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 e6bcace5 2021-06-22 stsp #include <sys/queue.h>
20 f8b19efd 2021-10-13 stsp #include <sys/tree.h>
21 08736cf9 2021-06-23 stsp #include <sys/uio.h>
22 e6bcace5 2021-06-22 stsp #include <sys/stat.h>
23 211cfef0 2022-01-05 stsp #include <sys/time.h>
25 e3f86256 2022-02-18 naddy #include <endian.h>
26 08736cf9 2021-06-23 stsp #include <stdint.h>
27 05118f5a 2021-06-22 stsp #include <imsg.h>
28 e6bcace5 2021-06-22 stsp #include <stdio.h>
29 e6bcace5 2021-06-22 stsp #include <stdlib.h>
30 e6bcace5 2021-06-22 stsp #include <string.h>
31 e6bcace5 2021-06-22 stsp #include <sha1.h>
32 211cfef0 2022-01-05 stsp #include <time.h>
33 e6bcace5 2021-06-22 stsp #include <limits.h>
34 e6bcace5 2021-06-22 stsp #include <zlib.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 f8b19efd 2021-10-13 stsp #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 08347b73 2021-10-14 stsp #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 211cfef0 2022-01-05 stsp #include "got_lib_ratelimit.h"
56 74881701 2021-10-15 stsp #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;
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 5060d5a1 2022-01-10 stsp unsigned char *delta_buf; /* if not encoded in delta cache file */
74 74881701 2021-10-15 stsp off_t delta_offset; /* offset in delta cache file */
75 5060d5a1 2022-01-10 stsp off_t delta_len; /* encoded delta length */
78 67fd6849 2022-02-13 stsp int have_reused_delta;
79 67fd6849 2022-02-13 stsp off_t reused_delta_offset; /* offset of delta in reused pack file */
80 67fd6849 2022-02-13 stsp 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 1d19226a 2021-10-13 stsp 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 5060d5a1 2022-01-10 stsp free(meta->delta_buf);
131 5060d5a1 2022-01-10 stsp meta->delta_buf = NULL;
132 67fd6849 2022-02-13 stsp free(meta->base_obj_id);
133 67fd6849 2022-02-13 stsp 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 5060d5a1 2022-01-10 stsp static off_t
166 e6bcace5 2021-06-22 stsp delta_size(struct got_delta_instruction *deltas, int ndeltas)
169 5060d5a1 2022-01-10 stsp 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 5060d5a1 2022-01-10 stsp static const struct got_error *
180 5060d5a1 2022-01-10 stsp append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
184 5060d5a1 2022-01-10 stsp if (*len + nseg >= *sz) {
185 5060d5a1 2022-01-10 stsp while (*len + nseg >= *sz)
186 5060d5a1 2022-01-10 stsp *sz += *sz / 2;
187 5060d5a1 2022-01-10 stsp n = realloc(*p, *sz);
188 5060d5a1 2022-01-10 stsp if (n == NULL)
189 5060d5a1 2022-01-10 stsp return got_error_from_errno("realloc");
192 5060d5a1 2022-01-10 stsp memcpy(*p + *len, seg, nseg);
193 5060d5a1 2022-01-10 stsp *len += nseg;
194 5060d5a1 2022-01-10 stsp return NULL;
197 5060d5a1 2022-01-10 stsp static const struct got_error *
198 5060d5a1 2022-01-10 stsp encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
199 5060d5a1 2022-01-10 stsp struct got_delta_instruction *deltas, int ndeltas,
200 5060d5a1 2022-01-10 stsp off_t delta_size, off_t base_size)
202 5060d5a1 2022-01-10 stsp const struct got_error *err;
203 5060d5a1 2022-01-10 stsp unsigned char buf[16], *bp;
205 5060d5a1 2022-01-10 stsp size_t len = 0;
207 5060d5a1 2022-01-10 stsp struct got_delta_instruction *d;
209 5060d5a1 2022-01-10 stsp m->delta_buf = malloc(delta_size);
210 5060d5a1 2022-01-10 stsp if (m->delta_buf == NULL)
211 5060d5a1 2022-01-10 stsp return got_error_from_errno("calloc");
213 5060d5a1 2022-01-10 stsp /* base object size */
214 5060d5a1 2022-01-10 stsp buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
215 5060d5a1 2022-01-10 stsp n = base_size >> GOT_DELTA_SIZE_SHIFT;
216 5060d5a1 2022-01-10 stsp for (i = 1; n > 0; i++) {
217 5060d5a1 2022-01-10 stsp buf[i - 1] |= GOT_DELTA_SIZE_MORE;
218 5060d5a1 2022-01-10 stsp buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
219 5060d5a1 2022-01-10 stsp n >>= GOT_DELTA_SIZE_SHIFT;
221 5060d5a1 2022-01-10 stsp err = append(&m->delta_buf, &len, &delta_size, buf, i);
223 5060d5a1 2022-01-10 stsp return err;
225 5060d5a1 2022-01-10 stsp /* target object size */
226 5060d5a1 2022-01-10 stsp buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
227 5060d5a1 2022-01-10 stsp n = o->size >> GOT_DELTA_SIZE_SHIFT;
228 5060d5a1 2022-01-10 stsp for (i = 1; n > 0; i++) {
229 5060d5a1 2022-01-10 stsp buf[i - 1] |= GOT_DELTA_SIZE_MORE;
230 5060d5a1 2022-01-10 stsp buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
231 5060d5a1 2022-01-10 stsp n >>= GOT_DELTA_SIZE_SHIFT;
233 5060d5a1 2022-01-10 stsp err = append(&m->delta_buf, &len, &delta_size, buf, i);
235 5060d5a1 2022-01-10 stsp return err;
237 5060d5a1 2022-01-10 stsp for (j = 0; j < ndeltas; j++) {
238 5060d5a1 2022-01-10 stsp d = &deltas[j];
239 5060d5a1 2022-01-10 stsp if (d->copy) {
240 5060d5a1 2022-01-10 stsp n = d->offset;
241 5060d5a1 2022-01-10 stsp bp = &buf[1];
242 5060d5a1 2022-01-10 stsp buf[0] = GOT_DELTA_BASE_COPY;
243 5060d5a1 2022-01-10 stsp for (i = 0; i < 4; i++) {
244 5060d5a1 2022-01-10 stsp /* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
245 5060d5a1 2022-01-10 stsp buf[0] |= 1 << i;
246 5060d5a1 2022-01-10 stsp *bp++ = n & 0xff;
248 5060d5a1 2022-01-10 stsp if (n == 0)
252 5060d5a1 2022-01-10 stsp n = d->len;
253 5060d5a1 2022-01-10 stsp if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
254 5060d5a1 2022-01-10 stsp /* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
255 5060d5a1 2022-01-10 stsp for (i = 0; i < 3 && n > 0; i++) {
256 5060d5a1 2022-01-10 stsp buf[0] |= 1 << (i + 4);
257 5060d5a1 2022-01-10 stsp *bp++ = n & 0xff;
261 5060d5a1 2022-01-10 stsp err = append(&m->delta_buf, &len, &delta_size,
262 5060d5a1 2022-01-10 stsp buf, bp - buf);
264 5060d5a1 2022-01-10 stsp return err;
265 5060d5a1 2022-01-10 stsp } else if (o->f == NULL) {
267 5060d5a1 2022-01-10 stsp while (n != d->len) {
268 5060d5a1 2022-01-10 stsp buf[0] = (d->len - n < 127) ? d->len - n : 127;
269 5060d5a1 2022-01-10 stsp err = append(&m->delta_buf, &len, &delta_size,
272 5060d5a1 2022-01-10 stsp return err;
273 5060d5a1 2022-01-10 stsp err = append(&m->delta_buf, &len, &delta_size,
274 5060d5a1 2022-01-10 stsp o->data + o->hdrlen + d->offset + n,
277 5060d5a1 2022-01-10 stsp return err;
278 5060d5a1 2022-01-10 stsp n += buf[0];
281 5060d5a1 2022-01-10 stsp char content[128];
283 5060d5a1 2022-01-10 stsp if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
284 5060d5a1 2022-01-10 stsp return got_error_from_errno("fseeko");
286 5060d5a1 2022-01-10 stsp while (n != d->len) {
287 5060d5a1 2022-01-10 stsp buf[0] = (d->len - n < 127) ? d->len - n : 127;
288 5060d5a1 2022-01-10 stsp err = append(&m->delta_buf, &len, &delta_size,
291 5060d5a1 2022-01-10 stsp return err;
292 5060d5a1 2022-01-10 stsp r = fread(content, 1, buf[0], o->f);
293 5060d5a1 2022-01-10 stsp if (r != buf[0])
294 5060d5a1 2022-01-10 stsp return got_ferror(o->f, GOT_ERR_IO);
295 5060d5a1 2022-01-10 stsp err = append(&m->delta_buf, &len, &delta_size,
296 5060d5a1 2022-01-10 stsp content, buf[0]);
298 5060d5a1 2022-01-10 stsp return err;
299 5060d5a1 2022-01-10 stsp n += buf[0];
304 5060d5a1 2022-01-10 stsp m->delta_len = len;
305 5060d5a1 2022-01-10 stsp return NULL;
308 74881701 2021-10-15 stsp static const struct got_error *
309 74881701 2021-10-15 stsp encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
310 74881701 2021-10-15 stsp struct got_delta_instruction *deltas, int ndeltas,
311 a319ca8c 2021-10-15 stsp off_t base_size, FILE *f)
313 a319ca8c 2021-10-15 stsp unsigned char buf[16], *bp;
317 a319ca8c 2021-10-15 stsp struct got_delta_instruction *d;
319 a319ca8c 2021-10-15 stsp /* base object size */
320 a319ca8c 2021-10-15 stsp buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
321 a319ca8c 2021-10-15 stsp n = base_size >> GOT_DELTA_SIZE_SHIFT;
322 a319ca8c 2021-10-15 stsp for (i = 1; n > 0; i++) {
323 a319ca8c 2021-10-15 stsp buf[i - 1] |= GOT_DELTA_SIZE_MORE;
324 a319ca8c 2021-10-15 stsp buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
325 a319ca8c 2021-10-15 stsp n >>= GOT_DELTA_SIZE_SHIFT;
327 a319ca8c 2021-10-15 stsp w = fwrite(buf, 1, i, f);
328 a319ca8c 2021-10-15 stsp if (w != i)
329 a319ca8c 2021-10-15 stsp return got_ferror(f, GOT_ERR_IO);
331 a319ca8c 2021-10-15 stsp /* target object size */
332 a319ca8c 2021-10-15 stsp buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
333 a319ca8c 2021-10-15 stsp n = o->size >> GOT_DELTA_SIZE_SHIFT;
334 a319ca8c 2021-10-15 stsp for (i = 1; n > 0; i++) {
335 a319ca8c 2021-10-15 stsp buf[i - 1] |= GOT_DELTA_SIZE_MORE;
336 a319ca8c 2021-10-15 stsp buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
337 a319ca8c 2021-10-15 stsp n >>= GOT_DELTA_SIZE_SHIFT;
339 a319ca8c 2021-10-15 stsp w = fwrite(buf, 1, i, f);
340 a319ca8c 2021-10-15 stsp if (w != i)
341 a319ca8c 2021-10-15 stsp return got_ferror(f, GOT_ERR_IO);
343 a319ca8c 2021-10-15 stsp for (j = 0; j < ndeltas; j++) {
344 a319ca8c 2021-10-15 stsp d = &deltas[j];
345 a319ca8c 2021-10-15 stsp if (d->copy) {
346 a319ca8c 2021-10-15 stsp n = d->offset;
347 a319ca8c 2021-10-15 stsp bp = &buf[1];
348 a319ca8c 2021-10-15 stsp buf[0] = GOT_DELTA_BASE_COPY;
349 a319ca8c 2021-10-15 stsp for (i = 0; i < 4; i++) {
350 a319ca8c 2021-10-15 stsp /* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
