/src/sleuthkit/tsk/fs/decmpfs.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | /* This file contains decompression routines used by APFS and HFS |
2 | | * It has one method derived from public domain ZLIB and others |
3 | | * that are TSK-specific. |
4 | | * |
5 | | * It would probably be cleaner to separate these into two files. |
6 | | */ |
7 | | |
8 | | #include <cstdint> |
9 | | #include <memory> |
10 | | #include <new> |
11 | | |
12 | | void error_detected(uint32_t errnum, const char* errstr, ...); |
13 | | void error_returned(const char* errstr, ...); |
14 | | |
15 | | #include "tsk/libtsk.h" |
16 | | #include "tsk_fs_i.h" |
17 | | #include "decmpfs.h" |
18 | | |
19 | | #ifdef HAVE_LIBZ |
20 | | #include <zlib.h> |
21 | | #endif |
22 | | |
23 | | #include "lzvn.h" |
24 | | #include "tsk_hfs.h" |
25 | | |
26 | | #ifdef HAVE_LIBZ |
27 | | |
28 | | /***************** ZLIB stuff *******************************/ |
29 | | |
30 | | /* The zlib_inflate method is adapted from the public domain |
31 | | * zpipe.c (part of zlib) at http://zlib.net/zpipe.c |
32 | | * |
33 | | * zpipe.c: example of proper use of zlib's inflate() and deflate() |
34 | | * Not copyrighted -- provided to the public domain |
35 | | * Version 1.4 11 December 2005 Mark Adler */ |
36 | | |
37 | 0 | #define CHUNK 16384 |
38 | | |
39 | | /* |
40 | | * Invokes the zlib library to inflate (uncompress) data. |
41 | | * |
42 | | * Returns and error code. Places the uncompressed data in a buffer supplied by the caller. Also |
43 | | * returns the uncompressed length, and the number of compressed bytes consumed. |
44 | | * |
45 | | * Will stop short of the end of compressed data, if a natural end of a compression unit is reached. Using |
46 | | * bytesConsumed, the caller can then advance the source pointer, and re-invoke the function. This will then |
47 | | * inflate the next following compression unit in the data stream. |
48 | | * |
49 | | * @param source - buffer of compressed data |
50 | | * @param sourceLen - length of the compressed data. |
51 | | * @param dest -- buffer to hold the uncompressed results |
52 | | * @param destLen -- length of the dest buffer |
53 | | * @param uncompressedLength -- return of the length of the uncompressed data found. |
54 | | * @param bytesConsumed -- return of the number of input bytes of compressed data used. |
55 | | * @return 0 on success, a negative number on error |
56 | | */ |
57 | | int |
58 | | zlib_inflate(char *source, uint64_t sourceLen, char *dest, uint64_t destLen, uint64_t * uncompressedLength, unsigned long *bytesConsumed) // this is unsigned long because that's what zlib uses. |
59 | 0 | { |
60 | |
|
61 | 0 | int ret; |
62 | 0 | unsigned have; |
63 | 0 | z_stream strm; |
64 | 0 | unsigned char in[CHUNK]; |
65 | 0 | unsigned char out[CHUNK]; |
66 | | |
67 | | // Some vars to help with copying bytes into "in" |
68 | 0 | char *srcPtr = source; |
69 | 0 | char *destPtr = dest; |
70 | 0 | uint64_t srcAvail = sourceLen; //uint64_t |
71 | 0 | uint64_t amtToCopy; |
72 | 0 | uint64_t copiedSoFar = 0; |
73 | | |
74 | | /* allocate inflate state */ |
75 | 0 | strm.zalloc = Z_NULL; |
76 | 0 | strm.zfree = Z_NULL; |
77 | 0 | strm.opaque = Z_NULL; |
78 | 0 | strm.avail_in = 0; |
79 | 0 | strm.next_in = Z_NULL; |
80 | 0 | ret = inflateInit(&strm); |
81 | 0 | if (ret != Z_OK) { |
82 | 0 | error_detected(TSK_ERR_FS_READ, |
83 | 0 | "zlib_inflate: failed to initialize inflation engine (%d)", |
84 | 0 | ret); |
85 | 0 | return ret; |
86 | 0 | } |
87 | | |
88 | | /* decompress until deflate stream ends or end of file */ |
89 | 0 | do { |
90 | | |
91 | | // Copy up to CHUNK bytes into "in" from source, advancing the pointer, and |
92 | | // setting strm.avail_in equal to the number of bytes copied. |
93 | 0 | if (srcAvail >= CHUNK) { |
94 | 0 | amtToCopy = CHUNK; |
95 | 0 | srcAvail -= CHUNK; |
96 | 0 | } |
97 | 0 | else { |
98 | 0 | amtToCopy = srcAvail; |
99 | 0 | srcAvail = 0; |
100 | 0 | } |
101 | | // wipe out any previous value, copy in the bytes, advance the pointer, record number of bytes. |
102 | 0 | memset(in, 0, CHUNK); |
103 | 0 | if (amtToCopy > SIZE_MAX || amtToCopy > UINT_MAX) { |
104 | 0 | error_detected(TSK_ERR_FS_READ, |
105 | 0 | "zlib_inflate: amtToCopy in one chunk is too large"); |
106 | 0 | return -100; |
107 | 0 | } |
108 | 0 | memcpy(in, srcPtr, (size_t) amtToCopy); // cast OK because of above test |
109 | 0 | srcPtr += amtToCopy; |
110 | 0 | strm.avail_in = (uInt) amtToCopy; // cast OK because of above test |
111 | |
|
112 | 0 | if (strm.avail_in == 0) |
113 | 0 | break; |
114 | 0 | strm.next_in = in; |
115 | | |
116 | | /* run inflate() on input until output buffer not full */ |
117 | 0 | do { |
118 | 0 | strm.avail_out = CHUNK; |
119 | 0 | strm.next_out = out; |
120 | 0 | ret = inflate(&strm, Z_NO_FLUSH); |
121 | 0 | if (ret == Z_NEED_DICT) |
122 | 0 | ret = Z_DATA_ERROR; // we don't have a custom dict |
123 | 0 | if (ret < 0 && ret != Z_BUF_ERROR) { // Z_BUF_ERROR is not fatal |
124 | 0 | error_detected(TSK_ERR_FS_READ, |
125 | 0 | " zlib_inflate: zlib returned error %d (%s)", ret, |
126 | 0 | strm.msg); |
127 | 0 | (void) inflateEnd(&strm); |
128 | 0 | return ret; |
129 | 0 | } |
130 | | |
131 | 0 | have = CHUNK - strm.avail_out; |
132 | | // Is there enough space in dest to copy the current chunk? |
133 | 0 | if (copiedSoFar + have > destLen) { |
134 | | // There is not enough space, so better return an error |
135 | 0 | error_detected(TSK_ERR_FS_READ, |
136 | 0 | " zlib_inflate: not enough space in inflation destination\n"); |
137 | 0 | (void) inflateEnd(&strm); |
138 | 0 | return -200; |
139 | 0 | } |
140 | | |
141 | | // Copy "have" bytes from out to destPtr, advance destPtr |
142 | 0 | memcpy(destPtr, out, have); |
143 | 0 | destPtr += have; |
144 | 0 | copiedSoFar += have; |
145 | |
|
146 | 0 | } while (strm.avail_out == 0 && ret != Z_STREAM_END); |
147 | | |
148 | | |
149 | | /* done when inflate() says it's done */ |
150 | 0 | } while (ret != Z_STREAM_END); |
151 | | |
152 | 0 | if (ret == Z_STREAM_END) |
153 | 0 | *uncompressedLength = copiedSoFar; |
154 | |
|
155 | 0 | *bytesConsumed = strm.total_in; |
156 | | /* clean up and return */ |
157 | 0 | (void) inflateEnd(&strm); |
