2 e6bcace5 2021-06-22 stsp * Copyright (c) 2020 Ori Bernstein
3 e6bcace5 2021-06-22 stsp * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
5 e6bcace5 2021-06-22 stsp * Permission to use, copy, modify, and distribute this software for any
6 e6bcace5 2021-06-22 stsp * purpose with or without fee is hereby granted, provided that the above
7 e6bcace5 2021-06-22 stsp * copyright notice and this permission notice appear in all copies.
9 e6bcace5 2021-06-22 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 e6bcace5 2021-06-22 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 e6bcace5 2021-06-22 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 e6bcace5 2021-06-22 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 e6bcace5 2021-06-22 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 e6bcace5 2021-06-22 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 e6bcace5 2021-06-22 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18 08736cf9 2021-06-23 stsp #include <sys/types.h>
19 08736cf9 2021-06-23 stsp #include <sys/uio.h>
20 e6bcace5 2021-06-22 stsp #include <sys/stat.h>
21 31ba2236 2022-01-05 thomas #include <sys/time.h>
23 08736cf9 2021-06-23 stsp #include <stdint.h>
24 e6bcace5 2021-06-22 stsp #include <stdio.h>
25 e6bcace5 2021-06-22 stsp #include <stdlib.h>
26 e6bcace5 2021-06-22 stsp #include <string.h>
27 31ba2236 2022-01-05 thomas #include <time.h>
28 db847d2c 2022-03-22 thomas #include <unistd.h>
29 e6bcace5 2021-06-22 stsp #include <limits.h>
30 e6bcace5 2021-06-22 stsp #include <zlib.h>
32 b347007e 2021-10-15 thomas #if defined(__FreeBSD__)
33 b347007e 2021-10-15 thomas #include <unistd.h>
36 e6bcace5 2021-06-22 stsp #include "got_error.h"
37 e6bcace5 2021-06-22 stsp #include "got_cancel.h"
38 e6bcace5 2021-06-22 stsp #include "got_object.h"
39 0be8fa4c 2021-10-15 thomas #include "got_path.h"
40 05118f5a 2021-06-22 stsp #include "got_reference.h"
41 05118f5a 2021-06-22 stsp #include "got_repository_admin.h"
42 e4d8ab47 2021-10-15 thomas #include "got_opentemp.h"
44 e6bcace5 2021-06-22 stsp #include "got_lib_deltify.h"
45 e6bcace5 2021-06-22 stsp #include "got_lib_delta.h"
46 e6bcace5 2021-06-22 stsp #include "got_lib_object.h"
47 e6bcace5 2021-06-22 stsp #include "got_lib_object_idset.h"
48 05118f5a 2021-06-22 stsp #include "got_lib_object_cache.h"
49 e6bcace5 2021-06-22 stsp #include "got_lib_deflate.h"
50 e6bcace5 2021-06-22 stsp #include "got_lib_pack.h"
51 05118f5a 2021-06-22 stsp #include "got_lib_privsep.h"
52 05118f5a 2021-06-22 stsp #include "got_lib_repository.h"
53 31ba2236 2022-01-05 thomas #include "got_lib_ratelimit.h"
55 b79280dd 2021-10-15 thomas #ifndef MIN
56 b79280dd 2021-10-15 thomas #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
60 e6bcace5 2021-06-22 stsp #define MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
63 e6bcace5 2021-06-22 stsp struct got_pack_meta {
64 e6bcace5 2021-06-22 stsp struct got_object_id id;
66 e6bcace5 2021-06-22 stsp int obj_type;
67 ab6186ae 2021-10-15 thomas off_t size;
68 e6bcace5 2021-06-22 stsp time_t mtime;
70 e6bcace5 2021-06-22 stsp /* The best delta we picked */
71 e6bcace5 2021-06-22 stsp struct got_pack_meta *head;
72 e6bcace5 2021-06-22 stsp struct got_pack_meta *prev;
73 3b6ceab7 2022-01-10 thomas unsigned char *delta_buf; /* if not encoded in delta cache file */
74 b79280dd 2021-10-15 thomas off_t delta_offset; /* offset in delta cache file */
75 3b6ceab7 2022-01-10 thomas off_t delta_len; /* encoded delta length */
78 f9c2e8e5 2022-02-13 thomas int have_reused_delta;
79 f9c2e8e5 2022-02-13 thomas off_t reused_delta_offset; /* offset of delta in reused pack file */
80 f9c2e8e5 2022-02-13 thomas struct got_object_id *base_obj_id;
82 e6bcace5 2021-06-22 stsp /* Only used for delta window */
83 e6bcace5 2021-06-22 stsp struct got_delta_table *dtab;
85 e6bcace5 2021-06-22 stsp /* Only used for writing offset deltas */
89 e6bcace5 2021-06-22 stsp struct got_pack_metavec {
90 e6bcace5 2021-06-22 stsp struct got_pack_meta **meta;
95 e6bcace5 2021-06-22 stsp static const struct got_error *
96 e6bcace5 2021-06-22 stsp alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
97 e6bcace5 2021-06-22 stsp const char *path, int obj_type, time_t mtime)
99 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
100 e6bcace5 2021-06-22 stsp struct got_pack_meta *m;
102 e6bcace5 2021-06-22 stsp *new = NULL;
104 e6bcace5 2021-06-22 stsp m = calloc(1, sizeof(*m));
105 e6bcace5 2021-06-22 stsp if (m == NULL)
106 b7e0a384 2021-10-15 thomas return got_error_from_errno("calloc");
108 e6bcace5 2021-06-22 stsp memcpy(&m->id, id, sizeof(m->id));
110 e6bcace5 2021-06-22 stsp m->path = strdup(path);
111 e6bcace5 2021-06-22 stsp if (m->path == NULL) {
112 e6bcace5 2021-06-22 stsp err = got_error_from_errno("strdup");
114 e6bcace5 2021-06-22 stsp return err;
117 e6bcace5 2021-06-22 stsp m->obj_type = obj_type;
118 e6bcace5 2021-06-22 stsp m->mtime = mtime;
120 e6bcace5 2021-06-22 stsp return NULL;
123 e6bcace5 2021-06-22 stsp static void
124 e6bcace5 2021-06-22 stsp clear_meta(struct got_pack_meta *meta)
126 e6bcace5 2021-06-22 stsp if (meta == NULL)
128 e6bcace5 2021-06-22 stsp free(meta->path);
129 e6bcace5 2021-06-22 stsp meta->path = NULL;
130 3b6ceab7 2022-01-10 thomas free(meta->delta_buf);
131 3b6ceab7 2022-01-10 thomas meta->delta_buf = NULL;
132 f9c2e8e5 2022-02-13 thomas free(meta->base_obj_id);
133 f9c2e8e5 2022-02-13 thomas meta->base_obj_id = NULL;
136 e6bcace5 2021-06-22 stsp static void
137 e6bcace5 2021-06-22 stsp free_nmeta(struct got_pack_meta **meta, int nmeta)
141 e6bcace5 2021-06-22 stsp for (i = 0; i < nmeta; i++)
142 e6bcace5 2021-06-22 stsp clear_meta(meta[i]);
143 e6bcace5 2021-06-22 stsp free(meta);
147 e6bcace5 2021-06-22 stsp delta_order_cmp(const void *pa, const void *pb)
149 e6bcace5 2021-06-22 stsp struct got_pack_meta *a, *b;
152 e6bcace5 2021-06-22 stsp a = *(struct got_pack_meta **)pa;
153 e6bcace5 2021-06-22 stsp b = *(struct got_pack_meta **)pb;
155 e6bcace5 2021-06-22 stsp if (a->obj_type != b->obj_type)
156 e6bcace5 2021-06-22 stsp return a->obj_type - b->obj_type;
157 e6bcace5 2021-06-22 stsp cmp = strcmp(a->path, b->path);
158 e6bcace5 2021-06-22 stsp if (cmp != 0)
159 e6bcace5 2021-06-22 stsp return cmp;
160 e6bcace5 2021-06-22 stsp if (a->mtime != b->mtime)
161 e6bcace5 2021-06-22 stsp return a->mtime - b->mtime;
162 e6bcace5 2021-06-22 stsp return got_object_id_cmp(&a->id, &b->id);
165 3b6ceab7 2022-01-10 thomas static off_t
166 e6bcace5 2021-06-22 stsp delta_size(struct got_delta_instruction *deltas, int ndeltas)
169 3b6ceab7 2022-01-10 thomas off_t size = 32;
170 e6bcace5 2021-06-22 stsp for (i = 0; i < ndeltas; i++) {
171 e6bcace5 2021-06-22 stsp if (deltas[i].copy)
172 e6bcace5 2021-06-22 stsp size += GOT_DELTA_SIZE_SHIFT;
174 e6bcace5 2021-06-22 stsp size += deltas[i].len + 1;
176 e6bcace5 2021-06-22 stsp return size;
179 3b6ceab7 2022-01-10 thomas static const struct got_error *
180 3b6ceab7 2022-01-10 thomas append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
184 3b6ceab7 2022-01-10 thomas if (*len + nseg >= *sz) {
185 3b6ceab7 2022-01-10 thomas while (*len + nseg >= *sz)
186 3b6ceab7 2022-01-10 thomas *sz += *sz / 2;
187 3b6ceab7 2022-01-10 thomas n = realloc(*p, *sz);
188 3b6ceab7 2022-01-10 thomas if (n == NULL)
189 3b6ceab7 2022-01-10 thomas return got_error_from_errno("realloc");
192 3b6ceab7 2022-01-10 thomas memcpy(*p + *len, seg, nseg);
193 3b6ceab7 2022-01-10 thomas *len += nseg;
194 3b6ceab7 2022-01-10 thomas return NULL;
197 3b6ceab7 2022-01-10 thomas static const struct got_error *
198 3b6ceab7 2022-01-10 thomas encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
199 3b6ceab7 2022-01-10 thomas struct got_delta_instruction *deltas, int ndeltas,
200 3b6ceab7 2022-01-10 thomas off_t delta_size, off_t base_size)
202 3b6ceab7 2022-01-10 thomas const struct got_error *err;
203 3b6ceab7 2022-01-10 thomas unsigned char buf[16], *bp;
204 3b6ceab7 2022-01-10 thomas int i, j;
205 3b6ceab7 2022-01-10 thomas size_t len = 0;
207 3b6ceab7 2022-01-10 thomas struct got_delta_instruction *d;
209 3b6ceab7 2022-01-10 thomas m->delta_buf = malloc(delta_size);
210 3b6ceab7 2022-01-10 thomas if (m->delta_buf == NULL)
211 3b6ceab7 2022-01-10 thomas return got_error_from_errno("calloc");
213 3b6ceab7 2022-01-10 thomas /* base object size */
214 3b6ceab7 2022-01-10 thomas buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
215 3b6ceab7 2022-01-10 thomas n = base_size >> GOT_DELTA_SIZE_SHIFT;
216 3b6ceab7 2022-01-10 thomas for (i = 1; n > 0; i++) {
217 3b6ceab7 2022-01-10 thomas buf[i - 1] |= GOT_DELTA_SIZE_MORE;
218 3b6ceab7 2022-01-10 thomas buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
219 3b6ceab7 2022-01-10 thomas n >>= GOT_DELTA_SIZE_SHIFT;
221 3b6ceab7 2022-01-10 thomas err = append(&m->delta_buf, &len, &delta_size, buf, i);
223 3b6ceab7 2022-01-10 thomas return err;
225 3b6ceab7 2022-01-10 thomas /* target object size */
226 3b6ceab7 2022-01-10 thomas buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
227 3b6ceab7 2022-01-10 thomas n = o->size >> GOT_DELTA_SIZE_SHIFT;
228 3b6ceab7 2022-01-10 thomas for (i = 1; n > 0; i++) {
229 3b6ceab7 2022-01-10 thomas buf[i - 1] |= GOT_DELTA_SIZE_MORE;
230 3b6ceab7 2022-01-10 thomas buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
231 3b6ceab7 2022-01-10 thomas n >>= GOT_DELTA_SIZE_SHIFT;
233 3b6ceab7 2022-01-10 thomas err = append(&m->delta_buf, &len, &delta_size, buf, i);
235 3b6ceab7 2022-01-10 thomas return err;
237 3b6ceab7 2022-01-10 thomas for (j = 0; j < ndeltas; j++) {
238 3b6ceab7 2022-01-10 thomas d = &deltas[j];
239 3b6ceab7 2022-01-10 thomas if (d->copy) {
240 3b6ceab7 2022-01-10 thomas n = d->offset;
241 3b6ceab7 2022-01-10 thomas bp = &buf[1];
242 3b6ceab7 2022-01-10 thomas buf[0] = GOT_DELTA_BASE_COPY;
243 3b6ceab7 2022-01-10 thomas for (i = 0; i < 4; i++) {
244 3b6ceab7 2022-01-10 thomas /* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
245 3b6ceab7 2022-01-10 thomas buf[0] |= 1 << i;
246 3b6ceab7 2022-01-10 thomas *bp++ = n & 0xff;
248 3b6ceab7 2022-01-10 thomas if (n == 0)
252 3b6ceab7 2022-01-10 thomas n = d->len;
253 3b6ceab7 2022-01-10 thomas if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
254 3b6ceab7 2022-01-10 thomas /* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
