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>
24 e93fb944 2022-05-10 stsp #include <sys/mman.h>
26 e3f86256 2022-02-18 naddy #include <endian.h>
27 e93fb944 2022-05-10 stsp #include <errno.h>
28 08736cf9 2021-06-23 stsp #include <stdint.h>
29 05118f5a 2021-06-22 stsp #include <imsg.h>
30 e6bcace5 2021-06-22 stsp #include <stdio.h>
31 e6bcace5 2021-06-22 stsp #include <stdlib.h>
32 e6bcace5 2021-06-22 stsp #include <string.h>
33 e6bcace5 2021-06-22 stsp #include <sha1.h>
34 211cfef0 2022-01-05 stsp #include <time.h>
35 bfc73a47 2022-03-19 naddy #include <unistd.h>
36 e6bcace5 2021-06-22 stsp #include <limits.h>
37 e6bcace5 2021-06-22 stsp #include <zlib.h>
39 e6bcace5 2021-06-22 stsp #include "got_error.h"
40 e6bcace5 2021-06-22 stsp #include "got_cancel.h"
41 e6bcace5 2021-06-22 stsp #include "got_object.h"
42 f8b19efd 2021-10-13 stsp #include "got_path.h"
43 05118f5a 2021-06-22 stsp #include "got_reference.h"
44 05118f5a 2021-06-22 stsp #include "got_repository_admin.h"
45 08347b73 2021-10-14 stsp #include "got_opentemp.h"
47 e6bcace5 2021-06-22 stsp #include "got_lib_deltify.h"
48 e6bcace5 2021-06-22 stsp #include "got_lib_delta.h"
49 e6bcace5 2021-06-22 stsp #include "got_lib_object.h"
50 e6bcace5 2021-06-22 stsp #include "got_lib_object_idset.h"
51 05118f5a 2021-06-22 stsp #include "got_lib_object_cache.h"
52 e6bcace5 2021-06-22 stsp #include "got_lib_deflate.h"
53 e6bcace5 2021-06-22 stsp #include "got_lib_pack.h"
54 05118f5a 2021-06-22 stsp #include "got_lib_privsep.h"
55 05118f5a 2021-06-22 stsp #include "got_lib_repository.h"
56 211cfef0 2022-01-05 stsp #include "got_lib_ratelimit.h"
57 2d9e6abf 2022-05-04 stsp #include "got_lib_inflate.h"
59 f8174ca5 2022-05-20 stsp #include "murmurhash2.h"
62 74881701 2021-10-15 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
66 e6bcace5 2021-06-22 stsp #define MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
69 70f8f24d 2022-04-14 stsp #ifndef nitems
70 70f8f24d 2022-04-14 stsp #define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
73 e6bcace5 2021-06-22 stsp struct got_pack_meta {
74 e6bcace5 2021-06-22 stsp struct got_object_id id;
75 f8174ca5 2022-05-20 stsp uint32_t path_hash;
76 e6bcace5 2021-06-22 stsp int obj_type;
78 e6bcace5 2021-06-22 stsp time_t mtime;
80 e6bcace5 2021-06-22 stsp /* The best delta we picked */
81 e6bcace5 2021-06-22 stsp struct got_pack_meta *head;
82 e6bcace5 2021-06-22 stsp struct got_pack_meta *prev;
83 2d9e6abf 2022-05-04 stsp unsigned char *delta_buf; /* if encoded in memory (compressed) */
84 2d9e6abf 2022-05-04 stsp off_t delta_offset; /* offset in delta cache file (compressed) */
85 5060d5a1 2022-01-10 stsp off_t delta_len; /* encoded delta length */
86 2d9e6abf 2022-05-04 stsp off_t delta_compressed_len; /* encoded+compressed delta length */
89 67fd6849 2022-02-13 stsp int have_reused_delta;
90 67fd6849 2022-02-13 stsp off_t reused_delta_offset; /* offset of delta in reused pack file */
91 67fd6849 2022-02-13 stsp struct got_object_id *base_obj_id;
93 e6bcace5 2021-06-22 stsp /* Only used for delta window */
94 e6bcace5 2021-06-22 stsp struct got_delta_table *dtab;
96 e6bcace5 2021-06-22 stsp /* Only used for writing offset deltas */
100 e6bcace5 2021-06-22 stsp struct got_pack_metavec {
101 e6bcace5 2021-06-22 stsp struct got_pack_meta **meta;
103 e6bcace5 2021-06-22 stsp int metasz;
106 e6bcace5 2021-06-22 stsp static const struct got_error *
107 e6bcace5 2021-06-22 stsp alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
108 e6bcace5 2021-06-22 stsp const char *path, int obj_type, time_t mtime)
