1 // SPDX-License-Identifier: GPL-2.0
2
3 /*
4 * Important notes about in-place decompression
5 *
6 * At least on x86, the kernel is decompressed in place: the compressed data
7 * is placed to the end of the output buffer, and the decompressor overwrites
8 * most of the compressed data. There must be enough safety margin to
9 * guarantee that the write position is always behind the read position.
10 *
11 * The safety margin for ZSTD with a 128 KB block size is calculated below.
12 * Note that the margin with ZSTD is bigger than with GZIP or XZ!
13 *
14 * The worst case for in-place decompression is that the beginning of
15 * the file is compressed extremely well, and the rest of the file is
16 * uncompressible. Thus, we must look for worst-case expansion when the
17 * compressor is encoding uncompressible data.
18 *
19 * The structure of the .zst file in case of a compressed kernel is as follows.
20 * Maximum sizes (as bytes) of the fields are in parenthesis.
21 *
22 * Frame Header: (18)
23 * Blocks: (N)
24 * Checksum: (4)
25 *
26 * The frame header and checksum overhead is at most 22 bytes.
27 *
28 * ZSTD stores the data in blocks. Each block has a header whose size is
29 * a 3 bytes. After the block header, there is up to 128 KB of payload.
30 * The maximum uncompressed size of the payload is 128 KB. The minimum
31 * uncompressed size of the payload is never less than the payload size
32 * (excluding the block header).
33 *
34 * The assumption, that the uncompressed size of the payload is never
35 * smaller than the payload itself, is valid only when talking about
36 * the payload as a whole. It is possible that the payload has parts where
37 * the decompressor consumes more input than it produces output. Calculating
38 * the worst case for this would be tricky. Instead of trying to do that,
39 * let's simply make sure that the decompressor never overwrites any bytes
40 * of the payload which it is currently reading.
41 *
42 * Now we have enough information to calculate the safety margin. We need
43 * - 22 bytes for the .zst file format headers;
44 * - 3 bytes per every 128 KiB of uncompressed size (one block header per
45 * block); and
46 * - 128 KiB (biggest possible zstd block size) to make sure that the
47 * decompressor never overwrites anything from the block it is currently
48 * reading.
49 *
50 * We get the following formula:
51 *
52 * safety_margin = 22 + uncompressed_size * 3 / 131072 + 131072
53 * <= 22 + (uncompressed_size >> 15) + 131072
54 */
55
56 /*
57 * Preboot environments #include "path/to/decompress_unzstd.c".
58 * All of the source files we depend on must be #included.
59 * zstd's only source dependency is xxhash, which has no source
60 * dependencies.
61 *
62 * When UNZSTD_PREBOOT is defined we declare __decompress(), which is
63 * used for kernel decompression, instead of unzstd().
64 *
65 * Define __DISABLE_EXPORTS in preboot environments to prevent symbols
66 * from xxhash and zstd from being exported by the EXPORT_SYMBOL macro.
67 */
68 #ifdef STATIC
69 # define UNZSTD_PREBOOT
70 # include "xxhash.c"
71 # include "zstd/decompress_sources.h"
72 #else
73 #include <linux/decompress/unzstd.h>
74 #endif
75
76 #include <linux/decompress/mm.h>
77 #include <linux/kernel.h>
78 #include <linux/zstd.h>
79
80 /* 128MB is the maximum window size supported by zstd. */
81 #define ZSTD_WINDOWSIZE_MAX (1 << ZSTD_WINDOWLOG_MAX)
82 /*
83 * Size of the input and output buffers in multi-call mode.
84 * Pick a larger size because it isn't used during kernel decompression,
85 * since that is single pass, and we have to allocate a large buffer for
86 * zstd's window anyway. The larger size speeds up initramfs decompression.
