 
1571   if (dict->pos == dict->end)
1572     dict->pos = 0;
1573 
1574   memcpy(b->out + b->out_pos, dict->buf + dict->start,
1575       copy_size);
1576 
1577   dict->start = dict->pos;
1578   b->out_pos += copy_size;
1579   return copy_size;
1580 }
1581 
1582 /*****************
1583  * Range decoder *
1584  *****************/
1585 
1586 /* Reset the range decoder. */
rc_reset(struct rc_dec * rc)1587 static void rc_reset(struct rc_dec *rc)
1588 {
1589   rc->range = (uint32_t)-1;
1590   rc->code = 0;
1591   rc->init_bytes_left = RC_INIT_BYTES;
1592 }
1593 
1594 /*
1595  * Read the first five initial bytes into rc->code if they haven't been
1596  * read already. (Yes, the first byte gets completely ignored.)
1597  */
rc_read_init(struct rc_dec * rc,struct xz_buf * b)1598 static int rc_read_init(struct rc_dec *rc, struct xz_buf *b)
1599 {
1600   while (rc->init_bytes_left > 0) {
1601     if (b->in_pos == b->in_size) return 0;
1602 
1603     rc->code = (rc->code << 8) + b->in[b->in_pos++];
1604     --rc->init_bytes_left;
1605   }
1606 
1607   return 1;
1608 }
1609 
1610 /* Return true if there may not be enough input for the next decoding loop. */
rc_limit_exceeded(const struct rc_dec * rc)1611 static inline int rc_limit_exceeded(const struct rc_dec *rc)
1612 {
1613   return rc->in_pos > rc->in_limit;
1614 }
1615 
1616 /*
1617  * Return true if it is possible (from point of view of range decoder) that
1618  * we have reached the end of the LZMA chunk.
1619  */
rc_is_finished(const struct rc_dec * rc)1620 static inline int rc_is_finished(const struct rc_dec *rc)
1621 {
1622   return rc->code == 0;
1623 }
1624 
1625 /* Read the next input byte if needed. */
rc_normalize(struct rc_dec * rc)1626 static inline void rc_normalize(struct rc_dec *rc)
1627 {
1628   if (rc->range < RC_TOP_VALUE) {
1629     rc->range <<= RC_SHIFT_BITS;
1630     rc->code = (rc->code << RC_SHIFT_BITS) + rc->in[rc->in_pos++];
1631   }
1632 }
1633 
1634 /*
1635  * Decode one bit. In some versions, this function has been splitted in three
1636  * functions so that the compiler is supposed to be able to more easily avoid
1637  * an extra branch. In this particular version of the LZMA decoder, this
1638  * doesn't seem to be a good idea (tested with GCC 3.3.6, 3.4.6, and 4.3.3
1639  * on x86). Using a non-splitted version results in nicer looking code too.
1640  *
1641  * NOTE: This must return an int. Do not make it return a bool or the speed
1642  * of the code generated by GCC 3.x decreases 10-15 %. (GCC 4.3 doesn't care,
1643  * and it generates 10-20 % faster code than GCC 3.x from this file anyway.)
1644  */
rc_bit(struct rc_dec * rc,uint16_t * prob)1645 static inline int rc_bit(struct rc_dec *rc, uint16_t *prob)
1646 {
1647   uint32_t bound;
1648   int bit;
1649 
1650   rc_normalize(rc);
1651   bound = (rc->range >> RC_BIT_MODEL_TOTAL_BITS) * *prob;
1652   if (rc->code < bound) {
1653     rc->range = bound;
1654     *prob += (RC_BIT_MODEL_TOTAL - *prob) >> RC_MOVE_BITS;
1655     bit = 0;
1656   } else {
1657     rc->range -= bound;
1658     rc->code -= bound;
1659     *prob -= *prob >> RC_MOVE_BITS;
1660     bit = 1;
1661   }
1662 
1663   return bit;
1664 }
1665 
1666 /* Decode a bittree starting from the most significant bit. */
rc_bittree(struct rc_dec * rc,uint16_t * probs,uint32_t limit)1667 static inline uint32_t rc_bittree(struct rc_dec *rc,
1668              uint16_t *probs, uint32_t limit)
1669 {
1670   uint32_t symbol = 1;
1671 
1672   do {
1673     if (rc_bit(rc, &probs[symbol]))
1674       symbol = (symbol << 1) + 1;
1675     else
1676       symbol <<= 1;
1677   } while (symbol < limit);
1678 
1679   return symbol;
1680 }
1681 
1682 /* Decode a bittree starting from the least significant bit. */
rc_bittree_reverse(struct rc_dec * rc,uint16_t * probs,uint32_t * dest,uint32_t limit)1683 static inline void rc_bittree_reverse(struct rc_dec *rc,
1684                  uint16_t *probs,
1685                  uint32_t *dest, uint32_t limit)
1686 {
1687   uint32_t symbol = 1;
1688   uint32_t i = 0;
1689 
1690   do {
1691     if (rc_bit(rc, &probs[symbol])) {
1692       symbol = (symbol << 1) + 1;
1693       *dest += 1 << i;
1694     } else {
1695       symbol <<= 1;
1696     }
1697   } while (++i < limit);
1698 }
1699 
1700 /* Decode direct bits (fixed fifty-fifty probability) */
rc_direct(struct rc_dec * rc,uint32_t * dest,uint32_t limit)1701 static inline void rc_direct(struct rc_dec *rc, uint32_t *dest, uint32_t limit)
1702 {
1703   uint32_t mask;
1704 
1705   do {
1706     rc_normalize(rc);
1707     rc->range >>= 1;
1708     rc->code -= rc->range;
1709     mask = (uint32_t)0 - (rc->code >> 31);
1710     rc->code += rc->range & mask;
1711     *dest = (*dest << 1) + (mask + 1);
1712   } while (--limit > 0);
1713 }
1714 
1715 /********
1716  * LZMA *
1717  ********/
1718 
1719 /* Get pointer to literal coder probability array. */
lzma_literal_probs(struct xz_dec_lzma2 * s)1720 static uint16_t *lzma_literal_probs(struct xz_dec_lzma2 *s)
1721 {
1722   uint32_t prev_byte = dict_get(&s->dict, 0);
1723   uint32_t low = prev_byte >> (8 - s->lzma.lc);
1724   uint32_t high = (s->dict.pos & s->lzma.literal_pos_mask) << s->lzma.lc;
1725   return s->lzma.literal[low + high];
1726 }
1727 
1728 /* Decode a literal (one 8-bit byte) */
lzma_literal(struct xz_dec_lzma2 * s)1729 static void lzma_literal(struct xz_dec_lzma2 *s)
1730 {
1731   uint16_t *probs;
1732   uint32_t symbol;
1733   uint32_t match_byte;
1734   uint32_t match_bit;
1735   uint32_t offset;
1736   uint32_t i;
1737 
1738   probs = lzma_literal_probs(s);
1739 
1740   if (lzma_state_is_literal(s->lzma.state)) {
1741     symbol = rc_bittree(&s->rc, probs, 0x100);
1742   } else {
1743     symbol = 1;
1744     match_byte = dict_get(&s->dict, s->lzma.rep0) << 1;
1745     offset = 0x100;
1746 
1747     do {
1748       match_bit = match_byte & offset;
1749       match_byte <<= 1;
1750       i = offset + match_bit + symbol;
1751 
1752       if (rc_bit(&s->rc, &probs[i])) {
1753         symbol = (symbol << 1) + 1;
1754         offset &= match_bit;
1755       } else {
1756         symbol <<= 1;
1757         offset &= ~match_bit;
1758       }
