]> git.proxmox.com Git - ceph.git/blob - ceph/src/zstd/contrib/seekable_format/zstdseek_compress.c
update sources to ceph Nautilus 14.2.1
[ceph.git] / ceph / src / zstd / contrib / seekable_format / zstdseek_compress.c
1 /*
2 * Copyright (c) 2017-present, Facebook, Inc.
3 * All rights reserved.
4 *
5 * This source code is licensed under both the BSD-style license (found in the
6 * LICENSE file in the root directory of this source tree) and the GPLv2 (found
7 * in the COPYING file in the root directory of this source tree).
8 */
9
10 #include <stdlib.h> /* malloc, free */
11
12 #define XXH_STATIC_LINKING_ONLY
13 #define XXH_NAMESPACE ZSTD_
14 #include "xxhash.h"
15
16 #define ZSTD_STATIC_LINKING_ONLY
17 #include "zstd.h"
18 #include "zstd_errors.h"
19 #include "mem.h"
20 #include "zstd_seekable.h"
21
22 #define CHECK_Z(f) { size_t const ret = (f); if (ret != 0) return ret; }
23
24 #undef ERROR
25 #define ERROR(name) ((size_t)-ZSTD_error_##name)
26
27 #undef MIN
28 #undef MAX
29 #define MIN(a, b) ((a) < (b) ? (a) : (b))
30 #define MAX(a, b) ((a) > (b) ? (a) : (b))
31
32 typedef struct {
33 U32 cSize;
34 U32 dSize;
35 U32 checksum;
36 } framelogEntry_t;
37
38 struct ZSTD_frameLog_s {
39 framelogEntry_t* entries;
40 U32 size;
41 U32 capacity;
42
43 int checksumFlag;
44
45 /* for use when streaming out the seek table */
46 U32 seekTablePos;
47 U32 seekTableIndex;
48 } framelog_t;
49
50 struct ZSTD_seekable_CStream_s {
51 ZSTD_CStream* cstream;
52 ZSTD_frameLog framelog;
53
54 U32 frameCSize;
55 U32 frameDSize;
56
57 XXH64_state_t xxhState;
58
59 U32 maxFrameSize;
60
61 int writingSeekTable;
62 };
63
64 size_t ZSTD_seekable_frameLog_allocVec(ZSTD_frameLog* fl)
65 {
66 /* allocate some initial space */
67 size_t const FRAMELOG_STARTING_CAPACITY = 16;
68 fl->entries = (framelogEntry_t*)malloc(
69 sizeof(framelogEntry_t) * FRAMELOG_STARTING_CAPACITY);
70 if (fl->entries == NULL) return ERROR(memory_allocation);
71 fl->capacity = FRAMELOG_STARTING_CAPACITY;
72
73 return 0;
74 }
75
76 size_t ZSTD_seekable_frameLog_freeVec(ZSTD_frameLog* fl)
77 {
78 if (fl != NULL) free(fl->entries);
79 return 0;
80 }
81
82 ZSTD_frameLog* ZSTD_seekable_createFrameLog(int checksumFlag)
83 {
84 ZSTD_frameLog* fl = malloc(sizeof(ZSTD_frameLog));
85 if (fl == NULL) return NULL;
86
87 if (ZSTD_isError(ZSTD_seekable_frameLog_allocVec(fl))) {
88 free(fl);
89 return NULL;
90 }
91
92 fl->checksumFlag = checksumFlag;
93 fl->seekTablePos = 0;
94 fl->seekTableIndex = 0;
95 fl->size = 0;
96
97 return fl;
98 }
99
100 size_t ZSTD_seekable_freeFrameLog(ZSTD_frameLog* fl)
101 {
102 ZSTD_seekable_frameLog_freeVec(fl);
103 free(fl);
104 return 0;
105 }
106
107 ZSTD_seekable_CStream* ZSTD_seekable_createCStream()
108 {
109 ZSTD_seekable_CStream* zcs = malloc(sizeof(ZSTD_seekable_CStream));
110
111 if (zcs == NULL) return NULL;
112
113 memset(zcs, 0, sizeof(*zcs));
114
115 zcs->cstream = ZSTD_createCStream();
116 if (zcs->cstream == NULL) goto failed1;
117
118 if (ZSTD_isError(ZSTD_seekable_frameLog_allocVec(&zcs->framelog))) goto failed2;
119
120 return zcs;
121
122 failed2:
123 ZSTD_freeCStream(zcs->cstream);
124 failed1:
125 free(zcs);
126 return NULL;
127 }
128
129 size_t ZSTD_seekable_freeCStream(ZSTD_seekable_CStream* zcs)
130 {
131 if (zcs == NULL) return 0; /* support free on null */
132 ZSTD_freeCStream(zcs->cstream);
133 ZSTD_seekable_frameLog_freeVec(&zcs->framelog);
134 free(zcs);
135
136 return 0;
137 }
138
139 size_t ZSTD_seekable_initCStream(ZSTD_seekable_CStream* zcs,