87 */
88 #define ZSTD_IOBUF_SIZE (1 << 17)
89
90 static int INIT handle_zstd_error(size_t ret, void (*error)(char *x))
91 {
92 const zstd_error_code err = zstd_get_error_code(ret);
93
94 if (!zstd_is_error(ret))
95 return 0;
96
97 /*
98 * zstd_get_error_name() cannot be used because error takes a char *
99 * not a const char *
100 */
101 switch (err) {
102 case ZSTD_error_memory_allocation:
103 error("ZSTD decompressor ran out of memory");
104 break;
105 case ZSTD_error_prefix_unknown:
106 error("Input is not in the ZSTD format (wrong magic bytes)");
107 break;
108 case ZSTD_error_dstSize_tooSmall:
109 case ZSTD_error_corruption_detected:
110 case ZSTD_error_checksum_wrong:
111 error("ZSTD-compressed data is corrupt");
112 break;
113 default:
114 error("ZSTD-compressed data is probably corrupt");
115 break;
116 }
117 return -1;
118 }
119
120 /*
121 * Handle the case where we have the entire input and output in one segment.
122 * We can allocate less memory (no circular buffer for the sliding window),
123 * and avoid some memcpy() calls.
124 */
125 static int INIT decompress_single(const u8 *in_buf, long in_len, u8 *out_buf,
126 long out_len, long *in_pos,
127 void (*error)(char *x))
128 {
129 const size_t wksp_size = zstd_dctx_workspace_bound();
130 void *wksp = large_malloc(wksp_size);
131 zstd_dctx *dctx = zstd_init_dctx(wksp, wksp_size);
132 int err;
133 size_t ret;
134
135 if (dctx == NULL) {
136 error("Out of memory while allocating zstd_dctx");
137 err = -1;
138 goto out;
139 }
140 /*
141 * Find out how large the frame actually is, there may be junk at
142 * the end of the frame that zstd_decompress_dctx() can't handle.
143 */
144 ret = zstd_find_frame_compressed_size(in_buf, in_len);
145 err = handle_zstd_error(ret, error);
146 if (err)
147 goto out;
148 in_len = (long)ret;
149
150 ret = zstd_decompress_dctx(dctx, out_buf, out_len, in_buf, in_len);
151 err = handle_zstd_error(ret, error);
152 if (err)
153 goto out;
154
155 if (in_pos != NULL)
156 *in_pos = in_len;
157
158 err = 0;
159 out:
160 if (wksp != NULL)
161 large_free(wksp);
162 return err;
163 }
164
165 static int INIT __unzstd(unsigned char *in_buf, long in_len,
166 long (*fill)(void*, unsigned long),
167 long (*flush)(void*, unsigned long),
168 unsigned char *out_buf, long out_len,
169 long *in_pos,
170 void (*error)(char *x))
171 {
172 zstd_in_buffer in;
173 zstd_out_buffer out;
174 zstd_frame_header header;
175 void *in_allocated = NULL;
176 void *out_allocated = NULL;
177 void *wksp = NULL;
178 size_t wksp_size;
179 zstd_dstream *dstream;
180 int err;
181 size_t ret;
182
183 /*
184 * ZSTD decompression code won't be happy if the buffer size is so big
185 * that its end address overflows. When the size is not provided, make
186 * it as big as possible without having the end address overflow.
187 */
188 if (out_len == 0)
189 out_len = UINTPTR_MAX - (uintptr_t)out_buf;
190
191 if (fill == NULL && flush == NULL)
192 /*
193 * We can decompress faster and with less memory when we have a
194 * single chunk.
195 */
196 return decompress_single(in_buf, in_len, out_buf, out_len,
197 in_pos, error);
198
199 /*
200 * If in_buf is not provided, we must be using fill(), so allocate
201 * a large enough buffer. If it is provided, it must be at least
202 * ZSTD_IOBUF_SIZE large.
203 */
204 if (in_buf == NULL) {
205 in_allocated = large_malloc(ZSTD_IOBUF_SIZE);
206 if (in_allocated == NULL) {
207 error("Out of memory while allocating input buffer");
208 err = -1;
209 goto out;
210 }
211 in_buf = in_allocated;
212 in_len = 0;
213 }
214 /* Read the first chunk, since we need to decode the frame header. */
215 if (fill != NULL)
216 in_len = fill(in_buf, ZSTD_IOBUF_SIZE);
217 if (in_len < 0) {
218 error("ZSTD-compressed data is truncated");
219 err = -1;
220 goto out;
221 }
222 /* Set the first non-empty input buffer. */
223 in.src = in_buf;
224 in.pos = 0;
225 in.size = in_len;
226 /* Allocate the output buffer if we are using flush(). */
227 if (flush != NULL) {
228 out_allocated = large_malloc(ZSTD_IOBUF_SIZE);
229 if (out_allocated == NULL) {
230 error("Out of memory while allocating output buffer");
231 err = -1;
232 goto out;
233 }
234 out_buf = out_allocated;
235 out_len = ZSTD_IOBUF_SIZE;
236 }
237 /* Set the output buffer. */
238 out.dst = out_buf;
239 out.pos = 0;
240 out.size = out_len;
241
242 /*
243 * We need to know the window size to allocate the zstd_dstream.
