1 /* 2 * XZ decompressor 3 * 4 * Authors: Lasse Collin <lasse.collin@tukaani.org> 5 * Igor Pavlov <https://7-zip.org/> 6 * 7 * This file has been put into the public domain. 8 * You can do whatever you want with this file. 9 */ 10 11 #ifndef XZ_H 12 #define XZ_H 13 14 #ifdef __KERNEL__ 15 # include <linux/stddef.h> 16 # include <linux/types.h> 17 #else 18 #ifdef __FreeBSD__ 19 # include <sys/stddef.h> 20 # include <sys/types.h> 21 #else 22 # include <stddef.h> 23 # include <stdint.h> 24 #endif 25 #endif 26 27 #ifdef __cplusplus 28 extern "C" { 29 #endif 30 31 /* In Linux, this is used to make extern functions static when needed. */ 32 #ifndef XZ_EXTERN 33 # define XZ_EXTERN extern 34 #endif 35 36 /** 37 * enum xz_mode - Operation mode 38 * 39 * @XZ_SINGLE: Single-call mode. This uses less RAM than 40 * multi-call modes, because the LZMA2 41 * dictionary doesn't need to be allocated as 42 * part of the decoder state. All required data 43 * structures are allocated at initialization, 44 * so xz_dec_run() cannot return XZ_MEM_ERROR. 45 * @XZ_PREALLOC: Multi-call mode with preallocated LZMA2 46 * dictionary buffer. All data structures are 47 * allocated at initialization, so xz_dec_run() 48 * cannot return XZ_MEM_ERROR. 49 * @XZ_DYNALLOC: Multi-call mode. The LZMA2 dictionary is 50 * allocated once the required size has been 51 * parsed from the stream headers. If the 52 * allocation fails, xz_dec_run() will return 53 * XZ_MEM_ERROR. 54 * 55 * It is possible to enable support only for a subset of the above 56 * modes at compile time by defining XZ_DEC_SINGLE, XZ_DEC_PREALLOC, 57 * or XZ_DEC_DYNALLOC. The xz_dec kernel module is always compiled 58 * with support for all operation modes, but the preboot code may 59 * be built with fewer features to minimize code size. 60 */ 61 enum xz_mode { 62 XZ_SINGLE, 63 XZ_PREALLOC, 64 XZ_DYNALLOC 65 }; 66 67 /** 68 * enum xz_ret - Return codes 69 * @XZ_OK: Everything is OK so far. More input or more 70 * output space is required to continue. This 71 * return code is possible only in multi-call mode 72 * (XZ_PREALLOC or XZ_DYNALLOC). 73 * @XZ_STREAM_END: Operation finished successfully. 74 * @XZ_UNSUPPORTED_CHECK: Integrity check type is not supported. Decoding 75 * is still possible in multi-call mode by simply 76 * calling xz_dec_run() again. 77 * Note that this return value is used only if 78 * XZ_DEC_ANY_CHECK was defined at build time, 79 * which is not used in the kernel. Unsupported 80 * check types return XZ_OPTIONS_ERROR if 81 * XZ_DEC_ANY_CHECK was not defined at build time. 82 * @XZ_MEM_ERROR: Allocating memory failed. This return code is 83 * possible only if the decoder was initialized 84 * with XZ_DYNALLOC. The amount of memory that was 85 * tried to be allocated was no more than the 86 * dict_max argument given to xz_dec_init(). 87 * @XZ_MEMLIMIT_ERROR: A bigger LZMA2 dictionary would be needed than 88 * allowed by the dict_max argument given to 89 * xz_dec_init(). This return value is possible 90 * only in multi-call mode (XZ_PREALLOC or 91 * XZ_DYNALLOC); the single-call mode (XZ_SINGLE) 92 * ignores the dict_max argument. 93 * @XZ_FORMAT_ERROR: File format was not recognized (wrong magic 94 * bytes). 95 * @XZ_OPTIONS_ERROR: This implementation doesn't support the requested 96 * compression options. In the decoder this means 97 * that the header CRC32 matches, but the header 98 * itself specifies something that we don't support. 99 * @XZ_DATA_ERROR: Compressed data is corrupt. 100 * @XZ_BUF_ERROR: Cannot make any progress. Details are slightly 101 * different between multi-call and single-call 102 * mode; more information below. 103 * 104 * In multi-call mode, XZ_BUF_ERROR is returned when two consecutive calls 105 * to XZ code cannot consume any input and cannot produce any new output. 