158 | 0 | return ret == Z_STREAM_END ? Z_OK : Z_DATA_ERROR; |
159 | 0 | } |
160 | | |
161 | | #endif |
162 | | |
163 | | |
164 | | |
165 | | /********************* TSK STUFF **********************/ |
166 | | |
167 | | /* |
168 | | * The Sleuth Kit |
169 | | * |
170 | | * Brian Carrier [carrier <at> sleuthkit [dot] org] |
171 | | * Copyright (c) 2019-2020 Brian Carrier. All Rights reserved |
172 | | * Copyright (c) 2018-2019 BlackBag Technologies. All Rights reserved |
173 | | * |
174 | | * This software is distributed under the Common Public License 1.0 |
175 | | */ |
176 | | |
177 | | typedef struct { |
178 | | uint32_t offset; |
179 | | uint32_t length; |
180 | | } CMP_OFFSET_ENTRY; |
181 | | |
182 | | /** |
183 | | * \internal |
184 | | * Reads the ZLIB compression block table from the attribute. |
185 | | * |
186 | | * @param rAtttr the attribute to read |
187 | | * @param tableSizeOut size of block table |
188 | | * @param tableOffsetOut the offset of the block table in the resource fork |
189 | | * @return 1 on success, 0 on error |
190 | | */ |
191 | | std::unique_ptr<CMP_OFFSET_ENTRY[]> |
192 | | decmpfs_read_zlib_block_table( |
193 | | const TSK_FS_ATTR *rAttr, |
194 | | uint32_t* tableSizeOut, |
195 | | uint32_t* tableOffsetOut) |
196 | 0 | { |
197 | 0 | ssize_t attrReadResult; |
198 | 0 | hfs_resource_fork_header rfHeader; |
199 | 0 | uint32_t dataOffset; |
200 | 0 | uint32_t offsetTableOffset; |
201 | 0 | char fourBytes[4]; // Size of the offset table, little endian |
202 | 0 | uint32_t tableSize; // Size of the offset table |
203 | 0 | size_t indx; |
204 | | |
205 | | // Read the resource fork header |
206 | 0 | attrReadResult = tsk_fs_attr_read(rAttr, 0, (char *) &rfHeader, |
207 | 0 | sizeof(hfs_resource_fork_header), TSK_FS_FILE_READ_FLAG_NONE); |
208 | 0 | if (attrReadResult != sizeof(hfs_resource_fork_header)) { |
209 | 0 | error_returned |
210 | 0 | (" %s: trying to read the resource fork header", __func__); |
211 | 0 | return nullptr; |
212 | 0 | } |
213 | | |
214 | | // Begin to parse the resource fork. For now, we just need the data offset. |
215 | 0 | dataOffset = tsk_getu32(TSK_BIG_ENDIAN, rfHeader.dataOffset); |
216 | | |
217 | | // The resource's data begins with an offset table, which defines blocks |
218 | | // of (optionally) zlib-compressed data (so that the OS can do file seeks |
219 | | // efficiently; each uncompressed block is 64KB). |
220 | 0 | offsetTableOffset = dataOffset + 4; |
221 | | |
222 | | // read 4 bytes, the number of table entries, little endian |
223 | 0 | attrReadResult = |
224 | 0 | tsk_fs_attr_read(rAttr, offsetTableOffset, fourBytes, 4, |
225 | 0 | TSK_FS_FILE_READ_FLAG_NONE); |
226 | 0 | if (attrReadResult != 4) { |
227 | 0 | error_returned |
228 | 0 | (" %s: trying to read the offset table size, " |
229 | 0 | "return value of %u should have been 4", __func__, attrReadResult); |
230 | 0 | return nullptr; |
231 | 0 | } |
232 | 0 | tableSize = tsk_getu32(TSK_LIT_ENDIAN, fourBytes); |
233 | |
|
234 | 0 | if (tableSize <= 0) { |
235 | 0 | error_returned |
236 | 0 | (" %s: table size is zero", __func__); |
237 | 0 | return nullptr; |
238 | 0 | } |
239 | | |
240 | | // Each table entry is 8 bytes long |
241 | 0 | std::unique_ptr<char[]> offsetTableData{new(std::nothrow) char[tableSize * 8]}; |
242 | 0 | if (!offsetTableData) { |
243 | 0 | error_returned |
244 | 0 | (" %s: space for the offset table raw data", __func__); |
245 | 0 | return nullptr; |
246 | 0 | } |
247 | | |
248 | 0 | std::unique_ptr<CMP_OFFSET_ENTRY[]> offsetTable{new(std::nothrow) CMP_OFFSET_ENTRY[tableSize]}; |
249 | 0 | if (!offsetTable) { |
250 | 0 | error_returned |
251 | 0 | (" %s: space for the offset table", __func__); |
252 | 0 | return nullptr; |
253 | 0 | } |
254 | | |
255 | 0 | attrReadResult = tsk_fs_attr_read(rAttr, offsetTableOffset + 4, |
256 | 0 | offsetTableData.get(), tableSize * 8, TSK_FS_FILE_READ_FLAG_NONE); |
257 | 0 | if (attrReadResult != (ssize_t) tableSize * 8) { |
258 | 0 | error_returned |
259 | 0 | (" %s: reading in the compression offset table, " |
260 | 0 | "return value %u should have been %u", __func__, attrReadResult, |
261 | 0 | tableSize * 8); |
262 | 0 | return nullptr; |
263 | 0 | } |
264 | | |
265 | 0 | for (indx = 0; indx < tableSize; ++indx) { |
266 | 0 | offsetTable[indx].offset = |
267 | 0 | tsk_getu32(TSK_LIT_ENDIAN, offsetTableData.get() + indx * 8); |
268 | 0 | offsetTable[indx].length = |
269 | 0 | tsk_getu32(TSK_LIT_ENDIAN, offsetTableData.get() + indx * 8 + 4); |
270 | 0 | } |
271 | |
|
272 | 0 | *tableSizeOut = tableSize; |
273 | 0 | *tableOffsetOut = offsetTableOffset; |
274 | 0 | return offsetTable; |
275 | 0 | } |
276 | | |
277 | | /** |
278 | | * \internal |
279 | | * Reads the LZVN compression block table from the attribute. |
280 | | * |
281 | | * @param rAtttr the attribute to read |
282 | | * @param offsetTableOut block table |
283 | | * @param tableSizeOut size of block table |
284 | | * @param tableOffsetOut the offset of the block table in the resource fork |
285 | | * @return 1 on success, 0 on error |
286 | | */ |
287 | | std::unique_ptr<CMP_OFFSET_ENTRY[]> |
288 | | decmpfs_read_lzvn_block_table( |
289 | | const TSK_FS_ATTR *rAttr, |
290 | | uint32_t* tableSizeOut, |
291 | | uint32_t* tableOffsetOut) |
292 | 0 | { |
293 | | // The offset table is a sequence of 4-byte offsets of compressed |
294 | | // blocks. The first 4 bytes is thus the offset of the first block, |
295 | | // but also 4 times the number of entries in the table. |
296 | 0 | char fourBytes[4]; |
297 | 0 | ssize_t attrReadResult = tsk_fs_attr_read(rAttr, 0, fourBytes, 4, |
298 | 0 | TSK_FS_FILE_READ_FLAG_NONE); |
299 | 0 | if (attrReadResult != 4) { |
300 | 0 | error_returned |
301 | 0 | (" %s: trying to read the offset table size, " |
302 | 0 | "return value of %u should have been 4", __func__, attrReadResult); |
303 | 0 | return nullptr; |
304 | 0 | } |
305 | | |
306 | 0 | const uint32_t tableDataSize = tsk_getu32(TSK_LIT_ENDIAN, fourBytes); |
307 | |
|
308 | 0 | if (tableDataSize <= 0) { |
309 | 0 | error_returned |
310 | 0 | (" %s: table size is zero", __func__); |
311 | 0 | return nullptr; |
312 | 0 | } |
313 | | |
314 | 0 | std::unique_ptr<char[]> offsetTableData(new(std::nothrow) char[tableDataSize]); |