255 3b6ceab7 2022-01-10 thomas for (i = 0; i < 3 && n > 0; i++) {
256 3b6ceab7 2022-01-10 thomas buf[0] |= 1 << (i + 4);
257 3b6ceab7 2022-01-10 thomas *bp++ = n & 0xff;
261 3b6ceab7 2022-01-10 thomas err = append(&m->delta_buf, &len, &delta_size,
262 3b6ceab7 2022-01-10 thomas buf, bp - buf);
264 3b6ceab7 2022-01-10 thomas return err;
265 3b6ceab7 2022-01-10 thomas } else if (o->f == NULL) {
267 3b6ceab7 2022-01-10 thomas while (n != d->len) {
268 3b6ceab7 2022-01-10 thomas buf[0] = (d->len - n < 127) ? d->len - n : 127;
269 3b6ceab7 2022-01-10 thomas err = append(&m->delta_buf, &len, &delta_size,
272 3b6ceab7 2022-01-10 thomas return err;
273 3b6ceab7 2022-01-10 thomas err = append(&m->delta_buf, &len, &delta_size,
274 3b6ceab7 2022-01-10 thomas o->data + o->hdrlen + d->offset + n,
277 3b6ceab7 2022-01-10 thomas return err;
278 3b6ceab7 2022-01-10 thomas n += buf[0];
281 3b6ceab7 2022-01-10 thomas char content[128];
282 3b6ceab7 2022-01-10 thomas size_t r;
283 3b6ceab7 2022-01-10 thomas if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
284 3b6ceab7 2022-01-10 thomas return got_error_from_errno("fseeko");
286 3b6ceab7 2022-01-10 thomas while (n != d->len) {
287 3b6ceab7 2022-01-10 thomas buf[0] = (d->len - n < 127) ? d->len - n : 127;
288 3b6ceab7 2022-01-10 thomas err = append(&m->delta_buf, &len, &delta_size,
291 3b6ceab7 2022-01-10 thomas return err;
292 3b6ceab7 2022-01-10 thomas r = fread(content, 1, buf[0], o->f);
293 3b6ceab7 2022-01-10 thomas if (r != buf[0])
294 3b6ceab7 2022-01-10 thomas return got_ferror(o->f, GOT_ERR_IO);
295 3b6ceab7 2022-01-10 thomas err = append(&m->delta_buf, &len, &delta_size,
296 3b6ceab7 2022-01-10 thomas content, buf[0]);
298 3b6ceab7 2022-01-10 thomas return err;
299 3b6ceab7 2022-01-10 thomas n += buf[0];
304 3b6ceab7 2022-01-10 thomas m->delta_len = len;
305 3b6ceab7 2022-01-10 thomas return NULL;
308 b79280dd 2021-10-15 thomas static const struct got_error *
309 b79280dd 2021-10-15 thomas encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
310 b79280dd 2021-10-15 thomas struct got_delta_instruction *deltas, int ndeltas,
311 d0772de9 2021-10-15 thomas off_t base_size, FILE *f)
313 d0772de9 2021-10-15 thomas unsigned char buf[16], *bp;
314 d0772de9 2021-10-15 thomas int i, j;
316 d0772de9 2021-10-15 thomas size_t w;
317 d0772de9 2021-10-15 thomas struct got_delta_instruction *d;
319 d0772de9 2021-10-15 thomas /* base object size */
320 d0772de9 2021-10-15 thomas buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
321 d0772de9 2021-10-15 thomas n = base_size >> GOT_DELTA_SIZE_SHIFT;
322 d0772de9 2021-10-15 thomas for (i = 1; n > 0; i++) {
323 d0772de9 2021-10-15 thomas buf[i - 1] |= GOT_DELTA_SIZE_MORE;
324 d0772de9 2021-10-15 thomas buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
325 d0772de9 2021-10-15 thomas n >>= GOT_DELTA_SIZE_SHIFT;
327 d0772de9 2021-10-15 thomas w = fwrite(buf, 1, i, f);
328 d0772de9 2021-10-15 thomas if (w != i)
329 d0772de9 2021-10-15 thomas return got_ferror(f, GOT_ERR_IO);
331 d0772de9 2021-10-15 thomas /* target object size */
332 d0772de9 2021-10-15 thomas buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
333 d0772de9 2021-10-15 thomas n = o->size >> GOT_DELTA_SIZE_SHIFT;
334 d0772de9 2021-10-15 thomas for (i = 1; n > 0; i++) {
335 d0772de9 2021-10-15 thomas buf[i - 1] |= GOT_DELTA_SIZE_MORE;
336 d0772de9 2021-10-15 thomas buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
337 d0772de9 2021-10-15 thomas n >>= GOT_DELTA_SIZE_SHIFT;
339 d0772de9 2021-10-15 thomas w = fwrite(buf, 1, i, f);
340 d0772de9 2021-10-15 thomas if (w != i)
341 d0772de9 2021-10-15 thomas return got_ferror(f, GOT_ERR_IO);
343 d0772de9 2021-10-15 thomas for (j = 0; j < ndeltas; j++) {
344 d0772de9 2021-10-15 thomas d = &deltas[j];
345 d0772de9 2021-10-15 thomas if (d->copy) {
346 d0772de9 2021-10-15 thomas n = d->offset;
347 d0772de9 2021-10-15 thomas bp = &buf[1];
348 d0772de9 2021-10-15 thomas buf[0] = GOT_DELTA_BASE_COPY;
349 d0772de9 2021-10-15 thomas for (i = 0; i < 4; i++) {
350 d0772de9 2021-10-15 thomas /* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
351 d0772de9 2021-10-15 thomas buf[0] |= 1 << i;
352 d0772de9 2021-10-15 thomas *bp++ = n & 0xff;
354 d0772de9 2021-10-15 thomas if (n == 0)
358 d0772de9 2021-10-15 thomas n = d->len;
359 d0772de9 2021-10-15 thomas if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
360 d0772de9 2021-10-15 thomas /* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
361 d0772de9 2021-10-15 thomas for (i = 0; i < 3 && n > 0; i++) {
362 d0772de9 2021-10-15 thomas buf[0] |= 1 << (i + 4);
363 d0772de9 2021-10-15 thomas *bp++ = n & 0xff;
367 d0772de9 2021-10-15 thomas w = fwrite(buf, 1, bp - buf, f);
368 d0772de9 2021-10-15 thomas if (w != bp - buf)
369 d0772de9 2021-10-15 thomas return got_ferror(f, GOT_ERR_IO);
370 2b0ae357 2022-01-10 thomas } else if (o->f == NULL) {
372 2b0ae357 2022-01-10 thomas while (n != d->len) {
373 2b0ae357 2022-01-10 thomas buf[0] = (d->len - n < 127) ? d->len - n : 127;
374 2b0ae357 2022-01-10 thomas w = fwrite(buf, 1, 1, f);
375 2b0ae357 2022-01-10 thomas if (w != 1)
376 2b0ae357 2022-01-10 thomas return got_ferror(f, GOT_ERR_IO);
377 2b0ae357 2022-01-10 thomas w = fwrite(o->data + o->hdrlen + d->offset + n,
378 2b0ae357 2022-01-10 thomas 1, buf[0], f);
379 2b0ae357 2022-01-10 thomas if (w != buf[0])
380 2b0ae357 2022-01-10 thomas return got_ferror(f, GOT_ERR_IO);
381 2b0ae357 2022-01-10 thomas n += buf[0];
384 d0772de9 2021-10-15 thomas char content[128];
385 d0772de9 2021-10-15 thomas size_t r;
386 d0772de9 2021-10-15 thomas if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
387 d0772de9 2021-10-15 thomas return got_error_from_errno("fseeko");
389 d0772de9 2021-10-15 thomas while (n != d->len) {
390 d0772de9 2021-10-15 thomas buf[0] = (d->len - n < 127) ? d->len - n : 127;
391 d0772de9 2021-10-15 thomas w = fwrite(buf, 1, 1, f);
392 d0772de9 2021-10-15 thomas if (w != 1)
393 d0772de9 2021-10-15 thomas return got_ferror(f, GOT_ERR_IO);
394 d0772de9 2021-10-15 thomas r = fread(content, 1, buf[0], o->f);
395 d0772de9 2021-10-15 thomas if (r != buf[0])
396 d0772de9 2021-10-15 thomas return got_ferror(o->f, GOT_ERR_IO);
397 d0772de9 2021-10-15 thomas w = fwrite(content, 1, buf[0], f);
398 d0772de9 2021-10-15 thomas if (w != buf[0])
399 d0772de9 2021-10-15 thomas return got_ferror(f, GOT_ERR_IO);
400 d0772de9 2021-10-15 thomas n += buf[0];
405 3b6ceab7 2022-01-10 thomas m->delta_len = ftello(f) - m->delta_offset;
406 d0772de9 2021-10-15 thomas return NULL;
409 31ba2236 2022-01-05 thomas static const struct got_error *
410 31ba2236 2022-01-05 thomas report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
411 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
412 85220b0e 2022-03-22 thomas off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
413 85220b0e 2022-03-22 thomas int nobj_written)
415 31ba2236 2022-01-05 thomas const struct got_error *err;
416 31ba2236 2022-01-05 thomas int elapsed;
418 31ba2236 2022-01-05 thomas if (progress_cb == NULL)
419 31ba2236 2022-01-05 thomas return NULL;
421 31ba2236 2022-01-05 thomas err = got_ratelimit_check(&elapsed, rl);
422 31ba2236 2022-01-05 thomas if (err || !elapsed)
423 31ba2236 2022-01-05 thomas return err;
425 85220b0e 2022-03-22 thomas return progress_cb(progress_arg, ncolored, nfound, ntrees,
426 85220b0e 2022-03-22 thomas packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
429 e6bcace5 2021-06-22 stsp static const struct got_error *
430 f9c2e8e5 2022-02-13 thomas add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
432 f9c2e8e5 2022-02-13 thomas if (v->nmeta == v->metasz){
433 f9c2e8e5 2022-02-13 thomas size_t newsize = 2 * v->metasz;
434 f9c2e8e5 2022-02-13 thomas struct got_pack_meta **new;
435 f9c2e8e5 2022-02-13 thomas new = reallocarray(v->meta, newsize, sizeof(*new));
436 f9c2e8e5 2022-02-13 thomas if (new == NULL)
437 f9c2e8e5 2022-02-13 thomas return got_error_from_errno("reallocarray");
438 f9c2e8e5 2022-02-13 thomas v->meta = new;
439 f9c2e8e5 2022-02-13 thomas v->metasz = newsize;
442 f9c2e8e5 2022-02-13 thomas v->meta[v->nmeta++] = m;
443 f9c2e8e5 2022-02-13 thomas return NULL;
446 f9c2e8e5 2022-02-13 thomas static const struct got_error *
447 f9c2e8e5 2022-02-13 thomas reuse_delta(int idx, struct got_pack_meta *m, struct got_pack_metavec *v,
448 f9c2e8e5 2022-02-13 thomas struct got_object_idset *idset, struct got_pack *pack,
449 f9c2e8e5 2022-02-13 thomas struct got_packidx *packidx, int delta_cache_fd,
450 f9c2e8e5 2022-02-13 thomas struct got_repository *repo)
452 f9c2e8e5 2022-02-13 thomas const struct got_error *err = NULL;
453 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *base = NULL;
454 f9c2e8e5 2022-02-13 thomas struct got_object_id *base_obj_id = NULL;
455 f9c2e8e5 2022-02-13 thomas off_t delta_len = 0, delta_offset = 0, delta_cache_offset = 0;
456 f9c2e8e5 2022-02-13 thomas uint64_t base_size, result_size;
458 f9c2e8e5 2022-02-13 thomas if (m->have_reused_delta)
459 f9c2e8e5 2022-02-13 thomas return NULL;
461 f9c2e8e5 2022-02-13 thomas err = got_object_read_raw_delta(&base_size, &result_size, &delta_len,
462 f9c2e8e5 2022-02-13 thomas &delta_offset, &delta_cache_offset, &base_obj_id, delta_cache_fd,
463 f9c2e8e5 2022-02-13 thomas packidx, idx, &m->id, repo);
465 f9c2e8e5 2022-02-13 thomas return err;
467 f9c2e8e5 2022-02-13 thomas if (delta_offset + delta_len < delta_offset)
468 f9c2e8e5 2022-02-13 thomas return got_error(GOT_ERR_BAD_PACKFILE);
470 f9c2e8e5 2022-02-13 thomas base = got_object_idset_get(idset, base_obj_id);
471 f9c2e8e5 2022-02-13 thomas if (base == NULL)
472 f9c2e8e5 2022-02-13 thomas goto done;
474 f9c2e8e5 2022-02-13 thomas m->delta_len = delta_len;
475 f9c2e8e5 2022-02-13 thomas m->delta_offset = delta_cache_offset;
476 f9c2e8e5 2022-02-13 thomas m->prev = base;
477 f9c2e8e5 2022-02-13 thomas m->size = result_size;
478 f9c2e8e5 2022-02-13 thomas m->have_reused_delta = 1;
479 f9c2e8e5 2022-02-13 thomas m->reused_delta_offset = delta_offset;
480 f9c2e8e5 2022-02-13 thomas m->base_obj_id = base_obj_id;
481 f9c2e8e5 2022-02-13 thomas base_obj_id = NULL;
482 f9c2e8e5 2022-02-13 thomas err = add_meta(m, v);
484 f9c2e8e5 2022-02-13 thomas free(base_obj_id);
485 f9c2e8e5 2022-02-13 thomas return err;
