blob: 259d732f3908505697dc9408433abed48b07f65c [file] [log] [blame]
b.liue9582032025-04-17 19:18:16 +08001/*
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#include <linux/stddef.h>
15#include <linux/types.h>
16
17/* In Linux, this is used to make extern functions static when needed. */
18#ifndef XZ_EXTERN
19# define XZ_EXTERN extern
20#endif
21
22/**
23 * enum xz_mode - Operation mode
24 *
25 * @XZ_SINGLE: Single-call mode. This uses less RAM than
26 * multi-call modes, because the LZMA2
27 * dictionary doesn't need to be allocated as
28 * part of the decoder state. All required data
29 * structures are allocated at initialization,
30 * so xz_dec_run() cannot return XZ_MEM_ERROR.
31 * @XZ_PREALLOC: Multi-call mode with preallocated LZMA2
32 * dictionary buffer. All data structures are
33 * allocated at initialization, so xz_dec_run()
34 * cannot return XZ_MEM_ERROR.
35 * @XZ_DYNALLOC: Multi-call mode. The LZMA2 dictionary is
36 * allocated once the required size has been
37 * parsed from the stream headers. If the
38 * allocation fails, xz_dec_run() will return
39 * XZ_MEM_ERROR.
40 *
41 * It is possible to enable support only for a subset of the above
42 * modes at compile time by defining XZ_DEC_SINGLE, XZ_DEC_PREALLOC,
43 * or XZ_DEC_DYNALLOC. The xz_dec kernel module is always compiled
44 * with support for all operation modes, but the preboot code may
45 * be built with fewer features to minimize code size.
46 */
47enum xz_mode {
48 XZ_SINGLE,
49 XZ_PREALLOC,
50 XZ_DYNALLOC
51};
52
53/**
54 * enum xz_ret - Return codes
55 * @XZ_OK: Everything is OK so far. More input or more
56 * output space is required to continue. This
57 * return code is possible only in multi-call mode
58 * (XZ_PREALLOC or XZ_DYNALLOC).
59 * @XZ_STREAM_END: Operation finished successfully.
60 * @XZ_UNSUPPORTED_CHECK: Integrity check type is not supported. Decoding
61 * is still possible in multi-call mode by simply
62 * calling xz_dec_run() again.
63 * Note that this return value is used only if
64 * XZ_DEC_ANY_CHECK was defined at build time,
65 * which is not used in the kernel. Unsupported
66 * check types return XZ_OPTIONS_ERROR if
67 * XZ_DEC_ANY_CHECK was not defined at build time.
68 * @XZ_MEM_ERROR: Allocating memory failed. This return code is
69 * possible only if the decoder was initialized
70 * with XZ_DYNALLOC. The amount of memory that was
71 * tried to be allocated was no more than the
72 * dict_max argument given to xz_dec_init().
73 * @XZ_MEMLIMIT_ERROR: A bigger LZMA2 dictionary would be needed than
74 * allowed by the dict_max argument given to
75 * xz_dec_init(). This return value is possible
76 * only in multi-call mode (XZ_PREALLOC or
77 * XZ_DYNALLOC); the single-call mode (XZ_SINGLE)
78 * ignores the dict_max argument.
79 * @XZ_FORMAT_ERROR: File format was not recognized (wrong magic
80 * bytes).
81 * @XZ_OPTIONS_ERROR: This implementation doesn't support the requested
82 * compression options. In the decoder this means
83 * that the header CRC32 matches, but the header
84 * itself specifies something that we don't support.
85 * @XZ_DATA_ERROR: Compressed data is corrupt.
86 * @XZ_BUF_ERROR: Cannot make any progress. Details are slightly
87 * different between multi-call and single-call
88 * mode; more information below.
89 *
90 * In multi-call mode, XZ_BUF_ERROR is returned when two consecutive calls
91 * to XZ code cannot consume any input and cannot produce any new output.
92 * This happens when there is no new input available, or the output buffer
93 * is full while at least one output byte is still pending. Assuming your
94 * code is not buggy, you can get this error only when decoding a compressed
95 * stream that is truncated or otherwise corrupt.