315 | 0 | if (!offsetTableData) { |
316 | 0 | error_returned |
317 | 0 | (" %s: space for the offset table raw data", __func__); |
318 | 0 | return nullptr; |
319 | 0 | } |
320 | | |
321 | | // Size of the offset table |
322 | | // table entries are 4 bytes, last entry is end of data |
323 | 0 | const uint32_t tableSize = tableDataSize / 4 - 1; |
324 | |
|
325 | 0 | std::unique_ptr<CMP_OFFSET_ENTRY[]> offsetTable(new(std::nothrow) CMP_OFFSET_ENTRY[tableSize]); |
326 | 0 | if (!offsetTable) { |
327 | 0 | error_returned |
328 | 0 | (" %s: space for the offset table", __func__); |
329 | 0 | return nullptr; |
330 | 0 | } |
331 | | |
332 | 0 | attrReadResult = tsk_fs_attr_read(rAttr, 0, |
333 | 0 | offsetTableData.get(), tableDataSize, TSK_FS_FILE_READ_FLAG_NONE); |
334 | 0 | if (attrReadResult != (ssize_t) tableDataSize) { |
335 | 0 | error_returned |
336 | 0 | (" %s: reading in the compression offset table, " |
337 | 0 | "return value %u should have been %u", __func__, attrReadResult, |
338 | 0 | tableDataSize); |
339 | 0 | return nullptr; |
340 | 0 | } |
341 | | |
342 | 0 | uint32_t a = tableDataSize; |
343 | 0 | uint32_t b; |
344 | 0 | size_t i; |
345 | |
|
346 | 0 | for (i = 0; i < tableSize; ++i) { |
347 | 0 | b = tsk_getu32(TSK_LIT_ENDIAN, offsetTableData.get() + 4*(i+1)); |
348 | 0 | offsetTable[i].offset = a; |
349 | 0 | offsetTable[i].length = b - a; |
350 | 0 | a = b; |
351 | 0 | } |
352 | |
|
353 | 0 | *tableSizeOut = tableSize; |
354 | 0 | *tableOffsetOut = 0; |
355 | 0 | return offsetTable; |
356 | 0 | } |
357 | | |
358 | | /** |
359 | | * \internal |
360 | | * "Decompress" a block which was stored uncompressed. |
361 | | * |
362 | | * @param rawBuf the compressed data |
363 | | * @param len length of the compressed data |
364 | | * @param uncBuf the decompressed data |
365 | | * @param uncLen length of the decompressed data |
366 | | * @return 1 on success, 0 on error |
367 | | */ |
368 | 0 | static int decmpfs_decompress_noncompressed_block(char* rawBuf, uint32_t len, char* uncBuf, uint64_t* uncLen) { |
369 | | // actually an uncompressed block of data; just copy |
370 | 0 | if (tsk_verbose) |
371 | 0 | tsk_fprintf(stderr, |
372 | 0 | "%s: Copying an uncompressed compression unit\n", __func__); |
373 | |
|
374 | 0 | if ((len - 1) > COMPRESSION_UNIT_SIZE) { |
375 | 0 | error_detected(TSK_ERR_FS_READ, |
376 | 0 | "%s: uncompressed block length %u is longer " |
377 | 0 | "than compression unit size %u", __func__, len - 1, |
378 | 0 | COMPRESSION_UNIT_SIZE); |
379 | 0 | return 0; |
380 | 0 | } |
381 | 0 | memcpy(uncBuf, rawBuf + 1, len - 1); |
382 | 0 | *uncLen = len - 1; |
383 | 0 | return 1; |
384 | 0 | } |
385 | | |
386 | | |
387 | | #ifdef HAVE_LIBZ |
388 | | /** |
389 | | * \internal |
390 | | * Decompress a block which was stored with ZLIB. |
391 | | * |
392 | | * @param rawBuf the compressed data |
393 | | * @param len length of the compressed data |
394 | | * @param uncBuf the decompressed data |
395 | | * @param uncLen length of the decompressed data |
396 | | * @return 1 on success, 0 on error |
397 | | */ |
398 | | static int decmpfs_decompress_zlib_block(char* rawBuf, uint32_t len, char* uncBuf, uint64_t* uncLen) |
399 | 0 | { |
400 | | // see if this block is compressed |
401 | 0 | if (len > 0 && (rawBuf[0] & 0x0F) != 0x0F) { |
402 | | // Uncompress the chunk of data |
403 | 0 | if (tsk_verbose) |
404 | 0 | tsk_fprintf(stderr, |
405 | 0 | "%s: Inflating the compression unit\n", __func__); |
406 | |
|
407 | 0 | unsigned long bytesConsumed; |
408 | 0 | int infResult = zlib_inflate(rawBuf, (uint64_t) len, |
409 | 0 | uncBuf, (uint64_t) COMPRESSION_UNIT_SIZE, |
410 | 0 | uncLen, &bytesConsumed); |
411 | 0 | if (infResult != 0) { |
412 | 0 | error_returned |
413 | 0 | (" %s: zlib inflation (uncompression) failed", |
414 | 0 | __func__, infResult); |
415 | 0 | return 0; |
416 | 0 | } |
417 | | |
418 | 0 | if (bytesConsumed != len) { |
419 | 0 | error_detected(TSK_ERR_FS_READ, |
420 | 0 | " %s, decompressor did not consume the whole compressed data", |
421 | 0 | __func__); |
422 | 0 | return 0; |
423 | 0 | } |
424 | | |
425 | 0 | return 1; |
426 | 0 | } |
427 | 0 | else { |
428 | | // actually an uncompressed block of data; just copy |
429 | 0 | return decmpfs_decompress_noncompressed_block(rawBuf, len, uncBuf, uncLen); |
430 | 0 | } |
431 | 0 | } |
432 | | #endif |
433 | | |
434 | | /** |
435 | | * \internal |
436 | | * Decompress a block which was stored with LZVN. |
437 | | * |
438 | | * @param rawBuf the compressed data |
439 | | * @param len length of the compressed data |
440 | | * @param uncBuf the decompressed data |
441 | | * @param uncLen length of the decompressed data |
442 | | * @return 1 on success, 0 on error |
443 | | */ |
444 | | static int decmpfs_decompress_lzvn_block(char* rawBuf, uint32_t len, char* uncBuf, uint64_t* uncLen) |
445 | 0 | { |
446 | | // see if this block is compressed |
447 | 0 | if (len > 0 && rawBuf[0] != 0x06) { |
448 | 0 | *uncLen = lzvn_decode_buffer(uncBuf, COMPRESSION_UNIT_SIZE, rawBuf, len); |
449 | 0 | return 1; // apparently this can't fail |
450 | 0 | } |
451 | 0 | else { |
452 | | // actually an uncompressed block of data; just copy |
453 | 0 | return decmpfs_decompress_noncompressed_block(rawBuf, len, uncBuf, uncLen); |
454 | 0 | } |
455 | 0 | } |
456 | | |
457 | | /** |
458 | | * \internal |
459 | | * Decompress a block. |
460 | | * |
461 | | * @param rAttr the attribute to read |
462 | | * @param rawBuf the compressed data |
463 | | * @param uncBuf the decompressed data |
464 | | * @param offsetTable table of compressed block offsets |
465 | | * @param offsetTableSize size of table of compressed block offsets |
466 | | * @param offsetTableOffset offset of table of compressed block offsets |
467 | | * @param indx index of block to read |
468 | | * @param decompress_block pointer to decompression function |
469 | | * @return decompressed size on success, -1 on error |
470 | | */ |
471 | | static ssize_t read_and_decompress_block( |
472 | | const TSK_FS_ATTR* rAttr, |
473 | | char* rawBuf, |
474 | | char* uncBuf, |
475 | | const CMP_OFFSET_ENTRY* offsetTable, |
476 | | [[maybe_unused]] uint32_t offsetTableSize, |
477 | | uint32_t offsetTableOffset, |
478 | | size_t indx, |