488 f9c2e8e5 2022-02-13 thomas static const struct got_error *
489 f9c2e8e5 2022-02-13 thomas find_pack_for_reuse(struct got_packidx **best_packidx,
490 f9c2e8e5 2022-02-13 thomas struct got_repository *repo)
492 de47d040 2022-03-22 thomas const struct got_error *err = NULL;
493 f9c2e8e5 2022-02-13 thomas struct got_pathlist_entry *pe;
494 f9c2e8e5 2022-02-13 thomas const char *best_packidx_path = NULL;
495 f9c2e8e5 2022-02-13 thomas int nobj_max = 0;
497 f9c2e8e5 2022-02-13 thomas *best_packidx = NULL;
499 de47d040 2022-03-22 thomas TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
500 f9c2e8e5 2022-02-13 thomas const char *path_packidx = pe->path;
501 f9c2e8e5 2022-02-13 thomas struct got_packidx *packidx;
502 f9c2e8e5 2022-02-13 thomas int nobj;
504 f9c2e8e5 2022-02-13 thomas err = got_repo_get_packidx(&packidx, path_packidx, repo);
508 f9c2e8e5 2022-02-13 thomas nobj = be32toh(packidx->hdr.fanout_table[0xff]);
509 f9c2e8e5 2022-02-13 thomas if (nobj > nobj_max) {
510 f9c2e8e5 2022-02-13 thomas best_packidx_path = path_packidx;
511 f9c2e8e5 2022-02-13 thomas nobj_max = nobj;
515 f9c2e8e5 2022-02-13 thomas if (best_packidx_path) {
516 f9c2e8e5 2022-02-13 thomas err = got_repo_get_packidx(best_packidx, best_packidx_path,
520 f9c2e8e5 2022-02-13 thomas return err;
523 f9c2e8e5 2022-02-13 thomas struct search_deltas_arg {
524 f9c2e8e5 2022-02-13 thomas struct got_packidx *packidx;
525 f9c2e8e5 2022-02-13 thomas struct got_pack *pack;
526 f9c2e8e5 2022-02-13 thomas struct got_object_idset *idset;
527 f9c2e8e5 2022-02-13 thomas struct got_pack_metavec *v;
528 f9c2e8e5 2022-02-13 thomas int delta_cache_fd;
529 f9c2e8e5 2022-02-13 thomas struct got_repository *repo;
530 f9c2e8e5 2022-02-13 thomas got_pack_progress_cb progress_cb;
531 f9c2e8e5 2022-02-13 thomas void *progress_arg;
532 f9c2e8e5 2022-02-13 thomas struct got_ratelimit *rl;
533 f9c2e8e5 2022-02-13 thomas got_cancel_cb cancel_cb;
534 f9c2e8e5 2022-02-13 thomas void *cancel_arg;
535 85220b0e 2022-03-22 thomas int ncolored;
536 85220b0e 2022-03-22 thomas int nfound;
537 85220b0e 2022-03-22 thomas int ntrees;
538 f9c2e8e5 2022-02-13 thomas int ncommits;
541 f9c2e8e5 2022-02-13 thomas static const struct got_error *
542 f9c2e8e5 2022-02-13 thomas search_delta_for_object(struct got_object_id *id, void *data, void *arg)
544 f9c2e8e5 2022-02-13 thomas const struct got_error *err;
545 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *m = data;
546 f9c2e8e5 2022-02-13 thomas struct search_deltas_arg *a = arg;
547 f9c2e8e5 2022-02-13 thomas int obj_idx;
548 f9c2e8e5 2022-02-13 thomas struct got_object *obj = NULL;
550 f9c2e8e5 2022-02-13 thomas if (a->cancel_cb) {
551 f9c2e8e5 2022-02-13 thomas err = (*a->cancel_cb)(a->cancel_arg);
553 f9c2e8e5 2022-02-13 thomas return err;
556 f9c2e8e5 2022-02-13 thomas if (!got_repo_check_packidx_bloom_filter(a->repo,
557 f9c2e8e5 2022-02-13 thomas a->packidx->path_packidx, id))
558 f9c2e8e5 2022-02-13 thomas return NULL;
560 f9c2e8e5 2022-02-13 thomas obj_idx = got_packidx_get_object_idx(a->packidx, id);
561 f9c2e8e5 2022-02-13 thomas if (obj_idx == -1)
562 f9c2e8e5 2022-02-13 thomas return NULL;
565 f9c2e8e5 2022-02-13 thomas * Opening and closing an object just to check its flags
566 f9c2e8e5 2022-02-13 thomas * is a bit expensive. We could have an imsg which requests
567 f9c2e8e5 2022-02-13 thomas * plain type/size information for an object without doing
568 f9c2e8e5 2022-02-13 thomas * work such as traversing the object's entire delta chain
569 f9c2e8e5 2022-02-13 thomas * to find the base object type, and other such info which
570 f9c2e8e5 2022-02-13 thomas * we don't really need here.
572 f9c2e8e5 2022-02-13 thomas err = got_object_open_from_packfile(&obj, &m->id, a->pack,
573 f9c2e8e5 2022-02-13 thomas a->packidx, obj_idx, a->repo);
575 f9c2e8e5 2022-02-13 thomas return err;
577 f9c2e8e5 2022-02-13 thomas if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
578 f9c2e8e5 2022-02-13 thomas reuse_delta(obj_idx, m, a->v, a->idset, a->pack, a->packidx,
579 f9c2e8e5 2022-02-13 thomas a->delta_cache_fd, a->repo);
581 f9c2e8e5 2022-02-13 thomas goto done;
582 f9c2e8e5 2022-02-13 thomas err = report_progress(a->progress_cb, a->progress_arg, a->rl,
583 85220b0e 2022-03-22 thomas a->ncolored, a->nfound, a->ntrees, 0L, a->ncommits,
584 85220b0e 2022-03-22 thomas got_object_idset_num_elements(a->idset), a->v->nmeta, 0);
587 f9c2e8e5 2022-02-13 thomas got_object_close(obj);
588 f9c2e8e5 2022-02-13 thomas return err;
591 f9c2e8e5 2022-02-13 thomas static const struct got_error *
592 f9c2e8e5 2022-02-13 thomas search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
593 85220b0e 2022-03-22 thomas int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
594 85220b0e 2022-03-22 thomas struct got_repository *repo,
595 f9c2e8e5 2022-02-13 thomas got_pack_progress_cb progress_cb, void *progress_arg,
596 f9c2e8e5 2022-02-13 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
598 f9c2e8e5 2022-02-13 thomas const struct got_error *err = NULL;
599 f9c2e8e5 2022-02-13 thomas char *path_packfile = NULL;
600 f9c2e8e5 2022-02-13 thomas struct got_packidx *packidx;
601 f9c2e8e5 2022-02-13 thomas struct got_pack *pack;
602 f9c2e8e5 2022-02-13 thomas struct search_deltas_arg sda;
604 f9c2e8e5 2022-02-13 thomas err = find_pack_for_reuse(&packidx, repo);
606 f9c2e8e5 2022-02-13 thomas return err;
608 f9c2e8e5 2022-02-13 thomas if (packidx == NULL)
609 f9c2e8e5 2022-02-13 thomas return NULL;
611 f9c2e8e5 2022-02-13 thomas err = got_packidx_get_packfile_path(&path_packfile,
612 f9c2e8e5 2022-02-13 thomas packidx->path_packidx);
614 f9c2e8e5 2022-02-13 thomas return err;
616 f9c2e8e5 2022-02-13 thomas pack = got_repo_get_cached_pack(repo, path_packfile);
617 f9c2e8e5 2022-02-13 thomas if (pack == NULL) {
618 f9c2e8e5 2022-02-13 thomas err = got_repo_cache_pack(&pack, repo, path_packfile, packidx);
620 f9c2e8e5 2022-02-13 thomas goto done;
623 f9c2e8e5 2022-02-13 thomas sda.packidx = packidx;
624 f9c2e8e5 2022-02-13 thomas sda.pack = pack;
625 f9c2e8e5 2022-02-13 thomas sda.idset = idset;
626 f9c2e8e5 2022-02-13 thomas sda.v = v;
627 f9c2e8e5 2022-02-13 thomas sda.delta_cache_fd = delta_cache_fd;
628 f9c2e8e5 2022-02-13 thomas sda.repo = repo;
629 f9c2e8e5 2022-02-13 thomas sda.progress_cb = progress_cb;
630 f9c2e8e5 2022-02-13 thomas sda.progress_arg = progress_arg;
631 f9c2e8e5 2022-02-13 thomas sda.rl = rl;
632 f9c2e8e5 2022-02-13 thomas sda.cancel_cb = cancel_cb;
633 f9c2e8e5 2022-02-13 thomas sda.cancel_arg = cancel_arg;
634 85220b0e 2022-03-22 thomas sda.ncolored = ncolored;
635 85220b0e 2022-03-22 thomas sda.nfound = nfound;
636 85220b0e 2022-03-22 thomas sda.ntrees = ntrees;
637 f9c2e8e5 2022-02-13 thomas sda.ncommits = ncommits;
638 f9c2e8e5 2022-02-13 thomas err = got_object_idset_for_each(idset, search_delta_for_object, &sda);
640 f9c2e8e5 2022-02-13 thomas free(path_packfile);
641 f9c2e8e5 2022-02-13 thomas return err;
644 f9c2e8e5 2022-02-13 thomas static const struct got_error *
645 85220b0e 2022-03-22 thomas pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
646 85220b0e 2022-03-22 thomas int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
647 85220b0e 2022-03-22 thomas struct got_repository *repo,
648 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
649 31ba2236 2022-01-05 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
651 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
652 e6bcace5 2021-06-22 stsp struct got_pack_meta *m = NULL, *base = NULL;
653 e6bcace5 2021-06-22 stsp struct got_raw_object *raw = NULL, *base_raw = NULL;
654 b79280dd 2021-10-15 thomas struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
655 3b6ceab7 2022-01-10 thomas int i, j, ndeltas, best_ndeltas;
656 3b6ceab7 2022-01-10 thomas off_t size, best_size;
657 959daf23 2021-10-29 thomas const int max_base_candidates = 3;
658 510885f7 2022-01-10 thomas size_t delta_memsize = 0;
659 510885f7 2022-01-10 thomas const size_t max_delta_memsize = 25 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
660 ecf9545f 2021-10-15 thomas int outfd = -1;
662 e6bcace5 2021-06-22 stsp qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
663 e6bcace5 2021-06-22 stsp for (i = 0; i < nmeta; i++) {
664 e6bcace5 2021-06-22 stsp if (cancel_cb) {
665 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
669 31ba2236 2022-01-05 thomas err = report_progress(progress_cb, progress_arg, rl,
670 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
671 85220b0e 2022-03-22 thomas nreused + i, 0);
673 31ba2236 2022-01-05 thomas goto done;
674 e6bcace5 2021-06-22 stsp m = meta[i];
676 e6bcace5 2021-06-22 stsp if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
677 e6bcace5 2021-06-22 stsp m->obj_type == GOT_OBJ_TYPE_TAG)
680 f8bb1d3e 2021-10-17 thomas err = got_object_raw_open(&raw, &outfd, repo, &m->id);
683 ab6186ae 2021-10-15 thomas m->size = raw->size;
685 2b0ae357 2022-01-10 thomas if (raw->f == NULL) {
686 2b0ae357 2022-01-10 thomas err = got_deltify_init_mem(&m->dtab, raw->data,
687 2b0ae357 2022-01-10 thomas raw->hdrlen, raw->size + raw->hdrlen);
689 2b0ae357 2022-01-10 thomas err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
690 2b0ae357 2022-01-10 thomas raw->size + raw->hdrlen);
695 e6bcace5 2021-06-22 stsp if (i > max_base_candidates) {
696 e6bcace5 2021-06-22 stsp struct got_pack_meta *n = NULL;
697 e6bcace5 2021-06-22 stsp n = meta[i - (max_base_candidates + 1)];
698 e6bcace5 2021-06-22 stsp got_deltify_free(n->dtab);
699 e6bcace5 2021-06-22 stsp n->dtab = NULL;
702 b79280dd 2021-10-15 thomas best_size = raw->size;
703 b79280dd 2021-10-15 thomas best_ndeltas = 0;
704 e6bcace5 2021-06-22 stsp for (j = MAX(0, i - max_base_candidates); j < i; j++) {
705 e6bcace5 2021-06-22 stsp if (cancel_cb) {
706 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
710 e6bcace5 2021-06-22 stsp base = meta[j];
711 e6bcace5 2021-06-22 stsp /* long chains make unpacking slow, avoid such bases */
712 62a05ae9 2021-10-29 thomas if (base->nchain >= 128 ||
713 e6bcace5 2021-06-22 stsp base->obj_type != m->obj_type)
716 8ab9215c 2021-10-15 thomas err = got_object_raw_open(&base_raw, &outfd, repo,
717 f8bb1d3e 2021-10-17 thomas &base->id);
721 2b0ae357 2022-01-10 thomas if (raw->f == NULL && base_raw->f == NULL) {
