1 /* 2 * XZ decompressor 3 * 4 * Authors: Lasse Collin <lasse.collin@tukaani.org> 5 * Igor Pavlov <http://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 #include "xz_hal_plat.h" 15 16 #ifdef __cplusplus 17 extern "C" { 18 #endif 19 20 /* In Linux, this is used to make extern functions static when needed. */ 21 #ifndef XZ_EXTERN 22 #define XZ_EXTERN extern 23 #endif 24 25 /** 26 * enum xz_mode - Operation mode 27 * 28 * @XZ_SINGLE: Single-call mode. This uses less RAM than 29 * than multi-call modes, because the LZMA2 30 * dictionary doesn't need to be allocated as 31 * part of the decoder state. All required data 32 * structures are allocated at initialization, 33 * so xz_dec_run() cannot return XZ_MEM_ERROR. 34 * @XZ_PREALLOC: Multi-call mode with preallocated LZMA2 35 * dictionary buffer. All data structures are 36 * allocated at initialization, so xz_dec_run() 37 * cannot return XZ_MEM_ERROR. 38 * @XZ_DYNALLOC: Multi-call mode. The LZMA2 dictionary is 39 * allocated once the required size has been 40 * parsed from the stream headers. If the 41 * allocation fails, xz_dec_run() will return 42 * XZ_MEM_ERROR. 43 * 44 * It is possible to enable support only for a subset of the above 45 * modes at compile time by defining XZ_DEC_SINGLE, XZ_DEC_PREALLOC, 46 * or XZ_DEC_DYNALLOC. The xz_dec kernel module is always compiled 47 * with support for all operation modes, but the preboot code may 48 * be built with fewer features to minimize code size. 49 */ 50 enum xz_mode { 51 XZ_SINGLE, 52 XZ_PREALLOC, 53 XZ_DYNALLOC 54 }; 55 56 /** 57 * enum xz_ret - Return codes 58 * @XZ_OK: Everything is OK so far. More input or more 59 * output space is required to continue. This 60 * return code is possible only in multi-call mode 61 * (XZ_PREALLOC or XZ_DYNALLOC). 62 * @XZ_STREAM_END: Operation finished successfully. 63 * @XZ_UNSUPPORTED_CHECK: Integrity check type is not supported. Decoding 64 * is still possible in multi-call mode by simply 65 * calling xz_dec_run() again. 66 * Note that this return value is used only if 67 * XZ_DEC_ANY_CHECK was defined at build time, 68 * which is not used in the kernel. Unsupported 69 * check types return XZ_OPTIONS_ERROR if 70 * XZ_DEC_ANY_CHECK was not defined at build time. 71 * @XZ_MEM_ERROR: Allocating memory failed. This return code is 72 * possible only if the decoder was initialized 73 * with XZ_DYNALLOC. The amount of memory that was 74 * tried to be allocated was no more than the 75 * dict_max argument given to xz_dec_init(). 76 * @XZ_MEMLIMIT_ERROR: A bigger LZMA2 dictionary would be needed than 77 * allowed by the dict_max argument given to 78 * xz_dec_init(). This return value is possible 79 * only in multi-call mode (XZ_PREALLOC or 80 * XZ_DYNALLOC); the single-call mode (XZ_SINGLE) 81 * ignores the dict_max argument. 82 * @XZ_FORMAT_ERROR: File format was not recognized (wrong magic 83 * bytes). 84 * @XZ_OPTIONS_ERROR: This implementation doesn't support the requested 85 * compression options. In the decoder this means 86 * that the header CRC32 matches, but the header 87 * itself specifies something that we don't support. 88 * @XZ_DATA_ERROR: Compressed data is corrupt. 89 * @XZ_BUF_ERROR: Cannot make any progress. Details are slightly 90 * different between multi-call and single-call 91 * mode; more information below. 92 * 93 * In multi-call mode, XZ_BUF_ERROR is returned when two consecutive calls 94 * to XZ code cannot consume any input and cannot produce any new output. 95 * This happens when there is no new input available, or the output buffer 96 * is full while at least one output byte is still pending. Assuming your 97 * code is not buggy, you can get this error only when decoding a compressed 98 * stream that is truncated or otherwise corrupt. 