1759     } while (symbol < 0x100);
1760   }
1761 
1762   dict_put(&s->dict, (uint8_t)symbol);
1763   lzma_state_literal(&s->lzma.state);
1764 }
1765 
1766 /* Decode the length of the match into s->lzma.len. */
lzma_len(struct xz_dec_lzma2 * s,struct lzma_len_dec * l,uint32_t pos_state)1767 static void lzma_len(struct xz_dec_lzma2 *s, struct lzma_len_dec *l,
1768          uint32_t pos_state)
1769 {
1770   uint16_t *probs;
1771   uint32_t limit;
1772 
1773   if (!rc_bit(&s->rc, &l->choice)) {
1774     probs = l->low[pos_state];
1775     limit = LEN_LOW_SYMBOLS;
1776     s->lzma.len = MATCH_LEN_MIN;
1777   } else {
1778     if (!rc_bit(&s->rc, &l->choice2)) {
1779       probs = l->mid[pos_state];
1780       limit = LEN_MID_SYMBOLS;
1781       s->lzma.len = MATCH_LEN_MIN + LEN_LOW_SYMBOLS;
1782     } else {
1783       probs = l->high;
1784       limit = LEN_HIGH_SYMBOLS;
1785       s->lzma.len = MATCH_LEN_MIN + LEN_LOW_SYMBOLS
1786           + LEN_MID_SYMBOLS;
1787     }
1788   }
1789 
1790   s->lzma.len += rc_bittree(&s->rc, probs, limit) - limit;
1791 }
1792 
1793 /* Decode a match. The distance will be stored in s->lzma.rep0. */
lzma_match(struct xz_dec_lzma2 * s,uint32_t pos_state)1794 static void lzma_match(struct xz_dec_lzma2 *s, uint32_t pos_state)
1795 {
1796   uint16_t *probs;
1797   uint32_t dist_slot;
1798   uint32_t limit;
1799 
1800   lzma_state_match(&s->lzma.state);
1801 
1802   s->lzma.rep3 = s->lzma.rep2;
1803   s->lzma.rep2 = s->lzma.rep1;
1804   s->lzma.rep1 = s->lzma.rep0;
1805 
1806   lzma_len(s, &s->lzma.match_len_dec, pos_state);
1807 
1808   probs = s->lzma.dist_slot[lzma_get_dist_state(s->lzma.len)];
1809   dist_slot = rc_bittree(&s->rc, probs, DIST_SLOTS) - DIST_SLOTS;
1810 
1811   if (dist_slot < DIST_MODEL_START) {
1812     s->lzma.rep0 = dist_slot;
1813   } else {
1814     limit = (dist_slot >> 1) - 1;
1815     s->lzma.rep0 = 2 + (dist_slot & 1);
1816 
1817     if (dist_slot < DIST_MODEL_END) {
1818       s->lzma.rep0 <<= limit;
1819       probs = s->lzma.dist_special + s->lzma.rep0
1820           - dist_slot - 1;
1821       rc_bittree_reverse(&s->rc, probs,
1822           &s->lzma.rep0, limit);
1823     } else {
1824       rc_direct(&s->rc, &s->lzma.rep0, limit - ALIGN_BITS);
1825       s->lzma.rep0 <<= ALIGN_BITS;
1826       rc_bittree_reverse(&s->rc, s->lzma.dist_align,
1827           &s->lzma.rep0, ALIGN_BITS);
1828     }
1829   }
1830 }
1831 
1832 /*
1833  * Decode a repeated match. The distance is one of the four most recently
1834  * seen matches. The distance will be stored in s->lzma.rep0.
1835  */
lzma_rep_match(struct xz_dec_lzma2 * s,uint32_t pos_state)1836 static void lzma_rep_match(struct xz_dec_lzma2 *s, uint32_t pos_state)
1837 {
1838   uint32_t tmp;
1839 
1840   if (!rc_bit(&s->rc, &s->lzma.is_rep0[s->lzma.state])) {
1841     if (!rc_bit(&s->rc, &s->lzma.is_rep0_long[
1842         s->lzma.state][pos_state])) {
1843       lzma_state_short_rep(&s->lzma.state);
1844       s->lzma.len = 1;
1845       return;
1846     }
1847   } else {
1848     if (!rc_bit(&s->rc, &s->lzma.is_rep1[s->lzma.state])) {
1849       tmp = s->lzma.rep1;
1850     } else {
1851       if (!rc_bit(&s->rc, &s->lzma.is_rep2[s->lzma.state])) {
1852         tmp = s->lzma.rep2;
1853       } else {
1854         tmp = s->lzma.rep3;
1855         s->lzma.rep3 = s->lzma.rep2;
1856       }
1857 
1858       s->lzma.rep2 = s->lzma.rep1;
1859     }
1860 
1861     s->lzma.rep1 = s->lzma.rep0;
1862     s->lzma.rep0 = tmp;
1863   }
1864 
1865   lzma_state_long_rep(&s->lzma.state);
1866   lzma_len(s, &s->lzma.rep_len_dec, pos_state);
1867 }
1868 
1869 /* LZMA decoder core */
lzma_main(struct xz_dec_lzma2 * s)1870 static int lzma_main(struct xz_dec_lzma2 *s)
1871 {
1872   uint32_t pos_state;
1873 
1874   /*
1875    * If the dictionary was reached during the previous call, try to
1876    * finish the possibly pending repeat in the dictionary.
1877    */
1878   if (dict_has_space(&s->dict) && s->lzma.len > 0)
1879     dict_repeat(&s->dict, &s->lzma.len, s->lzma.rep0);
1880 
1881   /*
1882    * Decode more LZMA symbols. One iteration may consume up to
1883    * LZMA_IN_REQUIRED - 1 bytes.
1884    */
1885   while (dict_has_space(&s->dict) && !rc_limit_exceeded(&s->rc)) {
1886     pos_state = s->dict.pos & s->lzma.pos_mask;
1887 
1888     if (!rc_bit(&s->rc, &s->lzma.is_match[
1889         s->lzma.state][pos_state])) {
1890       lzma_literal(s);
1891     } else {
1892       if (rc_bit(&s->rc, &s->lzma.is_rep[s->lzma.state]))
1893         lzma_rep_match(s, pos_state);
1894       else
1895         lzma_match(s, pos_state);
1896 
1897       if (!dict_repeat(&s->dict, &s->lzma.len, s->lzma.rep0))
1898         return 0;
1899     }
1900   }
1901 
1902   /*
1903    * Having the range decoder always normalized when we are outside
1904    * this function makes it easier to correctly handle end of the chunk.
1905    */
1906   rc_normalize(&s->rc);
1907 
1908   return 1;
1909 }
1910 
1911 /*
1912  * Reset the LZMA decoder and range decoder state. Dictionary is nore reset
1913  * here, because LZMA state may be reset without resetting the dictionary.
1914  */
lzma_reset(struct xz_dec_lzma2 * s)1915 static void lzma_reset(struct xz_dec_lzma2 *s)
1916 {
1917   uint16_t *probs;
1918   size_t i;
1919 
1920   s->lzma.state = STATE_LIT_LIT;
1921   s->lzma.rep0 = 0;
1922   s->lzma.rep1 = 0;
1923   s->lzma.rep2 = 0;
1924   s->lzma.rep3 = 0;
1925 
1926   /*
1927    * All probabilities are initialized to the same value. This hack
1928    * makes the code smaller by avoiding a separate loop for each
1929    * probability array.
1930    *
1931    * This could be optimized so that only that part of literal
1932    * probabilities that are actually required. In the common case
1933    * we would write 12 KiB less.
1934    */
1935   probs = s->lzma.is_match[0];
1936   for (i = 0; i < PROBS_TOTAL; ++i)
1937     probs[i] = RC_BIT_MODEL_TOTAL / 2;
1938 
1939   rc_reset(&s->rc);
1940 }
1941 
1942 /*
1943  * Decode and validate LZMA properties (lc/lp/pb) and calculate the bit masks
1944  * from the decoded lp and pb values. On success, the LZMA decoder state is
1945  * reset and true is returned.