722 2b0ae357 2022-01-10 thomas err = got_deltify_mem_mem(&deltas, &ndeltas,
723 2b0ae357 2022-01-10 thomas raw->data, raw->hdrlen,
724 2b0ae357 2022-01-10 thomas raw->size + raw->hdrlen,
725 2b0ae357 2022-01-10 thomas base->dtab, base_raw->data,
726 2b0ae357 2022-01-10 thomas base_raw->hdrlen,
727 2b0ae357 2022-01-10 thomas base_raw->size + base_raw->hdrlen);
728 2b0ae357 2022-01-10 thomas } else if (raw->f == NULL) {
729 2b0ae357 2022-01-10 thomas err = got_deltify_mem_file(&deltas, &ndeltas,
730 2b0ae357 2022-01-10 thomas raw->data, raw->hdrlen,
731 2b0ae357 2022-01-10 thomas raw->size + raw->hdrlen,
732 2b0ae357 2022-01-10 thomas base->dtab, base_raw->f,
733 2b0ae357 2022-01-10 thomas base_raw->hdrlen,
734 2b0ae357 2022-01-10 thomas base_raw->size + base_raw->hdrlen);
735 2b0ae357 2022-01-10 thomas } else if (base_raw->f == NULL) {
736 2b0ae357 2022-01-10 thomas err = got_deltify_file_mem(&deltas, &ndeltas,
737 2b0ae357 2022-01-10 thomas raw->f, raw->hdrlen,
738 2b0ae357 2022-01-10 thomas raw->size + raw->hdrlen,
739 2b0ae357 2022-01-10 thomas base->dtab, base_raw->data,
740 2b0ae357 2022-01-10 thomas base_raw->hdrlen,
741 2b0ae357 2022-01-10 thomas base_raw->size + base_raw->hdrlen);
743 2b0ae357 2022-01-10 thomas err = got_deltify(&deltas, &ndeltas,
744 2b0ae357 2022-01-10 thomas raw->f, raw->hdrlen,
745 2b0ae357 2022-01-10 thomas raw->size + raw->hdrlen,
746 2b0ae357 2022-01-10 thomas base->dtab, base_raw->f, base_raw->hdrlen,
747 2b0ae357 2022-01-10 thomas base_raw->size + base_raw->hdrlen);
749 e6bcace5 2021-06-22 stsp got_object_raw_close(base_raw);
750 e6bcace5 2021-06-22 stsp base_raw = NULL;
754 e6bcace5 2021-06-22 stsp size = delta_size(deltas, ndeltas);
755 b79280dd 2021-10-15 thomas if (size + 32 < best_size){
757 e6bcace5 2021-06-22 stsp * if we already picked a best delta,
758 e6bcace5 2021-06-22 stsp * replace it.
760 b79280dd 2021-10-15 thomas best_size = size;
761 b79280dd 2021-10-15 thomas free(best_deltas);
762 b79280dd 2021-10-15 thomas best_deltas = deltas;
763 b79280dd 2021-10-15 thomas best_ndeltas = ndeltas;
764 b79280dd 2021-10-15 thomas deltas = NULL;
765 e6bcace5 2021-06-22 stsp m->nchain = base->nchain + 1;
766 e6bcace5 2021-06-22 stsp m->prev = base;
767 e6bcace5 2021-06-22 stsp m->head = base->head;
768 e6bcace5 2021-06-22 stsp if (m->head == NULL)
769 e6bcace5 2021-06-22 stsp m->head = base;
771 e6bcace5 2021-06-22 stsp free(deltas);
772 e6bcace5 2021-06-22 stsp deltas = NULL;
773 e6bcace5 2021-06-22 stsp ndeltas = 0;
777 b79280dd 2021-10-15 thomas if (best_ndeltas > 0) {
778 510885f7 2022-01-10 thomas if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
779 510885f7 2022-01-10 thomas delta_memsize + best_size <= max_delta_memsize) {
780 510885f7 2022-01-10 thomas delta_memsize += best_size;
781 3b6ceab7 2022-01-10 thomas err = encode_delta_in_mem(m, raw, best_deltas,
782 3b6ceab7 2022-01-10 thomas best_ndeltas, best_size, m->prev->size);
784 3b6ceab7 2022-01-10 thomas m->delta_offset = ftello(delta_cache);
787 f9c2e8e5 2022-02-13 thomas * Storing compressed delta data in the delta
788 f9c2e8e5 2022-02-13 thomas * cache file would probably be more efficient
789 f9c2e8e5 2022-02-13 thomas * than writing uncompressed delta data here
790 f9c2e8e5 2022-02-13 thomas * and compressing it while writing the pack
791 f9c2e8e5 2022-02-13 thomas * file. This would also allow for reusing
792 f9c2e8e5 2022-02-13 thomas * deltas in their compressed form.
794 3b6ceab7 2022-01-10 thomas err = encode_delta(m, raw, best_deltas,
795 3b6ceab7 2022-01-10 thomas best_ndeltas, m->prev->size, delta_cache);
797 b79280dd 2021-10-15 thomas free(best_deltas);
798 b79280dd 2021-10-15 thomas best_deltas = NULL;
799 b79280dd 2021-10-15 thomas best_ndeltas = 0;
801 b79280dd 2021-10-15 thomas goto done;
804 e6bcace5 2021-06-22 stsp got_object_raw_close(raw);
805 e6bcace5 2021-06-22 stsp raw = NULL;
808 e6bcace5 2021-06-22 stsp for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
809 e6bcace5 2021-06-22 stsp got_deltify_free(meta[i]->dtab);
810 e6bcace5 2021-06-22 stsp meta[i]->dtab = NULL;
813 e6bcace5 2021-06-22 stsp got_object_raw_close(raw);
814 e6bcace5 2021-06-22 stsp if (base_raw)
815 e6bcace5 2021-06-22 stsp got_object_raw_close(base_raw);
816 ecf9545f 2021-10-15 thomas if (outfd != -1 && close(outfd) == -1 && err == NULL)
817 ecf9545f 2021-10-15 thomas err = got_error_from_errno("close");
818 b79280dd 2021-10-15 thomas free(deltas);
819 b79280dd 2021-10-15 thomas free(best_deltas);
820 e6bcace5 2021-06-22 stsp return err;
823 e6bcace5 2021-06-22 stsp static const struct got_error *
824 05118f5a 2021-06-22 stsp search_packidx(int *found, struct got_object_id *id,
825 05118f5a 2021-06-22 stsp struct got_repository *repo)
827 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
828 05118f5a 2021-06-22 stsp struct got_packidx *packidx = NULL;
831 05118f5a 2021-06-22 stsp *found = 0;
833 05118f5a 2021-06-22 stsp err = got_repo_search_packidx(&packidx, &idx, repo, id);
834 05118f5a 2021-06-22 stsp if (err == NULL)
835 05118f5a 2021-06-22 stsp *found = 1; /* object is already packed */
836 05118f5a 2021-06-22 stsp else if (err->code == GOT_ERR_NO_OBJ)
837 05118f5a 2021-06-22 stsp err = NULL;
838 05118f5a 2021-06-22 stsp return err;
841 07165b17 2021-07-01 stsp static const int obj_types[] = {
842 07165b17 2021-07-01 stsp GOT_OBJ_TYPE_ANY,
843 07165b17 2021-07-01 stsp GOT_OBJ_TYPE_COMMIT,
844 07165b17 2021-07-01 stsp GOT_OBJ_TYPE_TREE,
845 07165b17 2021-07-01 stsp GOT_OBJ_TYPE_BLOB,
846 07165b17 2021-07-01 stsp GOT_OBJ_TYPE_TAG,
847 07165b17 2021-07-01 stsp GOT_OBJ_TYPE_OFFSET_DELTA,
848 07165b17 2021-07-01 stsp GOT_OBJ_TYPE_REF_DELTA
851 05118f5a 2021-06-22 stsp static const struct got_error *
852 f9c2e8e5 2022-02-13 thomas add_object(int want_meta, struct got_object_idset *idset,
853 e6bcace5 2021-06-22 stsp struct got_object_id *id, const char *path, int obj_type,
854 78a00876 2022-03-22 thomas time_t mtime, int loose_obj_only, struct got_repository *repo,
855 78a00876 2022-03-22 thomas int *ncolored, int *nfound, int *ntrees,
856 78a00876 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
857 78a00876 2022-03-22 thomas struct got_ratelimit *rl)
859 e6bcace5 2021-06-22 stsp const struct got_error *err;
860 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *m = NULL;
862 05118f5a 2021-06-22 stsp if (loose_obj_only) {
863 05118f5a 2021-06-22 stsp int is_packed;
864 05118f5a 2021-06-22 stsp err = search_packidx(&is_packed, id, repo);
866 05118f5a 2021-06-22 stsp return err;
867 a605f678 2022-03-22 thomas if (is_packed && want_meta)
868 05118f5a 2021-06-22 stsp return NULL;
871 f9c2e8e5 2022-02-13 thomas if (want_meta) {
872 f9c2e8e5 2022-02-13 thomas err = alloc_meta(&m, id, path, obj_type, mtime);
874 78a00876 2022-03-22 thomas return err;
876 78a00876 2022-03-22 thomas (*nfound)++;
877 78a00876 2022-03-22 thomas err = report_progress(progress_cb, progress_arg, rl,
878 78a00876 2022-03-22 thomas *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
880 f9c2e8e5 2022-02-13 thomas return err;
883 f9c2e8e5 2022-02-13 thomas return got_object_idset_add(idset, id, m);
886 e6bcace5 2021-06-22 stsp static const struct got_error *
887 f9c2e8e5 2022-02-13 thomas load_tree_entries(struct got_object_id_queue *ids, int want_meta,
888 e6bcace5 2021-06-22 stsp struct got_object_idset *idset, struct got_object_id *tree_id,
889 e6bcace5 2021-06-22 stsp const char *dpath, time_t mtime, struct got_repository *repo,
890 85220b0e 2022-03-22 thomas int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
891 85220b0e 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
892 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
894 e6bcace5 2021-06-22 stsp const struct got_error *err;
895 e6bcace5 2021-06-22 stsp struct got_tree_object *tree;
896 e6bcace5 2021-06-22 stsp char *p = NULL;
899 e6bcace5 2021-06-22 stsp err = got_object_open_as_tree(&tree, repo, tree_id);
901 85220b0e 2022-03-22 thomas return err;
903 85220b0e 2022-03-22 thomas (*ntrees)++;
904 85220b0e 2022-03-22 thomas err = report_progress(progress_cb, progress_arg, rl,
905 85220b0e 2022-03-22 thomas *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
907 e6bcace5 2021-06-22 stsp return err;
909 e6bcace5 2021-06-22 stsp for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
910 e6bcace5 2021-06-22 stsp struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
911 e6bcace5 2021-06-22 stsp struct got_object_id *id = got_tree_entry_get_id(e);
912 e6bcace5 2021-06-22 stsp mode_t mode = got_tree_entry_get_mode(e);
914 e6bcace5 2021-06-22 stsp if (cancel_cb) {
915 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
920 08511b5e 2021-09-24 thomas if (got_object_tree_entry_is_submodule(e) ||
921 e6bcace5 2021-06-22 stsp got_object_idset_contains(idset, id))
924 e6bcace5 2021-06-22 stsp if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
925 e6bcace5 2021-06-22 stsp got_tree_entry_get_name(e)) == -1) {
926 e6bcace5 2021-06-22 stsp err = got_error_from_errno("asprintf");
930 e6bcace5 2021-06-22 stsp if (S_ISDIR(mode)) {
931 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
932 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, id);
935 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(ids, qid, entry);
936 08511b5e 2021-09-24 thomas } else if (S_ISREG(mode) || S_ISLNK(mode)) {
937 f9c2e8e5 2022-02-13 thomas err = add_object(want_meta, idset, id, p,
938 78a00876 2022-03-22 thomas GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo,
939 78a00876 2022-03-22 thomas ncolored, nfound, ntrees,
940 78a00876 2022-03-22 thomas progress_cb, progress_arg, rl);
948 e6bcace5 2021-06-22 stsp got_object_tree_close(tree);
950 e6bcace5 2021-06-22 stsp return err;
953 e6bcace5 2021-06-22 stsp static const struct got_error *
954 f9c2e8e5 2022-02-13 thomas load_tree(int want_meta, struct got_object_idset *idset,
955 e6bcace5 2021-06-22 stsp struct got_object_id *tree_id, const char *dpath, time_t mtime,
956 85220b0e 2022-03-22 thomas struct got_repository *repo, int loose_obj_only,
957 85220b0e 2022-03-22 thomas int *ncolored, int *nfound, int *ntrees,
958 85220b0e 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
959 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
961 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
