1 /*
2  * Copyright (c) 2020 Ori Bernstein
3  * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include <sys/types.h>
19 
20 #include <assert.h>
21 #include <endian.h>
22 #include <errno.h>
23 #include <limits.h>
24 #include <stdint.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28 #include <sha1.h>
29 
30 #include "got_error.h"
31 
32 #include "got_lib_deltify.h"
33 
34 #ifndef MIN
35 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
36 #endif
37 
38 static uint32_t geartab[] = {
39     0x67ed26b7, 0x32da500c, 0x53d0fee0, 0xce387dc7, 0xcd406d90, 0x2e83a4d4,
40     0x9fc9a38d, 0xb67259dc, 0xca6b1722, 0x6d2ea08c, 0x235cea2e, 0x3149bb5f,
41     0x1beda787, 0x2a6b77d5, 0x2f22d9ac, 0x91fc0544, 0xe413acfa, 0x5a30ff7a,
42     0xad6fdde0, 0x444fd0f5, 0x7ad87864, 0x58c5ff05, 0x8d2ec336, 0x2371f853,
43     0x550f8572, 0x6aa448dd, 0x7c9ddbcf, 0x95221e14, 0x2a82ec33, 0xcbec5a78,
44     0xc6795a0d, 0x243995b7, 0x1c909a2f, 0x4fded51c, 0x635d334b, 0x0e2b9999,
45     0x2702968d, 0x856de1d5, 0x3325d60e, 0xeb6a7502, 0xec2a9844, 0x0905835a,
46     0xa1820375, 0xa4be5cab, 0x96a6c058, 0x2c2ccd70, 0xba40fce3, 0xd794c46b,
47     0x8fbae83e, 0xc3aa7899, 0x3d3ff8ed, 0xa0d42b5b, 0x571c0c97, 0xd2811516,
48     0xf7e7b96c, 0x4fd2fcbd, 0xe2fdec94, 0x282cc436, 0x78e8e95c, 0x80a3b613,
49     0xcfbee20c, 0xd4a32d1c, 0x2a12ff13, 0x6af82936, 0xe5630258, 0x8efa6a98,
50     0x294fb2d1, 0xdeb57086, 0x5f0fddb3, 0xeceda7ce, 0x4c87305f, 0x3a6d3307,
51     0xe22d2942, 0x9d060217, 0x1e42ed02, 0xb6f63b52, 0x4367f39f, 0x055cf262,
52     0x03a461b2, 0x5ef9e382, 0x386bc03a, 0x2a1e79c7, 0xf1a0058b, 0xd4d2dea9,
53     0x56baf37d, 0x5daff6cc, 0xf03a951d, 0xaef7de45, 0xa8f4581e, 0x3960b555,
54     0xffbfff6d, 0xbe702a23, 0x8f5b6d6f, 0x061739fb, 0x98696f47, 0x3fd596d4,
55     0x151eac6b, 0xa9fcc4f5, 0x69181a12, 0x3ac5a107, 0xb5198fe7, 0x96bcb1da,
56     0x1b5ddf8e, 0xc757d650, 0x65865c3a, 0x8fc0a41a, 0x87435536, 0x99eda6f2,
57     0x41874794, 0x29cff4e8, 0xb70efd9a, 0x3103f6e7, 0x84d2453b, 0x15a450bd,
58     0x74f49af1, 0x60f664b1, 0xa1c86935, 0xfdafbce1, 0xe36353e3, 0x5d9ba739,
59     0xbc0559ba, 0x708b0054, 0xd41d808c, 0xb2f31723, 0x9027c41f, 0xf136d165,
60     0xb5374b12, 0x9420a6ac, 0x273958b6, 0xe6c2fad0, 0xebdc1f21, 0xfb33af8b,
61     0xc71c25cd, 0xe9a2d8e5, 0xbeb38a50, 0xbceb7cc2, 0x4e4e73f0, 0xcd6c251d,
62     0xde4c032c, 0x4b04ac30, 0x725b8b21, 0x4eb8c33b, 0x20d07b75, 0x0567aa63,