479 | | int (*decompress_block)(char* rawBuf, |
480 | | uint32_t len, |
481 | | char* uncBuf, |
482 | | uint64_t* uncLen) |
483 | | ) |
484 | 0 | { |
485 | | // @@@ BC: Looks like we should have bounds checks that indx < offsetTableSize, but we should confirm |
486 | 0 | ssize_t attrReadResult; |
487 | 0 | uint32_t offset = offsetTableOffset + offsetTable[indx].offset; |
488 | 0 | uint32_t len = offsetTable[indx].length; |
489 | 0 | uint64_t uncLen; |
490 | |
|
491 | 0 | if (tsk_verbose) |
492 | 0 | tsk_fprintf(stderr, |
493 | 0 | "%s: Reading compression unit %d, length %d\n", |
494 | 0 | __func__, indx, len); |
495 | | |
496 | | /* Github #383 referenced that if len is 0, then the below code causes |
497 | | * problems. Added this check, but I don't have data to verify this on. |
498 | | * it looks like it should at least not crash, but it isn't clear if it |
499 | | * will also do the right thing and if should actually break here |
500 | | * instead. */ |
501 | 0 | if (len == 0) { |
502 | 0 | return 0; |
503 | 0 | } |
504 | | |
505 | 0 | if (len > COMPRESSION_UNIT_SIZE + 1) { |
506 | 0 | error_detected(TSK_ERR_FS_READ, |
507 | 0 | "%s: block size is too large: %u", __func__, len); |
508 | 0 | return -1; |
509 | 0 | } |
510 | | |
511 | | // Read in the block of compressed data |
512 | 0 | attrReadResult = tsk_fs_attr_read(rAttr, offset, |
513 | 0 | rawBuf, len, TSK_FS_FILE_READ_FLAG_NONE); |
514 | 0 | if (attrReadResult != (ssize_t) len) { |
515 | 0 | char msg[] = |
516 | 0 | "%s%s: reading in the compression offset table, " |
517 | 0 | "return value %u should have been %u"; |
518 | |
|
519 | 0 | if (attrReadResult < 0 ) { |
520 | 0 | error_returned(msg, " ", __func__, attrReadResult, len); |
521 | 0 | } |
522 | 0 | else { |
523 | 0 | error_detected(TSK_ERR_FS_READ, "", __func__, attrReadResult, len); |
524 | 0 | } |
525 | 0 | return -1; |
526 | 0 | } |
527 | | |
528 | 0 | if (!decompress_block(rawBuf, len, uncBuf, &uncLen)) { |
529 | 0 | return -1; |
530 | 0 | } |
531 | | /* |
532 | | // If size is a multiple of COMPRESSION_UNIT_SIZE, |
533 | | // expected uncompressed length is COMPRESSION_UNIT_SIZE |
534 | | const uint32_t expUncLen = indx == offsetTableSize - 1 ? |
535 | | ((rAttr->fs_file->meta->size - 1) % COMPRESSION_UNIT_SIZE) + 1 : |
536 | | COMPRESSION_UNIT_SIZE; |
537 | | |
538 | | if (uncLen != expUncLen) { |
539 | | error_detected(TSK_ERR_FS_READ, |
540 | | "%s: compressed block decompressed to %u bytes, " |
541 | | "should have been %u bytes", __func__, uncLen, expUncLen); |
542 | | return -1; |
543 | | } |
544 | | */ |
545 | | |
546 | | // There are now uncLen bytes of uncompressed data available from |
547 | | // this comp unit. |
548 | 0 | return (ssize_t)uncLen; |
549 | 0 | } |
550 | | |
551 | | /** |
552 | | * \internal |
553 | | * Attr walk callback function for compressed resources |
554 | | * |
555 | | * @param fs_attr the attribute to read |
556 | | * @param flags |
557 | | * @param a_action action callback |
558 | | * @param ptr context for the action callback |
559 | | * @param read_block_table pointer to block table read function |
560 | | * @param decompress_block pointer to decompression function |
561 | | * @return 0 on success, 1 on error |
562 | | */ |
563 | | static uint8_t |
564 | | decmpfs_attr_walk_compressed_rsrc( |
565 | | const TSK_FS_ATTR * fs_attr, |
566 | | [[maybe_unused]] int flags, |
567 | | TSK_FS_FILE_WALK_CB a_action, |
568 | | void *ptr, |
569 | | std::unique_ptr<CMP_OFFSET_ENTRY[]> (*read_block_table)( |
570 | | const TSK_FS_ATTR *rAttr, |
571 | | uint32_t* tableSizeOut, |
572 | | uint32_t* tableOffsetOut |
573 | | ), |
574 | | int (*decompress_block)( |
575 | | char* rawBuf, |
576 | | uint32_t len, |
577 | | char* uncBuf, |
578 | | uint64_t* uncLen |
579 | | ) |
580 | | ) |
581 | 0 | { |
582 | 0 | if (tsk_verbose) |
583 | 0 | tsk_fprintf(stderr, |
584 | 0 | "%s: Entered, because this is a compressed file with compressed data in the resource fork\n", __func__); |
585 | | |
586 | | // clean up any error messages that are lying around |
587 | 0 | tsk_error_reset(); |
588 | 0 | if ((fs_attr == NULL) || (fs_attr->fs_file == NULL) |
589 | 0 | || (fs_attr->fs_file->meta == NULL) |
590 | 0 | || (fs_attr->fs_file->fs_info == NULL)) { |
591 | 0 | tsk_error_set_errno(TSK_ERR_FS_ARG); |
592 | 0 | tsk_error_set_errstr("%s: Null arguments given\n", __func__); |
593 | 0 | return 1; |
594 | 0 | } |
595 | | |
596 | | // Check that the ATTR being read is the main DATA resource, 128-0, |
597 | | // because this is the only one that can be compressed in HFS+ |
598 | 0 | if ((fs_attr->id != HFS_FS_ATTR_ID_DATA) || |
599 | 0 | (fs_attr->type != TSK_FS_ATTR_TYPE_HFS_DATA)) { |
600 | 0 | error_detected(TSK_ERR_FS_ARG, |
601 | 0 | "%s: arg specified an attribute %u-%u that is not the data fork, " |
602 | 0 | "Only the data fork can be compressed.", __func__, fs_attr->type, |
603 | 0 | fs_attr->id); |
604 | 0 | return 1; |
605 | 0 | } |
606 | | |
607 | | /* This MUST be a compressed attribute */ |
608 | 0 | if (!(fs_attr->flags & TSK_FS_ATTR_COMP)) { |
609 | 0 | error_detected(TSK_ERR_FS_FWALK, |
610 | 0 | "%s: called with non-special attribute: %x", |
611 | 0 | __func__, fs_attr->flags); |
612 | 0 | return 1; |
613 | 0 | } |
614 | | |
615 | 0 | TSK_FS_INFO* fs = fs_attr->fs_file->fs_info; |
616 | 0 | TSK_FS_FILE* fs_file = fs_attr->fs_file; |
617 | | |
618 | | /******** Open the Resource Fork ***********/ |
619 | | |
620 | | // find the attribute for the resource fork |
621 | 0 | const TSK_FS_ATTR* rAttr = |
622 | 0 | tsk_fs_file_attr_get_type(fs_file, TSK_FS_ATTR_TYPE_HFS_RSRC, |
623 | 0 | HFS_FS_ATTR_ID_RSRC, FALSE); |
624 | 0 | if (rAttr == NULL) { |
625 | 0 | error_returned |
626 | 0 | (" %s: could not get the attribute for the resource fork of the file", __func__); |
627 | 0 | return 1; |
628 | 0 | } |
629 | | |
630 | 0 | uint32_t offsetTableOffset; |
631 | 0 | uint32_t offsetTableSize; // The number of table entries |
632 | | |
633 | | // read the offset table from the fork header |
634 | 0 | std::unique_ptr<CMP_OFFSET_ENTRY[]> offsetTable = read_block_table( |
635 | 0 | rAttr, &offsetTableSize, &offsetTableOffset |
636 | 0 | ); |
637 | 0 | if (!offsetTable) { |
638 | 0 | return 1; |
639 | 0 | } |
640 | | |