351 a319ca8c 2021-10-15 stsp buf[0] |= 1 << i;
352 a319ca8c 2021-10-15 stsp *bp++ = n & 0xff;
354 a319ca8c 2021-10-15 stsp if (n == 0)
358 a319ca8c 2021-10-15 stsp n = d->len;
359 a319ca8c 2021-10-15 stsp if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
360 a319ca8c 2021-10-15 stsp /* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
361 a319ca8c 2021-10-15 stsp for (i = 0; i < 3 && n > 0; i++) {
362 a319ca8c 2021-10-15 stsp buf[0] |= 1 << (i + 4);
363 a319ca8c 2021-10-15 stsp *bp++ = n & 0xff;
367 a319ca8c 2021-10-15 stsp w = fwrite(buf, 1, bp - buf, f);
368 a319ca8c 2021-10-15 stsp if (w != bp - buf)
369 a319ca8c 2021-10-15 stsp return got_ferror(f, GOT_ERR_IO);
370 64a8571e 2022-01-07 stsp } else if (o->f == NULL) {
372 64a8571e 2022-01-07 stsp while (n != d->len) {
373 64a8571e 2022-01-07 stsp buf[0] = (d->len - n < 127) ? d->len - n : 127;
374 64a8571e 2022-01-07 stsp w = fwrite(buf, 1, 1, f);
375 64a8571e 2022-01-07 stsp if (w != 1)
376 64a8571e 2022-01-07 stsp return got_ferror(f, GOT_ERR_IO);
377 64a8571e 2022-01-07 stsp w = fwrite(o->data + o->hdrlen + d->offset + n,
378 64a8571e 2022-01-07 stsp 1, buf[0], f);
379 64a8571e 2022-01-07 stsp if (w != buf[0])
380 64a8571e 2022-01-07 stsp return got_ferror(f, GOT_ERR_IO);
381 64a8571e 2022-01-07 stsp n += buf[0];
384 a319ca8c 2021-10-15 stsp char content[128];
386 a319ca8c 2021-10-15 stsp if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
387 a319ca8c 2021-10-15 stsp return got_error_from_errno("fseeko");
389 a319ca8c 2021-10-15 stsp while (n != d->len) {
390 a319ca8c 2021-10-15 stsp buf[0] = (d->len - n < 127) ? d->len - n : 127;
391 a319ca8c 2021-10-15 stsp w = fwrite(buf, 1, 1, f);
392 a319ca8c 2021-10-15 stsp if (w != 1)
393 a319ca8c 2021-10-15 stsp return got_ferror(f, GOT_ERR_IO);
394 a319ca8c 2021-10-15 stsp r = fread(content, 1, buf[0], o->f);
395 a319ca8c 2021-10-15 stsp if (r != buf[0])
396 a319ca8c 2021-10-15 stsp return got_ferror(o->f, GOT_ERR_IO);
397 a319ca8c 2021-10-15 stsp w = fwrite(content, 1, buf[0], f);
398 a319ca8c 2021-10-15 stsp if (w != buf[0])
399 a319ca8c 2021-10-15 stsp return got_ferror(f, GOT_ERR_IO);
400 a319ca8c 2021-10-15 stsp n += buf[0];
405 5060d5a1 2022-01-10 stsp m->delta_len = ftello(f) - m->delta_offset;
406 a319ca8c 2021-10-15 stsp return NULL;
409 211cfef0 2022-01-05 stsp static const struct got_error *
410 211cfef0 2022-01-05 stsp report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
411 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
412 b8af7c06 2022-03-15 stsp off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
413 b8af7c06 2022-03-15 stsp int nobj_written)
415 211cfef0 2022-01-05 stsp const struct got_error *err;
416 211cfef0 2022-01-05 stsp int elapsed;
418 211cfef0 2022-01-05 stsp if (progress_cb == NULL)
419 211cfef0 2022-01-05 stsp return NULL;
421 211cfef0 2022-01-05 stsp err = got_ratelimit_check(&elapsed, rl);
422 211cfef0 2022-01-05 stsp if (err || !elapsed)
423 211cfef0 2022-01-05 stsp return err;
425 b8af7c06 2022-03-15 stsp return progress_cb(progress_arg, ncolored, nfound, ntrees,
426 b8af7c06 2022-03-15 stsp packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
429 e6bcace5 2021-06-22 stsp static const struct got_error *
430 67fd6849 2022-02-13 stsp add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
432 67fd6849 2022-02-13 stsp if (v->nmeta == v->metasz){
433 67fd6849 2022-02-13 stsp size_t newsize = 2 * v->metasz;
434 67fd6849 2022-02-13 stsp struct got_pack_meta **new;
435 67fd6849 2022-02-13 stsp new = reallocarray(v->meta, newsize, sizeof(*new));
436 67fd6849 2022-02-13 stsp if (new == NULL)
437 67fd6849 2022-02-13 stsp return got_error_from_errno("reallocarray");
438 67fd6849 2022-02-13 stsp v->meta = new;
439 67fd6849 2022-02-13 stsp v->metasz = newsize;
442 67fd6849 2022-02-13 stsp v->meta[v->nmeta++] = m;
443 67fd6849 2022-02-13 stsp return NULL;
446 67fd6849 2022-02-13 stsp static const struct got_error *
447 67fd6849 2022-02-13 stsp reuse_delta(int idx, struct got_pack_meta *m, struct got_pack_metavec *v,
448 67fd6849 2022-02-13 stsp struct got_object_idset *idset, struct got_pack *pack,
449 67fd6849 2022-02-13 stsp struct got_packidx *packidx, int delta_cache_fd,
450 67fd6849 2022-02-13 stsp struct got_repository *repo)
452 67fd6849 2022-02-13 stsp const struct got_error *err = NULL;
453 67fd6849 2022-02-13 stsp struct got_pack_meta *base = NULL;
454 67fd6849 2022-02-13 stsp struct got_object_id *base_obj_id = NULL;
455 67fd6849 2022-02-13 stsp off_t delta_len = 0, delta_offset = 0, delta_cache_offset = 0;
456 67fd6849 2022-02-13 stsp uint64_t base_size, result_size;
458 67fd6849 2022-02-13 stsp if (m->have_reused_delta)
459 67fd6849 2022-02-13 stsp return NULL;
461 67fd6849 2022-02-13 stsp err = got_object_read_raw_delta(&base_size, &result_size, &delta_len,
462 67fd6849 2022-02-13 stsp &delta_offset, &delta_cache_offset, &base_obj_id, delta_cache_fd,
463 67fd6849 2022-02-13 stsp packidx, idx, &m->id, repo);
465 67fd6849 2022-02-13 stsp return err;
467 67fd6849 2022-02-13 stsp if (delta_offset + delta_len < delta_offset)
468 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_PACKFILE);
470 67fd6849 2022-02-13 stsp base = got_object_idset_get(idset, base_obj_id);
471 67fd6849 2022-02-13 stsp if (base == NULL)
474 67fd6849 2022-02-13 stsp m->delta_len = delta_len;
475 67fd6849 2022-02-13 stsp m->delta_offset = delta_cache_offset;
476 67fd6849 2022-02-13 stsp m->prev = base;
477 67fd6849 2022-02-13 stsp m->size = result_size;
478 67fd6849 2022-02-13 stsp m->have_reused_delta = 1;
479 67fd6849 2022-02-13 stsp m->reused_delta_offset = delta_offset;
480 67fd6849 2022-02-13 stsp m->base_obj_id = base_obj_id;
481 67fd6849 2022-02-13 stsp base_obj_id = NULL;
482 67fd6849 2022-02-13 stsp err = add_meta(m, v);
484 67fd6849 2022-02-13 stsp free(base_obj_id);
485 67fd6849 2022-02-13 stsp return err;
488 67fd6849 2022-02-13 stsp static const struct got_error *
489 67fd6849 2022-02-13 stsp find_pack_for_reuse(struct got_packidx **best_packidx,
490 67fd6849 2022-02-13 stsp struct got_repository *repo)
492 9b576444 2022-03-14 stsp const struct got_error *err = NULL;
493 67fd6849 2022-02-13 stsp struct got_pathlist_entry *pe;
494 67fd6849 2022-02-13 stsp const char *best_packidx_path = NULL;
495 67fd6849 2022-02-13 stsp int nobj_max = 0;
497 67fd6849 2022-02-13 stsp *best_packidx = NULL;
499 9b576444 2022-03-14 stsp TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
500 67fd6849 2022-02-13 stsp const char *path_packidx = pe->path;
501 67fd6849 2022-02-13 stsp struct got_packidx *packidx;
504 67fd6849 2022-02-13 stsp err = got_repo_get_packidx(&packidx, path_packidx, repo);
508 67fd6849 2022-02-13 stsp nobj = be32toh(packidx->hdr.fanout_table[0xff]);
509 67fd6849 2022-02-13 stsp if (nobj > nobj_max) {
510 67fd6849 2022-02-13 stsp best_packidx_path = path_packidx;
511 67fd6849 2022-02-13 stsp nobj_max = nobj;
515 67fd6849 2022-02-13 stsp if (best_packidx_path) {
516 67fd6849 2022-02-13 stsp err = got_repo_get_packidx(best_packidx, best_packidx_path,
520 67fd6849 2022-02-13 stsp return err;
523 67fd6849 2022-02-13 stsp struct search_deltas_arg {
524 67fd6849 2022-02-13 stsp struct got_packidx *packidx;
525 67fd6849 2022-02-13 stsp struct got_pack *pack;
526 67fd6849 2022-02-13 stsp struct got_object_idset *idset;
527 67fd6849 2022-02-13 stsp struct got_pack_metavec *v;
528 67fd6849 2022-02-13 stsp int delta_cache_fd;
529 67fd6849 2022-02-13 stsp struct got_repository *repo;
530 67fd6849 2022-02-13 stsp got_pack_progress_cb progress_cb;
531 67fd6849 2022-02-13 stsp void *progress_arg;
532 67fd6849 2022-02-13 stsp struct got_ratelimit *rl;
533 67fd6849 2022-02-13 stsp got_cancel_cb cancel_cb;
534 67fd6849 2022-02-13 stsp void *cancel_arg;
535 b8af7c06 2022-03-15 stsp int ncolored;
536 b8af7c06 2022-03-15 stsp int nfound;
537 b8af7c06 2022-03-15 stsp int ntrees;
538 67fd6849 2022-02-13 stsp int ncommits;
541 67fd6849 2022-02-13 stsp static const struct got_error *
542 67fd6849 2022-02-13 stsp search_delta_for_object(struct got_object_id *id, void *data, void *arg)
544 67fd6849 2022-02-13 stsp const struct got_error *err;
545 67fd6849 2022-02-13 stsp struct got_pack_meta *m = data;
546 67fd6849 2022-02-13 stsp struct search_deltas_arg *a = arg;
547 67fd6849 2022-02-13 stsp int obj_idx;
548 67fd6849 2022-02-13 stsp struct got_object *obj = NULL;
550 67fd6849 2022-02-13 stsp if (a->cancel_cb) {
551 67fd6849 2022-02-13 stsp err = (*a->cancel_cb)(a->cancel_arg);
553 67fd6849 2022-02-13 stsp return err;
556 67fd6849 2022-02-13 stsp if (!got_repo_check_packidx_bloom_filter(a->repo,
557 67fd6849 2022-02-13 stsp a->packidx->path_packidx, id))
558 67fd6849 2022-02-13 stsp return NULL;
560 67fd6849 2022-02-13 stsp obj_idx = got_packidx_get_object_idx(a->packidx, id);
561 67fd6849 2022-02-13 stsp if (obj_idx == -1)
562 67fd6849 2022-02-13 stsp return NULL;
565 67fd6849 2022-02-13 stsp * Opening and closing an object just to check its flags
566 67fd6849 2022-02-13 stsp * is a bit expensive. We could have an imsg which requests
567 67fd6849 2022-02-13 stsp * plain type/size information for an object without doing
568 67fd6849 2022-02-13 stsp * work such as traversing the object's entire delta chain
569 67fd6849 2022-02-13 stsp * to find the base object type, and other such info which
570 67fd6849 2022-02-13 stsp * we don't really need here.