63     0xb56b2bb7, 0xc1f5fd3a, 0xcafd35ca, 0x470dd4da, 0xfe4f94cd, 0xfb8de424,
64     0xe8dbcf40, 0xfe50a37a, 0x62db5b5d, 0xf32f4ab6, 0x2c4a8a51, 0x18473dc0,
65     0xfe0cbb6e, 0xfe399efd, 0xdf34ecc9, 0x6ccd5055, 0x46097073, 0x139135c2,
66     0x721c76f6, 0x1c6a94b4, 0x6eee014d, 0x8a508e02, 0x3da538f5, 0x280d394f,
67     0x5248a0c4, 0x3ce94c6c, 0x9a71ad3a, 0x8493dd05, 0xe43f0ab6, 0x18e4ed42,
68     0x6c5c0e09, 0x42b06ec9, 0x8d330343, 0xa45b6f59, 0x2a573c0c, 0xd7fd3de6,
69     0xeedeab68, 0x5c84dafc, 0xbbd1b1a8, 0xa3ce1ad1, 0x85b70bed, 0xb6add07f,
70     0xa531309c, 0x8f8ab852, 0x564de332, 0xeac9ed0c, 0x73da402c, 0x3ec52761,
71     0x43af2f4d, 0xd6ff45c8, 0x4c367462, 0xd553bd6a, 0x44724855, 0x3b2aa728,
72     0x56e5eb65, 0xeaf16173, 0x33fa42ff, 0xd714bb5d, 0xfbd0a3b9, 0xaf517134,
73     0x9416c8cd, 0x534cf94f, 0x548947c2, 0x34193569, 0x32f4389a, 0xfe7028bc,
74     0xed73b1ed, 0x9db95770, 0x468e3922, 0x0440c3cd, 0x60059a62, 0x33504562,
75     0x2b229fbd, 0x5174dca5, 0xf7028752, 0xd63c6aa8, 0x31276f38, 0x0646721c,
76     0xb0191da8, 0xe00e6de0, 0x9eac1a6e, 0x9f7628a5, 0xed6c06ea, 0x0bb8af15,
77     0xf119fb12, 0x38693c1c, 0x732bc0fe, 0x84953275, 0xb82ec888, 0x33a4f1b3,
78     0x3099835e, 0x028a8782, 0x5fdd51d7, 0xc6c717b3, 0xb06caf71, 0x17c8c111,
79     0x61bad754, 0x9fd03061, 0xe09df1af, 0x3bc9eb73, 0x85878413, 0x9889aaf2,
80     0x3f5a9e46, 0x42c9f01f, 0x9984a4f4, 0xd5de43cc, 0xd294daed, 0xbecba2d2,
81     0xf1f6e72c, 0x5551128a, 0x83af87e2, 0x6f0342ba,
82 };
83 
84 static uint64_t
85 hashblk(const unsigned char *p, off_t n)
86 {
87 	unsigned char buf[SHA1_DIGEST_LENGTH];
88 	uint64_t h;
89 	SHA1_CTX ctx;
90 	SHA1Init(&ctx);
91 	SHA1Update(&ctx, p, n);
92 	SHA1Final(buf, &ctx);
93 	memcpy(&h, buf, sizeof(h));
94 	return be64toh(h);
95 }
96 
97 static const struct got_error *
98 addblk(struct got_delta_table *dt, FILE *f, off_t file_offset0, off_t len,
99     off_t offset, uint64_t h)
100 {
101 	const struct got_error *err = NULL;
102 	int i;
103 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
104 	uint8_t buf2[GOT_DELTIFY_MAXCHUNK];
105 	size_t r = 0;
106 
107 	if (len == 0)
108 		return NULL;
109 
110 	i = h % dt->nalloc;
111 	while (dt->blocks[i].len != 0) {
112 		/*
113 		 * Avoid adding duplicate blocks.
114 		 * NB: A matching hash is insufficient for detecting equality.