1946  */
lzma_props(struct xz_dec_lzma2 * s,uint8_t props)1947 static int lzma_props(struct xz_dec_lzma2 *s, uint8_t props)
1948 {
1949   if (props > (4 * 5 + 4) * 9 + 8)
1950     return 0;
1951 
1952   s->lzma.pos_mask = 0;
1953   while (props >= 9 * 5) {
1954     props -= 9 * 5;
1955     ++s->lzma.pos_mask;
1956   }
1957 
1958   s->lzma.pos_mask = (1 << s->lzma.pos_mask) - 1;
1959 
1960   s->lzma.literal_pos_mask = 0;
1961   while (props >= 9) {
1962     props -= 9;
1963     ++s->lzma.literal_pos_mask;
1964   }
1965 
1966   s->lzma.lc = props;
1967 
1968   if (s->lzma.lc + s->lzma.literal_pos_mask > 4)
1969     return 0;
1970 
1971   s->lzma.literal_pos_mask = (1 << s->lzma.literal_pos_mask) - 1;
1972 
1973   lzma_reset(s);
1974 
1975   return 1;
1976 }
1977 
1978 /*********
1979  * LZMA2 *
1980  *********/
1981 
1982 /*
1983  * The LZMA decoder assumes that if the input limit (s->rc.in_limit) hasn't
1984  * been exceeded, it is safe to read up to LZMA_IN_REQUIRED bytes. This
1985  * wrapper function takes care of making the LZMA decoder's assumption safe.
1986  *
1987  * As long as there is plenty of input left to be decoded in the current LZMA
1988  * chunk, we decode directly from the caller-supplied input buffer until
1989  * there's LZMA_IN_REQUIRED bytes left. Those remaining bytes are copied into
1990  * s->temp.buf, which (hopefully) gets filled on the next call to this
1991  * function. We decode a few bytes from the temporary buffer so that we can
1992  * continue decoding from the caller-supplied input buffer again.
1993  */
lzma2_lzma(struct xz_dec_lzma2 * s,struct xz_buf * b)1994 static int lzma2_lzma(struct xz_dec_lzma2 *s, struct xz_buf *b)
1995 {
1996   size_t in_avail;
1997   uint32_t tmp;
1998 
1999   in_avail = b->in_size - b->in_pos;
2000   if (s->temp.size > 0 || s->lzma2.compressed == 0) {
2001     tmp = 2 * LZMA_IN_REQUIRED - s->temp.size;
2002     if (tmp > s->lzma2.compressed - s->temp.size)
2003       tmp = s->lzma2.compressed - s->temp.size;
2004     if (tmp > in_avail)
2005       tmp = in_avail;
2006 
2007     memcpy(s->temp.buf + s->temp.size, b->in + b->in_pos, tmp);
2008 
2009     if (s->temp.size + tmp == s->lzma2.compressed) {
2010       memset(s->temp.buf + s->temp.size + tmp, 0,
2011           sizeof(s->temp.buf)
2012             - s->temp.size - tmp);
2013       s->rc.in_limit = s->temp.size + tmp;
2014     } else if (s->temp.size + tmp < LZMA_IN_REQUIRED) {
2015       s->temp.size += tmp;
2016       b->in_pos += tmp;
2017       return 1;
2018     } else {
2019       s->rc.in_limit = s->temp.size + tmp - LZMA_IN_REQUIRED;
2020     }
2021 
2022     s->rc.in = s->temp.buf;
2023     s->rc.in_pos = 0;
2024 
2025     if (!lzma_main(s) || s->rc.in_pos > s->temp.size + tmp)
2026       return 0;
2027 
2028     s->lzma2.compressed -= s->rc.in_pos;
2029 
2030     if (s->rc.in_pos < s->temp.size) {
2031       s->temp.size -= s->rc.in_pos;
2032       memmove(s->temp.buf, s->temp.buf + s->rc.in_pos,
2033           s->temp.size);
2034       return 1;
2035     }
2036 
2037     b->in_pos += s->rc.in_pos - s->temp.size;
2038     s->temp.size = 0;
2039   }
2040 
2041   in_avail = b->in_size - b->in_pos;
2042   if (in_avail >= LZMA_IN_REQUIRED) {
2043     s->rc.in = b->in;
2044     s->rc.in_pos = b->in_pos;
2045 
2046     if (in_avail >= s->lzma2.compressed + LZMA_IN_REQUIRED)
2047       s->rc.in_limit = b->in_pos + s->lzma2.compressed;
2048     else
2049       s->rc.in_limit = b->in_size - LZMA_IN_REQUIRED;
2050 
2051     if (!lzma_main(s))
2052       return 0;
2053 
2054     in_avail = s->rc.in_pos - b->in_pos;
2055     if (in_avail > s->lzma2.compressed) return 0;
2056 
2057     s->lzma2.compressed -= in_avail;
2058     b->in_pos = s->rc.in_pos;
2059   }
2060 
2061   in_avail = b->in_size - b->in_pos;
2062   if (in_avail < LZMA_IN_REQUIRED) {
2063     if (in_avail > s->lzma2.compressed)
2064       in_avail = s->lzma2.compressed;
2065 
2066     memcpy(s->temp.buf, b->in + b->in_pos, in_avail);
2067     s->temp.size = in_avail;
2068     b->in_pos += in_avail;
2069   }
2070 
2071   return 1;
2072 }
2073 
2074 /*
2075  * Take care of the LZMA2 control layer, and forward the job of actual LZMA
2076  * decoding or copying of uncompressed chunks to other functions.
2077  */
xz_dec_lzma2_run(struct xz_dec_lzma2 * s,struct xz_buf * b)2078 enum xz_ret xz_dec_lzma2_run(struct xz_dec_lzma2 *s,
2079                struct xz_buf *b)
2080 {
2081   uint32_t tmp;
2082 
2083   while (b->in_pos < b->in_size || s->lzma2.sequence == SEQ_LZMA_RUN) {
2084     switch (s->lzma2.sequence) {
2085     case SEQ_CONTROL:
2086       /*
2087        * LZMA2 control byte
2088        *
2089        * Exact values:
2090        *   0x00   End marker
2091        *   0x01   Dictionary reset followed by
2092        *          an uncompressed chunk
2093        *   0x02   Uncompressed chunk (no dictionary reset)
2094        *
2095        * Highest three bits (s->control & 0xE0):
2096        *   0xE0   Dictionary reset, new properties and state
2097        *          reset, followed by LZMA compressed chunk
2098        *   0xC0   New properties and state reset, followed
2099        *          by LZMA compressed chunk (no dictionary
2100        *          reset)
2101        *   0xA0   State reset using old properties,
2102        *          followed by LZMA compressed chunk (no
2103        *          dictionary reset)
2104        *   0x80   LZMA chunk (no dictionary or state reset)
2105        *
2106        * For LZMA compressed chunks, the lowest five bits
2107        * (s->control & 1F) are the highest bits of the
2108        * uncompressed size (bits 16-20).
2109        *
2110        * A new LZMA2 stream must begin with a dictionary
2111        * reset. The first LZMA chunk must set new
2112        * properties and reset the LZMA state.
2113        *
2114        * Values that don't match anything described above
2115        * are invalid and we return XZ_DATA_ERROR.
