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