110 e6bcace5 2021-06-22 stsp struct got_pack_meta *m;
112 e6bcace5 2021-06-22 stsp *new = NULL;
114 e6bcace5 2021-06-22 stsp m = calloc(1, sizeof(*m));
115 e6bcace5 2021-06-22 stsp if (m == NULL)
116 1d19226a 2021-10-13 stsp return got_error_from_errno("calloc");
118 e6bcace5 2021-06-22 stsp memcpy(&m->id, id, sizeof(m->id));
120 f8174ca5 2022-05-20 stsp m->path_hash = murmurhash2(path, strlen(path), 0xd70af26a);
121 e6bcace5 2021-06-22 stsp m->obj_type = obj_type;
122 e6bcace5 2021-06-22 stsp m->mtime = mtime;
124 e6bcace5 2021-06-22 stsp return NULL;
127 e6bcace5 2021-06-22 stsp static void
128 e6bcace5 2021-06-22 stsp clear_meta(struct got_pack_meta *meta)
130 e6bcace5 2021-06-22 stsp if (meta == NULL)
132 f8174ca5 2022-05-20 stsp meta->path_hash = 0;
133 5060d5a1 2022-01-10 stsp free(meta->delta_buf);
134 5060d5a1 2022-01-10 stsp meta->delta_buf = NULL;
135 67fd6849 2022-02-13 stsp free(meta->base_obj_id);
136 67fd6849 2022-02-13 stsp meta->base_obj_id = NULL;
139 e6bcace5 2021-06-22 stsp static void
140 e6bcace5 2021-06-22 stsp free_nmeta(struct got_pack_meta **meta, int nmeta)
144 e6bcace5 2021-06-22 stsp for (i = 0; i < nmeta; i++)
145 e6bcace5 2021-06-22 stsp clear_meta(meta[i]);
146 e6bcace5 2021-06-22 stsp free(meta);
150 e6bcace5 2021-06-22 stsp delta_order_cmp(const void *pa, const void *pb)
152 e6bcace5 2021-06-22 stsp struct got_pack_meta *a, *b;
154 e6bcace5 2021-06-22 stsp a = *(struct got_pack_meta **)pa;
155 e6bcace5 2021-06-22 stsp b = *(struct got_pack_meta **)pb;
157 e6bcace5 2021-06-22 stsp if (a->obj_type != b->obj_type)
158 e6bcace5 2021-06-22 stsp return a->obj_type - b->obj_type;
159 f8174ca5 2022-05-20 stsp if (a->path_hash < b->path_hash)
161 f8174ca5 2022-05-20 stsp if (a->path_hash > b->path_hash)
163 611e8e31 2022-05-01 stsp if (a->mtime < b->mtime)
165 611e8e31 2022-05-01 stsp if (a->mtime > b->mtime)
167 e6bcace5 2021-06-22 stsp return got_object_id_cmp(&a->id, &b->id);
170 5060d5a1 2022-01-10 stsp static off_t
171 e6bcace5 2021-06-22 stsp delta_size(struct got_delta_instruction *deltas, int ndeltas)
174 5060d5a1 2022-01-10 stsp off_t size = 32;
175 e6bcace5 2021-06-22 stsp for (i = 0; i < ndeltas; i++) {
176 e6bcace5 2021-06-22 stsp if (deltas[i].copy)
177 e6bcace5 2021-06-22 stsp size += GOT_DELTA_SIZE_SHIFT;
179 e6bcace5 2021-06-22 stsp size += deltas[i].len + 1;
181 e6bcace5 2021-06-22 stsp return size;
184 5060d5a1 2022-01-10 stsp static const struct got_error *
185 5060d5a1 2022-01-10 stsp append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
189 5060d5a1 2022-01-10 stsp if (*len + nseg >= *sz) {
190 5060d5a1 2022-01-10 stsp while (*len + nseg >= *sz)
191 5060d5a1 2022-01-10 stsp *sz += *sz / 2;
192 5060d5a1 2022-01-10 stsp n = realloc(*p, *sz);
193 5060d5a1 2022-01-10 stsp if (n == NULL)
194 5060d5a1 2022-01-10 stsp return got_error_from_errno("realloc");
197 5060d5a1 2022-01-10 stsp memcpy(*p + *len, seg, nseg);
198 5060d5a1 2022-01-10 stsp *len += nseg;
199 5060d5a1 2022-01-10 stsp return NULL;
202 5060d5a1 2022-01-10 stsp static const struct got_error *
203 5060d5a1 2022-01-10 stsp encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
204 5060d5a1 2022-01-10 stsp struct got_delta_instruction *deltas, int ndeltas,
205 5060d5a1 2022-01-10 stsp off_t delta_size, off_t base_size)
207 5060d5a1 2022-01-10 stsp const struct got_error *err;
208 5060d5a1 2022-01-10 stsp unsigned char buf[16], *bp;
210 2d9e6abf 2022-05-04 stsp size_t len = 0, compressed_len;
211 2d9e6abf 2022-05-04 stsp off_t bufsize = delta_size;
213 5060d5a1 2022-01-10 stsp struct got_delta_instruction *d;
214 2d9e6abf 2022-05-04 stsp uint8_t *delta_buf;