99 * 100 * In single-call mode, XZ_BUF_ERROR is returned only when the output buffer 101 * is too small or the compressed input is corrupt in a way that makes the 102 * decoder produce more output than the caller expected. When it is 103 * (relatively) clear that the compressed input is truncated, XZ_DATA_ERROR 104 * is used instead of XZ_BUF_ERROR. 105 */ 106 enum xz_ret { 107 XZ_OK, 108 XZ_STREAM_END, 109 XZ_UNSUPPORTED_CHECK, 110 XZ_MEM_ERROR, 111 XZ_MEMLIMIT_ERROR, 112 XZ_FORMAT_ERROR, 113 XZ_OPTIONS_ERROR, 114 XZ_DATA_ERROR, 115 XZ_BUF_ERROR 116 }; 117 118 /** 119 * struct xz_buf - Passing input and output buffers to XZ code 120 * @in: Beginning of the input buffer. This may be NULL if and only 121 * if in_pos is equal to in_size. 122 * @in_pos: Current position in the input buffer. This must not exceed 123 * in_size. 124 * @in_size: Size of the input buffer 125 * @out: Beginning of the output buffer. This may be NULL if and only 126 * if out_pos is equal to out_size. 127 * @out_pos: Current position in the output buffer. This must not exceed 128 * out_size. 129 * @out_size: Size of the output buffer 130 * 131 * Only the contents of the output buffer from out[out_pos] onward, and 132 * the variables in_pos and out_pos are modified by the XZ code. 133 */ 134 struct xz_buf { 135 const uint8_t *in; 136 size_t in_pos; 137 size_t in_size; 138 139 uint8_t *out; 140 size_t out_pos; 141 size_t out_size; 142 }; 143 144 /** 145 * struct xz_dec - Opaque type to hold the XZ decoder state 146 */ 147 struct xz_dec; 148 149 /** 150 * xz_dec_init() - Allocate and initialize a XZ decoder state 151 * @mode: Operation mode 152 * @dict_max: Maximum size of the LZMA2 dictionary (history buffer) for 153 * multi-call decoding. This is ignored in single-call mode 154 * (mode == XZ_SINGLE). LZMA2 dictionary is always 2^n bytes 155 * or 2^n + 2^(n-1) bytes (the latter sizes are less common 156 * in practice), so other values for dict_max don't make sense. 157 * In the kernel, dictionary sizes of 64 KiB, 128 KiB, 256 KiB, 158 * 512 KiB, and 1 MiB are probably the only reasonable values, 159 * except for kernel and initramfs images where a bigger 160 * dictionary can be fine and useful. 161 * 162 * Single-call mode (XZ_SINGLE): xz_dec_run() decodes the whole stream at 163 * once. The caller must provide enough output space or the decoding will 164 * fail. The output space is used as the dictionary buffer, which is why 165 * there is no need to allocate the dictionary as part of the decoder's 166 * internal state. 167 * 168 * Because the output buffer is used as the workspace, streams encoded using 169 * a big dictionary are not a problem in single-call mode. It is enough that 170 * the output buffer is big enough to hold the actual uncompressed data; it 171 * can be smaller than the dictionary size stored in the stream headers. 172 * 173 * Multi-call mode with preallocated dictionary (XZ_PREALLOC): dict_max bytes 174 * of memory is preallocated for the LZMA2 dictionary. This way there is no 175 * risk that xz_dec_run() could run out of memory, since xz_dec_run() will 176 * never allocate any memory. Instead, if the preallocated dictionary is too 177 * small for decoding the given input stream, xz_dec_run() will return 178 * XZ_MEMLIMIT_ERROR. Thus, it is important to know what kind of data will be 179 * decoded to avoid allocating excessive amount of memory for the dictionary. 180 * 181 * Multi-call mode with dynamically allocated dictionary (XZ_DYNALLOC): 182 * dict_max specifies the maximum allowed dictionary size that xz_dec_run() 183 * may allocate once it has parsed the dictionary size from the stream 184 * headers. This way excessive allocations can be avoided while still 185 * limiting the maximum memory usage to a sane value to prevent running the 186 * system out of memory when decompressing streams from untrusted sources. 