115 		 * The hash can only detect inequality.
116 		 */
117 		if (len == dt->blocks[i].len && h == dt->blocks[i].hash) {
118 			if (r == 0) {
119 				if (fseeko(f, file_offset0 + offset, SEEK_SET) == -1)
120 					return got_error_from_errno("fseeko");
121 				r = fread(buf, 1, len, f);
122 				if (r != len) {
123 					if (!ferror(f))
124 						return NULL;
125 					return got_ferror(f, GOT_ERR_IO);
126 				}
127 			}
128 			if (fseeko(f, file_offset0 + dt->blocks[i].offset,
129 			    SEEK_SET) == -1)
130 				return got_error_from_errno("fseeko");
131 			if (fread(buf2, 1, len, f) != len)
132 				return got_ferror(f, GOT_ERR_IO);
133 			if (memcmp(buf, buf2, len) == 0)
134 				return NULL;
135 		}
136 
137 		i = (i + 1) % dt->nalloc;
138 	}
139 	assert(dt->blocks[i].len == 0);
140 	dt->blocks[i].len = len;
141 	dt->blocks[i].offset = offset;
142 	dt->blocks[i].hash = h;
143 	dt->nblocks++;
144 	if (dt->nalloc < dt->nblocks + 64) {
145 		struct got_delta_block *db;
146 		size_t old_size = dt->nalloc;
147 		db = dt->blocks;
148 		dt->blocks = calloc(dt->nalloc + 64,
149 		    sizeof(struct got_delta_block));
150 		if (dt->blocks == NULL) {
151 			err = got_error_from_errno("calloc");
152 			dt->blocks = db;
153 			return err;
154 		}
155 		dt->nalloc += 64;
156 		/*
157 		 * Recompute all block positions. Hash-based indices of blocks
158 		 * in the array depend on the allocated length of the array.
159 		 */
160 		dt->nblocks = 0;
161 		for (i = 0; i < old_size; i++) {
162 			if (db[i].len == 0)
163 				continue;
164 			err = addblk(dt, f, file_offset0, db[i].len,
165 			    db[i].offset, db[i].hash);
166 			if (err)
167 				break;
168 		}
169 		free(db);
170 	}
171 
172 	return err;
173 }
174 
175 static const struct got_error *
176 lookupblk(struct got_delta_block **block, struct got_delta_table *dt,
177     unsigned char *p, off_t len, FILE *basefile, off_t basefile_offset0)
178 {
179 	int i;
180 	uint64_t h;
181 	uint8_t buf[GOT_DELTIFY_MAXCHUNK];
182 	size_t r;
183 
184 	*block = NULL;
185 
186 	h = hashblk(p, len);
187 	for (i = h % dt->nalloc; dt->blocks[i].len != 0;
188 	     i = (i + 1) % dt->nalloc) {
189 		if (dt->blocks[i].hash != h ||
190 		    dt->blocks[i].len != len)
191 			continue;
192 		if (fseeko(basefile, basefile_offset0 + dt->blocks[i].offset,
193 		    SEEK_SET) == -1)
194 			return got_error_from_errno("fseeko");
195 		r = fread(buf, 1, len, basefile);
196 		if (r != len)
197 			return got_ferror(basefile, GOT_ERR_IO);
198 		if (memcmp(p, buf, len) == 0) {
199 			*block = &dt->blocks[i];
200 			break;
201 		}
202 	}
203 	return NULL;
204 }
205 
206 static const struct got_error *
207 nextblk(uint8_t *buf, off_t *blocklen, FILE *f)
208 {
209 	uint32_t gh;
210 	const unsigned char *p;
211 	size_t r;
212 	off_t pos = ftello(f);
213 
214 	*blocklen = 0;
215 
216 	r = fread(buf, 1, GOT_DELTIFY_MAXCHUNK, f);
217 	if (r == 0 && ferror(f))
218 		return got_ferror(f, GOT_ERR_IO);
219 	if (r < GOT_DELTIFY_MINCHUNK)
220 		return NULL; /* no more delta-worthy blocks left */
221 