216 2d9e6abf 2022-05-04 stsp delta_buf = malloc(bufsize);
217 2d9e6abf 2022-05-04 stsp if (delta_buf == NULL)
218 2d9e6abf 2022-05-04 stsp return got_error_from_errno("malloc");
220 5060d5a1 2022-01-10 stsp /* base object size */
221 5060d5a1 2022-01-10 stsp buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
222 5060d5a1 2022-01-10 stsp n = base_size >> GOT_DELTA_SIZE_SHIFT;
223 5060d5a1 2022-01-10 stsp for (i = 1; n > 0; i++) {
224 5060d5a1 2022-01-10 stsp buf[i - 1] |= GOT_DELTA_SIZE_MORE;
225 5060d5a1 2022-01-10 stsp buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
226 5060d5a1 2022-01-10 stsp n >>= GOT_DELTA_SIZE_SHIFT;
228 2d9e6abf 2022-05-04 stsp err = append(&delta_buf, &len, &bufsize, buf, i);
232 5060d5a1 2022-01-10 stsp /* target object size */
233 5060d5a1 2022-01-10 stsp buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
234 5060d5a1 2022-01-10 stsp n = o->size >> GOT_DELTA_SIZE_SHIFT;
235 5060d5a1 2022-01-10 stsp for (i = 1; n > 0; i++) {
236 5060d5a1 2022-01-10 stsp buf[i - 1] |= GOT_DELTA_SIZE_MORE;
237 5060d5a1 2022-01-10 stsp buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
238 5060d5a1 2022-01-10 stsp n >>= GOT_DELTA_SIZE_SHIFT;
240 2d9e6abf 2022-05-04 stsp err = append(&delta_buf, &len, &bufsize, buf, i);
244 5060d5a1 2022-01-10 stsp for (j = 0; j < ndeltas; j++) {
245 5060d5a1 2022-01-10 stsp d = &deltas[j];
246 5060d5a1 2022-01-10 stsp if (d->copy) {
247 5060d5a1 2022-01-10 stsp n = d->offset;
248 5060d5a1 2022-01-10 stsp bp = &buf[1];
249 5060d5a1 2022-01-10 stsp buf[0] = GOT_DELTA_BASE_COPY;
250 5060d5a1 2022-01-10 stsp for (i = 0; i < 4; i++) {
251 5060d5a1 2022-01-10 stsp /* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
252 5060d5a1 2022-01-10 stsp buf[0] |= 1 << i;
253 5060d5a1 2022-01-10 stsp *bp++ = n & 0xff;
255 5060d5a1 2022-01-10 stsp if (n == 0)
259 5060d5a1 2022-01-10 stsp n = d->len;
260 5060d5a1 2022-01-10 stsp if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
261 5060d5a1 2022-01-10 stsp /* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
262 5060d5a1 2022-01-10 stsp for (i = 0; i < 3 && n > 0; i++) {
263 5060d5a1 2022-01-10 stsp buf[0] |= 1 << (i + 4);
264 5060d5a1 2022-01-10 stsp *bp++ = n & 0xff;
268 2d9e6abf 2022-05-04 stsp err = append(&delta_buf, &len, &bufsize,
269 5060d5a1 2022-01-10 stsp buf, bp - buf);
272 5060d5a1 2022-01-10 stsp } else if (o->f == NULL) {
274 5060d5a1 2022-01-10 stsp while (n != d->len) {
275 5060d5a1 2022-01-10 stsp buf[0] = (d->len - n < 127) ? d->len - n : 127;
276 2d9e6abf 2022-05-04 stsp err = append(&delta_buf, &len, &bufsize,
280 2d9e6abf 2022-05-04 stsp err = append(&delta_buf, &len, &bufsize,
281 5060d5a1 2022-01-10 stsp o->data + o->hdrlen + d->offset + n,
285 5060d5a1 2022-01-10 stsp n += buf[0];
288 5060d5a1 2022-01-10 stsp char content[128];
290 2d9e6abf 2022-05-04 stsp if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
291 2d9e6abf 2022-05-04 stsp err = got_error_from_errno("fseeko");
295 5060d5a1 2022-01-10 stsp while (n != d->len) {
296 5060d5a1 2022-01-10 stsp buf[0] = (d->len - n < 127) ? d->len - n : 127;
297 2d9e6abf 2022-05-04 stsp err = append(&delta_buf, &len, &bufsize,
301 5060d5a1 2022-01-10 stsp r = fread(content, 1, buf[0], o->f);
302 2d9e6abf 2022-05-04 stsp if (r != buf[0]) {
303 2d9e6abf 2022-05-04 stsp err = got_ferror(o->f, GOT_ERR_IO);
306 2d9e6abf 2022-05-04 stsp err = append(&delta_buf, &len, &bufsize,
307 5060d5a1 2022-01-10 stsp content, buf[0]);
310 5060d5a1 2022-01-10 stsp n += buf[0];
315 2d9e6abf 2022-05-04 stsp err = got_deflate_to_mem_mmap(&m->delta_buf, &compressed_len,