2116        */
2117       tmp = b->in[b->in_pos++];
2118 
2119       if (tmp == 0x00)
2120         return XZ_STREAM_END;
2121 
2122       if (tmp >= 0xE0 || tmp == 0x01) {
2123         s->lzma2.need_props = 1;
2124         s->lzma2.need_dict_reset = 0;
2125         dict_reset(&s->dict);
2126       } else if (s->lzma2.need_dict_reset) {
2127         return XZ_DATA_ERROR;
2128       }
2129 
2130       if (tmp >= 0x80) {
2131         s->lzma2.uncompressed = (tmp & 0x1F) << 16;
2132         s->lzma2.sequence = SEQ_UNCOMPRESSED_1;
2133 
2134         if (tmp >= 0xC0) {
2135           /*
2136            * When there are new properties,
2137            * state reset is done at
2138            * SEQ_PROPERTIES.
2139            */
2140           s->lzma2.need_props = 0;
2141           s->lzma2.next_sequence
2142               = SEQ_PROPERTIES;
2143 
2144         } else if (s->lzma2.need_props) {
2145           return XZ_DATA_ERROR;
2146 
2147         } else {
2148           s->lzma2.next_sequence
2149               = SEQ_LZMA_PREPARE;
2150           if (tmp >= 0xA0)
2151             lzma_reset(s);
2152         }
2153       } else {
2154         if (tmp > 0x02)
2155           return XZ_DATA_ERROR;
2156 
2157         s->lzma2.sequence = SEQ_COMPRESSED_0;
2158         s->lzma2.next_sequence = SEQ_COPY;
2159       }
2160 
2161       break;
2162 
2163     case SEQ_UNCOMPRESSED_1:
2164       s->lzma2.uncompressed
2165           += (uint32_t)b->in[b->in_pos++] << 8;
2166       s->lzma2.sequence = SEQ_UNCOMPRESSED_2;
2167       break;
2168 
2169     case SEQ_UNCOMPRESSED_2:
2170       s->lzma2.uncompressed
2171           += (uint32_t)b->in[b->in_pos++] + 1;
2172       s->lzma2.sequence = SEQ_COMPRESSED_0;
2173       break;
2174 
2175     case SEQ_COMPRESSED_0:
2176       s->lzma2.compressed
2177           = (uint32_t)b->in[b->in_pos++] << 8;
2178       s->lzma2.sequence = SEQ_COMPRESSED_1;
2179       break;
2180 
2181     case SEQ_COMPRESSED_1:
2182       s->lzma2.compressed
2183           += (uint32_t)b->in[b->in_pos++] + 1;
2184       s->lzma2.sequence = s->lzma2.next_sequence;
2185       break;
2186 
2187     case SEQ_PROPERTIES:
2188       if (!lzma_props(s, b->in[b->in_pos++]))
2189         return XZ_DATA_ERROR;
2190 
2191       s->lzma2.sequence = SEQ_LZMA_PREPARE;
2192 
2193     case SEQ_LZMA_PREPARE:
2194       if (s->lzma2.compressed < RC_INIT_BYTES)
2195         return XZ_DATA_ERROR;
2196 
2197       if (!rc_read_init(&s->rc, b))
2198         return XZ_OK;
2199 
2200       s->lzma2.compressed -= RC_INIT_BYTES;
2201       s->lzma2.sequence = SEQ_LZMA_RUN;
2202 
2203     case SEQ_LZMA_RUN:
2204       /*
2205        * Set dictionary limit to indicate how much we want
2206        * to be encoded at maximum. Decode new data into the
2207        * dictionary. Flush the new data from dictionary to
2208        * b->out. Check if we finished decoding this chunk.
2209        * In case the dictionary got full but we didn't fill
2210        * the output buffer yet, we may run this loop
2211        * multiple times without changing s->lzma2.sequence.
2212        */
2213       dict_limit(&s->dict, min_t(size_t,
2214           b->out_size - b->out_pos,
2215           s->lzma2.uncompressed));
2216       if (!lzma2_lzma(s, b))
2217         return XZ_DATA_ERROR;
2218 
2219       s->lzma2.uncompressed -= dict_flush(&s->dict, b);
2220 
2221       if (s->lzma2.uncompressed == 0) {
2222         if (s->lzma2.compressed > 0 || s->lzma.len > 0
2223             || !rc_is_finished(&s->rc))
2224           return XZ_DATA_ERROR;
2225 
2226         rc_reset(&s->rc);
2227         s->lzma2.sequence = SEQ_CONTROL;
2228 
2229       } else if (b->out_pos == b->out_size
2230           || (b->in_pos == b->in_size
2231             && s->temp.size
2232             < s->lzma2.compressed)) {
2233         return XZ_OK;
2234       }
2235 
2236       break;
2237 
2238     case SEQ_COPY:
2239       dict_uncompressed(&s->dict, b, &s->lzma2.compressed);
2240       if (s->lzma2.compressed > 0)
2241         return XZ_OK;
2242 
2243       s->lzma2.sequence = SEQ_CONTROL;
2244       break;
2245     }
2246   }
2247 
2248   return XZ_OK;
2249 }
2250 
xz_dec_lzma2_create(uint32_t dict_max)2251 struct xz_dec_lzma2 *xz_dec_lzma2_create(uint32_t dict_max)
2252 {
2253   struct xz_dec_lzma2 *s = malloc(sizeof(*s));
2254   if (s == NULL)
2255     return NULL;
2256 
2257   s->dict.size_max = dict_max;
2258   s->dict.buf = NULL;
2259   s->dict.allocated = 0;
2260 
2261   return s;
2262 }
2263 
xz_dec_lzma2_reset(struct xz_dec_lzma2 * s,uint8_t props)2264 enum xz_ret xz_dec_lzma2_reset(struct xz_dec_lzma2 *s, uint8_t props)
2265 {
2266   /* This limits dictionary size to 3 GiB to keep parsing simpler. */
2267   if (props > 39)
2268     return XZ_OPTIONS_ERROR;
2269 
2270   s->dict.size = 2 + (props & 1);
2271   s->dict.size <<= (props >> 1) + 11;
2272 
2273   if (s->dict.size > s->dict.size_max)
2274     return XZ_MEMLIMIT_ERROR;
2275 
2276   s->dict.end = s->dict.size;
2277 
2278   if (s->dict.allocated < s->dict.size) {
2279     free(s->dict.buf);
2280     s->dict.buf = malloc(s->dict.size);
2281     if (s->dict.buf == NULL) {
2282       s->dict.allocated = 0;
2283       return XZ_MEM_ERROR;
2284     }
2285   }
2286 
2287   s->lzma.len = 0;
2288 
2289   s->lzma2.sequence = SEQ_CONTROL;
2290   s->lzma2.need_dict_reset = 1;
2291 
2292   s->temp.size = 0;
2293 
2294   return XZ_OK;
2295 }
2296 
2297 /*
2298  * .xz Stream decoder
2299  */
2300 
2301 
2302 // BEGIN xz_stream.h
2303 /*
2304  * Definitions for handling the .xz file format
2305  */
2306 
2307 /*
2308  * See the .xz file format specification at
2309  * http://tukaani.org/xz/xz-file-format.txt
2310  * to understand the container format.
2311  */
2312 
2313 #define STREAM_HEADER_SIZE 12
2314 
2315 #define HEADER_MAGIC "\3757zXZ"
2316 #define HEADER_MAGIC_SIZE 6
2317 
2318 #define FOOTER_MAGIC "YZ"
2319 #define FOOTER_MAGIC_SIZE 2
2320 
2321 /*
2322  * Variable-length integer can hold a 63-bit unsigned integer or a special
2323  * value indicating that the value is unknown.
2324  *
2325  * Experimental: vli_type can be defined to uint32_t to save a few bytes
2326  * in code size (no effect on speed). Doing so limits the uncompressed and
2327  * compressed size of the file to less than 256 MiB and may also weaken
2328  * error detection slightly.