96 *
97 * In single-call mode, XZ_BUF_ERROR is returned only when the output buffer
98 * is too small or the compressed input is corrupt in a way that makes the
99 * decoder produce more output than the caller expected. When it is
100 * (relatively) clear that the compressed input is truncated, XZ_DATA_ERROR
101 * is used instead of XZ_BUF_ERROR.
102 */
103enum xz_ret {
104 XZ_OK,
105 XZ_STREAM_END,
106 XZ_UNSUPPORTED_CHECK,
107 XZ_MEM_ERROR,
108 XZ_MEMLIMIT_ERROR,
109 XZ_FORMAT_ERROR,
110 XZ_OPTIONS_ERROR,
111 XZ_DATA_ERROR,
112 XZ_BUF_ERROR
113};
114
115/**
116 * struct xz_buf - Passing input and output buffers to XZ code
117 * @in: Beginning of the input buffer. This may be NULL if and only
118 * if in_pos is equal to in_size.
119 * @in_pos: Current position in the input buffer. This must not exceed
120 * in_size.
121 * @in_size: Size of the input buffer
122 * @out: Beginning of the output buffer. This may be NULL if and only
123 * if out_pos is equal to out_size.
124 * @out_pos: Current position in the output buffer. This must not exceed
125 * out_size.
126 * @out_size: Size of the output buffer
127 *
128 * Only the contents of the output buffer from out[out_pos] onward, and
129 * the variables in_pos and out_pos are modified by the XZ code.
130 */
131struct xz_buf {
132 const uint8_t *in;
133 size_t in_pos;
134 size_t in_size;
135
136 uint8_t *out;
137 size_t out_pos;
138 size_t out_size;
139};
140
141/**
142 * struct xz_dec - Opaque type to hold the XZ decoder state
143 */
144struct xz_dec;
145
146/**
147 * xz_dec_init() - Allocate and initialize a XZ decoder state
148 * @mode: Operation mode
149 * @dict_max: Maximum size of the LZMA2 dictionary (history buffer) for
150 * multi-call decoding. This is ignored in single-call mode
151 * (mode == XZ_SINGLE). LZMA2 dictionary is always 2^n bytes
152 * or 2^n + 2^(n-1) bytes (the latter sizes are less common
153 * in practice), so other values for dict_max don't make sense.
154 * In the kernel, dictionary sizes of 64 KiB, 128 KiB, 256 KiB,
155 * 512 KiB, and 1 MiB are probably the only reasonable values,
156 * except for kernel and initramfs images where a bigger
157 * dictionary can be fine and useful.
158 *
159 * Single-call mode (XZ_SINGLE): xz_dec_run() decodes the whole stream at
160 * once. The caller must provide enough output space or the decoding will
161 * fail. The output space is used as the dictionary buffer, which is why
162 * there is no need to allocate the dictionary as part of the decoder's
163 * internal state.
164 *
165 * Because the output buffer is used as the workspace, streams encoded using
166 * a big dictionary are not a problem in single-call mode. It is enough that
167 * the output buffer is big enough to hold the actual uncompressed data; it
168 * can be smaller than the dictionary size stored in the stream headers.
169 *
170 * Multi-call mode with preallocated dictionary (XZ_PREALLOC): dict_max bytes
171 * of memory is preallocated for the LZMA2 dictionary. This way there is no
172 * risk that xz_dec_run() could run out of memory, since xz_dec_run() will
173 * never allocate any memory. Instead, if the preallocated dictionary is too
174 * small for decoding the given input stream, xz_dec_run() will return
175 * XZ_MEMLIMIT_ERROR. Thus, it is important to know what kind of data will be
176 * decoded to avoid allocating excessive amount of memory for the dictionary.
177 *
178 * Multi-call mode with dynamically allocated dictionary (XZ_DYNALLOC):
179 * dict_max specifies the maximum allowed dictionary size that xz_dec_run()
180 * may allocate once it has parsed the dictionary size from the stream
181 * headers. This way excessive allocations can be avoided while still
182 * limiting the maximum memory usage to a sane value to prevent running the
183 * system out of memory when decompressing streams from untrusted sources.
184 *
185 * On success, xz_dec_init() returns a pointer to struct xz_dec, which is
186 * ready to be used with xz_dec_run(). If memory allocation fails,
187 * xz_dec_init() returns NULL.