641 | | // Allocate two buffers for the raw and uncompressed data |
642 | | /* Raw data can be COMPRESSION_UNIT_SIZE+1 if the data is not |
643 | | * compressed and there is a 1-byte flag that indicates that |
644 | | * the data is not compressed. */ |
645 | 0 | std::unique_ptr<char[]> rawBuf{new(std::nothrow) char[COMPRESSION_UNIT_SIZE + 1]}; |
646 | 0 | if (!rawBuf) { |
647 | 0 | error_returned |
648 | 0 | (" %s: buffers for reading and uncompressing", __func__); |
649 | 0 | return 1; |
650 | 0 | } |
651 | | |
652 | 0 | std::unique_ptr<char[]> uncBuf{new(std::nothrow) char[COMPRESSION_UNIT_SIZE]}; |
653 | 0 | if (!uncBuf) { |
654 | 0 | error_returned |
655 | 0 | (" %s: buffers for reading and uncompressing", __func__); |
656 | 0 | return 1; |
657 | 0 | } |
658 | | |
659 | 0 | TSK_OFF_T off = 0; // the offset in the uncompressed data stream consumed thus far |
660 | | |
661 | | // FOR entry in the table DO |
662 | 0 | for (size_t indx = 0; indx < offsetTableSize; ++indx) { |
663 | 0 | ssize_t uncLen; // uncompressed length |
664 | 0 | unsigned int blockSize; |
665 | 0 | uint64_t lumpSize; |
666 | 0 | uint64_t remaining; |
667 | 0 | char *lumpStart; |
668 | |
|
669 | 0 | switch ((uncLen = read_and_decompress_block( |
670 | 0 | rAttr, rawBuf.get(), uncBuf.get(), |
671 | 0 | offsetTable.get(), offsetTableSize, offsetTableOffset, indx, |
672 | 0 | decompress_block))) |
673 | 0 | { |
674 | 0 | case -1: |
675 | 0 | return 1; |
676 | 0 | case 0: |
677 | 0 | continue; |
678 | 0 | default: |
679 | 0 | break; |
680 | 0 | } |
681 | | |
682 | | // Call the a_action callback with "Lumps" |
683 | | // that are at most the block size. |
684 | 0 | blockSize = fs->block_size; |
685 | 0 | remaining = uncLen; |
686 | 0 | lumpStart = uncBuf.get(); |
687 | |
|
688 | 0 | while (remaining > 0) { |
689 | 0 | int retval; // action return value |
690 | 0 | lumpSize = remaining <= blockSize ? remaining : blockSize; |
691 | | |
692 | | // Apply the callback function |
693 | 0 | if (tsk_verbose) |
694 | 0 | tsk_fprintf(stderr, |
695 | 0 | "%s: Calling action on lump of size %" |
696 | 0 | PRIu64 " offset %" PRIu64 " in the compression unit\n", |
697 | 0 | __func__, lumpSize, uncLen - remaining); |
698 | 0 | if (lumpSize > SIZE_MAX) { |
699 | 0 | error_detected(TSK_ERR_FS_FWALK, |
700 | 0 | " %s: lumpSize is too large for the action", __func__); |
701 | 0 | return 1; |
702 | 0 | } |
703 | | |
704 | 0 | retval = a_action(fs_attr->fs_file, off, 0, lumpStart, |
705 | 0 | (size_t) lumpSize, // cast OK because of above test |
706 | 0 | TSK_FS_BLOCK_FLAG_COMP, ptr); |
707 | |
|
708 | 0 | if (retval == TSK_WALK_ERROR) { |
709 | 0 | error_detected(TSK_ERR_FS | 201, |
710 | 0 | "%s: callback returned an error", __func__); |
711 | 0 | return 1; |
712 | 0 | } |
713 | 0 | else if (retval == TSK_WALK_STOP) { |
714 | 0 | break; |
715 | 0 | } |
716 | | |
717 | | // Find the next lump |
718 | 0 | off += lumpSize; |
719 | 0 | remaining -= lumpSize; |
720 | 0 | lumpStart += lumpSize; |
721 | 0 | } |
722 | 0 | } |
723 | | |
724 | 0 | return 0; |
725 | 0 | } |
726 | | |
727 | | |
728 | | #ifdef HAVE_LIBZ |
729 | | /** |
730 | | * \internal |
731 | | * Attr walk callback function for ZLIB compressed resources |
732 | | * |
733 | | * @param fs_attr the attribute to read |
734 | | * @param flags |
735 | | * @param a_action action callback |
736 | | * @param ptr context for the action callback |
737 | | * @return 0 on success, 1 on error |
738 | | */ |
739 | | uint8_t |
740 | | decmpfs_attr_walk_zlib_rsrc(const TSK_FS_ATTR * fs_attr, |
741 | | int flags, TSK_FS_FILE_WALK_CB a_action, void *ptr) |
742 | 0 | { |
743 | 0 | return decmpfs_attr_walk_compressed_rsrc( |
744 | 0 | fs_attr, flags, a_action, ptr, |
745 | 0 | decmpfs_read_zlib_block_table, |
746 | 0 | decmpfs_decompress_zlib_block |
747 | 0 | ); |
748 | 0 | } |
749 | | #endif |
750 | | |
751 | | /** |
752 | | * \internal |
753 | | * Attr walk callback function for LZVN compressed resources |
754 | | * |
755 | | * @param fs_attr the attribute to read |
756 | | * @param flags |
757 | | * @param a_action action callback |
758 | | * @param ptr context for the action callback |
759 | | * @return 0 on success, 1 on error |
760 | | */ |
761 | | uint8_t |
762 | | decmpfs_attr_walk_lzvn_rsrc(const TSK_FS_ATTR * fs_attr, |
763 | | int flags, TSK_FS_FILE_WALK_CB a_action, void *ptr) |
764 | 0 | { |
765 | 0 | return decmpfs_attr_walk_compressed_rsrc( |
766 | 0 | fs_attr, flags, a_action, ptr, |
767 | 0 | decmpfs_read_lzvn_block_table, |
768 | 0 | decmpfs_decompress_lzvn_block |
769 | 0 | ); |
770 | 0 | } |
771 | | |
772 | | |
773 | | /** |
774 | | * \internal |
775 | | * Read a compressed resource |
776 | | * |
777 | | * @param fs_attr the attribute to read |
778 | | * @param a_offset the offset from which to read |
779 | | * @param a_buf the buffer into which to read |
780 | | * @param a_len the length of the buffer |
781 | | * @param read_block_table pointer to block table read function |
782 | | * @param decompress_block pointer to decompression function |
783 | | * @return number of bytes read or -1 on error (incl if offset is past EOF) |
784 | | */ |
785 | | static ssize_t |
786 | | decmpfs_file_read_compressed_rsrc( |
787 | | const TSK_FS_ATTR * a_fs_attr, |
788 | | TSK_OFF_T a_offset, |
789 | | char *a_buf, |
790 | | size_t a_len, |
791 | | std::unique_ptr<CMP_OFFSET_ENTRY[]> (*read_block_table)( |
792 | | const TSK_FS_ATTR *rAttr, |
793 | | uint32_t* tableSizeOut, |
794 | | uint32_t* tableOffsetOut), |
795 | | int (*decompress_block)( |
796 | | char* rawBuf, |
797 | | uint32_t len, |
798 | | char* uncBuf, |
799 | | uint64_t* uncLen) |
800 | | ) |
801 | 0 | { |
802 | 0 | TSK_FS_FILE *fs_file; |
803 | 0 | const TSK_FS_ATTR *rAttr; |
804 | 0 | uint32_t offsetTableOffset; |
805 | 0 | uint32_t offsetTableSize; // Size of the offset table |
806 | 0 | TSK_OFF_T indx; // index for looping over the offset table |
807 | 0 | TSK_OFF_T startUnit = 0; |
808 | 0 | uint32_t startUnitOffset = 0; |
809 | 0 | TSK_OFF_T endUnit = 0; |
810 | 0 | uint64_t bytesCopied; |
811 | |
|
812 | 0 | if (tsk_verbose) |
813 | 0 | tsk_fprintf(stderr, |
814 | 0 | "%s: called because this file is compressed, with data in the resource fork\n", __func__); |
815 | | |