2329  */
2330 typedef uint64_t vli_type;
2331 
2332 #define VLI_MAX ((vli_type)-1 / 2)
2333 #define VLI_UNKNOWN ((vli_type)-1)
2334 
2335 /* Maximum encoded size of a VLI */
2336 #define VLI_BYTES_MAX (sizeof(vli_type) * 8 / 7)
2337 
2338 /* Integrity Check types */
2339 enum xz_check {
2340   XZ_CHECK_NONE = 0,
2341   XZ_CHECK_CRC32 = 1,
2342   XZ_CHECK_CRC64 = 4,
2343   XZ_CHECK_SHA256 = 10
2344 };
2345 
2346 /* Maximum possible Check ID */
2347 #define XZ_CHECK_MAX 15
2348 // END xz_stream.h
2349 
2350 #define IS_CRC64(check_type) ((check_type) == XZ_CHECK_CRC64)
2351 
2352 /* Hash used to validate the Index field */
2353 struct xz_dec_hash {
2354   vli_type unpadded;
2355   vli_type uncompressed;
2356   uint32_t crc32;
2357 };
2358 
2359 struct xz_dec {
2360   /* Position in dec_main() */
2361   enum {
2362     SEQ_STREAM_HEADER,
2363     SEQ_BLOCK_START,
2364     SEQ_BLOCK_HEADER,
2365     SEQ_BLOCK_UNCOMPRESS,
2366     SEQ_BLOCK_PADDING,
2367     SEQ_BLOCK_CHECK,
2368     SEQ_INDEX,
2369     SEQ_INDEX_PADDING,
2370     SEQ_INDEX_CRC32,
2371     SEQ_STREAM_FOOTER
2372   } sequence;
2373 
2374   /* Position in variable-length integers and Check fields */
2375   uint32_t pos;
2376 
2377   /* Variable-length integer decoded by dec_vli() */
2378   vli_type vli;
2379 
2380   /* Saved in_pos and out_pos */
2381   size_t in_start;
2382   size_t out_start;
2383 
2384   /* CRC32 or CRC64 value in Block or CRC32 value in Index */
2385   uint64_t crc;
2386 
2387   /* Type of the integrity check calculated from uncompressed data */
2388   enum xz_check check_type;
2389 
2390   /*
2391    * True if the next call to xz_dec_run() is allowed to return
2392    * XZ_BUF_ERROR.
2393    */
2394   int allow_buf_error;
2395 
2396   /* Information stored in Block Header */
2397   struct {
2398     /*
2399      * Value stored in the Compressed Size field, or
2400      * VLI_UNKNOWN if Compressed Size is not present.
2401      */
2402     vli_type compressed;
2403 
2404     /*
2405      * Value stored in the Uncompressed Size field, or
2406      * VLI_UNKNOWN if Uncompressed Size is not present.
2407      */
2408     vli_type uncompressed;
2409 
2410     /* Size of the Block Header field */
2411     uint32_t size;
2412   } block_header;
2413 
2414   /* Information collected when decoding Blocks */
2415   struct {
2416     /* Observed compressed size of the current Block */
2417     vli_type compressed;
2418 
2419     /* Observed uncompressed size of the current Block */
2420     vli_type uncompressed;
2421 
2422     /* Number of Blocks decoded so far */
2423     vli_type count;
2424 
2425     /*
2426      * Hash calculated from the Block sizes. This is used to
2427      * validate the Index field.
2428      */
2429     struct xz_dec_hash hash;
2430   } block;
2431 
2432   /* Variables needed when verifying the Index field */
2433   struct {
2434     /* Position in dec_index() */
2435     enum {
2436       SEQ_INDEX_COUNT,
2437       SEQ_INDEX_UNPADDED,
2438       SEQ_INDEX_UNCOMPRESSED
2439     } sequence;
2440 
2441     /* Size of the Index in bytes */
2442     vli_type size;
2443 
2444     /* Number of Records (matches block.count in valid files) */
2445     vli_type count;
2446 
2447     /*
2448      * Hash calculated from the Records (matches block.hash in
2449      * valid files).
2450      */
2451     struct xz_dec_hash hash;
2452   } index;
2453 
2454   /*
2455    * Temporary buffer needed to hold Stream Header, Block Header,
2456    * and Stream Footer. The Block Header is the biggest (1 KiB)
2457    * so we reserve space according to that. buf[] has to be aligned
2458    * to a multiple of four bytes; the size_t variables before it
2459    * should guarantee this.
2460    */
2461   struct {
2462     size_t pos;
2463     size_t size;
2464     uint8_t buf[1024];
2465   } temp;
2466 
2467   struct xz_dec_lzma2 *lzma2;
2468 
2469 #ifdef XZ_DEC_BCJ
2470   struct xz_dec_bcj *bcj;
2471   int bcj_active;
2472 #endif
2473 };
2474 
2475 /* Sizes of the Check field with different Check IDs */
2476 static const uint8_t check_sizes[16] = {
2477   0,
2478   4, 4, 4,
2479   8, 8, 8,
2480   16, 16, 16,
2481   32, 32, 32,
2482   64, 64, 64
2483 };
2484 
2485 /*
2486  * Fill s->temp by copying data starting from b->in[b->in_pos]. Caller
2487  * must have set s->temp.pos to indicate how much data we are supposed
2488  * to copy into s->temp.buf. Return true once s->temp.pos has reached
2489  * s->temp.size.
2490  */
fill_temp(struct xz_dec * s,struct xz_buf * b)2491 static int fill_temp(struct xz_dec *s, struct xz_buf *b)
2492 {
2493   size_t copy_size = min_t(size_t,
2494       b->in_size - b->in_pos, s->temp.size - s->temp.pos);
2495 
2496   memcpy(s->temp.buf + s->temp.pos, b->in + b->in_pos, copy_size);
2497   b->in_pos += copy_size;
2498   s->temp.pos += copy_size;
2499 
2500   if (s->temp.pos == s->temp.size) {
2501     s->temp.pos = 0;
2502     return 1;
2503   }
2504 
2505   return 0;
2506 }
2507 
2508 /* Decode a variable-length integer (little-endian base-128 encoding) */
dec_vli(struct xz_dec * s,const uint8_t * in,size_t * in_pos,size_t in_size)2509 static enum xz_ret dec_vli(struct xz_dec *s, const uint8_t *in,
2510          size_t *in_pos, size_t in_size)
2511 {
2512   uint8_t byte;
2513 
2514   if (s->pos == 0)
2515     s->vli = 0;
2516 
2517   while (*in_pos < in_size) {
2518     byte = in[*in_pos];
2519     ++*in_pos;
2520 
2521     s->vli |= (vli_type)(byte & 0x7F) << s->pos;
2522 
2523     if ((byte & 0x80) == 0) {
2524       /* Don't allow non-minimal encodings. */
2525       if (byte == 0 && s->pos != 0)
2526         return XZ_DATA_ERROR;
2527 
2528       s->pos = 0;
2529       return XZ_STREAM_END;
2530     }
2531 
2532     s->pos += 7;
2533     if (s->pos == 7 * VLI_BYTES_MAX)
2534       return XZ_DATA_ERROR;
2535   }
2536 
2537   return XZ_OK;
2538 }
2539 
2540 /*
2541  * Decode the Compressed Data field from a Block. Update and validate
2542  * the observed compressed and uncompressed sizes of the Block so that
2543  * they don't exceed the values possibly stored in the Block Header
2544  * (validation assumes that no integer overflow occurs, since vli_type
2545  * is normally uint64_t). Update the CRC32 or CRC64 value if presence of
2546  * the CRC32 or CRC64 field was indicated in Stream Header.
2547  *
2548  * Once the decoding is finished, validate that the observed sizes match
2549  * the sizes possibly stored in the Block Header. Update the hash and
2550  * Block count, which are later used to validate the Index field.