572 67fd6849 2022-02-13 stsp err = got_object_open_from_packfile(&obj, &m->id, a->pack,
573 67fd6849 2022-02-13 stsp a->packidx, obj_idx, a->repo);
575 67fd6849 2022-02-13 stsp return err;
577 67fd6849 2022-02-13 stsp if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
578 67fd6849 2022-02-13 stsp reuse_delta(obj_idx, m, a->v, a->idset, a->pack, a->packidx,
579 67fd6849 2022-02-13 stsp a->delta_cache_fd, a->repo);
582 67fd6849 2022-02-13 stsp err = report_progress(a->progress_cb, a->progress_arg, a->rl,
583 b8af7c06 2022-03-15 stsp a->ncolored, a->nfound, a->ntrees, 0L, a->ncommits,
584 b8af7c06 2022-03-15 stsp got_object_idset_num_elements(a->idset), a->v->nmeta, 0);
587 67fd6849 2022-02-13 stsp got_object_close(obj);
588 67fd6849 2022-02-13 stsp return err;
591 67fd6849 2022-02-13 stsp static const struct got_error *
592 67fd6849 2022-02-13 stsp search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
593 b8af7c06 2022-03-15 stsp int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
594 b8af7c06 2022-03-15 stsp struct got_repository *repo,
595 67fd6849 2022-02-13 stsp got_pack_progress_cb progress_cb, void *progress_arg,
596 67fd6849 2022-02-13 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
598 67fd6849 2022-02-13 stsp const struct got_error *err = NULL;
599 67fd6849 2022-02-13 stsp char *path_packfile = NULL;
600 67fd6849 2022-02-13 stsp struct got_packidx *packidx;
601 67fd6849 2022-02-13 stsp struct got_pack *pack;
602 67fd6849 2022-02-13 stsp struct search_deltas_arg sda;
604 67fd6849 2022-02-13 stsp err = find_pack_for_reuse(&packidx, repo);
606 67fd6849 2022-02-13 stsp return err;
608 67fd6849 2022-02-13 stsp if (packidx == NULL)
609 67fd6849 2022-02-13 stsp return NULL;
611 67fd6849 2022-02-13 stsp err = got_packidx_get_packfile_path(&path_packfile,
612 67fd6849 2022-02-13 stsp packidx->path_packidx);
614 67fd6849 2022-02-13 stsp return err;
616 67fd6849 2022-02-13 stsp pack = got_repo_get_cached_pack(repo, path_packfile);
617 67fd6849 2022-02-13 stsp if (pack == NULL) {
618 67fd6849 2022-02-13 stsp err = got_repo_cache_pack(&pack, repo, path_packfile, packidx);
623 67fd6849 2022-02-13 stsp sda.packidx = packidx;
624 67fd6849 2022-02-13 stsp sda.pack = pack;
625 67fd6849 2022-02-13 stsp sda.idset = idset;
627 67fd6849 2022-02-13 stsp sda.delta_cache_fd = delta_cache_fd;
628 67fd6849 2022-02-13 stsp sda.repo = repo;
629 67fd6849 2022-02-13 stsp sda.progress_cb = progress_cb;
630 67fd6849 2022-02-13 stsp sda.progress_arg = progress_arg;
631 67fd6849 2022-02-13 stsp sda.rl = rl;
632 67fd6849 2022-02-13 stsp sda.cancel_cb = cancel_cb;
633 67fd6849 2022-02-13 stsp sda.cancel_arg = cancel_arg;
634 b8af7c06 2022-03-15 stsp sda.ncolored = ncolored;
635 b8af7c06 2022-03-15 stsp sda.nfound = nfound;
636 b8af7c06 2022-03-15 stsp sda.ntrees = ntrees;
637 67fd6849 2022-02-13 stsp sda.ncommits = ncommits;
638 67fd6849 2022-02-13 stsp err = got_object_idset_for_each(idset, search_delta_for_object, &sda);
640 67fd6849 2022-02-13 stsp free(path_packfile);
641 67fd6849 2022-02-13 stsp return err;
644 67fd6849 2022-02-13 stsp static const struct got_error *
645 b8af7c06 2022-03-15 stsp pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
646 b8af7c06 2022-03-15 stsp int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
647 b8af7c06 2022-03-15 stsp struct got_repository *repo,
648 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
649 211cfef0 2022-01-05 stsp 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 74881701 2021-10-15 stsp struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
655 5060d5a1 2022-01-10 stsp int i, j, ndeltas, best_ndeltas;
656 5060d5a1 2022-01-10 stsp off_t size, best_size;
657 4f4d853e 2021-10-24 stsp const int max_base_candidates = 3;
658 402a5ec1 2022-01-10 stsp size_t delta_memsize = 0;
659 402a5ec1 2022-01-10 stsp const size_t max_delta_memsize = 25 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
660 cc7a354a 2021-10-15 stsp 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 211cfef0 2022-01-05 stsp err = report_progress(progress_cb, progress_arg, rl,
670 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
671 b8af7c06 2022-03-15 stsp nreused + i, 0);
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 94dac27c 2021-10-15 stsp err = got_object_raw_open(&raw, &outfd, repo, &m->id);
683 600b755e 2021-10-14 stsp m->size = raw->size;
685 64a8571e 2022-01-07 stsp if (raw->f == NULL) {
686 64a8571e 2022-01-07 stsp err = got_deltify_init_mem(&m->dtab, raw->data,
687 64a8571e 2022-01-07 stsp raw->hdrlen, raw->size + raw->hdrlen);
689 64a8571e 2022-01-07 stsp err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
690 64a8571e 2022-01-07 stsp 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 74881701 2021-10-15 stsp best_size = raw->size;
703 74881701 2021-10-15 stsp 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 22edbce7 2021-10-24 stsp if (base->nchain >= 128 ||
713 e6bcace5 2021-06-22 stsp base->obj_type != m->obj_type)
716 d3c116bf 2021-10-15 stsp err = got_object_raw_open(&base_raw, &outfd, repo,
717 94dac27c 2021-10-15 stsp &base->id);
721 64a8571e 2022-01-07 stsp if (raw->f == NULL && base_raw->f == NULL) {
722 64a8571e 2022-01-07 stsp err = got_deltify_mem_mem(&deltas, &ndeltas,
723 64a8571e 2022-01-07 stsp raw->data, raw->hdrlen,
724 64a8571e 2022-01-07 stsp raw->size + raw->hdrlen,
725 64a8571e 2022-01-07 stsp base->dtab, base_raw->data,
726 64a8571e 2022-01-07 stsp base_raw->hdrlen,
727 64a8571e 2022-01-07 stsp base_raw->size + base_raw->hdrlen);
728 64a8571e 2022-01-07 stsp } else if (raw->f == NULL) {
729 64a8571e 2022-01-07 stsp err = got_deltify_mem_file(&deltas, &ndeltas,
730 64a8571e 2022-01-07 stsp raw->data, raw->hdrlen,
731 64a8571e 2022-01-07 stsp raw->size + raw->hdrlen,
732 64a8571e 2022-01-07 stsp base->dtab, base_raw->f,
733 64a8571e 2022-01-07 stsp base_raw->hdrlen,
734 64a8571e 2022-01-07 stsp base_raw->size + base_raw->hdrlen);
735 64a8571e 2022-01-07 stsp } else if (base_raw->f == NULL) {
736 64a8571e 2022-01-07 stsp err = got_deltify_file_mem(&deltas, &ndeltas,
737 64a8571e 2022-01-07 stsp raw->f, raw->hdrlen,
738 64a8571e 2022-01-07 stsp raw->size + raw->hdrlen,
739 64a8571e 2022-01-07 stsp base->dtab, base_raw->data,
740 64a8571e 2022-01-07 stsp base_raw->hdrlen,
741 64a8571e 2022-01-07 stsp base_raw->size + base_raw->hdrlen);
743 64a8571e 2022-01-07 stsp err = got_deltify(&deltas, &ndeltas,
744 64a8571e 2022-01-07 stsp raw->f, raw->hdrlen,
745 64a8571e 2022-01-07 stsp raw->size + raw->hdrlen,
746 64a8571e 2022-01-07 stsp base->dtab, base_raw->f, base_raw->hdrlen,
747 64a8571e 2022-01-07 stsp 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 74881701 2021-10-15 stsp 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 74881701 2021-10-15 stsp best_size = size;
761 74881701 2021-10-15 stsp free(best_deltas);
762 74881701 2021-10-15 stsp best_deltas = deltas;
763 74881701 2021-10-15 stsp best_ndeltas = ndeltas;
764 74881701 2021-10-15 stsp 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 74881701 2021-10-15 stsp if (best_ndeltas > 0) {
778 402a5ec1 2022-01-10 stsp if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
779 402a5ec1 2022-01-10 stsp delta_memsize + best_size <= max_delta_memsize) {
780 402a5ec1 2022-01-10 stsp delta_memsize += best_size;
781 5060d5a1 2022-01-10 stsp err = encode_delta_in_mem(m, raw, best_deltas,
782 5060d5a1 2022-01-10 stsp best_ndeltas, best_size, m->prev->size);
784 5060d5a1 2022-01-10 stsp m->delta_offset = ftello(delta_cache);
787 67fd6849 2022-02-13 stsp * Storing compressed delta data in the delta
788 67fd6849 2022-02-13 stsp * cache file would probably be more efficient
789 67fd6849 2022-02-13 stsp * than writing uncompressed delta data here
790 67fd6849 2022-02-13 stsp * and compressing it while writing the pack
791 67fd6849 2022-02-13 stsp * file. This would also allow for reusing
792 67fd6849 2022-02-13 stsp * deltas in their compressed form.