140 int compressionLevel,
141 int checksumFlag,
142 U32 maxFrameSize)
143 {
144 zcs->framelog.size = 0;
145 zcs->frameCSize = 0;
146 zcs->frameDSize = 0;
147
148 /* make sure maxFrameSize has a reasonable value */
149 if (maxFrameSize > ZSTD_SEEKABLE_MAX_FRAME_DECOMPRESSED_SIZE) {
150 return ERROR(compressionParameter_unsupported);
151 }
152
153 zcs->maxFrameSize = maxFrameSize
154 ? maxFrameSize
155 : ZSTD_SEEKABLE_MAX_FRAME_DECOMPRESSED_SIZE;
156
157 zcs->framelog.checksumFlag = checksumFlag;
158 if (zcs->framelog.checksumFlag) {
159 XXH64_reset(&zcs->xxhState, 0);
160 }
161
162 zcs->framelog.seekTablePos = 0;
163 zcs->framelog.seekTableIndex = 0;
164 zcs->writingSeekTable = 0;
165
166 return ZSTD_initCStream(zcs->cstream, compressionLevel);
167 }
168
169 size_t ZSTD_seekable_logFrame(ZSTD_frameLog* fl,
170 unsigned compressedSize,
171 unsigned decompressedSize,
172 unsigned checksum)
173 {
174 if (fl->size == ZSTD_SEEKABLE_MAXFRAMES)
175 return ERROR(frameIndex_tooLarge);
176
177 /* grow the buffer if required */
178 if (fl->size == fl->capacity) {
179 /* exponential size increase for constant amortized runtime */
180 size_t const newCapacity = fl->capacity * 2;
181 framelogEntry_t* const newEntries = realloc(fl->entries,
182 sizeof(framelogEntry_t) * newCapacity);
183
184 if (newEntries == NULL) return ERROR(memory_allocation);
185
186 fl->entries = newEntries;
187 fl->capacity = newCapacity;
188 }
189
190 fl->entries[fl->size] = (framelogEntry_t){
191 compressedSize, decompressedSize, checksum
192 };
193 fl->size++;
194
195 return 0;
196 }
197
198 size_t ZSTD_seekable_endFrame(ZSTD_seekable_CStream* zcs, ZSTD_outBuffer* output)
199 {
200 size_t const prevOutPos = output->pos;
201 /* end the frame */
202 size_t ret = ZSTD_endStream(zcs->cstream, output);
203
204 zcs->frameCSize += output->pos - prevOutPos;
205
206 /* need to flush before doing the rest */
207 if (ret) return ret;
208
209 /* frame done */
210
211 /* store the frame data for later */
212 ret = ZSTD_seekable_logFrame(
213 &zcs->framelog, zcs->frameCSize, zcs->frameDSize,
214 zcs->framelog.checksumFlag
215 ? XXH64_digest(&zcs->xxhState) & 0xFFFFFFFFU
216 : 0);
217 if (ret) return ret;
218
219 /* reset for the next frame */
220 zcs->frameCSize = 0;
221 zcs->frameDSize = 0;
222
223 ZSTD_resetCStream(zcs->cstream, 0);
224 if (zcs->framelog.checksumFlag)
225 XXH64_reset(&zcs->xxhState, 0);
226
227 return 0;
228 }
229
230 size_t ZSTD_seekable_compressStream(ZSTD_seekable_CStream* zcs, ZSTD_outBuffer* output, ZSTD_inBuffer* input)
231 {
232 const BYTE* const inBase = (const BYTE*) input->src + input->pos;
233 size_t inLen = input->size - input->pos;
234
235 inLen = MIN(inLen, (size_t)(zcs->maxFrameSize - zcs->frameDSize));
236
237 /* if we haven't finished flushing the last frame, don't start writing a new one */
238 if (inLen > 0) {
239 ZSTD_inBuffer inTmp = { inBase, inLen, 0 };
240 size_t const prevOutPos = output->pos;
241
242 size_t const ret = ZSTD_compressStream(zcs->cstream, output, &inTmp);
243
244 if (zcs->framelog.checksumFlag) {
245 XXH64_update(&zcs->xxhState, inBase, inTmp.pos);
246 }
247
248 zcs->frameCSize += output->pos - prevOutPos;
249 zcs->frameDSize += inTmp.pos;
250
251 input->pos += inTmp.pos;
252
253 if (ZSTD_isError(ret)) return ret;
254 }
255
256 if (zcs->maxFrameSize == zcs->frameDSize) {
257 /* log the frame and start over */
258 size_t const ret = ZSTD_seekable_endFrame(zcs, output);
259 if (ZSTD_isError(ret)) return ret;