816 | | // Reading zero bytes? OK at any offset, I say! |
817 | 0 | if (a_len == 0) |
818 | 0 | return 0; |
819 | | |
820 | 0 | if (a_offset < 0) { |
821 | 0 | error_detected(TSK_ERR_FS_ARG, |
822 | 0 | "%s: reading from file at a negative offset", |
823 | 0 | __func__); |
824 | 0 | return -1; |
825 | 0 | } |
826 | | |
827 | 0 | if (a_len > SIZE_MAX / 2) { |
828 | 0 | error_detected(TSK_ERR_FS_ARG, |
829 | 0 | "%s: trying to read more than SIZE_MAX/2 is not supported.", |
830 | 0 | __func__); |
831 | 0 | return -1; |
832 | 0 | } |
833 | | |
834 | 0 | if ((a_fs_attr == NULL) || (a_fs_attr->fs_file == NULL) |
835 | 0 | || (a_fs_attr->fs_file->meta == NULL) |
836 | 0 | || (a_fs_attr->fs_file->fs_info == NULL)) { |
837 | 0 | error_detected(TSK_ERR_FS_ARG, |
838 | 0 | "%s: NULL parameters passed", __func__); |
839 | 0 | return -1; |
840 | 0 | } |
841 | | |
842 | | // This should be a compressed file. If not, that's an error! |
843 | 0 | if (!(a_fs_attr->flags & TSK_FS_ATTR_COMP)) { |
844 | 0 | error_detected(TSK_ERR_FS_ARG, |
845 | 0 | "%s: called with non-special attribute: %x", |
846 | 0 | __func__, a_fs_attr->flags); |
847 | 0 | return -1; |
848 | 0 | } |
849 | | |
850 | | // Check that the ATTR being read is the main DATA resource, 4352-0, |
851 | | // because this is the only one that can be compressed in HFS+ |
852 | 0 | if ((a_fs_attr->id != HFS_FS_ATTR_ID_DATA) || |
853 | 0 | (a_fs_attr->type != TSK_FS_ATTR_TYPE_HFS_DATA)) { |
854 | 0 | error_detected(TSK_ERR_FS_ARG, |
855 | 0 | "%s: arg specified an attribute %u-%u that is not the data fork, " |
856 | 0 | "Only the data fork can be compressed.", __func__, |
857 | 0 | a_fs_attr->type, a_fs_attr->id); |
858 | 0 | return -1; |
859 | 0 | } |
860 | | |
861 | | /******** Open the Resource Fork ***********/ |
862 | | // The file |
863 | 0 | fs_file = a_fs_attr->fs_file; |
864 | | |
865 | | // find the attribute for the resource fork |
866 | 0 | rAttr = |
867 | 0 | tsk_fs_file_attr_get_type(fs_file, TSK_FS_ATTR_TYPE_HFS_RSRC, |
868 | 0 | HFS_FS_ATTR_ID_RSRC, FALSE); |
869 | 0 | if (rAttr == NULL) { |
870 | 0 | error_returned |
871 | 0 | (" %s: could not get the attribute for the resource fork of the file", __func__); |
872 | 0 | return -1; |
873 | 0 | } |
874 | | |
875 | | // read the offset table from the fork header |
876 | 0 | std::unique_ptr<CMP_OFFSET_ENTRY[]> offsetTable = read_block_table( |
877 | 0 | rAttr, &offsetTableSize, &offsetTableOffset |
878 | 0 | ); |
879 | 0 | if (!offsetTable) { |
880 | 0 | return -1; |
881 | 0 | } |
882 | | |
883 | | // Compute the range of compression units needed for the request |
884 | 0 | startUnit = a_offset / COMPRESSION_UNIT_SIZE; |
885 | 0 | startUnitOffset = a_offset % COMPRESSION_UNIT_SIZE; |
886 | 0 | endUnit = (a_offset + a_len - 1) / COMPRESSION_UNIT_SIZE; |
887 | |
|
888 | 0 | if (startUnit >= offsetTableSize || endUnit >= offsetTableSize) { |
889 | 0 | error_detected(TSK_ERR_FS_ARG, |
890 | 0 | "%s: range of bytes requested %lld - %lld falls past the " |
891 | 0 | "end of the uncompressed stream %llu\n", |
892 | 0 | __func__, a_offset, a_offset + a_len, |
893 | 0 | offsetTable[offsetTableSize-1].offset + |
894 | 0 | offsetTable[offsetTableSize-1].length); |
895 | 0 | return -1; |
896 | 0 | } |
897 | | |
898 | 0 | if (tsk_verbose) |
899 | 0 | tsk_fprintf(stderr, |
900 | 0 | "%s: reading compression units: %" PRIdOFF |
901 | 0 | " to %" PRIdOFF "\n", __func__, startUnit, endUnit); |
902 | 0 | bytesCopied = 0; |
903 | | |
904 | | // Allocate buffers for the raw and uncompressed data |
905 | | /* Raw data can be COMPRESSION_UNIT_SIZE+1 if the zlib data is not |
906 | | * compressed and there is a 1-byte flag that indicates that |
907 | | * the data is not compressed. */ |
908 | 0 | std::unique_ptr<char[]> rawBuf{new(std::nothrow) char[COMPRESSION_UNIT_SIZE + 1]}; |
909 | 0 | if (!rawBuf) { |
910 | 0 | error_returned |
911 | 0 | (" %s: buffers for reading and uncompressing", __func__); |
912 | 0 | return -1; |
913 | 0 | } |
914 | | |
915 | 0 | std::unique_ptr<char[]> uncBuf{new(std::nothrow) char[COMPRESSION_UNIT_SIZE]}; |
916 | 0 | if (!uncBuf) { |
917 | 0 | error_returned |
918 | 0 | (" %s: buffers for reading and uncompressing", __func__); |
919 | 0 | return -1; |
920 | 0 | } |
921 | | |
922 | | // Read from the indicated comp units |
923 | 0 | for (indx = startUnit; indx <= endUnit; ++indx) { |
924 | 0 | char *uncBufPtr = uncBuf.get(); |
925 | 0 | size_t bytesToCopy; |
926 | |
|
927 | 0 | const ssize_t ret = read_and_decompress_block( |
928 | 0 | rAttr, rawBuf.get(), uncBuf.get(), |
929 | 0 | offsetTable.get(), offsetTableSize, offsetTableOffset, (size_t)indx, |
930 | 0 | decompress_block |
931 | 0 | ); |
932 | |
|
933 | 0 | switch (ret) { |
934 | 0 | case -1: |
935 | 0 | return -1; |
936 | 0 | case 0: |
937 | 0 | continue; |
938 | 0 | default: |
939 | 0 | break; |
940 | 0 | } |
941 | | |
942 | 0 | uint64_t uncLen = ret; |
943 | | |
944 | | // If this is the first comp unit, then we must skip over the |
945 | | // startUnitOffset bytes. |
946 | 0 | if (indx == startUnit) { |
947 | 0 | uncLen -= startUnitOffset; |
948 | 0 | uncBufPtr += startUnitOffset; |
949 | 0 | } |
950 | | |
951 | | // How many bytes to copy from this compression unit? |
952 | |
|
953 | 0 | if (bytesCopied + uncLen < (uint64_t) a_len) // cast OK because a_len > 0 |
954 | 0 | bytesToCopy = (size_t) uncLen; // uncLen <= size of compression unit, which is small, so cast is OK |
955 | 0 | else |
956 | 0 | bytesToCopy = (size_t) (((uint64_t) a_len) - bytesCopied); // diff <= compression unit size, so cast is OK |
957 | | |
958 | | // Copy into the output buffer, and update bookkeeping. |
959 | 0 | memcpy(a_buf + bytesCopied, uncBufPtr, bytesToCopy); |
960 | 0 | bytesCopied += bytesToCopy; |
961 | 0 | } |
962 | | |
963 | | // Well, we don't know (without a lot of work) what the |
964 | | // true uncompressed size of the stream is. All we know is the "upper bound" which |
965 | | // assumes that all of the compression units expand to their full size. If we did |
966 | | // know the true size, then we could reject requests that go beyond the end of the |
967 | | // stream. Instead, we treat the stream as if it is padded out to the full size of |
968 | | // the last compression unit with zeros. |