106 * This happens when there is no new input available, or the output buffer 107 * is full while at least one output byte is still pending. Assuming your 108 * code is not buggy, you can get this error only when decoding a compressed 109 * stream that is truncated or otherwise corrupt. 110 * 111 * In single-call mode, XZ_BUF_ERROR is returned only when the output buffer 112 * is too small or the compressed input is corrupt in a way that makes the 113 * decoder produce more output than the caller expected. When it is 114 * (relatively) clear that the compressed input is truncated, XZ_DATA_ERROR 115 * is used instead of XZ_BUF_ERROR. 116 */ 117 enum xz_ret { 118 XZ_OK, 119 XZ_STREAM_END, 120 XZ_UNSUPPORTED_CHECK, 121 XZ_MEM_ERROR, 122 XZ_MEMLIMIT_ERROR, 123 XZ_FORMAT_ERROR, 124 XZ_OPTIONS_ERROR, 125 XZ_DATA_ERROR, 126 XZ_BUF_ERROR 127 }; 128 129 /** 130 * struct xz_buf - Passing input and output buffers to XZ code 131 * @in: Beginning of the input buffer. This may be NULL if and only 132 * if in_pos is equal to in_size. 133 * @in_pos: Current position in the input buffer. This must not exceed 134 * in_size. 135 * @in_size: Size of the input buffer 136 * @out: Beginning of the output buffer. This may be NULL if and only 137 * if out_pos is equal to out_size. 138 * @out_pos: Current position in the output buffer. This must not exceed 139 * out_size. 140 * @out_size: Size of the output buffer 141 * 142 * Only the contents of the output buffer from out[out_pos] onward, and 143 * the variables in_pos and out_pos are modified by the XZ code. 144 */ 145 struct xz_buf { 146 const uint8_t *in; 147 size_t in_pos; 148 size_t in_size; 149 150 uint8_t *out; 151 size_t out_pos; 152 size_t out_size; 153 }; 154 155 /** 156 * struct xz_dec - Opaque type to hold the XZ decoder state 157 */ 158 struct xz_dec; 159 160 /** 161 * xz_dec_init() - Allocate and initialize a XZ decoder state 162 * @mode: Operation mode 163 * @dict_max: Maximum size of the LZMA2 dictionary (history buffer) for 164 * multi-call decoding. This is ignored in single-call mode 165 * (mode == XZ_SINGLE). LZMA2 dictionary is always 2^n bytes 166 * or 2^n + 2^(n-1) bytes (the latter sizes are less common 167 * in practice), so other values for dict_max don't make sense. 168 * In the kernel, dictionary sizes of 64 KiB, 128 KiB, 256 KiB, 169 * 512 KiB, and 1 MiB are probably the only reasonable values, 170 * except for kernel and initramfs images where a bigger 171 * dictionary can be fine and useful. 172 * 173 * Single-call mode (XZ_SINGLE): xz_dec_run() decodes the whole stream at 174 * once. The caller must provide enough output space or the decoding will 175 * fail. The output space is used as the dictionary buffer, which is why 176 * there is no need to allocate the dictionary as part of the decoder's 177 * internal state. 178 * 179 * Because the output buffer is used as the workspace, streams encoded using 180 * a big dictionary are not a problem in single-call mode. It is enough that 181 * the output buffer is big enough to hold the actual uncompressed data; it 182 * can be smaller than the dictionary size stored in the stream headers. 183 * 184 * Multi-call mode with preallocated dictionary (XZ_PREALLOC): dict_max bytes 185 * of memory is preallocated for the LZMA2 dictionary. This way there is no 186 * risk that xz_dec_run() could run out of memory, since xz_dec_run() will 187 * never allocate any memory. Instead, if the preallocated dictionary is too 188 * small for decoding the given input stream, xz_dec_run() will return 189 * XZ_MEMLIMIT_ERROR. Thus, it is important to know what kind of data will be 190 * decoded to avoid allocating excessive amount of memory for the dictionary. 191 * 192 * Multi-call mode with dynamically allocated dictionary (XZ_DYNALLOC): 193 * dict_max specifies the maximum allowed dictionary size that xz_dec_run() 194 * may allocate once it has parsed the dictionary size from the stream 195 * headers. This way excessive allocations can be avoided while still 196 * limiting the maximum memory usage to a sane value to prevent running the 197 * system out of memory when decompressing streams from untrusted sources. 