187 * 188 * On success, xz_dec_init() returns a pointer to struct xz_dec, which is 189 * ready to be used with xz_dec_run(). If memory allocation fails, 190 * xz_dec_init() returns NULL. 191 */ 192 XZ_EXTERN struct xz_dec *xz_dec_init(enum xz_mode mode, uint32_t dict_max); 193 194 /** 195 * xz_dec_run() - Run the XZ decoder 196 * @s: Decoder state allocated using xz_dec_init() 197 * @b: Input and output buffers 198 * 199 * The possible return values depend on build options and operation mode. 200 * See enum xz_ret for details. 201 * 202 * Note that if an error occurs in single-call mode (return value is not 203 * XZ_STREAM_END), b->in_pos and b->out_pos are not modified and the 204 * contents of the output buffer from b->out[b->out_pos] onward are 205 * undefined. This is true even after XZ_BUF_ERROR, because with some filter 206 * chains, there may be a second pass over the output buffer, and this pass 207 * cannot be properly done if the output buffer is truncated. Thus, you 208 * cannot give the single-call decoder a too small buffer and then expect to 209 * get that amount valid data from the beginning of the stream. You must use 210 * the multi-call decoder if you don't want to uncompress the whole stream. 211 */ 212 XZ_EXTERN enum xz_ret xz_dec_run(struct xz_dec *s, struct xz_buf *b); 213 214 /** 215 * xz_dec_reset() - Reset an already allocated decoder state 216 * @s: Decoder state allocated using xz_dec_init() 217 * 218 * This function can be used to reset the multi-call decoder state without 219 * freeing and reallocating memory with xz_dec_end() and xz_dec_init(). 220 * 221 * In single-call mode, xz_dec_reset() is always called in the beginning of 222 * xz_dec_run(). Thus, explicit call to xz_dec_reset() is useful only in 223 * multi-call mode. 224 */ 225 XZ_EXTERN void xz_dec_reset(struct xz_dec *s); 226 227 /** 228 * xz_dec_end() - Free the memory allocated for the decoder state 229 * @s: Decoder state allocated using xz_dec_init(). If s is NULL, 230 * this function does nothing. 231 */ 232 XZ_EXTERN void xz_dec_end(struct xz_dec *s); 233 234 /* 235 * Standalone build (userspace build or in-kernel build for boot time use) 236 * needs a CRC32 implementation. For normal in-kernel use, kernel's own 237 * CRC32 module is used instead, and users of this module don't need to 238 * care about the functions below. 239 */ 240 #ifndef XZ_INTERNAL_CRC32 241 #ifdef __KERNEL__ 242 #define XZ_INTERNAL_CRC32 0 243 #else 244 #define XZ_INTERNAL_CRC32 1 245 #endif 246 #endif 247 248 /* 249 * If CRC64 support has been enabled with XZ_USE_CRC64, a CRC64 250 * implementation is needed too. 251 */ 252 #ifndef XZ_USE_CRC64 253 #undef XZ_INTERNAL_CRC64 254 #define XZ_INTERNAL_CRC64 0 255 #endif 256 #ifndef XZ_INTERNAL_CRC64 257 #ifdef __KERNEL__ 258 #error Using CRC64 in the kernel has not been implemented. 259 #else 260 #define XZ_INTERNAL_CRC64 1 261 #endif 262 #endif 263 264 #if XZ_INTERNAL_CRC32 265 /* 266 * This must be called before any other xz_* function to initialize 267 * the CRC32 lookup table. 268 */ 269 XZ_EXTERN void xz_crc32_init(void); 270 271 /* 272 * Update CRC32 value using the polynomial from IEEE-802.3. To start a new 273 * calculation, the third argument must be zero. To continue the calculation, 274 * the previously returned value is passed as the third argument. 275 */ 276 XZ_EXTERN uint32_t xz_crc32(const uint8_t *buf, size_t size, uint32_t crc); 277 #endif 278 279 #if XZ_INTERNAL_CRC64 280 /* 281 * This must be called before any other xz_* function (except xz_crc32_init()) 282 * to initialize the CRC64 lookup table. 283 */ 284 XZ_EXTERN void xz_crc64_init(void); 285 286 /* 287 * Update CRC64 value using the polynomial from ECMA-182. To start a new 288 * calculation, the third argument must be zero. To continue the calculation, 289 * the previously returned value is passed as the third argument. 290 */ 291 XZ_EXTERN uint64_t xz_crc64(const uint8_t *buf, size_t size, uint64_t crc); 292 #endif 293 294 #ifdef __cplusplus 295 } 296 #endif 297 298 #endif 299