316 2d9e6abf 2022-05-04 stsp NULL, NULL, delta_buf, 0, len);
320 5060d5a1 2022-01-10 stsp m->delta_len = len;
321 2d9e6abf 2022-05-04 stsp m->delta_compressed_len = compressed_len;
323 2d9e6abf 2022-05-04 stsp free(delta_buf);
324 2d9e6abf 2022-05-04 stsp return err;
327 74881701 2021-10-15 stsp static const struct got_error *
328 74881701 2021-10-15 stsp encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
329 74881701 2021-10-15 stsp struct got_delta_instruction *deltas, int ndeltas,
330 a319ca8c 2021-10-15 stsp off_t base_size, FILE *f)
332 2d9e6abf 2022-05-04 stsp const struct got_error *err;
333 a319ca8c 2021-10-15 stsp unsigned char buf[16], *bp;
336 2d9e6abf 2022-05-04 stsp struct got_deflate_buf zb;
337 a319ca8c 2021-10-15 stsp struct got_delta_instruction *d;
338 2d9e6abf 2022-05-04 stsp off_t delta_len = 0, compressed_len = 0;
340 2d9e6abf 2022-05-04 stsp err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
342 2d9e6abf 2022-05-04 stsp return err;
344 a319ca8c 2021-10-15 stsp /* base object size */
345 a319ca8c 2021-10-15 stsp buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
346 a319ca8c 2021-10-15 stsp n = base_size >> GOT_DELTA_SIZE_SHIFT;
347 a319ca8c 2021-10-15 stsp for (i = 1; n > 0; i++) {
348 a319ca8c 2021-10-15 stsp buf[i - 1] |= GOT_DELTA_SIZE_MORE;
349 a319ca8c 2021-10-15 stsp buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
350 a319ca8c 2021-10-15 stsp n >>= GOT_DELTA_SIZE_SHIFT;
353 2d9e6abf 2022-05-04 stsp err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
354 2d9e6abf 2022-05-04 stsp buf, 0, i, f, NULL);
357 2d9e6abf 2022-05-04 stsp delta_len += i;
359 a319ca8c 2021-10-15 stsp /* target object size */
360 a319ca8c 2021-10-15 stsp buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
361 a319ca8c 2021-10-15 stsp n = o->size >> GOT_DELTA_SIZE_SHIFT;
362 a319ca8c 2021-10-15 stsp for (i = 1; n > 0; i++) {
363 a319ca8c 2021-10-15 stsp buf[i - 1] |= GOT_DELTA_SIZE_MORE;
364 a319ca8c 2021-10-15 stsp buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
365 a319ca8c 2021-10-15 stsp n >>= GOT_DELTA_SIZE_SHIFT;
368 2d9e6abf 2022-05-04 stsp err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
369 2d9e6abf 2022-05-04 stsp buf, 0, i, f, NULL);
372 2d9e6abf 2022-05-04 stsp delta_len += i;
374 a319ca8c 2021-10-15 stsp for (j = 0; j < ndeltas; j++) {
375 a319ca8c 2021-10-15 stsp d = &deltas[j];
376 a319ca8c 2021-10-15 stsp if (d->copy) {
377 a319ca8c 2021-10-15 stsp n = d->offset;
378 a319ca8c 2021-10-15 stsp bp = &buf[1];
379 a319ca8c 2021-10-15 stsp buf[0] = GOT_DELTA_BASE_COPY;
380 a319ca8c 2021-10-15 stsp for (i = 0; i < 4; i++) {
381 a319ca8c 2021-10-15 stsp /* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
382 a319ca8c 2021-10-15 stsp buf[0] |= 1 << i;
383 a319ca8c 2021-10-15 stsp *bp++ = n & 0xff;
385 a319ca8c 2021-10-15 stsp if (n == 0)
388 a319ca8c 2021-10-15 stsp n = d->len;
389 a319ca8c 2021-10-15 stsp if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
390 a319ca8c 2021-10-15 stsp /* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
391 a319ca8c 2021-10-15 stsp for (i = 0; i < 3 && n > 0; i++) {
392 a319ca8c 2021-10-15 stsp buf[0] |= 1 << (i + 4);
393 a319ca8c 2021-10-15 stsp *bp++ = n & 0xff;
397 2d9e6abf 2022-05-04 stsp err = got_deflate_append_to_file_mmap(&zb,
398 2d9e6abf 2022-05-04 stsp &compressed_len, buf, 0, bp - buf, f, NULL);
401 2d9e6abf 2022-05-04 stsp delta_len += (bp - buf);
402 64a8571e 2022-01-07 stsp } else if (o->f == NULL) {
404 64a8571e 2022-01-07 stsp while (n != d->len) {
405 64a8571e 2022-01-07 stsp buf[0] = (d->len - n < 127) ? d->len - n : 127;