962 e6bcace5 2021-06-22 stsp struct got_object_id_queue tree_ids;
963 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
965 e6bcace5 2021-06-22 stsp if (got_object_idset_contains(idset, tree_id))
966 e6bcace5 2021-06-22 stsp return NULL;
968 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, tree_id);
970 e6bcace5 2021-06-22 stsp return err;
972 dbdddfee 2021-06-23 naddy STAILQ_INIT(&tree_ids);
973 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
975 dbdddfee 2021-06-23 naddy while (!STAILQ_EMPTY(&tree_ids)) {
976 e6bcace5 2021-06-22 stsp if (cancel_cb) {
977 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
982 dbdddfee 2021-06-23 naddy qid = STAILQ_FIRST(&tree_ids);
983 dbdddfee 2021-06-23 naddy STAILQ_REMOVE_HEAD(&tree_ids, entry);
985 e6bcace5 2021-06-22 stsp if (got_object_idset_contains(idset, qid->id)) {
986 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
990 f9c2e8e5 2022-02-13 thomas err = add_object(want_meta, idset, qid->id, dpath,
991 78a00876 2022-03-22 thomas GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo,
992 78a00876 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
994 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
998 f9c2e8e5 2022-02-13 thomas err = load_tree_entries(&tree_ids, want_meta, idset, qid->id,
999 85220b0e 2022-03-22 thomas dpath, mtime, repo, loose_obj_only, ncolored, nfound,
1000 85220b0e 2022-03-22 thomas ntrees, progress_cb, progress_arg, rl,
1001 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
1002 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1007 e6bcace5 2021-06-22 stsp got_object_id_queue_free(&tree_ids);
1008 e6bcace5 2021-06-22 stsp return err;
1011 e6bcace5 2021-06-22 stsp static const struct got_error *
1012 f9c2e8e5 2022-02-13 thomas load_commit(int want_meta, struct got_object_idset *idset,
1013 05118f5a 2021-06-22 stsp struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1014 85220b0e 2022-03-22 thomas int *ncolored, int *nfound, int *ntrees,
1015 85220b0e 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
1016 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1018 e6bcace5 2021-06-22 stsp const struct got_error *err;
1019 e6bcace5 2021-06-22 stsp struct got_commit_object *commit;
1021 e6bcace5 2021-06-22 stsp if (got_object_idset_contains(idset, id))
1022 e6bcace5 2021-06-22 stsp return NULL;
1024 05118f5a 2021-06-22 stsp if (loose_obj_only) {
1025 05118f5a 2021-06-22 stsp int is_packed;
1026 05118f5a 2021-06-22 stsp err = search_packidx(&is_packed, id, repo);
1028 05118f5a 2021-06-22 stsp return err;
1029 a605f678 2022-03-22 thomas if (is_packed && want_meta)
1030 05118f5a 2021-06-22 stsp return NULL;
1033 e6bcace5 2021-06-22 stsp err = got_object_open_as_commit(&commit, repo, id);
1035 e6bcace5 2021-06-22 stsp return err;
1037 f9c2e8e5 2022-02-13 thomas err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_COMMIT,
1038 05118f5a 2021-06-22 stsp got_object_commit_get_committer_time(commit),
1039 78a00876 2022-03-22 thomas loose_obj_only, repo,
1040 78a00876 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1042 e6bcace5 2021-06-22 stsp goto done;
1044 f9c2e8e5 2022-02-13 thomas err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1045 05118f5a 2021-06-22 stsp "", got_object_commit_get_committer_time(commit),
1046 85220b0e 2022-03-22 thomas repo, loose_obj_only, ncolored, nfound, ntrees,
1047 85220b0e 2022-03-22 thomas progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1049 e6bcace5 2021-06-22 stsp got_object_commit_close(commit);
1050 e6bcace5 2021-06-22 stsp return err;
1053 05118f5a 2021-06-22 stsp static const struct got_error *
1054 f9c2e8e5 2022-02-13 thomas load_tag(int want_meta, struct got_object_idset *idset,
1055 05118f5a 2021-06-22 stsp struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1056 85220b0e 2022-03-22 thomas int *ncolored, int *nfound, int *ntrees,
1057 85220b0e 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
1058 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1060 05118f5a 2021-06-22 stsp const struct got_error *err;
1061 05118f5a 2021-06-22 stsp struct got_tag_object *tag = NULL;
1063 05118f5a 2021-06-22 stsp if (got_object_idset_contains(idset, id))
1064 05118f5a 2021-06-22 stsp return NULL;
1066 05118f5a 2021-06-22 stsp if (loose_obj_only) {
1067 05118f5a 2021-06-22 stsp int is_packed;
1068 05118f5a 2021-06-22 stsp err = search_packidx(&is_packed, id, repo);
1070 05118f5a 2021-06-22 stsp return err;
1071 a605f678 2022-03-22 thomas if (is_packed && want_meta)
1072 05118f5a 2021-06-22 stsp return NULL;
1075 05118f5a 2021-06-22 stsp err = got_object_open_as_tag(&tag, repo, id);
1077 05118f5a 2021-06-22 stsp return err;
1079 f9c2e8e5 2022-02-13 thomas err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1080 78a00876 2022-03-22 thomas got_object_tag_get_tagger_time(tag), loose_obj_only, repo,
1081 78a00876 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1083 05118f5a 2021-06-22 stsp goto done;
1085 05118f5a 2021-06-22 stsp switch (got_object_tag_get_object_type(tag)) {
1086 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_COMMIT:
1087 f9c2e8e5 2022-02-13 thomas err = load_commit(want_meta, idset,
1088 85220b0e 2022-03-22 thomas got_object_tag_get_object_id(tag), repo, loose_obj_only,
1089 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1090 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
1092 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_TREE:
1093 f9c2e8e5 2022-02-13 thomas err = load_tree(want_meta, idset,
1094 f9c2e8e5 2022-02-13 thomas got_object_tag_get_object_id(tag), "",
1095 85220b0e 2022-03-22 thomas got_object_tag_get_tagger_time(tag), repo, loose_obj_only,
1096 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1097 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
1104 05118f5a 2021-06-22 stsp got_object_tag_close(tag);
1105 05118f5a 2021-06-22 stsp return err;
1108 e6bcace5 2021-06-22 stsp enum findtwixt_color {
1109 e6bcace5 2021-06-22 stsp COLOR_KEEP = 0,
1110 e6bcace5 2021-06-22 stsp COLOR_DROP,
1111 e6bcace5 2021-06-22 stsp COLOR_BLANK,
1113 e6bcace5 2021-06-22 stsp static const int findtwixt_colors[] = {
1114 e6bcace5 2021-06-22 stsp COLOR_KEEP,
1115 e6bcace5 2021-06-22 stsp COLOR_DROP,
1116 e6bcace5 2021-06-22 stsp COLOR_BLANK
1119 e6bcace5 2021-06-22 stsp static const struct got_error *
1120 e6bcace5 2021-06-22 stsp queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1121 e6bcace5 2021-06-22 stsp int color, struct got_repository *repo)
1123 e6bcace5 2021-06-22 stsp const struct got_error *err;
1124 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
1126 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, id);
1128 e6bcace5 2021-06-22 stsp return err;
1130 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(ids, qid, entry);
1131 e6bcace5 2021-06-22 stsp qid->data = (void *)&findtwixt_colors[color];
1132 e6bcace5 2021-06-22 stsp return NULL;
1135 e6bcace5 2021-06-22 stsp static const struct got_error *
1136 e6bcace5 2021-06-22 stsp drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
1137 e6bcace5 2021-06-22 stsp struct got_object_id *id, struct got_repository *repo,
1138 e6bcace5 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1140 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1141 e6bcace5 2021-06-22 stsp struct got_commit_object *commit;
1142 e6bcace5 2021-06-22 stsp const struct got_object_id_queue *parents;
1143 e6bcace5 2021-06-22 stsp struct got_object_id_queue ids;
1144 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
1146 dbdddfee 2021-06-23 naddy STAILQ_INIT(&ids);
1148 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, id);
1150 e6bcace5 2021-06-22 stsp return err;
1151 dbdddfee 2021-06-23 naddy STAILQ_INSERT_HEAD(&ids, qid, entry);
1153 dbdddfee 2021-06-23 naddy while (!STAILQ_EMPTY(&ids)) {
1154 e6bcace5 2021-06-22 stsp if (cancel_cb) {
1155 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
1160 dbdddfee 2021-06-23 naddy qid = STAILQ_FIRST(&ids);
1161 dbdddfee 2021-06-23 naddy STAILQ_REMOVE_HEAD(&ids, entry);
1163 e6bcace5 2021-06-22 stsp if (got_object_idset_contains(drop, qid->id)) {
1164 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1168 07165b17 2021-07-01 stsp err = got_object_idset_add(drop, qid->id,
1169 07165b17 2021-07-01 stsp (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1170 e6bcace5 2021-06-22 stsp if (err) {
1171 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1175 e6bcace5 2021-06-22 stsp if (!got_object_idset_contains(keep, qid->id)) {
1176 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1180 e6bcace5 2021-06-22 stsp err = got_object_open_as_commit(&commit, repo, qid->id);
1181 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1185 e6bcace5 2021-06-22 stsp parents = got_object_commit_get_parent_ids(commit);
1186 e6bcace5 2021-06-22 stsp if (parents) {
1187 e6bcace5 2021-06-22 stsp err = got_object_id_queue_copy(parents, &ids);
1188 e6bcace5 2021-06-22 stsp if (err) {
1189 e6bcace5 2021-06-22 stsp got_object_commit_close(commit);
1193 e6bcace5 2021-06-22 stsp got_object_commit_close(commit);
1196 e6bcace5 2021-06-22 stsp got_object_id_queue_free(&ids);
1197 e6bcace5 2021-06-22 stsp return err;
1200 e6bcace5 2021-06-22 stsp struct append_id_arg {
1201 e6bcace5 2021-06-22 stsp struct got_object_id **array;
1205 e6bcace5 2021-06-22 stsp static const struct got_error *
1206 e6bcace5 2021-06-22 stsp append_id(struct got_object_id *id, void *data, void *arg)
1208 e6bcace5 2021-06-22 stsp struct append_id_arg *a = arg;
1210 e6bcace5 2021-06-22 stsp a->array[a->idx] = got_object_id_dup(id);
1211 e6bcace5 2021-06-22 stsp if (a->array[a->idx] == NULL)
1212 e6bcace5 2021-06-22 stsp return got_error_from_errno("got_object_id_dup");
1215 e6bcace5 2021-06-22 stsp return NULL;
1218 e6bcace5 2021-06-22 stsp static const struct got_error *
1219 85220b0e 2022-03-22 thomas findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1220 e6bcace5 2021-06-22 stsp struct got_object_id **head, int nhead,
1221 e6bcace5 2021-06-22 stsp struct got_object_id **tail, int ntail,
1222 e6bcace5 2021-06-22 stsp struct got_repository *repo,
1223 85220b0e 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
1224 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1226 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1227 e6bcace5 2021-06-22 stsp struct got_object_id_queue ids;