198 * 199 * On success, xz_dec_init() returns a pointer to struct xz_dec, which is 200 * ready to be used with xz_dec_run(). If memory allocation fails, 201 * xz_dec_init() returns NULL. 202 */ 203 XZ_EXTERN struct xz_dec *xz_dec_init(enum xz_mode mode, uint32_t dict_max); 204 205 /** 206 * xz_dec_run() - Run the XZ decoder for a single XZ stream 207 * @s: Decoder state allocated using xz_dec_init() 208 * @b: Input and output buffers 209 * 210 * The possible return values depend on build options and operation mode. 211 * See enum xz_ret for details. 212 * 213 * Note that if an error occurs in single-call mode (return value is not 214 * XZ_STREAM_END), b->in_pos and b->out_pos are not modified and the 215 * contents of the output buffer from b->out[b->out_pos] onward are 216 * undefined. This is true even after XZ_BUF_ERROR, because with some filter 217 * chains, there may be a second pass over the output buffer, and this pass 218 * cannot be properly done if the output buffer is truncated. Thus, you 219 * cannot give the single-call decoder a too small buffer and then expect to 220 * get that amount valid data from the beginning of the stream. You must use 221 * the multi-call decoder if you don't want to uncompress the whole stream. 222 * 223 * Use xz_dec_run() when XZ data is stored inside some other file format. 224 * The decoding will stop after one XZ stream has been decompresed. To 225 * decompress regular .xz files which might have multiple concatenated 226 * streams, use xz_dec_catrun() instead. 227 */ 228 XZ_EXTERN enum xz_ret xz_dec_run(struct xz_dec *s, struct xz_buf *b); 229 230 /** 231 * xz_dec_catrun() - Run the XZ decoder with support for concatenated streams 232 * @s: Decoder state allocated using xz_dec_init() 233 * @b: Input and output buffers 234 * @finish: This is an int instead of bool to avoid requiring stdbool.h. 235 * As long as more input might be coming, finish must be false. 236 * When the caller knows that it has provided all the input to 237 * the decoder (some possibly still in b->in), it must set finish 238 * to true. Only when finish is true can this function return 239 * XZ_STREAM_END to indicate successful decompression of the 240 * file. In single-call mode (XZ_SINGLE) finish is assumed to 241 * always be true; the caller-provided value is ignored. 242 * 243 * This is like xz_dec_run() except that this makes it easy to decode .xz 244 * files with multiple streams (multiple .xz files concatenated as is). 245 * The rarely-used Stream Padding feature is supported too, that is, there 246 * can be null bytes after or between the streams. The number of null bytes 247 * must be a multiple of four. 248 * 249 * When finish is false and b->in_pos == b->in_size, it is possible that 250 * XZ_BUF_ERROR isn't returned even when no progress is possible (XZ_OK is 251 * returned instead). This shouldn't matter because in this situation a 252 * reasonable caller will attempt to provide more input or set finish to 253 * true for the next xz_dec_catrun() call anyway. 254 * 255 * For any struct xz_dec that has been initialized for multi-call mode: 256 * Once decoding has been started with xz_dec_run() or xz_dec_catrun(), 257 * the same function must be used until xz_dec_reset() or xz_dec_end(). 258 * Switching between the two decoding functions without resetting results 259 * in undefined behavior. 260 * 261 * xz_dec_catrun() is only available if XZ_DEC_CONCATENATED was defined 262 * at compile time. 263 */ 264 XZ_EXTERN enum xz_ret xz_dec_catrun(struct xz_dec *s, struct xz_buf *b, 265 int finish); 266 267 /** 268 * xz_dec_reset() - Reset an already allocated decoder state 269 * @s: Decoder state allocated using xz_dec_init() 270 * 271 * This function can be used to reset the multi-call decoder state without 272 * freeing and reallocating memory with xz_dec_end() and xz_dec_init(). 