2551  */
dec_block(struct xz_dec * s,struct xz_buf * b)2552 static enum xz_ret dec_block(struct xz_dec *s, struct xz_buf *b)
2553 {
2554   enum xz_ret ret;
2555 
2556   s->in_start = b->in_pos;
2557   s->out_start = b->out_pos;
2558 
2559 #ifdef XZ_DEC_BCJ
2560   if (s->bcj_active)
2561     ret = xz_dec_bcj_run(s->bcj, s->lzma2, b);
2562   else
2563 #endif
2564     ret = xz_dec_lzma2_run(s->lzma2, b);
2565 
2566   s->block.compressed += b->in_pos - s->in_start;
2567   s->block.uncompressed += b->out_pos - s->out_start;
2568 
2569   /*
2570    * There is no need to separately check for VLI_UNKNOWN, since
2571    * the observed sizes are always smaller than VLI_UNKNOWN.
2572    */
2573   if (s->block.compressed > s->block_header.compressed
2574       || s->block.uncompressed
2575         > s->block_header.uncompressed)
2576     return XZ_DATA_ERROR;
2577 
2578   if (s->check_type == XZ_CHECK_CRC32)
2579     s->crc = xz_crc32(b->out + s->out_start,
2580         b->out_pos - s->out_start, s->crc);
2581   else if (s->check_type == XZ_CHECK_CRC64)
2582     s->crc = ~(s->crc);
2583     size_t size = b->out_pos - s->out_start;
2584     uint8_t *buf = b->out + s->out_start;
2585     while (size) {
2586       s->crc = xz_crc64_table[*buf++ ^ (s->crc & 0xFF)] ^ (s->crc >> 8);
2587       --size;
2588     }
2589     s->crc=~(s->crc);
2590 
2591   if (ret == XZ_STREAM_END) {
2592     if (s->block_header.compressed != VLI_UNKNOWN
2593         && s->block_header.compressed
2594           != s->block.compressed)
2595       return XZ_DATA_ERROR;
2596 
2597     if (s->block_header.uncompressed != VLI_UNKNOWN
2598         && s->block_header.uncompressed
2599           != s->block.uncompressed)
2600       return XZ_DATA_ERROR;
2601 
2602     s->block.hash.unpadded += s->block_header.size
2603         + s->block.compressed;
2604 
2605     s->block.hash.unpadded += check_sizes[s->check_type];
2606 
2607     s->block.hash.uncompressed += s->block.uncompressed;
2608     s->block.hash.crc32 = xz_crc32(
2609         (const uint8_t *)&s->block.hash,
2610         sizeof(s->block.hash), s->block.hash.crc32);
2611 
2612     ++s->block.count;
2613   }
2614 
2615   return ret;
2616 }
2617 
2618 /* Update the Index size and the CRC32 value. */
index_update(struct xz_dec * s,const struct xz_buf * b)2619 static void index_update(struct xz_dec *s, const struct xz_buf *b)
2620 {
2621   size_t in_used = b->in_pos - s->in_start;
2622   s->index.size += in_used;
2623   s->crc = xz_crc32(b->in + s->in_start, in_used, s->crc);
2624 }
2625 
2626 /*
2627  * Decode the Number of Records, Unpadded Size, and Uncompressed Size
2628  * fields from the Index field. That is, Index Padding and CRC32 are not
2629  * decoded by this function.
2630  *
2631  * This can return XZ_OK (more input needed), XZ_STREAM_END (everything
2632  * successfully decoded), or XZ_DATA_ERROR (input is corrupt).
2633  */
dec_index(struct xz_dec * s,struct xz_buf * b)2634 static enum xz_ret dec_index(struct xz_dec *s, struct xz_buf *b)
2635 {
2636   enum xz_ret ret;
2637 
2638   do {
2639     ret = dec_vli(s, b->in, &b->in_pos, b->in_size);
2640     if (ret != XZ_STREAM_END) {
2641       index_update(s, b);
2642       return ret;
2643     }
2644 
2645     switch (s->index.sequence) {
2646     case SEQ_INDEX_COUNT:
2647       s->index.count = s->vli;
2648 
2649       /*
2650        * Validate that the Number of Records field
2651        * indicates the same number of Records as
2652        * there were Blocks in the Stream.
2653        */
2654       if (s->index.count != s->block.count)
2655         return XZ_DATA_ERROR;
2656 
2657       s->index.sequence = SEQ_INDEX_UNPADDED;
2658       break;
2659 
2660     case SEQ_INDEX_UNPADDED:
2661       s->index.hash.unpadded += s->vli;
2662       s->index.sequence = SEQ_INDEX_UNCOMPRESSED;
2663       break;
2664 
2665     case SEQ_INDEX_UNCOMPRESSED:
2666       s->index.hash.uncompressed += s->vli;
2667       s->index.hash.crc32 = xz_crc32(
2668           (const uint8_t *)&s->index.hash,
2669           sizeof(s->index.hash),
2670           s->index.hash.crc32);
2671       --s->index.count;
2672       s->index.sequence = SEQ_INDEX_UNPADDED;
2673       break;
2674     }
2675   } while (s->index.count > 0);
2676 
2677   return XZ_STREAM_END;
2678 }
2679 
2680 /*
2681  * Validate that the next four or eight input bytes match the value
2682  * of s->crc. s->pos must be zero when starting to validate the first byte.
2683  * The "bits" argument allows using the same code for both CRC32 and CRC64.
2684  */
crc_validate(struct xz_dec * s,struct xz_buf * b,uint32_t bits)2685 static enum xz_ret crc_validate(struct xz_dec *s, struct xz_buf *b,
2686         uint32_t bits)
2687 {
2688   do {
2689     if (b->in_pos == b->in_size)
2690       return XZ_OK;
2691 
2692     if (((s->crc >> s->pos) & 0xFF) != b->in[b->in_pos++])
2693       return XZ_DATA_ERROR;
2694 
2695     s->pos += 8;
2696 
2697   } while (s->pos < bits);
2698 
2699   s->crc = 0;
2700   s->pos = 0;
2701 
2702   return XZ_STREAM_END;
2703 }
2704 
2705 /*
2706  * Skip over the Check field when the Check ID is not supported.
2707  * Returns true once the whole Check field has been skipped over.
2708  */
check_skip(struct xz_dec * s,struct xz_buf * b)2709 static int check_skip(struct xz_dec *s, struct xz_buf *b)
2710 {
2711   while (s->pos < check_sizes[s->check_type]) {
2712     if (b->in_pos == b->in_size) return 0;
2713 
2714     ++b->in_pos;
2715     ++s->pos;
2716   }
2717 
2718   s->pos = 0;
2719 
2720   return 1;
2721 }
2722 
2723 /* Decode the Stream Header field (the first 12 bytes of the .xz Stream). */
dec_stream_header(struct xz_dec * s)2724 static enum xz_ret dec_stream_header(struct xz_dec *s)
2725 {
2726   if (!memeq(s->temp.buf, HEADER_MAGIC, HEADER_MAGIC_SIZE))
2727     return XZ_FORMAT_ERROR;
2728 
2729   if (xz_crc32(s->temp.buf + HEADER_MAGIC_SIZE, 2, 0)
2730       != get_le32(s->temp.buf + HEADER_MAGIC_SIZE + 2))
2731     return XZ_DATA_ERROR;
2732 
2733   if (s->temp.buf[HEADER_MAGIC_SIZE] != 0)
2734     return XZ_OPTIONS_ERROR;
2735 
2736   /*
2737    * Of integrity checks, we support none (Check ID = 0),
2738    * CRC32 (Check ID = 1), and optionally CRC64 (Check ID = 4).