1228 e6bcace5 2021-06-22 stsp struct got_object_idset *keep, *drop;
1229 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
1230 05118f5a 2021-06-22 stsp int i, ncolor, nkeep, obj_type;
1232 dbdddfee 2021-06-23 naddy STAILQ_INIT(&ids);
1233 e6bcace5 2021-06-22 stsp *res = NULL;
1234 e6bcace5 2021-06-22 stsp *nres = 0;
1235 85220b0e 2022-03-22 thomas *ncolored = 0;
1237 e6bcace5 2021-06-22 stsp keep = got_object_idset_alloc();
1238 e6bcace5 2021-06-22 stsp if (keep == NULL)
1239 e6bcace5 2021-06-22 stsp return got_error_from_errno("got_object_idset_alloc");
1241 e6bcace5 2021-06-22 stsp drop = got_object_idset_alloc();
1242 e6bcace5 2021-06-22 stsp if (drop == NULL) {
1243 e6bcace5 2021-06-22 stsp err = got_error_from_errno("got_object_idset_alloc");
1244 e6bcace5 2021-06-22 stsp goto done;
1247 05118f5a 2021-06-22 stsp for (i = 0; i < nhead; i++) {
1248 05118f5a 2021-06-22 stsp struct got_object_id *id = head[i];
1249 05118f5a 2021-06-22 stsp if (id == NULL)
1251 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
1253 05118f5a 2021-06-22 stsp return err;
1254 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_COMMIT)
1256 05118f5a 2021-06-22 stsp err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
1258 05118f5a 2021-06-22 stsp goto done;
1260 05118f5a 2021-06-22 stsp for (i = 0; i < ntail; i++) {
1261 05118f5a 2021-06-22 stsp struct got_object_id *id = tail[i];
1262 05118f5a 2021-06-22 stsp if (id == NULL)
1264 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
1266 05118f5a 2021-06-22 stsp return err;
1267 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_COMMIT)
1269 05118f5a 2021-06-22 stsp err = queue_commit_id(&ids, id, COLOR_DROP, repo);
1271 05118f5a 2021-06-22 stsp goto done;
1274 dbdddfee 2021-06-23 naddy while (!STAILQ_EMPTY(&ids)) {
1275 e6bcace5 2021-06-22 stsp int qcolor;
1276 dbdddfee 2021-06-23 naddy qid = STAILQ_FIRST(&ids);
1277 e6bcace5 2021-06-22 stsp qcolor = *((int *)qid->data);
1279 e6bcace5 2021-06-22 stsp if (got_object_idset_contains(drop, qid->id))
1280 e6bcace5 2021-06-22 stsp ncolor = COLOR_DROP;
1281 e6bcace5 2021-06-22 stsp else if (got_object_idset_contains(keep, qid->id))
1282 e6bcace5 2021-06-22 stsp ncolor = COLOR_KEEP;
1284 e6bcace5 2021-06-22 stsp ncolor = COLOR_BLANK;
1286 85220b0e 2022-03-22 thomas (*ncolored)++;
1287 85220b0e 2022-03-22 thomas err = report_progress(progress_cb, progress_arg, rl,
1288 85220b0e 2022-03-22 thomas *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1289 85220b0e 2022-03-22 thomas if (err)
1290 85220b0e 2022-03-22 thomas goto done;
1292 e6bcace5 2021-06-22 stsp if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
1293 e6bcace5 2021-06-22 stsp qcolor == COLOR_KEEP)) {
1294 dbdddfee 2021-06-23 naddy STAILQ_REMOVE_HEAD(&ids, entry);
1295 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1299 e6bcace5 2021-06-22 stsp if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
1300 e6bcace5 2021-06-22 stsp err = drop_commit(keep, drop, qid->id, repo,
1301 e6bcace5 2021-06-22 stsp cancel_cb, cancel_arg);
1303 e6bcace5 2021-06-22 stsp goto done;
1304 e6bcace5 2021-06-22 stsp } else if (ncolor == COLOR_BLANK) {
1305 e6bcace5 2021-06-22 stsp struct got_commit_object *commit;
1306 e6bcace5 2021-06-22 stsp struct got_object_id *id;
1307 e6bcace5 2021-06-22 stsp const struct got_object_id_queue *parents;
1308 e6bcace5 2021-06-22 stsp struct got_object_qid *pid;
1310 e6bcace5 2021-06-22 stsp id = got_object_id_dup(qid->id);
1311 e6bcace5 2021-06-22 stsp if (id == NULL) {
1312 e6bcace5 2021-06-22 stsp err = got_error_from_errno("got_object_id_dup");
1313 e6bcace5 2021-06-22 stsp goto done;
1315 e6bcace5 2021-06-22 stsp if (qcolor == COLOR_KEEP)
1316 07165b17 2021-07-01 stsp err = got_object_idset_add(keep, id,
1317 07165b17 2021-07-01 stsp (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1319 07165b17 2021-07-01 stsp err = got_object_idset_add(drop, id,
1320 07165b17 2021-07-01 stsp (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1321 e6bcace5 2021-06-22 stsp if (err) {
1323 e6bcace5 2021-06-22 stsp goto done;
1326 e6bcace5 2021-06-22 stsp err = got_object_open_as_commit(&commit, repo, id);
1327 e6bcace5 2021-06-22 stsp if (err) {
1329 e6bcace5 2021-06-22 stsp goto done;
1331 e6bcace5 2021-06-22 stsp parents = got_object_commit_get_parent_ids(commit);
1332 e6bcace5 2021-06-22 stsp if (parents) {
1333 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(pid, parents, entry) {
1334 e6bcace5 2021-06-22 stsp err = queue_commit_id(&ids, pid->id,
1335 e6bcace5 2021-06-22 stsp qcolor, repo);
1336 e6bcace5 2021-06-22 stsp if (err) {
1338 e6bcace5 2021-06-22 stsp goto done;
1342 e6bcace5 2021-06-22 stsp got_object_commit_close(commit);
1343 e6bcace5 2021-06-22 stsp commit = NULL;
1345 e6bcace5 2021-06-22 stsp /* should not happen */
1346 e6bcace5 2021-06-22 stsp err = got_error_fmt(GOT_ERR_NOT_IMPL,
1347 e6bcace5 2021-06-22 stsp "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
1348 e6bcace5 2021-06-22 stsp goto done;
1351 dbdddfee 2021-06-23 naddy STAILQ_REMOVE_HEAD(&ids, entry);
1352 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1355 e6bcace5 2021-06-22 stsp nkeep = got_object_idset_num_elements(keep);
1356 e6bcace5 2021-06-22 stsp if (nkeep > 0) {
1357 e6bcace5 2021-06-22 stsp struct append_id_arg arg;
1358 e6bcace5 2021-06-22 stsp arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1359 e6bcace5 2021-06-22 stsp if (arg.array == NULL) {
1360 e6bcace5 2021-06-22 stsp err = got_error_from_errno("calloc");
1361 e6bcace5 2021-06-22 stsp goto done;
1363 e6bcace5 2021-06-22 stsp arg.idx = 0;
1364 e6bcace5 2021-06-22 stsp err = got_object_idset_for_each(keep, append_id, &arg);
1365 e6bcace5 2021-06-22 stsp if (err) {
1366 e6bcace5 2021-06-22 stsp free(arg.array);
1367 e6bcace5 2021-06-22 stsp goto done;
1369 e6bcace5 2021-06-22 stsp *res = arg.array;
1370 e6bcace5 2021-06-22 stsp *nres = nkeep;
1373 e6bcace5 2021-06-22 stsp got_object_idset_free(keep);
1374 e6bcace5 2021-06-22 stsp got_object_idset_free(drop);
1375 e6bcace5 2021-06-22 stsp got_object_id_queue_free(&ids);
1376 e6bcace5 2021-06-22 stsp return err;
1379 e6bcace5 2021-06-22 stsp static const struct got_error *
1380 85220b0e 2022-03-22 thomas load_object_ids(int *ncolored, int *nfound, int *ntrees,
1381 85220b0e 2022-03-22 thomas struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1382 e6bcace5 2021-06-22 stsp struct got_object_id **ours, int nours, struct got_repository *repo,
1383 85220b0e 2022-03-22 thomas int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
1384 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1386 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1387 e6bcace5 2021-06-22 stsp struct got_object_id **ids = NULL;
1388 05118f5a 2021-06-22 stsp int i, nobj = 0, obj_type;
1390 85220b0e 2022-03-22 thomas *ncolored = 0;
1391 85220b0e 2022-03-22 thomas *nfound = 0;
1392 85220b0e 2022-03-22 thomas *ntrees = 0;
1394 85220b0e 2022-03-22 thomas err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1395 85220b0e 2022-03-22 thomas repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1396 e6bcace5 2021-06-22 stsp if (err || nobj == 0)
1397 e6bcace5 2021-06-22 stsp goto done;
1399 e6bcace5 2021-06-22 stsp for (i = 0; i < ntheirs; i++) {
1400 05118f5a 2021-06-22 stsp struct got_object_id *id = theirs[i];
1401 05118f5a 2021-06-22 stsp if (id == NULL)
1403 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
1405 05118f5a 2021-06-22 stsp return err;
1406 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_COMMIT)
1408 85220b0e 2022-03-22 thomas err = load_commit(0, idset, id, repo, loose_obj_only,
1409 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1410 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
1412 05118f5a 2021-06-22 stsp goto done;
1415 05118f5a 2021-06-22 stsp for (i = 0; i < ntheirs; i++) {
1416 f4a2ff2d 2021-07-01 stsp struct got_object_id *id = theirs[i];
1417 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *m;
1418 05118f5a 2021-06-22 stsp if (id == NULL)
1420 f9c2e8e5 2022-02-13 thomas m = got_object_idset_get(idset, id);
1421 f9c2e8e5 2022-02-13 thomas if (m == NULL) {
1422 07165b17 2021-07-01 stsp err = got_object_get_type(&obj_type, repo, id);
1424 07165b17 2021-07-01 stsp goto done;
1426 f9c2e8e5 2022-02-13 thomas obj_type = m->obj_type;
1427 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_TAG)
1429 85220b0e 2022-03-22 thomas err = load_tag(0, idset, id, repo, loose_obj_only,
1430 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1431 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
1433 05118f5a 2021-06-22 stsp goto done;
1436 eca70f98 2021-09-03 stsp for (i = 0; i < nobj; i++) {
1437 85220b0e 2022-03-22 thomas err = load_commit(1, idset, ids[i], repo, loose_obj_only,
1438 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1439 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
1441 eca70f98 2021-09-03 stsp goto done;
1444 05118f5a 2021-06-22 stsp for (i = 0; i < nours; i++) {
1445 05118f5a 2021-06-22 stsp struct got_object_id *id = ours[i];
1446 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *m;
1447 05118f5a 2021-06-22 stsp if (id == NULL)
1449 f9c2e8e5 2022-02-13 thomas m = got_object_idset_get(idset, id);
1450 f9c2e8e5 2022-02-13 thomas if (m == NULL) {
1451 07165b17 2021-07-01 stsp err = got_object_get_type(&obj_type, repo, id);
1453 07165b17 2021-07-01 stsp goto done;
1455 f9c2e8e5 2022-02-13 thomas obj_type = m->obj_type;
1456 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_TAG)
1458 85220b0e 2022-03-22 thomas err = load_tag(1, idset, id, repo, loose_obj_only,
1459 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1460 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
1462 e6bcace5 2021-06-22 stsp goto done;
1465 e6bcace5 2021-06-22 stsp for (i = 0; i < nobj; i++) {
1466 e6bcace5 2021-06-22 stsp free(ids[i]);
1468 e6bcace5 2021-06-22 stsp free(ids);
1469 e6bcace5 2021-06-22 stsp return err;
1472 e6bcace5 2021-06-22 stsp const struct got_error *
1473 e6bcace5 2021-06-22 stsp hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1477 e6bcace5 2021-06-22 stsp SHA1Update(ctx, buf, len);
1478 e6bcace5 2021-06-22 stsp n = fwrite(buf, 1, len, f);