273 * 274 * In single-call mode, xz_dec_reset() is always called in the beginning of 275 * xz_dec_run(). Thus, explicit call to xz_dec_reset() is useful only in 276 * multi-call mode. 277 */ 278 XZ_EXTERN void xz_dec_reset(struct xz_dec *s); 279 280 /** 281 * xz_dec_end() - Free the memory allocated for the decoder state 282 * @s: Decoder state allocated using xz_dec_init(). If s is NULL, 283 * this function does nothing. 284 */ 285 XZ_EXTERN void xz_dec_end(struct xz_dec *s); 286 287 /* 288 * Decompressor for MicroLZMA, an LZMA variant with a very minimal header. 289 * See xz_dec_microlzma_alloc() below for details. 290 * 291 * These functions aren't used or available in preboot code and thus aren't 292 * marked with XZ_EXTERN. This avoids warnings about static functions that 293 * are never defined. 294 */ 295 /** 296 * struct xz_dec_microlzma - Opaque type to hold the MicroLZMA decoder state 297 */ 298 struct xz_dec_microlzma; 299 300 /** 301 * xz_dec_microlzma_alloc() - Allocate memory for the MicroLZMA decoder 302 * @mode XZ_SINGLE or XZ_PREALLOC 303 * @dict_size LZMA dictionary size. This must be at least 4 KiB and 304 * at most 3 GiB. 305 * 306 * In contrast to xz_dec_init(), this function only allocates the memory 307 * and remembers the dictionary size. xz_dec_microlzma_reset() must be used 308 * before calling xz_dec_microlzma_run(). 309 * 310 * The amount of allocated memory is a little less than 30 KiB with XZ_SINGLE. 311 * With XZ_PREALLOC also a dictionary buffer of dict_size bytes is allocated. 312 * 313 * On success, xz_dec_microlzma_alloc() returns a pointer to 314 * struct xz_dec_microlzma. If memory allocation fails or 315 * dict_size is invalid, NULL is returned. 316 * 317 * The compressed format supported by this decoder is a raw LZMA stream 318 * whose first byte (always 0x00) has been replaced with bitwise-negation 319 * of the LZMA properties (lc/lp/pb) byte. For example, if lc/lp/pb is 320 * 3/0/2, the first byte is 0xA2. This way the first byte can never be 0x00. 321 * Just like with LZMA2, lc + lp <= 4 must be true. The LZMA end-of-stream 322 * marker must not be used. The unused values are reserved for future use. 323 * This MicroLZMA header format was created for use in EROFS but may be used 324 * by others too. 325 */ 326 extern struct xz_dec_microlzma *xz_dec_microlzma_alloc(enum xz_mode mode, 327 uint32_t dict_size); 328 329 /** 330 * xz_dec_microlzma_reset() - Reset the MicroLZMA decoder state 331 * @s Decoder state allocated using xz_dec_microlzma_alloc() 332 * @comp_size Compressed size of the input stream 333 * @uncomp_size Uncompressed size of the input stream. A value smaller 334 * than the real uncompressed size of the input stream can 335 * be specified if uncomp_size_is_exact is set to false. 336 * uncomp_size can never be set to a value larger than the 337 * expected real uncompressed size because it would eventually 338 * result in XZ_DATA_ERROR. 339 * @uncomp_size_is_exact This is an int instead of bool to avoid 340 * requiring stdbool.h. This should normally be set to true. 341 * When this is set to false, error detection is weaker. 342 */ 343 extern void xz_dec_microlzma_reset(struct xz_dec_microlzma *s, 344 uint32_t comp_size, uint32_t uncomp_size, 345 int uncomp_size_is_exact); 346 347 /** 348 * xz_dec_microlzma_run() - Run the MicroLZMA decoder 349 * @s Decoder state initialized using xz_dec_microlzma_reset() 350 * @b: Input and output buffers 351 * 352 * This works similarly to xz_dec_run() with a few important differences. 