969 | | |
970 | | // Have we read and copied all of the bytes requested? |
971 | 0 | if (bytesCopied < a_len) { |
972 | | // set the remaining bytes to zero |
973 | 0 | memset(a_buf + bytesCopied, 0, a_len - (size_t) bytesCopied); // cast OK because diff must be < compression unit size |
974 | 0 | } |
975 | |
|
976 | 0 | return (ssize_t) bytesCopied; // cast OK, cannot be greater than a_len which cannot be greater than SIZE_MAX/2 (rounded down). |
977 | 0 | } |
978 | | |
979 | | |
980 | | #ifdef HAVE_LIBZ |
981 | | /** |
982 | | * \internal |
983 | | * Read a ZLIB compressed resource |
984 | | * |
985 | | * @param fs_attr the attribute to read |
986 | | * @param a_offset the offset from which to read |
987 | | * @param a_buf the buffer into which to read |
988 | | * @param a_len the length of the buffer |
989 | | * @return number of bytes read or -1 on error (incl if offset is past EOF) |
990 | | */ |
991 | | ssize_t |
992 | | decmpfs_file_read_zlib_rsrc(const TSK_FS_ATTR * a_fs_attr, |
993 | | TSK_OFF_T a_offset, char *a_buf, size_t a_len) |
994 | 0 | { |
995 | 0 | return decmpfs_file_read_compressed_rsrc( |
996 | 0 | a_fs_attr, a_offset, a_buf, a_len, |
997 | 0 | decmpfs_read_zlib_block_table, |
998 | 0 | decmpfs_decompress_zlib_block |
999 | 0 | ); |
1000 | 0 | } |
1001 | | #endif |
1002 | | |
1003 | | /** |
1004 | | * Read an LZVN compressed resource |
1005 | | * |
1006 | | * @param fs_attr the attribute to read |
1007 | | * @param a_offset the offset from which to read |
1008 | | * @param a_buf the buffer into which to read |
1009 | | * @param a_len the length of the buffer |
1010 | | * @return number of bytes read or -1 on error (incl if offset is past EOF) |
1011 | | */ |
1012 | | ssize_t |
1013 | | decmpfs_file_read_lzvn_rsrc(const TSK_FS_ATTR * a_fs_attr, |
1014 | | TSK_OFF_T a_offset, char *a_buf, size_t a_len) |
1015 | 0 | { |
1016 | 0 | return decmpfs_file_read_compressed_rsrc( |
1017 | 0 | a_fs_attr, a_offset, a_buf, a_len, |
1018 | 0 | decmpfs_read_lzvn_block_table, |
1019 | 0 | decmpfs_decompress_lzvn_block |
1020 | 0 | ); |
1021 | 0 | } |
1022 | | |
1023 | | /** |
1024 | | * \internal |
1025 | | * "Decompress" an uncompressed attr |
1026 | | * |
1027 | | * HFS+ compression schemes allow for some blocks to be stored uncompressed. |
1028 | | * |
1029 | | * @param rawBuf source buffer |
1030 | | * @param rawSize size of source buffer |
1031 | | * @param uncSize expected uncompressed size |
1032 | | * @param dstBuf destination buffer |
1033 | | * @param dstSize size of destination buffer |
1034 | | * @return 1 |
1035 | | */ |
1036 | | static int decmpfs_decompress_noncompressed_attr( |
1037 | | char* rawBuf, |
1038 | | [[maybe_unused]] uint32_t rawSize, |
1039 | | uint64_t uncSize, |
1040 | | char** dstBuf, |
1041 | | uint64_t* dstSize) |
1042 | 0 | { |
1043 | 0 | if (tsk_verbose) |
1044 | 0 | tsk_fprintf(stderr, |
1045 | 0 | "%s: Leading byte, 0x%02x, indicates that the data is not really compressed.\n" |
1046 | 0 | "%s: Loading the default DATA attribute.", __func__, rawBuf[0], __func__); |
1047 | |
|
1048 | 0 | *dstBuf = rawBuf + 1; // + 1 indicator byte |
1049 | 0 | *dstSize = uncSize; |
1050 | 0 | return 1; |
1051 | 0 | } |
1052 | | |
1053 | | bool decmpfs_is_compressed_zlib_attr( |
1054 | | const char* rawBuf, |
1055 | | [[maybe_unused]] uint32_t rawSize) |
1056 | 0 | { |
1057 | | // ZLIB blocks cannot start with 0xF as the low nibble, so that's used |
1058 | | // as the flag for noncompressed blocks |
1059 | 0 | return (rawBuf[0] & 0x0F) != 0x0F; |
1060 | 0 | } |
1061 | | |
1062 | | /** |
1063 | | * \internal |
1064 | | * Decompress a ZLIB compressed attr |
1065 | | * |
1066 | | * @param rawBuf source buffer |
1067 | | * @param rawSize size of source buffer |
1068 | | * @param uncSize expected uncompressed size |
1069 | | * @param dstSize size of destination buffer |
1070 | | * @return 1 on success, 0 on error |
1071 | | */ |
1072 | | std::unique_ptr<char[]> decmpfs_decompress_zlib_attr( |
1073 | | char* rawBuf, |
1074 | | uint32_t rawSize, |
1075 | | uint64_t uncSize, |
1076 | | uint64_t* dstSize) |
1077 | 0 | { |
1078 | 0 | #ifdef HAVE_LIBZ |
1079 | 0 | uint64_t uLen; |
1080 | 0 | unsigned long bytesConsumed; |
1081 | 0 | int infResult; |
1082 | |
|
1083 | 0 | if (tsk_verbose) |
1084 | 0 | tsk_fprintf(stderr, |
1085 | 0 | "%s: Uncompressing (inflating) data.", __func__); |
1086 | | // Uncompress the remainder of the attribute, and load as 128-0 |
1087 | | // Note: cast is OK because uncSize will be quite modest, < 4000. |
1088 | | |
1089 | | // add some extra space |
1090 | 0 | std::unique_ptr<char[]> uncBuf{new(std::nothrow) char[uncSize + 100]}; |
1091 | 0 | if (!uncBuf) { |
1092 | 0 | error_returned |
1093 | 0 | (" - %s, space for the uncompressed attr", __func__); |
1094 | 0 | return nullptr; |
1095 | 0 | } |
1096 | | |
1097 | 0 | infResult = zlib_inflate(rawBuf, (uint64_t) rawSize, |
1098 | 0 | uncBuf.get(), (uint64_t) (uncSize + 100), |
1099 | 0 | &uLen, &bytesConsumed); |
1100 | 0 | if (infResult != 0) { |
1101 | 0 | error_returned |
1102 | 0 | (" %s, zlib could not uncompress attr", __func__); |
1103 | 0 | return nullptr; |
1104 | 0 | } |
1105 | | |
1106 | 0 | if (bytesConsumed != rawSize) { |
1107 | 0 | error_detected(TSK_ERR_FS_READ, |
1108 | 0 | " %s, decompressor did not consume the whole compressed data", |
1109 | 0 | __func__); |
1110 | 0 | return nullptr; |
1111 | 0 | } |
1112 | | |
1113 | 0 | *dstSize = uncSize; |
1114 | 0 | return uncBuf; |
1115 | | #else |
1116 | | // ZLIB compression library is not available, so we will load a |
1117 | | // zero-length default DATA attribute. Without this, icat may |
1118 | | // misbehave. |
1119 | | |
1120 | | if (tsk_verbose) |
1121 | | tsk_fprintf(stderr, |
1122 | | "%s: ZLIB not available, so loading an empty default DATA attribute.\n", __func__); |
1123 | | |
1124 | | // Dummy array is one byte long, so the ptr is not null, but we set the |
1125 | | // length to zero bytes, so it is never read. |
1126 | | *dstSize = 0; |
1127 | | return std::unique_ptr<char[]>{new(std::nothrow) char[1]}; |
1128 | | #endif |
1129 | 0 | } |
1130 | | |
1131 | | bool decmpfs_is_compressed_lzvn_attr( |
1132 | | const char* rawBuf, |
1133 | | [[maybe_unused]] uint32_t rawSize) |