406 2d9e6abf 2022-05-04 stsp err = got_deflate_append_to_file_mmap(&zb,
407 2d9e6abf 2022-05-04 stsp &compressed_len, buf, 0, 1, f, NULL);
410 2d9e6abf 2022-05-04 stsp delta_len++;
411 2d9e6abf 2022-05-04 stsp err = got_deflate_append_to_file_mmap(&zb,
412 2d9e6abf 2022-05-04 stsp &compressed_len,
413 2d9e6abf 2022-05-04 stsp o->data + o->hdrlen + d->offset + n, 0,
414 2d9e6abf 2022-05-04 stsp buf[0], f, NULL);
417 2d9e6abf 2022-05-04 stsp delta_len += buf[0];
418 64a8571e 2022-01-07 stsp n += buf[0];
421 a319ca8c 2021-10-15 stsp char content[128];
423 2d9e6abf 2022-05-04 stsp if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
424 2d9e6abf 2022-05-04 stsp err = got_error_from_errno("fseeko");
428 a319ca8c 2021-10-15 stsp while (n != d->len) {
429 a319ca8c 2021-10-15 stsp buf[0] = (d->len - n < 127) ? d->len - n : 127;
430 2d9e6abf 2022-05-04 stsp err = got_deflate_append_to_file_mmap(&zb,
431 2d9e6abf 2022-05-04 stsp &compressed_len, buf, 0, 1, f, NULL);
434 2d9e6abf 2022-05-04 stsp delta_len++;
435 a319ca8c 2021-10-15 stsp r = fread(content, 1, buf[0], o->f);
436 2d9e6abf 2022-05-04 stsp if (r != buf[0]) {
437 2d9e6abf 2022-05-04 stsp err = got_ferror(o->f, GOT_ERR_IO);
440 2d9e6abf 2022-05-04 stsp err = got_deflate_append_to_file_mmap(&zb,
441 2d9e6abf 2022-05-04 stsp &compressed_len, content, 0, buf[0], f,
445 2d9e6abf 2022-05-04 stsp delta_len += buf[0];
446 a319ca8c 2021-10-15 stsp n += buf[0];
451 2d9e6abf 2022-05-04 stsp err = got_deflate_flush(&zb, f, NULL, &compressed_len);
455 2d9e6abf 2022-05-04 stsp /* sanity check */
456 2d9e6abf 2022-05-04 stsp if (compressed_len != ftello(f) - m->delta_offset) {
457 2d9e6abf 2022-05-04 stsp err = got_error(GOT_ERR_COMPRESSION);
461 2d9e6abf 2022-05-04 stsp m->delta_len = delta_len;
462 2d9e6abf 2022-05-04 stsp m->delta_compressed_len = compressed_len;
464 2d9e6abf 2022-05-04 stsp got_deflate_end(&zb);
465 2d9e6abf 2022-05-04 stsp return err;
468 211cfef0 2022-01-05 stsp static const struct got_error *
469 211cfef0 2022-01-05 stsp report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
470 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
471 b8af7c06 2022-03-15 stsp off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
472 b8af7c06 2022-03-15 stsp int nobj_written)
474 211cfef0 2022-01-05 stsp const struct got_error *err;
475 211cfef0 2022-01-05 stsp int elapsed;
477 211cfef0 2022-01-05 stsp if (progress_cb == NULL)
478 211cfef0 2022-01-05 stsp return NULL;
480 211cfef0 2022-01-05 stsp err = got_ratelimit_check(&elapsed, rl);
481 211cfef0 2022-01-05 stsp if (err || !elapsed)
482 211cfef0 2022-01-05 stsp return err;
484 b8af7c06 2022-03-15 stsp return progress_cb(progress_arg, ncolored, nfound, ntrees,
485 b8af7c06 2022-03-15 stsp packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
488 e6bcace5 2021-06-22 stsp static const struct got_error *
489 67fd6849 2022-02-13 stsp add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
491 67fd6849 2022-02-13 stsp if (v->nmeta == v->metasz){
492 67fd6849 2022-02-13 stsp size_t newsize = 2 * v->metasz;
493 67fd6849 2022-02-13 stsp struct got_pack_meta **new;
494 67fd6849 2022-02-13 stsp new = reallocarray(v->meta, newsize, sizeof(*new));
495 67fd6849 2022-02-13 stsp if (new == NULL)
496 67fd6849 2022-02-13 stsp return got_error_from_errno("reallocarray");
497 67fd6849 2022-02-13 stsp v->meta = new;
498 67fd6849 2022-02-13 stsp v->metasz = newsize;
501 67fd6849 2022-02-13 stsp v->meta[v->nmeta++] = m;
502 67fd6849 2022-02-13 stsp return NULL;
505 67fd6849 2022-02-13 stsp static const struct got_error *