794 5060d5a1 2022-01-10 stsp err = encode_delta(m, raw, best_deltas,
795 5060d5a1 2022-01-10 stsp best_ndeltas, m->prev->size, delta_cache);
797 74881701 2021-10-15 stsp free(best_deltas);
798 74881701 2021-10-15 stsp best_deltas = NULL;
799 74881701 2021-10-15 stsp best_ndeltas = 0;
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 cc7a354a 2021-10-15 stsp if (outfd != -1 && close(outfd) == -1 && err == NULL)
817 cc7a354a 2021-10-15 stsp err = got_error_from_errno("close");
818 74881701 2021-10-15 stsp free(deltas);
819 74881701 2021-10-15 stsp 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 67fd6849 2022-02-13 stsp 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 05118f5a 2021-06-22 stsp time_t mtime, int loose_obj_only, struct got_repository *repo)
856 e6bcace5 2021-06-22 stsp const struct got_error *err;
857 67fd6849 2022-02-13 stsp struct got_pack_meta *m = NULL;
859 05118f5a 2021-06-22 stsp if (loose_obj_only) {
860 05118f5a 2021-06-22 stsp int is_packed;
861 05118f5a 2021-06-22 stsp err = search_packidx(&is_packed, id, repo);
863 05118f5a 2021-06-22 stsp return err;
864 05118f5a 2021-06-22 stsp if (is_packed)
865 05118f5a 2021-06-22 stsp return NULL;
868 67fd6849 2022-02-13 stsp if (want_meta) {
869 67fd6849 2022-02-13 stsp err = alloc_meta(&m, id, path, obj_type, mtime);
871 67fd6849 2022-02-13 stsp return err;
874 67fd6849 2022-02-13 stsp return got_object_idset_add(idset, id, m);
877 e6bcace5 2021-06-22 stsp static const struct got_error *
878 67fd6849 2022-02-13 stsp load_tree_entries(struct got_object_id_queue *ids, int want_meta,
879 e6bcace5 2021-06-22 stsp struct got_object_idset *idset, struct got_object_id *tree_id,
880 e6bcace5 2021-06-22 stsp const char *dpath, time_t mtime, struct got_repository *repo,
881 b8af7c06 2022-03-15 stsp int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
882 b8af7c06 2022-03-15 stsp got_pack_progress_cb progress_cb, void *progress_arg,
883 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
885 e6bcace5 2021-06-22 stsp const struct got_error *err;
886 e6bcace5 2021-06-22 stsp struct got_tree_object *tree;
887 e6bcace5 2021-06-22 stsp char *p = NULL;
890 e6bcace5 2021-06-22 stsp err = got_object_open_as_tree(&tree, repo, tree_id);
892 b8af7c06 2022-03-15 stsp return err;
894 b8af7c06 2022-03-15 stsp (*ntrees)++;
895 b8af7c06 2022-03-15 stsp err = report_progress(progress_cb, progress_arg, rl,
896 b8af7c06 2022-03-15 stsp *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
898 e6bcace5 2021-06-22 stsp return err;
900 e6bcace5 2021-06-22 stsp for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
901 e6bcace5 2021-06-22 stsp struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
902 e6bcace5 2021-06-22 stsp struct got_object_id *id = got_tree_entry_get_id(e);
903 e6bcace5 2021-06-22 stsp mode_t mode = got_tree_entry_get_mode(e);
905 e6bcace5 2021-06-22 stsp if (cancel_cb) {
906 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
911 3af9de88 2021-09-22 stsp if (got_object_tree_entry_is_submodule(e) ||
912 e6bcace5 2021-06-22 stsp got_object_idset_contains(idset, id))
915 e6bcace5 2021-06-22 stsp if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
916 e6bcace5 2021-06-22 stsp got_tree_entry_get_name(e)) == -1) {
917 e6bcace5 2021-06-22 stsp err = got_error_from_errno("asprintf");
921 e6bcace5 2021-06-22 stsp if (S_ISDIR(mode)) {
922 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
923 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, id);
926 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(ids, qid, entry);
927 3af9de88 2021-09-22 stsp } else if (S_ISREG(mode) || S_ISLNK(mode)) {
928 67fd6849 2022-02-13 stsp err = add_object(want_meta, idset, id, p,
929 67fd6849 2022-02-13 stsp GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo);
932 b8af7c06 2022-03-15 stsp (*nfound)++;
933 b8af7c06 2022-03-15 stsp err = report_progress(progress_cb, progress_arg, rl,
934 b8af7c06 2022-03-15 stsp *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
942 e6bcace5 2021-06-22 stsp got_object_tree_close(tree);
944 e6bcace5 2021-06-22 stsp return err;
947 e6bcace5 2021-06-22 stsp static const struct got_error *
948 67fd6849 2022-02-13 stsp load_tree(int want_meta, struct got_object_idset *idset,
949 e6bcace5 2021-06-22 stsp struct got_object_id *tree_id, const char *dpath, time_t mtime,
950 b8af7c06 2022-03-15 stsp struct got_repository *repo, int loose_obj_only,
951 b8af7c06 2022-03-15 stsp int *ncolored, int *nfound, int *ntrees,
952 b8af7c06 2022-03-15 stsp got_pack_progress_cb progress_cb, void *progress_arg,
953 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
955 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
956 e6bcace5 2021-06-22 stsp struct got_object_id_queue tree_ids;
957 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
959 e6bcace5 2021-06-22 stsp if (got_object_idset_contains(idset, tree_id))
960 e6bcace5 2021-06-22 stsp return NULL;
962 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, tree_id);
964 e6bcace5 2021-06-22 stsp return err;
966 dbdddfee 2021-06-23 naddy STAILQ_INIT(&tree_ids);
967 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
969 dbdddfee 2021-06-23 naddy while (!STAILQ_EMPTY(&tree_ids)) {
970 e6bcace5 2021-06-22 stsp if (cancel_cb) {
971 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
976 dbdddfee 2021-06-23 naddy qid = STAILQ_FIRST(&tree_ids);
977 dbdddfee 2021-06-23 naddy STAILQ_REMOVE_HEAD(&tree_ids, entry);
979 e6bcace5 2021-06-22 stsp if (got_object_idset_contains(idset, qid->id)) {
980 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
984 67fd6849 2022-02-13 stsp err = add_object(want_meta, idset, qid->id, dpath,
985 67fd6849 2022-02-13 stsp GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo);
987 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
991 b8af7c06 2022-03-15 stsp (*nfound)++;
992 b8af7c06 2022-03-15 stsp err = report_progress(progress_cb, progress_arg, rl,
993 b8af7c06 2022-03-15 stsp *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
997 67fd6849 2022-02-13 stsp err = load_tree_entries(&tree_ids, want_meta, idset, qid->id,
998 b8af7c06 2022-03-15 stsp dpath, mtime, repo, loose_obj_only, ncolored, nfound,
999 b8af7c06 2022-03-15 stsp ntrees, progress_cb, progress_arg, rl,
1000 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1001 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1006 e6bcace5 2021-06-22 stsp got_object_id_queue_free(&tree_ids);
1007 e6bcace5 2021-06-22 stsp return err;
1010 e6bcace5 2021-06-22 stsp static const struct got_error *
1011 67fd6849 2022-02-13 stsp load_commit(int want_meta, struct got_object_idset *idset,
1012 05118f5a 2021-06-22 stsp struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1013 b8af7c06 2022-03-15 stsp int *ncolored, int *nfound, int *ntrees,
1014 b8af7c06 2022-03-15 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1015 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1017 e6bcace5 2021-06-22 stsp const struct got_error *err;
1018 e6bcace5 2021-06-22 stsp struct got_commit_object *commit;
1020 e6bcace5 2021-06-22 stsp if (got_object_idset_contains(idset, id))
1021 e6bcace5 2021-06-22 stsp return NULL;
1023 05118f5a 2021-06-22 stsp if (loose_obj_only) {
1024 05118f5a 2021-06-22 stsp int is_packed;
1025 05118f5a 2021-06-22 stsp err = search_packidx(&is_packed, id, repo);
1027 05118f5a 2021-06-22 stsp return err;
1028 05118f5a 2021-06-22 stsp if (is_packed)
1029 05118f5a 2021-06-22 stsp return NULL;
1032 e6bcace5 2021-06-22 stsp err = got_object_open_as_commit(&commit, repo, id);
1034 e6bcace5 2021-06-22 stsp return err;
1036 67fd6849 2022-02-13 stsp err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_COMMIT,
1037 05118f5a 2021-06-22 stsp got_object_commit_get_committer_time(commit),
1038 05118f5a 2021-06-22 stsp loose_obj_only, repo);
1040 e6bcace5 2021-06-22 stsp goto done;
1042 b8af7c06 2022-03-15 stsp (*nfound)++;
1043 b8af7c06 2022-03-15 stsp err = report_progress(progress_cb, progress_arg, rl,
1044 b8af7c06 2022-03-15 stsp *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
1046 b8af7c06 2022-03-15 stsp goto done;
1048 67fd6849 2022-02-13 stsp err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1049 05118f5a 2021-06-22 stsp "", got_object_commit_get_committer_time(commit),
1050 b8af7c06 2022-03-15 stsp repo, loose_obj_only, ncolored, nfound, ntrees,
1051 b8af7c06 2022-03-15 stsp progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1053 e6bcace5 2021-06-22 stsp got_object_commit_close(commit);
1054 e6bcace5 2021-06-22 stsp return err;
1057 05118f5a 2021-06-22 stsp static const struct got_error *
1058 67fd6849 2022-02-13 stsp load_tag(int want_meta, struct got_object_idset *idset,
1059 05118f5a 2021-06-22 stsp struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1060 b8af7c06 2022-03-15 stsp int *ncolored, int *nfound, int *ntrees,