188 */
189XZ_EXTERN struct xz_dec *xz_dec_init(enum xz_mode mode, uint32_t dict_max);
190
191/**
192 * xz_dec_run() - Run the XZ decoder
193 * @s: Decoder state allocated using xz_dec_init()
194 * @b: Input and output buffers
195 *
196 * The possible return values depend on build options and operation mode.
197 * See enum xz_ret for details.
198 *
199 * Note that if an error occurs in single-call mode (return value is not
200 * XZ_STREAM_END), b->in_pos and b->out_pos are not modified and the
201 * contents of the output buffer from b->out[b->out_pos] onward are
202 * undefined. This is true even after XZ_BUF_ERROR, because with some filter
203 * chains, there may be a second pass over the output buffer, and this pass
204 * cannot be properly done if the output buffer is truncated. Thus, you
205 * cannot give the single-call decoder a too small buffer and then expect to
206 * get that amount valid data from the beginning of the stream. You must use
207 * the multi-call decoder if you don't want to uncompress the whole stream.
208 */
209XZ_EXTERN enum xz_ret xz_dec_run(struct xz_dec *s, struct xz_buf *b);
210
211/**
212 * xz_dec_reset() - Reset an already allocated decoder state
213 * @s: Decoder state allocated using xz_dec_init()
214 *
215 * This function can be used to reset the multi-call decoder state without
216 * freeing and reallocating memory with xz_dec_end() and xz_dec_init().
217 *
218 * In single-call mode, xz_dec_reset() is always called in the beginning of
219 * xz_dec_run(). Thus, explicit call to xz_dec_reset() is useful only in
220 * multi-call mode.
221 */
222XZ_EXTERN void xz_dec_reset(struct xz_dec *s);
223
224/**
225 * xz_dec_end() - Free the memory allocated for the decoder state
226 * @s: Decoder state allocated using xz_dec_init(). If s is NULL,
227 * this function does nothing.
228 */
229XZ_EXTERN void xz_dec_end(struct xz_dec *s);
230
231/*
232 * Standalone build (userspace build or in-kernel build for boot time use)
233 * needs a CRC32 implementation. For normal in-kernel use, kernel's own
234 * CRC32 module is used instead, and users of this module don't need to
235 * care about the functions below.
236 */
237#ifndef XZ_INTERNAL_CRC32
238# ifdef __KERNEL__
239# define XZ_INTERNAL_CRC32 0
240# else
241# define XZ_INTERNAL_CRC32 1
242# endif
243#endif
244
245/*
246 * If CRC64 support has been enabled with XZ_USE_CRC64, a CRC64
247 * implementation is needed too.
248 */
249#ifndef XZ_USE_CRC64
250# undef XZ_INTERNAL_CRC64
251# define XZ_INTERNAL_CRC64 0
252#endif
253#ifndef XZ_INTERNAL_CRC64
254# ifdef __KERNEL__
255# error Using CRC64 in the kernel has not been implemented.
256# else
257# define XZ_INTERNAL_CRC64 1
258# endif
259#endif
260
261#if XZ_INTERNAL_CRC32
262/*
263 * This must be called before any other xz_* function to initialize
264 * the CRC32 lookup table.
265 */
266XZ_EXTERN void xz_crc32_init(void);
267
268/*
269 * Update CRC32 value using the polynomial from IEEE-802.3. To start a new
270 * calculation, the third argument must be zero. To continue the calculation,
271 * the previously returned value is passed as the third argument.
272 */
273XZ_EXTERN uint32_t xz_crc32(const uint8_t *buf, size_t size, uint32_t crc);
274#endif
275
276#if XZ_INTERNAL_CRC64
277/*
278 * This must be called before any other xz_* function (except xz_crc32_init())
279 * to initialize the CRC64 lookup table.
280 */
281XZ_EXTERN void xz_crc64_init(void);
282
283/*
284 * Update CRC64 value using the polynomial from ECMA-182. To start a new
285 * calculation, the third argument must be zero. To continue the calculation,
286 * the previously returned value is passed as the third argument.
287 */
288XZ_EXTERN uint64_t xz_crc64(const uint8_t *buf, size_t size, uint64_t crc);
289#endif
290
291#ifdef __cplusplus
292}
293#endif
294
295#endif