2739    * However, if XZ_DEC_ANY_CHECK is defined, we will accept other
2740    * check types too, but then the check won't be verified and
2741    * a warning (XZ_UNSUPPORTED_CHECK) will be given.
2742    */
2743   s->check_type = s->temp.buf[HEADER_MAGIC_SIZE + 1];
2744 
2745   if (s->check_type > XZ_CHECK_MAX)
2746     return XZ_OPTIONS_ERROR;
2747 
2748   if (s->check_type > XZ_CHECK_CRC32 && !IS_CRC64(s->check_type))
2749     return XZ_UNSUPPORTED_CHECK;
2750 
2751   return XZ_OK;
2752 }
2753 
2754 /* Decode the Stream Footer field (the last 12 bytes of the .xz Stream) */
dec_stream_footer(struct xz_dec * s)2755 static enum xz_ret dec_stream_footer(struct xz_dec *s)
2756 {
2757   if (!memeq(s->temp.buf + 10, FOOTER_MAGIC, FOOTER_MAGIC_SIZE))
2758     return XZ_DATA_ERROR;
2759 
2760   if (xz_crc32(s->temp.buf + 4, 6, 0) != get_le32(s->temp.buf))
2761     return XZ_DATA_ERROR;
2762 
2763   /*
2764    * Validate Backward Size. Note that we never added the size of the
2765    * Index CRC32 field to s->index.size, thus we use s->index.size / 4
2766    * instead of s->index.size / 4 - 1.
2767    */
2768   if ((s->index.size >> 2) != get_le32(s->temp.buf + 4))
2769     return XZ_DATA_ERROR;
2770 
2771   if (s->temp.buf[8] != 0 || s->temp.buf[9] != s->check_type)
2772     return XZ_DATA_ERROR;
2773 
2774   /*
2775    * Use XZ_STREAM_END instead of XZ_OK to be more convenient
2776    * for the caller.
2777    */
2778   return XZ_STREAM_END;
2779 }
2780 
2781 /* Decode the Block Header and initialize the filter chain. */
dec_block_header(struct xz_dec * s)2782 static enum xz_ret dec_block_header(struct xz_dec *s)
2783 {
2784   enum xz_ret ret;
2785 
2786   /*
2787    * Validate the CRC32. We know that the temp buffer is at least
2788    * eight bytes so this is safe.
2789    */
2790   s->temp.size -= 4;
2791   if (xz_crc32(s->temp.buf, s->temp.size, 0)
2792       != get_le32(s->temp.buf + s->temp.size))
2793     return XZ_DATA_ERROR;
2794 
2795   s->temp.pos = 2;
2796 
2797   /*
2798    * Catch unsupported Block Flags. We support only one or two filters
2799    * in the chain, so we catch that with the same test.
2800    */
2801 #ifdef XZ_DEC_BCJ
2802   if (s->temp.buf[1] & 0x3E)
2803 #else
2804   if (s->temp.buf[1] & 0x3F)
2805 #endif
2806     return XZ_OPTIONS_ERROR;
2807 
2808   /* Compressed Size */
2809   if (s->temp.buf[1] & 0x40) {
2810     if (dec_vli(s, s->temp.buf, &s->temp.pos, s->temp.size)
2811           != XZ_STREAM_END)
2812       return XZ_DATA_ERROR;
2813 
2814     s->block_header.compressed = s->vli;
2815   } else {
2816     s->block_header.compressed = VLI_UNKNOWN;
2817   }
2818 
2819   /* Uncompressed Size */
2820   if (s->temp.buf[1] & 0x80) {
2821     if (dec_vli(s, s->temp.buf, &s->temp.pos, s->temp.size)
2822         != XZ_STREAM_END)
2823       return XZ_DATA_ERROR;
2824 
2825     s->block_header.uncompressed = s->vli;
2826   } else {
2827     s->block_header.uncompressed = VLI_UNKNOWN;
2828   }
2829 
2830 #ifdef XZ_DEC_BCJ
2831   /* If there are two filters, the first one must be a BCJ filter. */
2832   s->bcj_active = s->temp.buf[1] & 0x01;
2833   if (s->bcj_active) {
2834     if (s->temp.size - s->temp.pos < 2)
2835       return XZ_OPTIONS_ERROR;
2836 
2837     ret = xz_dec_bcj_reset(s->bcj, s->temp.buf[s->temp.pos++]);
2838     if (ret != XZ_OK)
2839       return ret;
2840 
2841     /*
2842      * We don't support custom start offset,
2843      * so Size of Properties must be zero.
2844      */
2845     if (s->temp.buf[s->temp.pos++] != 0x00)
2846       return XZ_OPTIONS_ERROR;
2847   }
2848 #endif
2849 
2850   /* Valid Filter Flags always take at least two bytes. */
2851   if (s->temp.size - s->temp.pos < 2)
2852     return XZ_DATA_ERROR;
2853 
2854   /* Filter ID = LZMA2 */
2855   if (s->temp.buf[s->temp.pos++] != 0x21)
2856     return XZ_OPTIONS_ERROR;
2857 
2858   /* Size of Properties = 1-byte Filter Properties */
2859   if (s->temp.buf[s->temp.pos++] != 0x01)
2860     return XZ_OPTIONS_ERROR;
2861 
2862   /* Filter Properties contains LZMA2 dictionary size. */
2863   if (s->temp.size - s->temp.pos < 1)
2864     return XZ_DATA_ERROR;
2865 
2866   ret = xz_dec_lzma2_reset(s->lzma2, s->temp.buf[s->temp.pos++]);
2867   if (ret != XZ_OK)
2868     return ret;
2869 
2870   /* The rest must be Header Padding. */
2871   while (s->temp.pos < s->temp.size)
2872     if (s->temp.buf[s->temp.pos++] != 0x00)
2873       return XZ_OPTIONS_ERROR;
2874 
2875   s->temp.pos = 0;
2876   s->block.compressed = 0;
2877   s->block.uncompressed = 0;
2878 
2879   return XZ_OK;
2880 }
2881 
dec_main(struct xz_dec * s,struct xz_buf * b)2882 static enum xz_ret dec_main(struct xz_dec *s, struct xz_buf *b)
2883 {
2884   enum xz_ret ret;
2885 
2886   /*
2887    * Store the start position for the case when we are in the middle
2888    * of the Index field.
2889    */
2890   s->in_start = b->in_pos;
2891 
2892   for (;;) {
2893     switch (s->sequence) {
2894     case SEQ_STREAM_HEADER:
2895       /*
2896        * Stream Header is copied to s->temp, and then
2897        * decoded from there. This way if the caller
2898        * gives us only little input at a time, we can
2899        * still keep the Stream Header decoding code
2900        * simple. Similar approach is used in many places
2901        * in this file.
2902        */
2903       if (!fill_temp(s, b))
2904         return XZ_OK;
2905 
2906       /*
2907        * If dec_stream_header() returns
2908        * XZ_UNSUPPORTED_CHECK, it is still possible
2909        * to continue decoding if working in multi-call
2910        * mode. Thus, update s->sequence before calling
2911        * dec_stream_header().
2912        */
2913       s->sequence = SEQ_BLOCK_START;
2914 
2915       ret = dec_stream_header(s);
2916       if (ret != XZ_OK)
2917         return ret;
2918 
2919     case SEQ_BLOCK_START:
2920       /* We need one byte of input to continue. */
2921       if (b->in_pos == b->in_size)
2922         return XZ_OK;
2923 
2924       /* See if this is the beginning of the Index field. */
2925       if (b->in[b->in_pos] == 0) {
2926         s->in_start = b->in_pos++;
2927         s->sequence = SEQ_INDEX;
2928         break;
2929       }
2930 
2931       /*
2932        * Calculate the size of the Block Header and
2933        * prepare to decode it.