1479 e6bcace5 2021-06-22 stsp if (n != len)
1480 e6bcace5 2021-06-22 stsp return got_ferror(f, GOT_ERR_IO);
1481 e6bcace5 2021-06-22 stsp return NULL;
1484 e6bcace5 2021-06-22 stsp static void
1485 e6bcace5 2021-06-22 stsp putbe32(char *b, uint32_t n)
1487 e6bcace5 2021-06-22 stsp b[0] = n >> 24;
1488 e6bcace5 2021-06-22 stsp b[1] = n >> 16;
1489 e6bcace5 2021-06-22 stsp b[2] = n >> 8;
1490 e6bcace5 2021-06-22 stsp b[3] = n >> 0;
1493 e6bcace5 2021-06-22 stsp static int
1494 e6bcace5 2021-06-22 stsp write_order_cmp(const void *pa, const void *pb)
1496 e6bcace5 2021-06-22 stsp struct got_pack_meta *a, *b, *ahd, *bhd;
1498 e6bcace5 2021-06-22 stsp a = *(struct got_pack_meta **)pa;
1499 e6bcace5 2021-06-22 stsp b = *(struct got_pack_meta **)pb;
1500 e6bcace5 2021-06-22 stsp ahd = (a->head == NULL) ? a : a->head;
1501 e6bcace5 2021-06-22 stsp bhd = (b->head == NULL) ? b : b->head;
1502 e6bcace5 2021-06-22 stsp if (ahd->mtime != bhd->mtime)
1503 e6bcace5 2021-06-22 stsp return bhd->mtime - ahd->mtime;
1504 e6bcace5 2021-06-22 stsp if (ahd != bhd)
1505 e6bcace5 2021-06-22 stsp return (uintptr_t)bhd - (uintptr_t)ahd;
1506 e6bcace5 2021-06-22 stsp if (a->nchain != b->nchain)
1507 e6bcace5 2021-06-22 stsp return a->nchain - b->nchain;
1508 e6bcace5 2021-06-22 stsp return a->mtime - b->mtime;
1511 f9c2e8e5 2022-02-13 thomas static int
1512 f9c2e8e5 2022-02-13 thomas reuse_write_order_cmp(const void *pa, const void *pb)
1514 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *a, *b;
1516 f9c2e8e5 2022-02-13 thomas a = *(struct got_pack_meta **)pa;
1517 f9c2e8e5 2022-02-13 thomas b = *(struct got_pack_meta **)pb;
1519 f9c2e8e5 2022-02-13 thomas if (a->reused_delta_offset < b->reused_delta_offset)
1520 f9c2e8e5 2022-02-13 thomas return -1;
1521 f9c2e8e5 2022-02-13 thomas if (a->reused_delta_offset > b->reused_delta_offset)
1522 f9c2e8e5 2022-02-13 thomas return 1;
1523 f9c2e8e5 2022-02-13 thomas return 0;
1526 e6bcace5 2021-06-22 stsp static const struct got_error *
1527 e6bcace5 2021-06-22 stsp packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1531 e6bcace5 2021-06-22 stsp *hdrlen = 0;
1533 e6bcace5 2021-06-22 stsp hdr[0] = obj_type << 4;
1534 e6bcace5 2021-06-22 stsp hdr[0] |= len & 0xf;
1535 e6bcace5 2021-06-22 stsp len >>= 4;
1536 e6bcace5 2021-06-22 stsp for (i = 1; len != 0; i++){
1537 e6bcace5 2021-06-22 stsp if (i >= bufsize)
1538 e6bcace5 2021-06-22 stsp return got_error(GOT_ERR_NO_SPACE);
1539 e6bcace5 2021-06-22 stsp hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1540 e6bcace5 2021-06-22 stsp hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1541 e6bcace5 2021-06-22 stsp len >>= GOT_DELTA_SIZE_SHIFT;
1544 e6bcace5 2021-06-22 stsp *hdrlen = i;
1545 e6bcace5 2021-06-22 stsp return NULL;
1548 e6bcace5 2021-06-22 stsp static int
1549 e6bcace5 2021-06-22 stsp packoff(char *hdr, off_t off)
1552 e6bcace5 2021-06-22 stsp char rbuf[8];
1554 e6bcace5 2021-06-22 stsp rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1555 e6bcace5 2021-06-22 stsp for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1556 e6bcace5 2021-06-22 stsp rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1557 e6bcace5 2021-06-22 stsp GOT_DELTA_SIZE_MORE;
1561 e6bcace5 2021-06-22 stsp while (i > 0)
1562 e6bcace5 2021-06-22 stsp hdr[j++] = rbuf[--i];
1566 e6bcace5 2021-06-22 stsp static const struct got_error *
1567 3b6ceab7 2022-01-10 thomas deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1568 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *m)
1570 3b6ceab7 2022-01-10 thomas const struct got_error *err;
1571 3b6ceab7 2022-01-10 thomas char buf[32];
1574 f9c2e8e5 2022-02-13 thomas if (m->prev->off != 0) {
1575 3b6ceab7 2022-01-10 thomas err = packhdr(&nh, buf, sizeof(buf),
1576 3b6ceab7 2022-01-10 thomas GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1577 3b6ceab7 2022-01-10 thomas if (err)
1578 3b6ceab7 2022-01-10 thomas return err;
1579 3b6ceab7 2022-01-10 thomas nh += packoff(buf + nh, m->off - m->prev->off);
1580 3b6ceab7 2022-01-10 thomas err = hwrite(packfile, buf, nh, ctx);
1581 3b6ceab7 2022-01-10 thomas if (err)
1582 3b6ceab7 2022-01-10 thomas return err;
1583 3b6ceab7 2022-01-10 thomas *packfile_size += nh;
1584 3b6ceab7 2022-01-10 thomas } else {
1585 3b6ceab7 2022-01-10 thomas err = packhdr(&nh, buf, sizeof(buf),
1586 3b6ceab7 2022-01-10 thomas GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1587 3b6ceab7 2022-01-10 thomas if (err)
1588 3b6ceab7 2022-01-10 thomas return err;
1589 3b6ceab7 2022-01-10 thomas err = hwrite(packfile, buf, nh, ctx);
1590 3b6ceab7 2022-01-10 thomas if (err)
1591 3b6ceab7 2022-01-10 thomas return err;
1592 3b6ceab7 2022-01-10 thomas *packfile_size += nh;
1593 3b6ceab7 2022-01-10 thomas err = hwrite(packfile, m->prev->id.sha1,
1594 3b6ceab7 2022-01-10 thomas sizeof(m->prev->id.sha1), ctx);
1595 3b6ceab7 2022-01-10 thomas if (err)
1596 3b6ceab7 2022-01-10 thomas return err;
1597 3b6ceab7 2022-01-10 thomas *packfile_size += sizeof(m->prev->id.sha1);
1600 3b6ceab7 2022-01-10 thomas return NULL;
1603 3b6ceab7 2022-01-10 thomas static const struct got_error *
1604 f9c2e8e5 2022-02-13 thomas write_packed_object(off_t *packfile_size, FILE *packfile,
1605 f9c2e8e5 2022-02-13 thomas FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1606 f9c2e8e5 2022-02-13 thomas SHA1_CTX *ctx, struct got_repository *repo)
1608 f9c2e8e5 2022-02-13 thomas const struct got_error *err = NULL;
1609 f9c2e8e5 2022-02-13 thomas struct got_deflate_checksum csum;
1610 f9c2e8e5 2022-02-13 thomas char buf[32];
1612 f9c2e8e5 2022-02-13 thomas struct got_raw_object *raw = NULL;
1613 f9c2e8e5 2022-02-13 thomas off_t outlen;
1615 f9c2e8e5 2022-02-13 thomas csum.output_sha1 = ctx;
1616 f9c2e8e5 2022-02-13 thomas csum.output_crc = NULL;
1618 f9c2e8e5 2022-02-13 thomas m->off = ftello(packfile);
1619 f9c2e8e5 2022-02-13 thomas if (m->delta_len == 0) {
1620 f9c2e8e5 2022-02-13 thomas err = got_object_raw_open(&raw, outfd, repo, &m->id);
1621 f9c2e8e5 2022-02-13 thomas if (err)
1622 f9c2e8e5 2022-02-13 thomas goto done;
1623 f9c2e8e5 2022-02-13 thomas err = packhdr(&nh, buf, sizeof(buf),
1624 f9c2e8e5 2022-02-13 thomas m->obj_type, raw->size);
1625 f9c2e8e5 2022-02-13 thomas if (err)
1626 f9c2e8e5 2022-02-13 thomas goto done;
1627 f9c2e8e5 2022-02-13 thomas err = hwrite(packfile, buf, nh, ctx);
1628 f9c2e8e5 2022-02-13 thomas if (err)
1629 f9c2e8e5 2022-02-13 thomas goto done;
1630 f9c2e8e5 2022-02-13 thomas *packfile_size += nh;
1631 f9c2e8e5 2022-02-13 thomas if (raw->f == NULL) {
1632 f9c2e8e5 2022-02-13 thomas err = got_deflate_to_file_mmap(&outlen,
1633 f9c2e8e5 2022-02-13 thomas raw->data + raw->hdrlen, 0, raw->size,
1634 f9c2e8e5 2022-02-13 thomas packfile, &csum);
1635 f9c2e8e5 2022-02-13 thomas if (err)
1636 f9c2e8e5 2022-02-13 thomas goto done;
1637 f9c2e8e5 2022-02-13 thomas } else {
1638 f9c2e8e5 2022-02-13 thomas if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1639 f9c2e8e5 2022-02-13 thomas == -1) {
1640 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("fseeko");
1641 f9c2e8e5 2022-02-13 thomas goto done;
1643 f9c2e8e5 2022-02-13 thomas err = got_deflate_to_file(&outlen, raw->f,
1644 f9c2e8e5 2022-02-13 thomas raw->size, packfile, &csum);
1645 f9c2e8e5 2022-02-13 thomas if (err)
1646 f9c2e8e5 2022-02-13 thomas goto done;
1648 f9c2e8e5 2022-02-13 thomas *packfile_size += outlen;
1649 f9c2e8e5 2022-02-13 thomas got_object_raw_close(raw);
1650 f9c2e8e5 2022-02-13 thomas raw = NULL;
1651 f9c2e8e5 2022-02-13 thomas } else if (m->delta_buf) {
1652 f9c2e8e5 2022-02-13 thomas err = deltahdr(packfile_size, ctx, packfile, m);
1653 f9c2e8e5 2022-02-13 thomas if (err)
1654 f9c2e8e5 2022-02-13 thomas goto done;
1655 f9c2e8e5 2022-02-13 thomas err = got_deflate_to_file_mmap(&outlen,
1656 f9c2e8e5 2022-02-13 thomas m->delta_buf, 0, m->delta_len, packfile, &csum);
1657 f9c2e8e5 2022-02-13 thomas if (err)
1658 f9c2e8e5 2022-02-13 thomas goto done;
1659 f9c2e8e5 2022-02-13 thomas *packfile_size += outlen;
1660 f9c2e8e5 2022-02-13 thomas free(m->delta_buf);
1661 f9c2e8e5 2022-02-13 thomas m->delta_buf = NULL;
1662 f9c2e8e5 2022-02-13 thomas } else {
1663 f9c2e8e5 2022-02-13 thomas if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1664 f9c2e8e5 2022-02-13 thomas == -1) {
1665 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("fseeko");
1666 f9c2e8e5 2022-02-13 thomas goto done;
1668 f9c2e8e5 2022-02-13 thomas err = deltahdr(packfile_size, ctx, packfile, m);
1669 f9c2e8e5 2022-02-13 thomas if (err)
1670 f9c2e8e5 2022-02-13 thomas goto done;
1671 f9c2e8e5 2022-02-13 thomas err = got_deflate_to_file(&outlen, delta_cache,
1672 f9c2e8e5 2022-02-13 thomas m->delta_len, packfile, &csum);
1673 f9c2e8e5 2022-02-13 thomas if (err)
1674 f9c2e8e5 2022-02-13 thomas goto done;
1675 f9c2e8e5 2022-02-13 thomas *packfile_size += outlen;
1678 f9c2e8e5 2022-02-13 thomas if (raw)
1679 f9c2e8e5 2022-02-13 thomas got_object_raw_close(raw);
1680 f9c2e8e5 2022-02-13 thomas return err;
1683 f9c2e8e5 2022-02-13 thomas static const struct got_error *
1684 b79280dd 2021-10-15 thomas genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1685 f9c2e8e5 2022-02-13 thomas struct got_pack_meta **deltify, int ndeltify,
1686 f9c2e8e5 2022-02-13 thomas struct got_pack_meta **reuse, int nreuse,
1687 85220b0e 2022-03-22 thomas int ncolored, int nfound, int ntrees, int nours,
1688 85220b0e 2022-03-22 thomas struct got_repository *repo,
1689 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1690 31ba2236 2022-01-05 thomas struct got_ratelimit *rl,
1691 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1693 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1695 e6bcace5 2021-06-22 stsp SHA1_CTX ctx;
1696 e6bcace5 2021-06-22 stsp struct got_pack_meta *m;
1697 e4d8ab47 2021-10-15 thomas char buf[32];
1698 e8f02263 2022-01-23 thomas size_t n;
1699 f9c2e8e5 2022-02-13 thomas off_t packfile_size = 0;
1700 ecf9545f 2021-10-15 thomas int outfd = -1;
1702 e6bcace5 2021-06-22 stsp SHA1Init(&ctx);
1704 e6bcace5 2021-06-22 stsp err = hwrite(packfile, "PACK", 4, &ctx);
1706 e6bcace5 2021-06-22 stsp return err;
1707 e6bcace5 2021-06-22 stsp putbe32(buf, GOT_PACKFILE_VERSION);
1708 e6bcace5 2021-06-22 stsp err = hwrite(packfile, buf, 4, &ctx);
1710 e6bcace5 2021-06-22 stsp goto done;