260
261 /* get the client ready for the next frame */
262 return (size_t)zcs->maxFrameSize;
263 }
264
265 return (size_t)(zcs->maxFrameSize - zcs->frameDSize);
266 }
267
268 static inline size_t ZSTD_seekable_seekTableSize(const ZSTD_frameLog* fl)
269 {
270 size_t const sizePerFrame = 8 + (fl->checksumFlag?4:0);
271 size_t const seekTableLen = ZSTD_skippableHeaderSize +
272 sizePerFrame * fl->size +
273 ZSTD_seekTableFooterSize;
274
275 return seekTableLen;
276 }
277
278 static inline size_t ZSTD_stwrite32(ZSTD_frameLog* fl,
279 ZSTD_outBuffer* output, U32 const value,
280 U32 const offset)
281 {
282 if (fl->seekTablePos < offset + 4) {
283 BYTE tmp[4]; /* so that we can work with buffers too small to write a whole word to */
284 size_t const lenWrite =
285 MIN(output->size - output->pos, offset + 4 - fl->seekTablePos);
286 MEM_writeLE32(tmp, value);
287 memcpy((BYTE*)output->dst + output->pos,
288 tmp + (fl->seekTablePos - offset), lenWrite);
289 output->pos += lenWrite;
290 fl->seekTablePos += lenWrite;
291
292 if (lenWrite < 4) return ZSTD_seekable_seekTableSize(fl) - fl->seekTablePos;
293 }
294 return 0;
295 }
296
297 size_t ZSTD_seekable_writeSeekTable(ZSTD_frameLog* fl, ZSTD_outBuffer* output)
298 {
299 /* seekTableIndex: the current index in the table and
300 * seekTableSize: the amount of the table written so far
301 *
302 * This function is written this way so that if it has to return early
303 * because of a small buffer, it can keep going where it left off.
304 */
305
306 size_t const sizePerFrame = 8 + (fl->checksumFlag?4:0);
307 size_t const seekTableLen = ZSTD_seekable_seekTableSize(fl);
308
309 CHECK_Z(ZSTD_stwrite32(fl, output, ZSTD_MAGIC_SKIPPABLE_START | 0xE, 0));
310 CHECK_Z(ZSTD_stwrite32(fl, output, seekTableLen - ZSTD_skippableHeaderSize,
311 4));
312
313 while (fl->seekTableIndex < fl->size) {
314 CHECK_Z(ZSTD_stwrite32(fl, output,
315 fl->entries[fl->seekTableIndex].cSize,
316 ZSTD_skippableHeaderSize +
317 sizePerFrame * fl->seekTableIndex + 0));
318
319 CHECK_Z(ZSTD_stwrite32(fl, output,
320 fl->entries[fl->seekTableIndex].dSize,
321 ZSTD_skippableHeaderSize +
322 sizePerFrame * fl->seekTableIndex + 4));
323
324 if (fl->checksumFlag) {
325 CHECK_Z(ZSTD_stwrite32(
326 fl, output, fl->entries[fl->seekTableIndex].checksum,
327 ZSTD_skippableHeaderSize +
328 sizePerFrame * fl->seekTableIndex + 8));
329 }
330
331 fl->seekTableIndex++;
332 }
333
334 CHECK_Z(ZSTD_stwrite32(fl, output, fl->size,
335 seekTableLen - ZSTD_seekTableFooterSize));
336
337 if (output->size - output->pos < 1) return seekTableLen - fl->seekTablePos;
338 if (fl->seekTablePos < seekTableLen - 4) {
339 BYTE sfd = 0;
340 sfd |= (fl->checksumFlag) << 7;
341
342 ((BYTE*)output->dst)[output->pos] = sfd;
343 output->pos++;
344 fl->seekTablePos++;
345 }
346
347 CHECK_Z(ZSTD_stwrite32(fl, output, ZSTD_SEEKABLE_MAGICNUMBER,
348 seekTableLen - 4));
349
350 if (fl->seekTablePos != seekTableLen) return ERROR(GENERIC);
351 return 0;
352 }
353
354 size_t ZSTD_seekable_endStream(ZSTD_seekable_CStream* zcs, ZSTD_outBuffer* output)
355 {
356 if (!zcs->writingSeekTable && zcs->frameDSize) {
357 const size_t endFrame = ZSTD_seekable_endFrame(zcs, output);
358 if (ZSTD_isError(endFrame)) return endFrame;
359 /* return an accurate size hint */
360 if (endFrame) return endFrame + ZSTD_seekable_seekTableSize(&zcs->framelog);
361 }
362
363 zcs->writingSeekTable = 1;
364
365 return ZSTD_seekable_writeSeekTable(&zcs->framelog, output);
366 }