353 * Only the differences are documented here. 354 * 355 * The only possible return values are XZ_OK, XZ_STREAM_END, and 356 * XZ_DATA_ERROR. This function cannot return XZ_BUF_ERROR: if no progress 357 * is possible due to lack of input data or output space, this function will 358 * keep returning XZ_OK. Thus, the calling code must be written so that it 359 * will eventually provide input and output space matching (or exceeding) 360 * comp_size and uncomp_size arguments given to xz_dec_microlzma_reset(). 361 * If the caller cannot do this (for example, if the input file is truncated 362 * or otherwise corrupt), the caller must detect this error by itself to 363 * avoid an infinite loop. 364 * 365 * If the compressed data seems to be corrupt, XZ_DATA_ERROR is returned. 366 * This can happen also when incorrect dictionary, uncompressed, or 367 * compressed sizes have been specified. 368 * 369 * With XZ_PREALLOC only: As an extra feature, b->out may be NULL to skip over 370 * uncompressed data. This way the caller doesn't need to provide a temporary 371 * output buffer for the bytes that will be ignored. 372 * 373 * With XZ_SINGLE only: In contrast to xz_dec_run(), the return value XZ_OK 374 * is also possible and thus XZ_SINGLE is actually a limited multi-call mode. 375 * After XZ_OK the bytes decoded so far may be read from the output buffer. 376 * It is possible to continue decoding but the variables b->out and b->out_pos 377 * MUST NOT be changed by the caller. Increasing the value of b->out_size is 378 * allowed to make more output space available; one doesn't need to provide 379 * space for the whole uncompressed data on the first call. The input buffer 380 * may be changed normally like with XZ_PREALLOC. This way input data can be 381 * provided from non-contiguous memory. 382 */ 383 extern enum xz_ret xz_dec_microlzma_run(struct xz_dec_microlzma *s, 384 struct xz_buf *b); 385 386 /** 387 * xz_dec_microlzma_end() - Free the memory allocated for the decoder state 388 * @s: Decoder state allocated using xz_dec_microlzma_alloc(). 389 * If s is NULL, this function does nothing. 390 */ 391 extern void xz_dec_microlzma_end(struct xz_dec_microlzma *s); 392 393 /* 394 * Standalone build (userspace build or in-kernel build for boot time use) 395 * needs a CRC32 implementation. For normal in-kernel use, kernel's own 396 * CRC32 module is used instead, and users of this module don't need to 397 * care about the functions below. 398 */ 399 #ifndef XZ_INTERNAL_CRC32 400 # ifdef __KERNEL__ 401 # define XZ_INTERNAL_CRC32 0 402 # else 403 # define XZ_INTERNAL_CRC32 1 404 # endif 405 #endif 406 407 /* 408 * If CRC64 support has been enabled with XZ_USE_CRC64, a CRC64 409 * implementation is needed too. 410 */ 411 #ifndef XZ_USE_CRC64 412 # undef XZ_INTERNAL_CRC64 413 # define XZ_INTERNAL_CRC64 0 414 #endif 415 #ifndef XZ_INTERNAL_CRC64 416 # ifdef __KERNEL__ 417 # error Using CRC64 in the kernel has not been implemented. 418 # else 419 # define XZ_INTERNAL_CRC64 1 420 # endif 421 #endif 422 423 #if XZ_INTERNAL_CRC32 424 /* 425 * This must be called before any other xz_* function to initialize 426 * the CRC32 lookup table. 427 */ 428 XZ_EXTERN void xz_crc32_init(void); 429 430 /* 431 * Update CRC32 value using the polynomial from IEEE-802.3. To start a new 432 * calculation, the third argument must be zero. To continue the calculation, 433 * the previously returned value is passed as the third argument. 434 */ 435 XZ_EXTERN uint32_t xz_crc32(const uint8_t *buf, size_t size, uint32_t crc); 436 #endif 437 438 #if XZ_INTERNAL_CRC64 439 /* 440 * This must be called before any other xz_* function (except xz_crc32_init()) 441 * to initialize the CRC64 lookup table. 442 */ 443 XZ_EXTERN void xz_crc64_init(void); 444 445 /* 446 * Update CRC64 value using the polynomial from ECMA-182. To start a new 447 * calculation, the third argument must be zero. To continue the calculation, 448 * the previously returned value is passed as the third argument. 449 */ 450 XZ_EXTERN uint64_t xz_crc64(const uint8_t *buf, size_t size, uint64_t crc); 451 #endif 452 453 #ifdef __cplusplus 454 } 455 #endif 456 457 #endif 458