506 67fd6849 2022-02-13 stsp find_pack_for_reuse(struct got_packidx **best_packidx,
507 67fd6849 2022-02-13 stsp struct got_repository *repo)
509 9b576444 2022-03-14 stsp const struct got_error *err = NULL;
510 67fd6849 2022-02-13 stsp struct got_pathlist_entry *pe;
511 67fd6849 2022-02-13 stsp const char *best_packidx_path = NULL;
512 67fd6849 2022-02-13 stsp int nobj_max = 0;
514 67fd6849 2022-02-13 stsp *best_packidx = NULL;
516 9b576444 2022-03-14 stsp TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
517 67fd6849 2022-02-13 stsp const char *path_packidx = pe->path;
518 67fd6849 2022-02-13 stsp struct got_packidx *packidx;
521 67fd6849 2022-02-13 stsp err = got_repo_get_packidx(&packidx, path_packidx, repo);
525 67fd6849 2022-02-13 stsp nobj = be32toh(packidx->hdr.fanout_table[0xff]);
526 67fd6849 2022-02-13 stsp if (nobj > nobj_max) {
527 67fd6849 2022-02-13 stsp best_packidx_path = path_packidx;
528 67fd6849 2022-02-13 stsp nobj_max = nobj;
532 67fd6849 2022-02-13 stsp if (best_packidx_path) {
533 67fd6849 2022-02-13 stsp err = got_repo_get_packidx(best_packidx, best_packidx_path,
537 67fd6849 2022-02-13 stsp return err;
540 fae7e038 2022-05-07 stsp struct send_id_arg {
541 fae7e038 2022-05-07 stsp struct imsgbuf *ibuf;
542 fae7e038 2022-05-07 stsp struct got_object_id *ids[GOT_IMSG_OBJ_ID_LIST_MAX_NIDS];
543 fae7e038 2022-05-07 stsp size_t nids;
546 67fd6849 2022-02-13 stsp static const struct got_error *
547 fae7e038 2022-05-07 stsp send_id(struct got_object_id *id, void *data, void *arg)
549 fae7e038 2022-05-07 stsp const struct got_error *err = NULL;
550 fae7e038 2022-05-07 stsp struct send_id_arg *a = arg;
552 fae7e038 2022-05-07 stsp a->ids[a->nids++] = id;
554 fae7e038 2022-05-07 stsp if (a->nids >= GOT_IMSG_OBJ_ID_LIST_MAX_NIDS) {
555 fae7e038 2022-05-07 stsp err = got_privsep_send_object_idlist(a->ibuf, a->ids, a->nids);
557 67fd6849 2022-02-13 stsp return err;
558 fae7e038 2022-05-07 stsp a->nids = 0;
561 fae7e038 2022-05-07 stsp return NULL;
564 fae7e038 2022-05-07 stsp static const struct got_error *
565 fae7e038 2022-05-07 stsp recv_reused_delta(struct got_imsg_reused_delta *delta,
566 fae7e038 2022-05-07 stsp struct got_object_idset *idset, struct got_pack_metavec *v)
568 fae7e038 2022-05-07 stsp struct got_pack_meta *m, *base;
570 fae7e038 2022-05-07 stsp if (delta->delta_offset + delta->delta_size < delta->delta_offset ||
571 fae7e038 2022-05-07 stsp delta->delta_offset +
572 fae7e038 2022-05-07 stsp delta->delta_compressed_size < delta->delta_offset)
573 fae7e038 2022-05-07 stsp return got_error(GOT_ERR_BAD_PACKFILE);
575 fae7e038 2022-05-07 stsp m = got_object_idset_get(idset, &delta->id);
576 fae7e038 2022-05-07 stsp if (m == NULL)
577 fae7e038 2022-05-07 stsp return got_error(GOT_ERR_NO_OBJ);
579 fae7e038 2022-05-07 stsp base = got_object_idset_get(idset, &delta->base_id);
580 fae7e038 2022-05-07 stsp if (base == NULL)
581 fae7e038 2022-05-07 stsp return got_error(GOT_ERR_NO_OBJ);
583 fae7e038 2022-05-07 stsp m->delta_len = delta->delta_size;
584 fae7e038 2022-05-07 stsp m->delta_compressed_len = delta->delta_compressed_size;
585 fae7e038 2022-05-07 stsp m->delta_offset = delta->delta_out_offset;
586 fae7e038 2022-05-07 stsp m->prev = base;
587 fae7e038 2022-05-07 stsp m->size = delta->result_size;
588 fae7e038 2022-05-07 stsp m->have_reused_delta = 1;
589 fae7e038 2022-05-07 stsp m->reused_delta_offset = delta->delta_offset;
590 fae7e038 2022-05-07 stsp m->base_obj_id = got_object_id_dup(&delta->base_id);
591 fae7e038 2022-05-07 stsp if (m->base_obj_id == NULL)
592 fae7e038 2022-05-07 stsp return got_error_from_errno("got_object_id_dup");