1061 b8af7c06 2022-03-15 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1062 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1064 05118f5a 2021-06-22 stsp const struct got_error *err;
1065 05118f5a 2021-06-22 stsp struct got_tag_object *tag = NULL;
1067 05118f5a 2021-06-22 stsp if (got_object_idset_contains(idset, id))
1068 05118f5a 2021-06-22 stsp return NULL;
1070 05118f5a 2021-06-22 stsp if (loose_obj_only) {
1071 05118f5a 2021-06-22 stsp int is_packed;
1072 05118f5a 2021-06-22 stsp err = search_packidx(&is_packed, id, repo);
1074 05118f5a 2021-06-22 stsp return err;
1075 05118f5a 2021-06-22 stsp if (is_packed)
1076 05118f5a 2021-06-22 stsp return NULL;
1079 05118f5a 2021-06-22 stsp err = got_object_open_as_tag(&tag, repo, id);
1081 05118f5a 2021-06-22 stsp return err;
1083 67fd6849 2022-02-13 stsp err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1084 05118f5a 2021-06-22 stsp got_object_tag_get_tagger_time(tag),
1085 05118f5a 2021-06-22 stsp loose_obj_only, repo);
1087 05118f5a 2021-06-22 stsp goto done;
1089 b8af7c06 2022-03-15 stsp (*nfound)++;
1090 b8af7c06 2022-03-15 stsp err = report_progress(progress_cb, progress_arg, rl,
1091 b8af7c06 2022-03-15 stsp *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
1093 b8af7c06 2022-03-15 stsp goto done;
1095 05118f5a 2021-06-22 stsp switch (got_object_tag_get_object_type(tag)) {
1096 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_COMMIT:
1097 67fd6849 2022-02-13 stsp err = load_commit(want_meta, idset,
1098 b8af7c06 2022-03-15 stsp got_object_tag_get_object_id(tag), repo, loose_obj_only,
1099 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1100 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1102 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_TREE:
1103 67fd6849 2022-02-13 stsp err = load_tree(want_meta, idset,
1104 67fd6849 2022-02-13 stsp got_object_tag_get_object_id(tag), "",
1105 b8af7c06 2022-03-15 stsp got_object_tag_get_tagger_time(tag), repo, loose_obj_only,
1106 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1107 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1114 05118f5a 2021-06-22 stsp got_object_tag_close(tag);
1115 05118f5a 2021-06-22 stsp return err;
1118 e6bcace5 2021-06-22 stsp enum findtwixt_color {
1119 e6bcace5 2021-06-22 stsp COLOR_KEEP = 0,
1120 e6bcace5 2021-06-22 stsp COLOR_DROP,
1121 e6bcace5 2021-06-22 stsp COLOR_BLANK,
1123 e6bcace5 2021-06-22 stsp static const int findtwixt_colors[] = {
1124 e6bcace5 2021-06-22 stsp COLOR_KEEP,
1125 e6bcace5 2021-06-22 stsp COLOR_DROP,
1126 e6bcace5 2021-06-22 stsp COLOR_BLANK
1129 e6bcace5 2021-06-22 stsp static const struct got_error *
1130 e6bcace5 2021-06-22 stsp queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1131 e6bcace5 2021-06-22 stsp int color, struct got_repository *repo)
1133 e6bcace5 2021-06-22 stsp const struct got_error *err;
1134 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
1136 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, id);
1138 e6bcace5 2021-06-22 stsp return err;
1140 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(ids, qid, entry);
1141 e6bcace5 2021-06-22 stsp qid->data = (void *)&findtwixt_colors[color];
1142 e6bcace5 2021-06-22 stsp return NULL;
1145 e6bcace5 2021-06-22 stsp static const struct got_error *
1146 e6bcace5 2021-06-22 stsp drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
1147 e6bcace5 2021-06-22 stsp struct got_object_id *id, struct got_repository *repo,
1148 e6bcace5 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1150 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1151 e6bcace5 2021-06-22 stsp struct got_commit_object *commit;
1152 e6bcace5 2021-06-22 stsp const struct got_object_id_queue *parents;
1153 e6bcace5 2021-06-22 stsp struct got_object_id_queue ids;
1154 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
1156 dbdddfee 2021-06-23 naddy STAILQ_INIT(&ids);
1158 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, id);
1160 e6bcace5 2021-06-22 stsp return err;
1161 dbdddfee 2021-06-23 naddy STAILQ_INSERT_HEAD(&ids, qid, entry);
1163 dbdddfee 2021-06-23 naddy while (!STAILQ_EMPTY(&ids)) {
1164 e6bcace5 2021-06-22 stsp if (cancel_cb) {
1165 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
1170 dbdddfee 2021-06-23 naddy qid = STAILQ_FIRST(&ids);
1171 dbdddfee 2021-06-23 naddy STAILQ_REMOVE_HEAD(&ids, entry);
1173 e6bcace5 2021-06-22 stsp if (got_object_idset_contains(drop, qid->id)) {
1174 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1178 07165b17 2021-07-01 stsp err = got_object_idset_add(drop, qid->id,
1179 07165b17 2021-07-01 stsp (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1180 e6bcace5 2021-06-22 stsp if (err) {
1181 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1185 e6bcace5 2021-06-22 stsp if (!got_object_idset_contains(keep, qid->id)) {
1186 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1190 e6bcace5 2021-06-22 stsp err = got_object_open_as_commit(&commit, repo, qid->id);
1191 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1195 e6bcace5 2021-06-22 stsp parents = got_object_commit_get_parent_ids(commit);
1196 e6bcace5 2021-06-22 stsp if (parents) {
1197 e6bcace5 2021-06-22 stsp err = got_object_id_queue_copy(parents, &ids);
1198 e6bcace5 2021-06-22 stsp if (err) {
1199 e6bcace5 2021-06-22 stsp got_object_commit_close(commit);
1203 e6bcace5 2021-06-22 stsp got_object_commit_close(commit);
1206 e6bcace5 2021-06-22 stsp got_object_id_queue_free(&ids);
1207 e6bcace5 2021-06-22 stsp return err;
1210 e6bcace5 2021-06-22 stsp struct append_id_arg {
1211 e6bcace5 2021-06-22 stsp struct got_object_id **array;
1215 e6bcace5 2021-06-22 stsp static const struct got_error *
1216 e6bcace5 2021-06-22 stsp append_id(struct got_object_id *id, void *data, void *arg)
1218 e6bcace5 2021-06-22 stsp struct append_id_arg *a = arg;
1220 e6bcace5 2021-06-22 stsp a->array[a->idx] = got_object_id_dup(id);
1221 e6bcace5 2021-06-22 stsp if (a->array[a->idx] == NULL)
1222 e6bcace5 2021-06-22 stsp return got_error_from_errno("got_object_id_dup");
1225 e6bcace5 2021-06-22 stsp return NULL;
1228 e6bcace5 2021-06-22 stsp static const struct got_error *
1229 b8af7c06 2022-03-15 stsp findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1230 e6bcace5 2021-06-22 stsp struct got_object_id **head, int nhead,
1231 e6bcace5 2021-06-22 stsp struct got_object_id **tail, int ntail,
1232 e6bcace5 2021-06-22 stsp struct got_repository *repo,
1233 b8af7c06 2022-03-15 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1234 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1236 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1237 e6bcace5 2021-06-22 stsp struct got_object_id_queue ids;
1238 e6bcace5 2021-06-22 stsp struct got_object_idset *keep, *drop;
1239 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
1240 05118f5a 2021-06-22 stsp int i, ncolor, nkeep, obj_type;
1242 dbdddfee 2021-06-23 naddy STAILQ_INIT(&ids);
1243 e6bcace5 2021-06-22 stsp *res = NULL;
1244 e6bcace5 2021-06-22 stsp *nres = 0;
1245 b8af7c06 2022-03-15 stsp *ncolored = 0;
1247 e6bcace5 2021-06-22 stsp keep = got_object_idset_alloc();
1248 e6bcace5 2021-06-22 stsp if (keep == NULL)
1249 e6bcace5 2021-06-22 stsp return got_error_from_errno("got_object_idset_alloc");
1251 e6bcace5 2021-06-22 stsp drop = got_object_idset_alloc();
1252 e6bcace5 2021-06-22 stsp if (drop == NULL) {
1253 e6bcace5 2021-06-22 stsp err = got_error_from_errno("got_object_idset_alloc");
1254 e6bcace5 2021-06-22 stsp goto done;
1257 05118f5a 2021-06-22 stsp for (i = 0; i < nhead; i++) {
1258 05118f5a 2021-06-22 stsp struct got_object_id *id = head[i];
1259 05118f5a 2021-06-22 stsp if (id == NULL)
1261 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
1263 05118f5a 2021-06-22 stsp return err;
1264 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_COMMIT)
1266 05118f5a 2021-06-22 stsp err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
1268 05118f5a 2021-06-22 stsp goto done;
1270 05118f5a 2021-06-22 stsp for (i = 0; i < ntail; i++) {
1271 05118f5a 2021-06-22 stsp struct got_object_id *id = tail[i];
1272 05118f5a 2021-06-22 stsp if (id == NULL)
1274 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
1276 05118f5a 2021-06-22 stsp return err;
1277 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_COMMIT)
1279 05118f5a 2021-06-22 stsp err = queue_commit_id(&ids, id, COLOR_DROP, repo);
1281 05118f5a 2021-06-22 stsp goto done;
1284 dbdddfee 2021-06-23 naddy while (!STAILQ_EMPTY(&ids)) {
1285 e6bcace5 2021-06-22 stsp int qcolor;
1286 dbdddfee 2021-06-23 naddy qid = STAILQ_FIRST(&ids);
1287 e6bcace5 2021-06-22 stsp qcolor = *((int *)qid->data);
1289 e6bcace5 2021-06-22 stsp if (got_object_idset_contains(drop, qid->id))
1290 e6bcace5 2021-06-22 stsp ncolor = COLOR_DROP;
1291 e6bcace5 2021-06-22 stsp else if (got_object_idset_contains(keep, qid->id))
1292 e6bcace5 2021-06-22 stsp ncolor = COLOR_KEEP;