1711 f9c2e8e5 2022-02-13 thomas putbe32(buf, ndeltify + nreuse);
1712 e6bcace5 2021-06-22 stsp err = hwrite(packfile, buf, 4, &ctx);
1714 e6bcace5 2021-06-22 stsp goto done;
1716 f9c2e8e5 2022-02-13 thomas qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1717 f9c2e8e5 2022-02-13 thomas write_order_cmp);
1718 f9c2e8e5 2022-02-13 thomas for (i = 0; i < ndeltify; i++) {
1719 31ba2236 2022-01-05 thomas err = report_progress(progress_cb, progress_arg, rl,
1720 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, packfile_size, nours,
1721 85220b0e 2022-03-22 thomas ndeltify + nreuse, ndeltify + nreuse, i);
1722 31ba2236 2022-01-05 thomas if (err)
1723 31ba2236 2022-01-05 thomas goto done;
1724 f9c2e8e5 2022-02-13 thomas m = deltify[i];
1725 f9c2e8e5 2022-02-13 thomas err = write_packed_object(&packfile_size, packfile,
1726 f9c2e8e5 2022-02-13 thomas delta_cache, m, &outfd, &ctx, repo);
1727 f9c2e8e5 2022-02-13 thomas if (err)
1728 f9c2e8e5 2022-02-13 thomas goto done;
1731 f9c2e8e5 2022-02-13 thomas qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1732 f9c2e8e5 2022-02-13 thomas reuse_write_order_cmp);
1733 f9c2e8e5 2022-02-13 thomas for (i = 0; i < nreuse; i++) {
1734 f9c2e8e5 2022-02-13 thomas err = report_progress(progress_cb, progress_arg, rl,
1735 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, packfile_size, nours,
1736 85220b0e 2022-03-22 thomas ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1737 f9c2e8e5 2022-02-13 thomas if (err)
1738 f9c2e8e5 2022-02-13 thomas goto done;
1739 f9c2e8e5 2022-02-13 thomas m = reuse[i];
1740 f9c2e8e5 2022-02-13 thomas err = write_packed_object(&packfile_size, packfile,
1741 f9c2e8e5 2022-02-13 thomas delta_cache, m, &outfd, &ctx, repo);
1742 f9c2e8e5 2022-02-13 thomas if (err)
1743 f9c2e8e5 2022-02-13 thomas goto done;
1746 e6bcace5 2021-06-22 stsp SHA1Final(pack_sha1, &ctx);
1747 e6bcace5 2021-06-22 stsp n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1748 e6bcace5 2021-06-22 stsp if (n != SHA1_DIGEST_LENGTH)
1749 e6bcace5 2021-06-22 stsp err = got_ferror(packfile, GOT_ERR_IO);
1750 05118f5a 2021-06-22 stsp packfile_size += SHA1_DIGEST_LENGTH;
1751 dc7edd42 2021-08-22 stsp packfile_size += sizeof(struct got_packfile_hdr);
1752 31ba2236 2022-01-05 thomas if (progress_cb) {
1753 85220b0e 2022-03-22 thomas err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1754 85220b0e 2022-03-22 thomas packfile_size, nours, ndeltify + nreuse,
1755 85220b0e 2022-03-22 thomas ndeltify + nreuse, ndeltify + nreuse);
1756 31ba2236 2022-01-05 thomas if (err)
1757 31ba2236 2022-01-05 thomas goto done;
1760 ecf9545f 2021-10-15 thomas if (outfd != -1 && close(outfd) == -1 && err == NULL)
1761 ecf9545f 2021-10-15 thomas err = got_error_from_errno("close");
1762 e6bcace5 2021-06-22 stsp return err;
1765 f9c2e8e5 2022-02-13 thomas static const struct got_error *
1766 f9c2e8e5 2022-02-13 thomas remove_unused_object(struct got_object_idset_element *entry, void *arg)
1768 f9c2e8e5 2022-02-13 thomas struct got_object_idset *idset = arg;
1770 f9c2e8e5 2022-02-13 thomas if (got_object_idset_get_element_data(entry) == NULL)
1771 f9c2e8e5 2022-02-13 thomas got_object_idset_remove_element(idset, entry);
1773 f9c2e8e5 2022-02-13 thomas return NULL;
1776 f9c2e8e5 2022-02-13 thomas static const struct got_error *
1777 f9c2e8e5 2022-02-13 thomas remove_reused_object(struct got_object_idset_element *entry, void *arg)
1779 f9c2e8e5 2022-02-13 thomas struct got_object_idset *idset = arg;
1780 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *m;
1782 f9c2e8e5 2022-02-13 thomas m = got_object_idset_get_element_data(entry);
1783 f9c2e8e5 2022-02-13 thomas if (m->have_reused_delta)
1784 f9c2e8e5 2022-02-13 thomas got_object_idset_remove_element(idset, entry);
1786 f9c2e8e5 2022-02-13 thomas return NULL;
1789 f9c2e8e5 2022-02-13 thomas static const struct got_error *
1790 f9c2e8e5 2022-02-13 thomas add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1792 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *m = data;
1793 f9c2e8e5 2022-02-13 thomas struct got_pack_metavec *v = arg;
1795 f9c2e8e5 2022-02-13 thomas return add_meta(m, v);
1798 e6bcace5 2021-06-22 stsp const struct got_error *
1799 e6bcace5 2021-06-22 stsp got_pack_create(uint8_t *packsha1, FILE *packfile,
1800 e6bcace5 2021-06-22 stsp struct got_object_id **theirs, int ntheirs,
1801 e6bcace5 2021-06-22 stsp struct got_object_id **ours, int nours,
1802 f8a36e22 2021-08-26 stsp struct got_repository *repo, int loose_obj_only, int allow_empty,
1803 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1804 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1806 e6bcace5 2021-06-22 stsp const struct got_error *err;
1807 f9c2e8e5 2022-02-13 thomas int delta_cache_fd = -1;
1808 b79280dd 2021-10-15 thomas FILE *delta_cache = NULL;
1809 f9c2e8e5 2022-02-13 thomas struct got_object_idset *idset;
1810 31ba2236 2022-01-05 thomas struct got_ratelimit rl;
1811 f9c2e8e5 2022-02-13 thomas struct got_pack_metavec deltify, reuse;
1812 85220b0e 2022-03-22 thomas int ncolored = 0, nfound = 0, ntrees = 0;
1814 f9c2e8e5 2022-02-13 thomas memset(&deltify, 0, sizeof(deltify));
1815 f9c2e8e5 2022-02-13 thomas memset(&reuse, 0, sizeof(reuse));
1817 31ba2236 2022-01-05 thomas got_ratelimit_init(&rl, 0, 500);
1819 f9c2e8e5 2022-02-13 thomas idset = got_object_idset_alloc();
1820 f9c2e8e5 2022-02-13 thomas if (idset == NULL)
1821 f9c2e8e5 2022-02-13 thomas return got_error_from_errno("got_object_idset_alloc");
1823 85220b0e 2022-03-22 thomas err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1824 85220b0e 2022-03-22 thomas ntheirs, ours, nours, repo, loose_obj_only,
1825 85220b0e 2022-03-22 thomas progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1827 e6bcace5 2021-06-22 stsp return err;
1829 f9c2e8e5 2022-02-13 thomas err = got_object_idset_for_each_element(idset,
1830 f9c2e8e5 2022-02-13 thomas remove_unused_object, idset);
1831 f9c2e8e5 2022-02-13 thomas if (err)
1832 f9c2e8e5 2022-02-13 thomas goto done;
1834 f9a643e8 2022-02-13 thomas if (progress_cb) {
1835 85220b0e 2022-03-22 thomas err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1836 85220b0e 2022-03-22 thomas 0L, nours, got_object_idset_num_elements(idset), 0, 0);
1837 f9a643e8 2022-02-13 thomas if (err)
1838 f9a643e8 2022-02-13 thomas goto done;
1841 f9c2e8e5 2022-02-13 thomas if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1842 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_CANNOT_PACK);
1843 05118f5a 2021-06-22 stsp goto done;
1846 f9c2e8e5 2022-02-13 thomas delta_cache_fd = got_opentempfd();
1847 f9c2e8e5 2022-02-13 thomas if (delta_cache_fd == -1) {
1848 b79280dd 2021-10-15 thomas err = got_error_from_errno("got_opentemp");
1849 b79280dd 2021-10-15 thomas goto done;
1852 f9c2e8e5 2022-02-13 thomas reuse.metasz = 64;
1853 f9c2e8e5 2022-02-13 thomas reuse.meta = calloc(reuse.metasz,
1854 f9c2e8e5 2022-02-13 thomas sizeof(struct got_pack_meta *));
1855 f9c2e8e5 2022-02-13 thomas if (reuse.meta == NULL) {
1856 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("calloc");
1857 f9c2e8e5 2022-02-13 thomas goto done;
1860 85220b0e 2022-03-22 thomas err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
1861 85220b0e 2022-03-22 thomas ntrees, nours, repo, progress_cb, progress_arg, &rl,
1862 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
1863 f9c2e8e5 2022-02-13 thomas if (err)
1864 f9c2e8e5 2022-02-13 thomas goto done;
1865 f9c2e8e5 2022-02-13 thomas if (reuse.nmeta > 0) {
1866 f9c2e8e5 2022-02-13 thomas err = got_object_idset_for_each_element(idset,
1867 f9c2e8e5 2022-02-13 thomas remove_reused_object, idset);
1868 f9c2e8e5 2022-02-13 thomas if (err)
1869 f9c2e8e5 2022-02-13 thomas goto done;
1872 f9c2e8e5 2022-02-13 thomas delta_cache = fdopen(delta_cache_fd, "a+");
1873 f9c2e8e5 2022-02-13 thomas if (delta_cache == NULL) {
1874 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("fdopen");
1875 f9c2e8e5 2022-02-13 thomas goto done;
1877 f9c2e8e5 2022-02-13 thomas delta_cache_fd = -1;
1879 f9c2e8e5 2022-02-13 thomas if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1880 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("fseeko");
1881 f9c2e8e5 2022-02-13 thomas goto done;
1884 f9c2e8e5 2022-02-13 thomas deltify.meta = calloc(got_object_idset_num_elements(idset),
1885 f9c2e8e5 2022-02-13 thomas sizeof(struct got_pack_meta *));
1886 f9c2e8e5 2022-02-13 thomas if (deltify.meta == NULL) {
1887 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("calloc");
1888 f9c2e8e5 2022-02-13 thomas goto done;
1890 f9c2e8e5 2022-02-13 thomas deltify.metasz = got_object_idset_num_elements(idset);
1892 f9c2e8e5 2022-02-13 thomas err = got_object_idset_for_each(idset, add_meta_idset_cb, &deltify);
1893 f9c2e8e5 2022-02-13 thomas if (err)
1894 f9c2e8e5 2022-02-13 thomas goto done;
1895 f9c2e8e5 2022-02-13 thomas if (deltify.nmeta > 0) {
1896 85220b0e 2022-03-22 thomas err = pick_deltas(deltify.meta, deltify.nmeta, ncolored,
1897 85220b0e 2022-03-22 thomas nfound, ntrees, nours, reuse.nmeta, delta_cache, repo,
1898 31ba2236 2022-01-05 thomas progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1900 f8a36e22 2021-08-26 stsp goto done;
1901 b79280dd 2021-10-15 thomas if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1902 b79280dd 2021-10-15 thomas err = got_error_from_errno("fseeko");
1903 b79280dd 2021-10-15 thomas goto done;
1907 f9c2e8e5 2022-02-13 thomas err = genpack(packsha1, packfile, delta_cache, deltify.meta,
1908 85220b0e 2022-03-22 thomas deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1909 85220b0e 2022-03-22 thomas nours, repo, progress_cb, progress_arg, &rl,
1910 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
1912 e6bcace5 2021-06-22 stsp goto done;
1914 f9c2e8e5 2022-02-13 thomas free_nmeta(deltify.meta, deltify.nmeta);
1915 f9c2e8e5 2022-02-13 thomas free_nmeta(reuse.meta, reuse.nmeta);
1916 f9c2e8e5 2022-02-13 thomas got_object_idset_free(idset);
1917 f9c2e8e5 2022-02-13 thomas if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1918 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("close");
1919 b79280dd 2021-10-15 thomas if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1920 b79280dd 2021-10-15 thomas err = got_error_from_errno("fclose");
1921 e6bcace5 2021-06-22 stsp return err;