1134 | 0 | { |
1135 | | // LZVN blocks cannot start with 0x06, so that's used as the flag for |
1136 | | // noncompressed blocks |
1137 | 0 | return rawBuf[0] != 0x06; |
1138 | 0 | } |
1139 | | |
1140 | | /** |
1141 | | * \internal |
1142 | | * Decompress an LZVN compressed attr |
1143 | | * |
1144 | | * @param rawBuf source buffer |
1145 | | * @param rawSize size of source buffer |
1146 | | * @param uncSize expected uncompressed size |
1147 | | * @param dstSize size of destination buffer |
1148 | | * @return 1 on success, 0 on error |
1149 | | */ |
1150 | | std::unique_ptr<char[]> decmpfs_decompress_lzvn_attr( |
1151 | | char* rawBuf, |
1152 | | uint32_t rawSize, |
1153 | | uint64_t uncSize, |
1154 | | uint64_t* dstSize) |
1155 | 0 | { |
1156 | 0 | std::unique_ptr<char[]> uncBuf{new(std::nothrow) char[uncSize]}; |
1157 | 0 | *dstSize = lzvn_decode_buffer(uncBuf.get(), uncSize, rawBuf, rawSize); |
1158 | 0 | return uncBuf; |
1159 | 0 | } |
1160 | | |
1161 | | /** |
1162 | | * \internal |
1163 | | * Read a compressed attr |
1164 | | * |
1165 | | * @param fs_file the file |
1166 | | * @param cmpType compression type |
1167 | | * @param buffer destination buffer |
1168 | | * @param attributeLength length of the attribute |
1169 | | * @param uncSize uncompressed size |
1170 | | * @param decompress_attr pointer to the decompression function |
1171 | | * @return 1 on success, 0 on error |
1172 | | */ |
1173 | | static int |
1174 | | decmpfs_file_read_compressed_attr( |
1175 | | TSK_FS_FILE* fs_file, |
1176 | | uint8_t cmpType, |
1177 | | char* buffer, |
1178 | | TSK_OFF_T attributeLength, |
1179 | | uint64_t uncSize, |
1180 | | bool (*is_compressed)( |
1181 | | const char* rawBuf, |
1182 | | uint32_t rawSize |
1183 | | ), |
1184 | | std::unique_ptr<char[]> (*decompress_attr)( |
1185 | | char* rawBuf, |
1186 | | uint32_t rawSize, |
1187 | | uint64_t uncSize, |
1188 | | uint64_t* dstSize |
1189 | | ) |
1190 | | ) |
1191 | 0 | { |
1192 | | // Data is inline. We will load the uncompressed data as a |
1193 | | // resident attribute. |
1194 | 0 | if (tsk_verbose) |
1195 | 0 | tsk_fprintf(stderr, |
1196 | 0 | "%s: Compressed data is inline in the attribute, will load this as the default DATA attribute.\n", __func__); |
1197 | |
|
1198 | 0 | if (attributeLength <= 16) { |
1199 | 0 | if (tsk_verbose) |
1200 | 0 | tsk_fprintf(stderr, |
1201 | 0 | "%s: WARNING, Compression Record of type %u is not followed by" |
1202 | 0 | " compressed data. No data will be loaded into the DATA" |
1203 | 0 | " attribute.\n", __func__, cmpType); |
1204 | | |
1205 | | // oddly, this is not actually considered an error |
1206 | 0 | return 1; |
1207 | 0 | } |
1208 | | |
1209 | 0 | TSK_FS_ATTR *fs_attr_unc; |
1210 | | |
1211 | | // There is data following the compression record, as there should be. |
1212 | 0 | if ((fs_attr_unc = tsk_fs_attrlist_getnew( |
1213 | 0 | fs_file->meta->attr, TSK_FS_ATTR_RES)) == NULL) |
1214 | 0 | { |
1215 | 0 | error_returned(" - %s, FS_ATTR for uncompressed data", __func__); |
1216 | 0 | return 0; |
1217 | 0 | } |
1218 | | |
1219 | 0 | char* dstBuf = nullptr; |
1220 | 0 | std::unique_ptr<char[]> dstBufStore; |
1221 | 0 | uint64_t dstSize; |
1222 | |
|
1223 | 0 | if (is_compressed(buffer + 16, attributeLength - 16)) { |
1224 | 0 | dstBufStore = decompress_attr( |
1225 | 0 | buffer + 16, attributeLength - 16, uncSize, &dstSize |
1226 | 0 | ); |
1227 | 0 | if (!dstBufStore) { |
1228 | 0 | return 0; |
1229 | 0 | } |
1230 | 0 | dstBuf = dstBufStore.get(); |
1231 | 0 | } |
1232 | 0 | else { |
1233 | 0 | if (!decmpfs_decompress_noncompressed_attr(buffer + 16, attributeLength - 16, uncSize, &dstBuf, &dstSize)) { |
1234 | 0 | return 0; |
1235 | 0 | } |
1236 | 0 | } |
1237 | | |
1238 | 0 | if (dstSize != uncSize) { |
1239 | 0 | error_detected(TSK_ERR_FS_READ, |
1240 | 0 | " %s, actual uncompressed size not equal to the size in the compression record", __func__); |
1241 | 0 | return 0; |
1242 | 0 | } |
1243 | | |
1244 | 0 | if (tsk_verbose) |
1245 | 0 | tsk_fprintf(stderr, |
1246 | 0 | "%s: Loading decompressed data as default DATA attribute.", |
1247 | 0 | __func__); |
1248 | | |
1249 | | // Load the remainder of the attribute as 128-0 |
1250 | | // set the details in the fs_attr structure. |
1251 | | // Note, we are loading this as a RESIDENT attribute. |
1252 | 0 | if (tsk_fs_attr_set_str(fs_file, |
1253 | 0 | fs_attr_unc, "DECOMP", |
1254 | 0 | TSK_FS_ATTR_TYPE_HFS_DATA, |
1255 | 0 | TSK_FS_ATTR_ID_DEFAULT, |
1256 | 0 | dstBuf, |
1257 | 0 | dstSize)) |
1258 | 0 | { |
1259 | 0 | error_returned(" - %s", __func__); |
1260 | 0 | return 0; |
1261 | 0 | } |
1262 | | |
1263 | 0 | return 1; |
1264 | 0 | } |
1265 | | |
1266 | | /** |
1267 | | * \internal |
1268 | | * Read a ZLIB compressed attr |
1269 | | * |
1270 | | * @param fs_file the file |
1271 | | * @param buffer destination buffer |
1272 | | * @param attributeLength length of the attribute |
1273 | | * @param uncSize uncompressed size |
1274 | | * @return 1 on success, 0 on error |
1275 | | */ |
1276 | | int decmpfs_file_read_zlib_attr(TSK_FS_FILE* fs_file, |
1277 | | char* buffer, |
1278 | | TSK_OFF_T attributeLength, |
1279 | | uint64_t uncSize) |
1280 | 0 | { |
1281 | 0 | return decmpfs_file_read_compressed_attr( |
1282 | 0 | fs_file, DECMPFS_TYPE_ZLIB_ATTR, |
1283 | 0 | buffer, attributeLength, uncSize, |
1284 | 0 | decmpfs_is_compressed_zlib_attr, |
1285 | 0 | decmpfs_decompress_zlib_attr |
1286 | 0 | ); |
1287 | 0 | } |
1288 | | |
1289 | | /** |
1290 | | * \internal |
1291 | | * Read an LZVN compressed attr |
1292 | | * |
1293 | | * @param fs_file the file |
1294 | | * @param buffer destination buffer |
1295 | | * @param attributeLength length of the attribute |
1296 | | * @param uncSize uncompressed size |
1297 | | * @return 1 on success, 0 on error |
1298 | | */ |
1299 | | int decmpfs_file_read_lzvn_attr(TSK_FS_FILE* fs_file, |
1300 | | char* buffer, |
1301 | | TSK_OFF_T attributeLength, |
1302 | | uint64_t uncSize) |
1303 | 0 | { |
1304 | 0 | return decmpfs_file_read_compressed_attr( |
1305 | 0 | fs_file, DECMPFS_TYPE_LZVN_ATTR, |
1306 | 0 | buffer, attributeLength, uncSize, |
1307 | 0 | decmpfs_is_compressed_lzvn_attr, |
1308 | 0 | decmpfs_decompress_lzvn_attr |
1309 | 0 | ); |
1310 | 0 | } |