1294 e6bcace5 2021-06-22 stsp ncolor = COLOR_BLANK;
1296 b8af7c06 2022-03-15 stsp (*ncolored)++;
1297 b8af7c06 2022-03-15 stsp err = report_progress(progress_cb, progress_arg, rl,
1298 b8af7c06 2022-03-15 stsp *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1300 b8af7c06 2022-03-15 stsp goto done;
1302 e6bcace5 2021-06-22 stsp if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
1303 e6bcace5 2021-06-22 stsp qcolor == COLOR_KEEP)) {
1304 dbdddfee 2021-06-23 naddy STAILQ_REMOVE_HEAD(&ids, entry);
1305 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1309 e6bcace5 2021-06-22 stsp if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
1310 e6bcace5 2021-06-22 stsp err = drop_commit(keep, drop, qid->id, repo,
1311 e6bcace5 2021-06-22 stsp cancel_cb, cancel_arg);
1313 e6bcace5 2021-06-22 stsp goto done;
1314 e6bcace5 2021-06-22 stsp } else if (ncolor == COLOR_BLANK) {
1315 e6bcace5 2021-06-22 stsp struct got_commit_object *commit;
1316 e6bcace5 2021-06-22 stsp struct got_object_id *id;
1317 e6bcace5 2021-06-22 stsp const struct got_object_id_queue *parents;
1318 e6bcace5 2021-06-22 stsp struct got_object_qid *pid;
1320 e6bcace5 2021-06-22 stsp id = got_object_id_dup(qid->id);
1321 e6bcace5 2021-06-22 stsp if (id == NULL) {
1322 e6bcace5 2021-06-22 stsp err = got_error_from_errno("got_object_id_dup");
1323 e6bcace5 2021-06-22 stsp goto done;
1325 e6bcace5 2021-06-22 stsp if (qcolor == COLOR_KEEP)
1326 07165b17 2021-07-01 stsp err = got_object_idset_add(keep, id,
1327 07165b17 2021-07-01 stsp (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1329 07165b17 2021-07-01 stsp err = got_object_idset_add(drop, id,
1330 07165b17 2021-07-01 stsp (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1331 e6bcace5 2021-06-22 stsp if (err) {
1333 e6bcace5 2021-06-22 stsp goto done;
1336 e6bcace5 2021-06-22 stsp err = got_object_open_as_commit(&commit, repo, id);
1337 e6bcace5 2021-06-22 stsp if (err) {
1339 e6bcace5 2021-06-22 stsp goto done;
1341 e6bcace5 2021-06-22 stsp parents = got_object_commit_get_parent_ids(commit);
1342 e6bcace5 2021-06-22 stsp if (parents) {
1343 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(pid, parents, entry) {
1344 e6bcace5 2021-06-22 stsp err = queue_commit_id(&ids, pid->id,
1345 e6bcace5 2021-06-22 stsp qcolor, repo);
1346 e6bcace5 2021-06-22 stsp if (err) {
1348 e6bcace5 2021-06-22 stsp goto done;
1352 e6bcace5 2021-06-22 stsp got_object_commit_close(commit);
1353 e6bcace5 2021-06-22 stsp commit = NULL;
1355 e6bcace5 2021-06-22 stsp /* should not happen */
1356 e6bcace5 2021-06-22 stsp err = got_error_fmt(GOT_ERR_NOT_IMPL,
1357 e6bcace5 2021-06-22 stsp "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
1358 e6bcace5 2021-06-22 stsp goto done;
1361 dbdddfee 2021-06-23 naddy STAILQ_REMOVE_HEAD(&ids, entry);
1362 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1365 e6bcace5 2021-06-22 stsp nkeep = got_object_idset_num_elements(keep);
1366 e6bcace5 2021-06-22 stsp if (nkeep > 0) {
1367 e6bcace5 2021-06-22 stsp struct append_id_arg arg;
1368 e6bcace5 2021-06-22 stsp arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1369 e6bcace5 2021-06-22 stsp if (arg.array == NULL) {
1370 e6bcace5 2021-06-22 stsp err = got_error_from_errno("calloc");
1371 e6bcace5 2021-06-22 stsp goto done;
1373 e6bcace5 2021-06-22 stsp arg.idx = 0;
1374 e6bcace5 2021-06-22 stsp err = got_object_idset_for_each(keep, append_id, &arg);
1375 e6bcace5 2021-06-22 stsp if (err) {
1376 e6bcace5 2021-06-22 stsp free(arg.array);
1377 e6bcace5 2021-06-22 stsp goto done;
1379 e6bcace5 2021-06-22 stsp *res = arg.array;
1380 e6bcace5 2021-06-22 stsp *nres = nkeep;
1383 e6bcace5 2021-06-22 stsp got_object_idset_free(keep);
1384 e6bcace5 2021-06-22 stsp got_object_idset_free(drop);
1385 e6bcace5 2021-06-22 stsp got_object_id_queue_free(&ids);
1386 e6bcace5 2021-06-22 stsp return err;
1389 e6bcace5 2021-06-22 stsp static const struct got_error *
1390 b8af7c06 2022-03-15 stsp load_object_ids(int *ncolored, int *nfound, int *ntrees,
1391 b8af7c06 2022-03-15 stsp struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1392 e6bcace5 2021-06-22 stsp struct got_object_id **ours, int nours, struct got_repository *repo,
1393 b8af7c06 2022-03-15 stsp int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
1394 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1396 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1397 e6bcace5 2021-06-22 stsp struct got_object_id **ids = NULL;
1398 05118f5a 2021-06-22 stsp int i, nobj = 0, obj_type;
1400 b8af7c06 2022-03-15 stsp *ncolored = 0;
1401 b8af7c06 2022-03-15 stsp *nfound = 0;
1402 b8af7c06 2022-03-15 stsp *ntrees = 0;
1404 b8af7c06 2022-03-15 stsp err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1405 b8af7c06 2022-03-15 stsp repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1406 e6bcace5 2021-06-22 stsp if (err || nobj == 0)
1407 e6bcace5 2021-06-22 stsp goto done;
1409 e6bcace5 2021-06-22 stsp for (i = 0; i < ntheirs; i++) {
1410 05118f5a 2021-06-22 stsp struct got_object_id *id = theirs[i];
1411 05118f5a 2021-06-22 stsp if (id == NULL)
1413 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
1415 05118f5a 2021-06-22 stsp return err;
1416 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_COMMIT)
1418 b8af7c06 2022-03-15 stsp err = load_commit(0, idset, id, repo, loose_obj_only,
1419 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1420 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1422 05118f5a 2021-06-22 stsp goto done;
1425 05118f5a 2021-06-22 stsp for (i = 0; i < ntheirs; i++) {
1426 f4a2ff2d 2021-07-01 stsp struct got_object_id *id = theirs[i];
1427 67fd6849 2022-02-13 stsp struct got_pack_meta *m;
1428 05118f5a 2021-06-22 stsp if (id == NULL)
1430 67fd6849 2022-02-13 stsp m = got_object_idset_get(idset, id);
1431 67fd6849 2022-02-13 stsp if (m == NULL) {
1432 07165b17 2021-07-01 stsp err = got_object_get_type(&obj_type, repo, id);
1434 07165b17 2021-07-01 stsp goto done;
1436 67fd6849 2022-02-13 stsp obj_type = m->obj_type;
1437 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_TAG)
1439 b8af7c06 2022-03-15 stsp err = load_tag(0, idset, id, repo, loose_obj_only,
1440 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1441 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1443 05118f5a 2021-06-22 stsp goto done;
1446 eca70f98 2021-09-03 stsp for (i = 0; i < nobj; i++) {
1447 b8af7c06 2022-03-15 stsp err = load_commit(1, idset, ids[i], repo, loose_obj_only,
1448 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1449 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1451 eca70f98 2021-09-03 stsp goto done;
1454 05118f5a 2021-06-22 stsp for (i = 0; i < nours; i++) {
1455 05118f5a 2021-06-22 stsp struct got_object_id *id = ours[i];
1456 67fd6849 2022-02-13 stsp struct got_pack_meta *m;
1457 05118f5a 2021-06-22 stsp if (id == NULL)
1459 67fd6849 2022-02-13 stsp m = got_object_idset_get(idset, id);
1460 67fd6849 2022-02-13 stsp if (m == NULL) {
1461 07165b17 2021-07-01 stsp err = got_object_get_type(&obj_type, repo, id);
1463 07165b17 2021-07-01 stsp goto done;
1465 67fd6849 2022-02-13 stsp obj_type = m->obj_type;
1466 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_TAG)
1468 b8af7c06 2022-03-15 stsp err = load_tag(1, idset, id, repo, loose_obj_only,
1469 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1470 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1472 e6bcace5 2021-06-22 stsp goto done;
1475 e6bcace5 2021-06-22 stsp for (i = 0; i < nobj; i++) {
1476 e6bcace5 2021-06-22 stsp free(ids[i]);
1478 e6bcace5 2021-06-22 stsp free(ids);
1479 e6bcace5 2021-06-22 stsp return err;
1482 e6bcace5 2021-06-22 stsp const struct got_error *
1483 e6bcace5 2021-06-22 stsp hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1487 e6bcace5 2021-06-22 stsp SHA1Update(ctx, buf, len);
1488 e6bcace5 2021-06-22 stsp n = fwrite(buf, 1, len, f);
1489 e6bcace5 2021-06-22 stsp if (n != len)
1490 e6bcace5 2021-06-22 stsp return got_ferror(f, GOT_ERR_IO);
1491 e6bcace5 2021-06-22 stsp return NULL;
1494 e6bcace5 2021-06-22 stsp static void
1495 e6bcace5 2021-06-22 stsp putbe32(char *b, uint32_t n)
1497 e6bcace5 2021-06-22 stsp b[0] = n >> 24;
1498 e6bcace5 2021-06-22 stsp b[1] = n >> 16;
1499 e6bcace5 2021-06-22 stsp b[2] = n >> 8;
1500 e6bcace5 2021-06-22 stsp b[3] = n >> 0;
1503 e6bcace5 2021-06-22 stsp static int
1504 e6bcace5 2021-06-22 stsp write_order_cmp(const void *pa, const void *pb)
1506 e6bcace5 2021-06-22 stsp struct got_pack_meta *a, *b, *ahd, *bhd;
1508 e6bcace5 2021-06-22 stsp a = *(struct got_pack_meta **)pa;
1509 e6bcace5 2021-06-22 stsp b = *(struct got_pack_meta **)pb;
1510 e6bcace5 2021-06-22 stsp ahd = (a->head == NULL) ? a : a->head;
1511 e6bcace5 2021-06-22 stsp bhd = (b->head == NULL) ? b : b->head;
1512 e6bcace5 2021-06-22 stsp if (ahd->mtime != bhd->mtime)
1513 e6bcace5 2021-06-22 stsp return bhd->mtime - ahd->mtime;