222 	/* Got a deltifiable block. Find the split-point where it ends. */
223 	p = buf + GOT_DELTIFY_MINCHUNK;
224 	gh = 0;
225 	while (p != buf + r) {
226 		gh = (gh << 1) + geartab[*p++];
227 		if ((gh & GOT_DELTIFY_SPLITMASK) == 0)
228 			break;
229 	}
230 
231 	*blocklen = (p - buf);
232 	if (fseeko(f, pos + *blocklen, SEEK_SET) == -1)
233 		return got_error_from_errno("fseeko");
234 
235 	return NULL;
236 }
237 
238 const struct got_error *
239 got_deltify_init(struct got_delta_table **dt, FILE *f, off_t fileoffset,
240     off_t filesize)
241 {
242 	const struct got_error *err = NULL;
243 	uint64_t h;
244 	const off_t offset0 = fileoffset;
245 
246 	*dt = calloc(1, sizeof(**dt));
247 	if (*dt == NULL)
248 		return got_error_from_errno("calloc");
249 
250 	(*dt)->nblocks = 0;
251 	(*dt)->nalloc = 128;
252 	(*dt)->blocks = calloc((*dt)->nalloc, sizeof(struct got_delta_block));
253 	if ((*dt)->blocks == NULL) {
254 		err = got_error_from_errno("calloc");
255 		goto done;
256 	}
257 
258 	if (fseeko(f, fileoffset, SEEK_SET) == -1)
259 		return got_error_from_errno("fseeko");
260 
261 	while (fileoffset < filesize) {
262 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
263 		off_t blocklen;
264 		err = nextblk(buf, &blocklen, f);
265 		if (err)
266 			goto done;
267 		if (blocklen == 0)
268 			break;
269 		h = hashblk(buf, blocklen);
270 		err = addblk(*dt, f, offset0, blocklen,
271 		    fileoffset - offset0, h);
272 		if (err)
273 			goto done;
274 		fileoffset += blocklen;
275 		if (fseeko(f, fileoffset, SEEK_SET) == -1)
276 			return got_error_from_errno("fseeko");
277 	}
278 done:
279 	if (err) {
280 		free((*dt)->blocks);
281 		free(*dt);
282 		*dt = NULL;
283 	}
284 
285 	return err;
286 }
287 
288 void
289 got_deltify_free(struct got_delta_table *dt)
290 {
291 	if (dt == NULL)
292 		return;
293 	free(dt->blocks);
294 	free(dt);
295 }
296 
297 static const struct got_error *
298 emitdelta(struct got_delta_instruction **deltas, size_t *nalloc, int *ndeltas,
299     const size_t alloc_chunk_size, int copy, off_t offset, off_t len)
300 {
301 	struct got_delta_instruction *d, *p;
302 
303 	if (*nalloc < *ndeltas + alloc_chunk_size) {
304 		p = reallocarray(*deltas, *nalloc + alloc_chunk_size,
305 		    sizeof(struct got_delta_instruction));
306 		if (p == NULL)
307 			return got_error_from_errno("realloc");
308 		*deltas = p;
309 		*nalloc += alloc_chunk_size;
310 	}
311 	*ndeltas += 1;
312 	d = &(*deltas)[*ndeltas - 1];
313 	d->copy = copy;
314 	d->offset = offset;
315 	d->len = len;
316 	return NULL;
317 }
318 
319 static const struct got_error *
320 stretchblk(FILE *basefile, off_t base_offset0, struct got_delta_block *block,
321     FILE *f, off_t filesize, off_t *blocklen)
322 {
323 	uint8_t basebuf[GOT_DELTIFY_MAXCHUNK], buf[GOT_DELTIFY_MAXCHUNK];
324 	size_t base_r, r, i;
325 	int buf_equal = 1;
326 
327 	if (fseeko(basefile, base_offset0 + block->offset + *blocklen,
328 	    SEEK_SET) == -1)
329 		return got_error_from_errno("fseeko");
330 
331 	while (buf_equal && *blocklen < (1 << 24) - 1) {
332 		base_r = fread(basebuf, 1, sizeof(basebuf), basefile);