594 fae7e038 2022-05-07 stsp return add_meta(m, v);
597 67fd6849 2022-02-13 stsp static const struct got_error *
598 67fd6849 2022-02-13 stsp search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
599 b8af7c06 2022-03-15 stsp int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
600 b8af7c06 2022-03-15 stsp struct got_repository *repo,
601 67fd6849 2022-02-13 stsp got_pack_progress_cb progress_cb, void *progress_arg,
602 67fd6849 2022-02-13 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
604 67fd6849 2022-02-13 stsp const struct got_error *err = NULL;
605 67fd6849 2022-02-13 stsp struct got_packidx *packidx;
606 67fd6849 2022-02-13 stsp struct got_pack *pack;
607 fae7e038 2022-05-07 stsp struct send_id_arg sia;
608 fae7e038 2022-05-07 stsp struct got_imsg_reused_delta deltas[GOT_IMSG_REUSED_DELTAS_MAX_NDELTAS];
609 fae7e038 2022-05-07 stsp size_t ndeltas, i;
611 67fd6849 2022-02-13 stsp err = find_pack_for_reuse(&packidx, repo);
613 67fd6849 2022-02-13 stsp return err;
615 67fd6849 2022-02-13 stsp if (packidx == NULL)
616 67fd6849 2022-02-13 stsp return NULL;
618 fae7e038 2022-05-07 stsp err = got_object_prepare_delta_reuse(&pack, packidx,
619 fae7e038 2022-05-07 stsp delta_cache_fd, repo);
621 67fd6849 2022-02-13 stsp return err;
623 fae7e038 2022-05-07 stsp memset(&sia, 0, sizeof(sia));
624 fae7e038 2022-05-07 stsp sia.ibuf = pack->privsep_child->ibuf;
625 fae7e038 2022-05-07 stsp err = got_object_idset_for_each(idset, send_id, &sia);
627 fae7e038 2022-05-07 stsp return err;
628 fae7e038 2022-05-07 stsp if (sia.nids > 0) {
629 fae7e038 2022-05-07 stsp err = got_privsep_send_object_idlist(pack->privsep_child->ibuf,
630 fae7e038 2022-05-07 stsp sia.ids, sia.nids);
632 fae7e038 2022-05-07 stsp return err;
634 fae7e038 2022-05-07 stsp err = got_privsep_send_object_idlist_done(pack->privsep_child->ibuf);
636 fae7e038 2022-05-07 stsp return err;
639 fae7e038 2022-05-07 stsp int done = 0;
641 fae7e038 2022-05-07 stsp if (cancel_cb) {
642 fae7e038 2022-05-07 stsp err = (*cancel_cb)(cancel_arg);
647 fae7e038 2022-05-07 stsp err = got_privsep_recv_reused_deltas(&done, deltas, &ndeltas,
648 fae7e038 2022-05-07 stsp pack->privsep_child->ibuf);
649 fae7e038 2022-05-07 stsp if (err || done)
652 fae7e038 2022-05-07 stsp for (i = 0; i < ndeltas; i++) {
653 fae7e038 2022-05-07 stsp struct got_imsg_reused_delta *delta = &deltas[i];
654 fae7e038 2022-05-07 stsp err = recv_reused_delta(delta, idset, v);
659 fae7e038 2022-05-07 stsp err = report_progress(progress_cb, progress_arg, rl,
660 fae7e038 2022-05-07 stsp ncolored, nfound, ntrees, 0L, ncommits,
661 fae7e038 2022-05-07 stsp got_object_idset_num_elements(idset), v->nmeta, 0);
666 67fd6849 2022-02-13 stsp return err;
669 67fd6849 2022-02-13 stsp static const struct got_error *
670 b8af7c06 2022-03-15 stsp pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
671 b8af7c06 2022-03-15 stsp int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
672 b8af7c06 2022-03-15 stsp struct got_repository *repo,
673 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
674 211cfef0 2022-01-05 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
676 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
677 e6bcace5 2021-06-22 stsp struct got_pack_meta *m = NULL, *base = NULL;
678 e6bcace5 2021-06-22 stsp struct got_raw_object *raw = NULL, *base_raw = NULL;
679 74881701 2021-10-15 stsp struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
680 5060d5a1 2022-01-10 stsp int i, j, ndeltas, best_ndeltas;
681 5060d5a1 2022-01-10 stsp off_t size, best_size;