1514 e6bcace5 2021-06-22 stsp if (ahd != bhd)
1515 e6bcace5 2021-06-22 stsp return (uintptr_t)bhd - (uintptr_t)ahd;
1516 e6bcace5 2021-06-22 stsp if (a->nchain != b->nchain)
1517 e6bcace5 2021-06-22 stsp return a->nchain - b->nchain;
1518 e6bcace5 2021-06-22 stsp return a->mtime - b->mtime;
1521 67fd6849 2022-02-13 stsp static int
1522 67fd6849 2022-02-13 stsp reuse_write_order_cmp(const void *pa, const void *pb)
1524 67fd6849 2022-02-13 stsp struct got_pack_meta *a, *b;
1526 67fd6849 2022-02-13 stsp a = *(struct got_pack_meta **)pa;
1527 67fd6849 2022-02-13 stsp b = *(struct got_pack_meta **)pb;
1529 67fd6849 2022-02-13 stsp if (a->reused_delta_offset < b->reused_delta_offset)
1530 67fd6849 2022-02-13 stsp return -1;
1531 67fd6849 2022-02-13 stsp if (a->reused_delta_offset > b->reused_delta_offset)
1536 e6bcace5 2021-06-22 stsp static const struct got_error *
1537 e6bcace5 2021-06-22 stsp packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1541 e6bcace5 2021-06-22 stsp *hdrlen = 0;
1543 e6bcace5 2021-06-22 stsp hdr[0] = obj_type << 4;
1544 e6bcace5 2021-06-22 stsp hdr[0] |= len & 0xf;
1545 e6bcace5 2021-06-22 stsp len >>= 4;
1546 e6bcace5 2021-06-22 stsp for (i = 1; len != 0; i++){
1547 e6bcace5 2021-06-22 stsp if (i >= bufsize)
1548 e6bcace5 2021-06-22 stsp return got_error(GOT_ERR_NO_SPACE);
1549 e6bcace5 2021-06-22 stsp hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1550 e6bcace5 2021-06-22 stsp hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1551 e6bcace5 2021-06-22 stsp len >>= GOT_DELTA_SIZE_SHIFT;
1554 e6bcace5 2021-06-22 stsp *hdrlen = i;
1555 e6bcace5 2021-06-22 stsp return NULL;
1558 e6bcace5 2021-06-22 stsp static int
1559 e6bcace5 2021-06-22 stsp packoff(char *hdr, off_t off)
1562 e6bcace5 2021-06-22 stsp char rbuf[8];
1564 e6bcace5 2021-06-22 stsp rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1565 e6bcace5 2021-06-22 stsp for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1566 e6bcace5 2021-06-22 stsp rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1567 e6bcace5 2021-06-22 stsp GOT_DELTA_SIZE_MORE;
1571 e6bcace5 2021-06-22 stsp while (i > 0)
1572 e6bcace5 2021-06-22 stsp hdr[j++] = rbuf[--i];
1576 e6bcace5 2021-06-22 stsp static const struct got_error *
1577 5060d5a1 2022-01-10 stsp deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1578 67fd6849 2022-02-13 stsp struct got_pack_meta *m)
1580 5060d5a1 2022-01-10 stsp const struct got_error *err;
1581 5060d5a1 2022-01-10 stsp char buf[32];
1584 67fd6849 2022-02-13 stsp if (m->prev->off != 0) {
1585 5060d5a1 2022-01-10 stsp err = packhdr(&nh, buf, sizeof(buf),
1586 5060d5a1 2022-01-10 stsp GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1588 5060d5a1 2022-01-10 stsp return err;
1589 5060d5a1 2022-01-10 stsp nh += packoff(buf + nh, m->off - m->prev->off);
1590 5060d5a1 2022-01-10 stsp err = hwrite(packfile, buf, nh, ctx);
1592 5060d5a1 2022-01-10 stsp return err;
1593 5060d5a1 2022-01-10 stsp *packfile_size += nh;
1595 5060d5a1 2022-01-10 stsp err = packhdr(&nh, buf, sizeof(buf),
1596 5060d5a1 2022-01-10 stsp GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1598 5060d5a1 2022-01-10 stsp return err;
1599 5060d5a1 2022-01-10 stsp err = hwrite(packfile, buf, nh, ctx);
1601 5060d5a1 2022-01-10 stsp return err;
1602 5060d5a1 2022-01-10 stsp *packfile_size += nh;
1603 5060d5a1 2022-01-10 stsp err = hwrite(packfile, m->prev->id.sha1,
1604 5060d5a1 2022-01-10 stsp sizeof(m->prev->id.sha1), ctx);
1606 5060d5a1 2022-01-10 stsp return err;
1607 5060d5a1 2022-01-10 stsp *packfile_size += sizeof(m->prev->id.sha1);
1610 5060d5a1 2022-01-10 stsp return NULL;
1613 5060d5a1 2022-01-10 stsp static const struct got_error *
1614 67fd6849 2022-02-13 stsp write_packed_object(off_t *packfile_size, FILE *packfile,
1615 67fd6849 2022-02-13 stsp FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1616 67fd6849 2022-02-13 stsp SHA1_CTX *ctx, struct got_repository *repo)
1618 67fd6849 2022-02-13 stsp const struct got_error *err = NULL;
1619 67fd6849 2022-02-13 stsp struct got_deflate_checksum csum;
1620 67fd6849 2022-02-13 stsp char buf[32];
1622 67fd6849 2022-02-13 stsp struct got_raw_object *raw = NULL;
1623 67fd6849 2022-02-13 stsp off_t outlen;
1625 67fd6849 2022-02-13 stsp csum.output_sha1 = ctx;
1626 67fd6849 2022-02-13 stsp csum.output_crc = NULL;
1628 67fd6849 2022-02-13 stsp m->off = ftello(packfile);
1629 67fd6849 2022-02-13 stsp if (m->delta_len == 0) {
1630 67fd6849 2022-02-13 stsp err = got_object_raw_open(&raw, outfd, repo, &m->id);
1632 67fd6849 2022-02-13 stsp goto done;
1633 67fd6849 2022-02-13 stsp err = packhdr(&nh, buf, sizeof(buf),
1634 67fd6849 2022-02-13 stsp m->obj_type, raw->size);
1636 67fd6849 2022-02-13 stsp goto done;
1637 67fd6849 2022-02-13 stsp err = hwrite(packfile, buf, nh, ctx);
1639 67fd6849 2022-02-13 stsp goto done;
1640 67fd6849 2022-02-13 stsp *packfile_size += nh;
1641 67fd6849 2022-02-13 stsp if (raw->f == NULL) {
1642 67fd6849 2022-02-13 stsp err = got_deflate_to_file_mmap(&outlen,
1643 67fd6849 2022-02-13 stsp raw->data + raw->hdrlen, 0, raw->size,
1644 67fd6849 2022-02-13 stsp packfile, &csum);
1646 67fd6849 2022-02-13 stsp goto done;
1648 67fd6849 2022-02-13 stsp if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1650 67fd6849 2022-02-13 stsp err = got_error_from_errno("fseeko");
1651 67fd6849 2022-02-13 stsp goto done;
1653 67fd6849 2022-02-13 stsp err = got_deflate_to_file(&outlen, raw->f,
1654 67fd6849 2022-02-13 stsp raw->size, packfile, &csum);
1656 67fd6849 2022-02-13 stsp goto done;
1658 67fd6849 2022-02-13 stsp *packfile_size += outlen;
1659 67fd6849 2022-02-13 stsp got_object_raw_close(raw);
1660 67fd6849 2022-02-13 stsp raw = NULL;
1661 67fd6849 2022-02-13 stsp } else if (m->delta_buf) {
1662 67fd6849 2022-02-13 stsp err = deltahdr(packfile_size, ctx, packfile, m);
1664 67fd6849 2022-02-13 stsp goto done;
1665 67fd6849 2022-02-13 stsp err = got_deflate_to_file_mmap(&outlen,
1666 67fd6849 2022-02-13 stsp m->delta_buf, 0, m->delta_len, packfile, &csum);
1668 67fd6849 2022-02-13 stsp goto done;
1669 67fd6849 2022-02-13 stsp *packfile_size += outlen;
1670 67fd6849 2022-02-13 stsp free(m->delta_buf);
1671 67fd6849 2022-02-13 stsp m->delta_buf = NULL;
1673 67fd6849 2022-02-13 stsp if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1675 67fd6849 2022-02-13 stsp err = got_error_from_errno("fseeko");
1676 67fd6849 2022-02-13 stsp goto done;
1678 67fd6849 2022-02-13 stsp err = deltahdr(packfile_size, ctx, packfile, m);
1680 67fd6849 2022-02-13 stsp goto done;
1681 67fd6849 2022-02-13 stsp err = got_deflate_to_file(&outlen, delta_cache,
1682 67fd6849 2022-02-13 stsp m->delta_len, packfile, &csum);
1684 67fd6849 2022-02-13 stsp goto done;
1685 67fd6849 2022-02-13 stsp *packfile_size += outlen;
1689 67fd6849 2022-02-13 stsp got_object_raw_close(raw);
1690 67fd6849 2022-02-13 stsp return err;
1693 67fd6849 2022-02-13 stsp static const struct got_error *
1694 74881701 2021-10-15 stsp genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1695 67fd6849 2022-02-13 stsp struct got_pack_meta **deltify, int ndeltify,
1696 67fd6849 2022-02-13 stsp struct got_pack_meta **reuse, int nreuse,
1697 b8af7c06 2022-03-15 stsp int ncolored, int nfound, int ntrees, int nours,
1698 b8af7c06 2022-03-15 stsp struct got_repository *repo,
1699 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1700 211cfef0 2022-01-05 stsp struct got_ratelimit *rl,
1701 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1703 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1705 e6bcace5 2021-06-22 stsp SHA1_CTX ctx;
1706 e6bcace5 2021-06-22 stsp struct got_pack_meta *m;
1707 08347b73 2021-10-14 stsp char buf[32];
1709 67fd6849 2022-02-13 stsp off_t packfile_size = 0;
1710 cc7a354a 2021-10-15 stsp int outfd = -1;
1712 e6bcace5 2021-06-22 stsp SHA1Init(&ctx);
1714 e6bcace5 2021-06-22 stsp err = hwrite(packfile, "PACK", 4, &ctx);
1716 e6bcace5 2021-06-22 stsp return err;
1717 e6bcace5 2021-06-22 stsp putbe32(buf, GOT_PACKFILE_VERSION);
1718 e6bcace5 2021-06-22 stsp err = hwrite(packfile, buf, 4, &ctx);
1720 e6bcace5 2021-06-22 stsp goto done;
1721 67fd6849 2022-02-13 stsp putbe32(buf, ndeltify + nreuse);
1722 e6bcace5 2021-06-22 stsp err = hwrite(packfile, buf, 4, &ctx);
1724 e6bcace5 2021-06-22 stsp goto done;
1726 67fd6849 2022-02-13 stsp qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1727 67fd6849 2022-02-13 stsp write_order_cmp);
1728 67fd6849 2022-02-13 stsp for (i = 0; i < ndeltify; i++) {
1729 211cfef0 2022-01-05 stsp err = report_progress(progress_cb, progress_arg, rl,
1730 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, packfile_size, nours,
1731 b8af7c06 2022-03-15 stsp ndeltify + nreuse, ndeltify + nreuse, i);
1733 211cfef0 2022-01-05 stsp goto done;
1734 67fd6849 2022-02-13 stsp m = deltify[i];
1735 67fd6849 2022-02-13 stsp err = write_packed_object(&packfile_size, packfile,