2934        */
2935       s->block_header.size
2936         = ((uint32_t)b->in[b->in_pos] + 1) * 4;
2937 
2938       s->temp.size = s->block_header.size;
2939       s->temp.pos = 0;
2940       s->sequence = SEQ_BLOCK_HEADER;
2941 
2942     case SEQ_BLOCK_HEADER:
2943       if (!fill_temp(s, b))
2944         return XZ_OK;
2945 
2946       ret = dec_block_header(s);
2947       if (ret != XZ_OK)
2948         return ret;
2949 
2950       s->sequence = SEQ_BLOCK_UNCOMPRESS;
2951 
2952     case SEQ_BLOCK_UNCOMPRESS:
2953       ret = dec_block(s, b);
2954       if (ret != XZ_STREAM_END)
2955         return ret;
2956 
2957       s->sequence = SEQ_BLOCK_PADDING;
2958 
2959     case SEQ_BLOCK_PADDING:
2960       /*
2961        * Size of Compressed Data + Block Padding
2962        * must be a multiple of four. We don't need
2963        * s->block.compressed for anything else
2964        * anymore, so we use it here to test the size
2965        * of the Block Padding field.
2966        */
2967       while (s->block.compressed & 3) {
2968         if (b->in_pos == b->in_size)
2969           return XZ_OK;
2970 
2971         if (b->in[b->in_pos++] != 0)
2972           return XZ_DATA_ERROR;
2973 
2974         ++s->block.compressed;
2975       }
2976 
2977       s->sequence = SEQ_BLOCK_CHECK;
2978 
2979     case SEQ_BLOCK_CHECK:
2980       if (s->check_type == XZ_CHECK_CRC32) {
2981         ret = crc_validate(s, b, 32);
2982         if (ret != XZ_STREAM_END)
2983           return ret;
2984       }
2985       else if (IS_CRC64(s->check_type)) {
2986         ret = crc_validate(s, b, 64);
2987         if (ret != XZ_STREAM_END)
2988           return ret;
2989       }
2990       else if (!check_skip(s, b)) {
2991         return XZ_OK;
2992       }
2993 
2994       s->sequence = SEQ_BLOCK_START;
2995       break;
2996 
2997     case SEQ_INDEX:
2998       ret = dec_index(s, b);
2999       if (ret != XZ_STREAM_END)
3000         return ret;
3001 
3002       s->sequence = SEQ_INDEX_PADDING;
3003 
3004     case SEQ_INDEX_PADDING:
3005       while ((s->index.size + (b->in_pos - s->in_start))
3006           & 3) {
3007         if (b->in_pos == b->in_size) {
3008           index_update(s, b);
3009           return XZ_OK;
3010         }
3011 
3012         if (b->in[b->in_pos++] != 0)
3013           return XZ_DATA_ERROR;
3014       }
3015 
3016       /* Finish the CRC32 value and Index size. */
3017       index_update(s, b);
3018 
3019       /* Compare the hashes to validate the Index field. */
3020       if (!memeq(&s->block.hash, &s->index.hash,
3021           sizeof(s->block.hash)))
3022         return XZ_DATA_ERROR;
3023 
3024       s->sequence = SEQ_INDEX_CRC32;
3025 
3026     case SEQ_INDEX_CRC32:
3027       ret = crc_validate(s, b, 32);
3028       if (ret != XZ_STREAM_END)
3029         return ret;
3030 
3031       s->temp.size = STREAM_HEADER_SIZE;
3032       s->sequence = SEQ_STREAM_FOOTER;
3033 
3034     case SEQ_STREAM_FOOTER:
3035       if (!fill_temp(s, b))
3036         return XZ_OK;
3037 
3038       return dec_stream_footer(s);
3039     }
3040   }
3041 
3042   /* Never reached */
3043 }
3044 
3045 /*
3046  * xz_dec_run() is a wrapper for dec_main() to handle some special cases in
3047  * multi-call and single-call decoding.
3048  *
3049  * In multi-call mode, we must return XZ_BUF_ERROR when it seems clear that we
3050  * are not going to make any progress anymore. This is to prevent the caller
3051  * from calling us infinitely when the input file is truncated or otherwise
3052  * corrupt. Since zlib-style API allows that the caller fills the input buffer
3053  * only when the decoder doesn't produce any new output, we have to be careful
3054  * to avoid returning XZ_BUF_ERROR too easily: XZ_BUF_ERROR is returned only
3055  * after the second consecutive call to xz_dec_run() that makes no progress.
3056  *
3057  * In single-call mode, if we couldn't decode everything and no error
3058  * occurred, either the input is truncated or the output buffer is too small.
3059  * Since we know that the last input byte never produces any output, we know
3060  * that if all the input was consumed and decoding wasn't finished, the file
3061  * must be corrupt. Otherwise the output buffer has to be too small or the
3062  * file is corrupt in a way that decoding it produces too big output.
3063  *
3064  * If single-call decoding fails, we reset b->in_pos and b->out_pos back to
3065  * their original values. This is because with some filter chains there won't
3066  * be any valid uncompressed data in the output buffer unless the decoding
3067  * actually succeeds (that's the price to pay of using the output buffer as
3068  * the workspace).
3069  */
xz_dec_run(struct xz_dec * s,struct xz_buf * b)3070 enum xz_ret xz_dec_run(struct xz_dec *s, struct xz_buf *b)
3071 {
3072   size_t in_start;
3073   size_t out_start;
3074   enum xz_ret ret;
3075 
3076   in_start = b->in_pos;
3077   out_start = b->out_pos;
3078   ret = dec_main(s, b);
3079 
3080   if (ret == XZ_OK && in_start == b->in_pos && out_start == b->out_pos) {
3081     if (s->allow_buf_error)
3082       ret = XZ_BUF_ERROR;
3083 
3084     s->allow_buf_error = 1;
3085   } else {
3086     s->allow_buf_error = 0;
3087   }
3088 
3089   return ret;
3090 }
3091 
xz_dec_init(uint32_t dict_max)3092 struct xz_dec *xz_dec_init(uint32_t dict_max)
3093 {
3094   struct xz_dec *s = malloc(sizeof(*s));
3095   if (!s)
3096     return NULL;
3097 
3098 #ifdef XZ_DEC_BCJ
3099   s->bcj = malloc(sizeof(*s->bcj));
3100   if (!s->bcj)
3101     goto error_bcj;
3102 #endif
3103 
3104   s->lzma2 = xz_dec_lzma2_create(dict_max);
3105   if (s->lzma2 == NULL)
3106     goto error_lzma2;
3107 
3108   xz_dec_reset(s);
3109   return s;
3110 
3111 error_lzma2:
3112 #ifdef XZ_DEC_BCJ
3113   free(s->bcj);
3114 error_bcj:
3115 #endif
3116   free(s);
3117   return NULL;
3118 }
3119 
xz_dec_reset(struct xz_dec * s)3120 void xz_dec_reset(struct xz_dec *s)
3121 {
3122   s->sequence = SEQ_STREAM_HEADER;
3123   s->allow_buf_error = 0;
3124   s->pos = 0;
3125   s->crc = 0;
3126   memset(&s->block, 0, sizeof(s->block));
3127   memset(&s->index, 0, sizeof(s->index));
3128   s->temp.pos = 0;
3129   s->temp.size = STREAM_HEADER_SIZE;
3130 }
3131 
xz_dec_end(struct xz_dec * s)3132 void xz_dec_end(struct xz_dec *s)
3133 {
3134   if (s != NULL) {
3135     free((s->lzma2)->dict.buf);
3136     free(s->lzma2);
3137 
3138 #ifdef XZ_DEC_BCJ
3139     free(s->bcj);
3140 #endif
3141     free(s);
3142   }
3143 }
3144