333 		if (base_r == 0) {
334 			if (ferror(basefile))
335 				return got_ferror(basefile, GOT_ERR_IO);
336 			break;
337 		}
338 		r = fread(buf, 1, sizeof(buf), f);
339 		if (r == 0) {
340 			if (ferror(f))
341 				return got_ferror(f, GOT_ERR_IO);
342 			break;
343 		}
344 		for (i = 0; i < MIN(base_r, r); i++) {
345 			if (buf[i] != basebuf[i]) {
346 				buf_equal = 0;
347 				break;
348 			}
349 			(*blocklen)++;
350 		}
351 	}
352 
353 	return NULL;
354 }
355 
356 const struct got_error *
357 got_deltify(struct got_delta_instruction **deltas, int *ndeltas,
358     FILE *f, off_t fileoffset, off_t filesize,
359     struct got_delta_table *dt, FILE *basefile,
360     off_t basefile_offset0, off_t basefile_size)
361 {
362 	const struct got_error *err = NULL;
363 	const off_t offset0 = fileoffset;
364 	size_t nalloc = 0;
365 	const size_t alloc_chunk_size = 64;
366 
367 	*deltas = NULL;
368 	*ndeltas = 0;
369 
370 	/*
371 	 * offset0 indicates where data to be deltified begins.
372 	 * For example, we want to avoid deltifying a Git object header at
373 	 * the beginning of the file.
374 	 */
375 	if (fseeko(f, offset0, SEEK_SET) == -1)
376 		return got_error_from_errno("fseeko");
377 
378 	*deltas = reallocarray(NULL, alloc_chunk_size,
379 	    sizeof(struct got_delta_instruction));
380 	if (*deltas == NULL)
381 		return got_error_from_errno("reallocarray");
382 	nalloc = alloc_chunk_size;
383 
384 	while (fileoffset < filesize) {
385 		uint8_t buf[GOT_DELTIFY_MAXCHUNK];
386 		off_t blocklen;
387 		struct got_delta_block *block;
388 		err = nextblk(buf, &blocklen, f);
389 		if (err)
390 			break;
391 		if (blocklen == 0) {
392 			/* Source remainder from the file itself. */
393 			if (fileoffset < filesize) {
394 				err = emitdelta(deltas, &nalloc, ndeltas,
395 				    alloc_chunk_size, 0, fileoffset - offset0,
396 				    filesize - fileoffset);
397 			}
398 			break;
399 		}
400 		err = lookupblk(&block, dt, buf, blocklen, basefile,
401 		    basefile_offset0);
402 		if (err)
403 			break;
404 		if (block != NULL) {
405 			/*
406 			 * We have found a matching block in the delta base.
407 			 * Attempt to stretch the block as far as possible and
408 			 * generate a copy instruction.
409 			 */
410 			err = stretchblk(basefile, basefile_offset0, block,
411 			    f, filesize, &blocklen);
412 			if (err)
413 				break;
414 			err = emitdelta(deltas, &nalloc, ndeltas,
415 			    alloc_chunk_size, 1, block->offset, blocklen);
416 			if (err)
417 				break;
418 		} else {
419 			/*
420 			 * No match.
421 			 * This block needs to be sourced from the file itself.
422 			 */
423 			err = emitdelta(deltas, &nalloc, ndeltas,
424 			    alloc_chunk_size, 0, fileoffset - offset0, blocklen);
425 			if (err)
426 				break;
427 		}
428 		fileoffset += blocklen;
429 		if (fseeko(f, fileoffset, SEEK_SET) == -1) {
430 			err = got_error_from_errno("fseeko");
431 			break;
432 		}
433 	}
434 
435 	if (err) {
436 		free(*deltas);
437 		*deltas = NULL;
438 		*ndeltas = 0;
439 	}
440 	return err;
441 }