1736 67fd6849 2022-02-13 stsp delta_cache, m, &outfd, &ctx, repo);
1738 67fd6849 2022-02-13 stsp goto done;
1741 67fd6849 2022-02-13 stsp qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1742 67fd6849 2022-02-13 stsp reuse_write_order_cmp);
1743 67fd6849 2022-02-13 stsp for (i = 0; i < nreuse; i++) {
1744 67fd6849 2022-02-13 stsp err = report_progress(progress_cb, progress_arg, rl,
1745 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, packfile_size, nours,
1746 b8af7c06 2022-03-15 stsp ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1748 67fd6849 2022-02-13 stsp goto done;
1749 67fd6849 2022-02-13 stsp m = reuse[i];
1750 67fd6849 2022-02-13 stsp err = write_packed_object(&packfile_size, packfile,
1751 67fd6849 2022-02-13 stsp delta_cache, m, &outfd, &ctx, repo);
1753 67fd6849 2022-02-13 stsp goto done;
1756 e6bcace5 2021-06-22 stsp SHA1Final(pack_sha1, &ctx);
1757 e6bcace5 2021-06-22 stsp n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1758 e6bcace5 2021-06-22 stsp if (n != SHA1_DIGEST_LENGTH)
1759 e6bcace5 2021-06-22 stsp err = got_ferror(packfile, GOT_ERR_IO);
1760 05118f5a 2021-06-22 stsp packfile_size += SHA1_DIGEST_LENGTH;
1761 dc7edd42 2021-08-22 stsp packfile_size += sizeof(struct got_packfile_hdr);
1762 211cfef0 2022-01-05 stsp if (progress_cb) {
1763 b8af7c06 2022-03-15 stsp err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1764 b8af7c06 2022-03-15 stsp packfile_size, nours, ndeltify + nreuse,
1765 b8af7c06 2022-03-15 stsp ndeltify + nreuse, ndeltify + nreuse);
1767 211cfef0 2022-01-05 stsp goto done;
1770 cc7a354a 2021-10-15 stsp if (outfd != -1 && close(outfd) == -1 && err == NULL)
1771 cc7a354a 2021-10-15 stsp err = got_error_from_errno("close");
1772 e6bcace5 2021-06-22 stsp return err;
1775 67fd6849 2022-02-13 stsp static const struct got_error *
1776 67fd6849 2022-02-13 stsp remove_unused_object(struct got_object_idset_element *entry, void *arg)
1778 67fd6849 2022-02-13 stsp struct got_object_idset *idset = arg;
1780 67fd6849 2022-02-13 stsp if (got_object_idset_get_element_data(entry) == NULL)
1781 67fd6849 2022-02-13 stsp got_object_idset_remove_element(idset, entry);
1783 67fd6849 2022-02-13 stsp return NULL;
1786 67fd6849 2022-02-13 stsp static const struct got_error *
1787 67fd6849 2022-02-13 stsp remove_reused_object(struct got_object_idset_element *entry, void *arg)
1789 67fd6849 2022-02-13 stsp struct got_object_idset *idset = arg;
1790 67fd6849 2022-02-13 stsp struct got_pack_meta *m;
1792 67fd6849 2022-02-13 stsp m = got_object_idset_get_element_data(entry);
1793 67fd6849 2022-02-13 stsp if (m->have_reused_delta)
1794 67fd6849 2022-02-13 stsp got_object_idset_remove_element(idset, entry);
1796 67fd6849 2022-02-13 stsp return NULL;
1799 67fd6849 2022-02-13 stsp static const struct got_error *
1800 67fd6849 2022-02-13 stsp add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1802 67fd6849 2022-02-13 stsp struct got_pack_meta *m = data;
1803 67fd6849 2022-02-13 stsp struct got_pack_metavec *v = arg;
1805 67fd6849 2022-02-13 stsp return add_meta(m, v);
1808 e6bcace5 2021-06-22 stsp const struct got_error *
1809 e6bcace5 2021-06-22 stsp got_pack_create(uint8_t *packsha1, FILE *packfile,
1810 e6bcace5 2021-06-22 stsp struct got_object_id **theirs, int ntheirs,
1811 e6bcace5 2021-06-22 stsp struct got_object_id **ours, int nours,
1812 f8a36e22 2021-08-26 stsp struct got_repository *repo, int loose_obj_only, int allow_empty,
1813 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1814 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1816 e6bcace5 2021-06-22 stsp const struct got_error *err;
1817 67fd6849 2022-02-13 stsp int delta_cache_fd = -1;
1818 74881701 2021-10-15 stsp FILE *delta_cache = NULL;
1819 67fd6849 2022-02-13 stsp struct got_object_idset *idset;
1820 211cfef0 2022-01-05 stsp struct got_ratelimit rl;
1821 67fd6849 2022-02-13 stsp struct got_pack_metavec deltify, reuse;
1822 b8af7c06 2022-03-15 stsp int ncolored = 0, nfound = 0, ntrees = 0;
1824 67fd6849 2022-02-13 stsp memset(&deltify, 0, sizeof(deltify));
1825 67fd6849 2022-02-13 stsp memset(&reuse, 0, sizeof(reuse));
1827 211cfef0 2022-01-05 stsp got_ratelimit_init(&rl, 0, 500);
1829 67fd6849 2022-02-13 stsp idset = got_object_idset_alloc();
1830 67fd6849 2022-02-13 stsp if (idset == NULL)
1831 67fd6849 2022-02-13 stsp return got_error_from_errno("got_object_idset_alloc");
1833 b8af7c06 2022-03-15 stsp err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1834 b8af7c06 2022-03-15 stsp ntheirs, ours, nours, repo, loose_obj_only,
1835 b8af7c06 2022-03-15 stsp progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1837 e6bcace5 2021-06-22 stsp return err;
1839 67fd6849 2022-02-13 stsp err = got_object_idset_for_each_element(idset,
1840 67fd6849 2022-02-13 stsp remove_unused_object, idset);
1842 67fd6849 2022-02-13 stsp goto done;
1844 28526235 2022-02-13 stsp if (progress_cb) {
1845 b8af7c06 2022-03-15 stsp err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1846 b8af7c06 2022-03-15 stsp 0L, nours, got_object_idset_num_elements(idset), 0, 0);
1848 28526235 2022-02-13 stsp goto done;
1851 67fd6849 2022-02-13 stsp if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1852 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_CANNOT_PACK);
1853 05118f5a 2021-06-22 stsp goto done;
1856 67fd6849 2022-02-13 stsp delta_cache_fd = got_opentempfd();
1857 67fd6849 2022-02-13 stsp if (delta_cache_fd == -1) {
1858 74881701 2021-10-15 stsp err = got_error_from_errno("got_opentemp");
1859 74881701 2021-10-15 stsp goto done;
1862 67fd6849 2022-02-13 stsp reuse.metasz = 64;
1863 67fd6849 2022-02-13 stsp reuse.meta = calloc(reuse.metasz,
1864 67fd6849 2022-02-13 stsp sizeof(struct got_pack_meta *));
1865 67fd6849 2022-02-13 stsp if (reuse.meta == NULL) {
1866 67fd6849 2022-02-13 stsp err = got_error_from_errno("calloc");
1867 67fd6849 2022-02-13 stsp goto done;
1870 b8af7c06 2022-03-15 stsp err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
1871 b8af7c06 2022-03-15 stsp ntrees, nours, repo, progress_cb, progress_arg, &rl,
1872 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1874 67fd6849 2022-02-13 stsp goto done;
1875 67fd6849 2022-02-13 stsp if (reuse.nmeta > 0) {
1876 67fd6849 2022-02-13 stsp err = got_object_idset_for_each_element(idset,
1877 67fd6849 2022-02-13 stsp remove_reused_object, idset);
1879 67fd6849 2022-02-13 stsp goto done;
1882 67fd6849 2022-02-13 stsp delta_cache = fdopen(delta_cache_fd, "a+");
1883 67fd6849 2022-02-13 stsp if (delta_cache == NULL) {
1884 67fd6849 2022-02-13 stsp err = got_error_from_errno("fdopen");
1885 67fd6849 2022-02-13 stsp goto done;
1887 67fd6849 2022-02-13 stsp delta_cache_fd = -1;
1889 67fd6849 2022-02-13 stsp if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1890 67fd6849 2022-02-13 stsp err = got_error_from_errno("fseeko");
1891 67fd6849 2022-02-13 stsp goto done;
1894 67fd6849 2022-02-13 stsp deltify.meta = calloc(got_object_idset_num_elements(idset),
1895 67fd6849 2022-02-13 stsp sizeof(struct got_pack_meta *));
1896 67fd6849 2022-02-13 stsp if (deltify.meta == NULL) {
1897 67fd6849 2022-02-13 stsp err = got_error_from_errno("calloc");
1898 67fd6849 2022-02-13 stsp goto done;
1900 67fd6849 2022-02-13 stsp deltify.metasz = got_object_idset_num_elements(idset);
1902 67fd6849 2022-02-13 stsp err = got_object_idset_for_each(idset, add_meta_idset_cb, &deltify);
1904 67fd6849 2022-02-13 stsp goto done;
1905 67fd6849 2022-02-13 stsp if (deltify.nmeta > 0) {
1906 b8af7c06 2022-03-15 stsp err = pick_deltas(deltify.meta, deltify.nmeta, ncolored,
1907 b8af7c06 2022-03-15 stsp nfound, ntrees, nours, reuse.nmeta, delta_cache, repo,
1908 211cfef0 2022-01-05 stsp progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1910 f8a36e22 2021-08-26 stsp goto done;
1911 74881701 2021-10-15 stsp if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1912 74881701 2021-10-15 stsp err = got_error_from_errno("fseeko");
1913 74881701 2021-10-15 stsp goto done;
1917 67fd6849 2022-02-13 stsp err = genpack(packsha1, packfile, delta_cache, deltify.meta,
1918 b8af7c06 2022-03-15 stsp deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1919 b8af7c06 2022-03-15 stsp nours, repo, progress_cb, progress_arg, &rl,
1920 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1922 e6bcace5 2021-06-22 stsp goto done;
1924 67fd6849 2022-02-13 stsp free_nmeta(deltify.meta, deltify.nmeta);
1925 67fd6849 2022-02-13 stsp free_nmeta(reuse.meta, reuse.nmeta);
1926 67fd6849 2022-02-13 stsp got_object_idset_free(idset);
1927 67fd6849 2022-02-13 stsp if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1928 67fd6849 2022-02-13 stsp err = got_error_from_errno("close");
1929 74881701 2021-10-15 stsp if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1930 74881701 2021-10-15 stsp err = got_error_from_errno("fclose");
1931 e6bcace5 2021-06-22 stsp return err;