244 * Since we are streaming, we need to allocate a buffer for the sliding
245 * window. The window size varies from 1 KB to ZSTD_WINDOWSIZE_MAX
246 * (8 MB), so it is important to use the actual value so as not to
247 * waste memory when it is smaller.
248 */
249 ret = zstd_get_frame_header(&header, in.src, in.size);
250 err = handle_zstd_error(ret, error);
251 if (err)
252 goto out;
253 if (ret != 0) {
254 error("ZSTD-compressed data has an incomplete frame header");
255 err = -1;
256 goto out;
257 }
258 if (header.windowSize > ZSTD_WINDOWSIZE_MAX) {
259 error("ZSTD-compressed data has too large a window size");
260 err = -1;
261 goto out;
262 }
263
264 /*
265 * Allocate the zstd_dstream now that we know how much memory is
266 * required.
267 */
268 wksp_size = zstd_dstream_workspace_bound(header.windowSize);
269 wksp = large_malloc(wksp_size);
270 dstream = zstd_init_dstream(header.windowSize, wksp, wksp_size);
271 if (dstream == NULL) {
272 error("Out of memory while allocating ZSTD_DStream");
273 err = -1;
274 goto out;
275 }
276
277 /*
278 * Decompression loop:
279 * Read more data if necessary (error if no more data can be read).
280 * Call the decompression function, which returns 0 when finished.
281 * Flush any data produced if using flush().
282 */
283 if (in_pos != NULL)
284 *in_pos = 0;
285 do {
286 /*
287 * If we need to reload data, either we have fill() and can
288 * try to get more data, or we don't and the input is truncated.
289 */
290 if (in.pos == in.size) {
291 if (in_pos != NULL)
292 *in_pos += in.pos;
293 in_len = fill ? fill(in_buf, ZSTD_IOBUF_SIZE) : -1;
294 if (in_len < 0) {
295 error("ZSTD-compressed data is truncated");
296 err = -1;
297 goto out;
298 }
299 in.pos = 0;
300 in.size = in_len;
301 }
302 /* Returns zero when the frame is complete. */
303 ret = zstd_decompress_stream(dstream, &out, &in);
304 err = handle_zstd_error(ret, error);
305 if (err)
306 goto out;
307 /* Flush all of the data produced if using flush(). */
308 if (flush != NULL && out.pos > 0) {
309 if (out.pos != flush(out.dst, out.pos)) {
310 error("Failed to flush()");
311 err = -1;
312 goto out;
313 }
314 out.pos = 0;
315 }
316 } while (ret != 0);
317
318 if (in_pos != NULL)
319 *in_pos += in.pos;
320
321 err = 0;
322 out:
323 if (in_allocated != NULL)
324 large_free(in_allocated);
325 if (out_allocated != NULL)
326 large_free(out_allocated);
327 if (wksp != NULL)
328 large_free(wksp);
329 return err;
330 }
331
332 #ifndef UNZSTD_PREBOOT
333 STATIC int INIT unzstd(unsigned char *buf, long len,
334 long (*fill)(void*, unsigned long),
335 long (*flush)(void*, unsigned long),
336 unsigned char *out_buf,
337 long *pos,
338 void (*error)(char *x))
339 {
340 return __unzstd(buf, len, fill, flush, out_buf, 0, pos, error);
341 }
342 #else
343 STATIC int INIT __decompress(unsigned char *buf, long len,
344 long (*fill)(void*, unsigned long),
345 long (*flush)(void*, unsigned long),
346 unsigned char *out_buf, long out_len,
347 long *pos,
348 void (*error)(char *x))
349 {
350 return __unzstd(buf, len, fill, flush, out_buf, out_len, pos, error);
351 }
352 #endif
353
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.