682 4f4d853e 2021-10-24 stsp const int max_base_candidates = 3;
683 402a5ec1 2022-01-10 stsp size_t delta_memsize = 0;
684 402a5ec1 2022-01-10 stsp const size_t max_delta_memsize = 25 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
685 cc7a354a 2021-10-15 stsp int outfd = -1;
687 e6bcace5 2021-06-22 stsp qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
688 e6bcace5 2021-06-22 stsp for (i = 0; i < nmeta; i++) {
689 e6bcace5 2021-06-22 stsp if (cancel_cb) {
690 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
694 211cfef0 2022-01-05 stsp err = report_progress(progress_cb, progress_arg, rl,
695 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
696 b8af7c06 2022-03-15 stsp nreused + i, 0);
699 e6bcace5 2021-06-22 stsp m = meta[i];
701 e6bcace5 2021-06-22 stsp if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
702 e6bcace5 2021-06-22 stsp m->obj_type == GOT_OBJ_TYPE_TAG)
705 94dac27c 2021-10-15 stsp err = got_object_raw_open(&raw, &outfd, repo, &m->id);
708 600b755e 2021-10-14 stsp m->size = raw->size;
710 64a8571e 2022-01-07 stsp if (raw->f == NULL) {
711 64a8571e 2022-01-07 stsp err = got_deltify_init_mem(&m->dtab, raw->data,
712 64a8571e 2022-01-07 stsp raw->hdrlen, raw->size + raw->hdrlen);
714 64a8571e 2022-01-07 stsp err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
715 64a8571e 2022-01-07 stsp raw->size + raw->hdrlen);
720 e6bcace5 2021-06-22 stsp if (i > max_base_candidates) {
721 e6bcace5 2021-06-22 stsp struct got_pack_meta *n = NULL;
722 e6bcace5 2021-06-22 stsp n = meta[i - (max_base_candidates + 1)];
723 e6bcace5 2021-06-22 stsp got_deltify_free(n->dtab);
724 e6bcace5 2021-06-22 stsp n->dtab = NULL;
727 74881701 2021-10-15 stsp best_size = raw->size;
728 74881701 2021-10-15 stsp best_ndeltas = 0;
729 e6bcace5 2021-06-22 stsp for (j = MAX(0, i - max_base_candidates); j < i; j++) {
730 e6bcace5 2021-06-22 stsp if (cancel_cb) {
731 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
735 e6bcace5 2021-06-22 stsp base = meta[j];
736 e6bcace5 2021-06-22 stsp /* long chains make unpacking slow, avoid such bases */
737 22edbce7 2021-10-24 stsp if (base->nchain >= 128 ||
738 e6bcace5 2021-06-22 stsp base->obj_type != m->obj_type)
741 d3c116bf 2021-10-15 stsp err = got_object_raw_open(&base_raw, &outfd, repo,
742 94dac27c 2021-10-15 stsp &base->id);
746 64a8571e 2022-01-07 stsp if (raw->f == NULL && base_raw->f == NULL) {
747 64a8571e 2022-01-07 stsp err = got_deltify_mem_mem(&deltas, &ndeltas,
748 64a8571e 2022-01-07 stsp raw->data, raw->hdrlen,
749 64a8571e 2022-01-07 stsp raw->size + raw->hdrlen,
750 64a8571e 2022-01-07 stsp base->dtab, base_raw->data,
751 64a8571e 2022-01-07 stsp base_raw->hdrlen,
752 64a8571e 2022-01-07 stsp base_raw->size + base_raw->hdrlen);
753 64a8571e 2022-01-07 stsp } else if (raw->f == NULL) {
754 64a8571e 2022-01-07 stsp err = got_deltify_mem_file(&deltas, &ndeltas,
755 64a8571e 2022-01-07 stsp raw->data, raw->hdrlen,
756 64a8571e 2022-01-07 stsp raw->size + raw->hdrlen,
757 64a8571e 2022-01-07 stsp base->dtab, base_raw->f,
758 64a8571e 2022-01-07 stsp base_raw->hdrlen,
759 64a8571e 2022-01-07 stsp base_raw->size + base_raw->hdrlen);
760 64a8571e 2022-01-07 stsp } else if (base_raw->f == NULL) {
761 64a8571e 2022-01-07 stsp err = got_deltify_file_mem(&deltas, &ndeltas,
762 64a8571e 2022-01-07 stsp raw->f, raw->hdrlen,
763 64a8571e 2022-01-07 stsp raw->size + raw->hdrlen,
764 64a8571e 2022-01-07 stsp base->dtab, base_raw->data,
765 64a8571e 2022-01-07 stsp base_raw->hdrlen,
766 64a8571e 2022-01-07 stsp base_raw->size + base_raw->hdrlen);
768 64a8571e 2022-01-07 stsp err = got_deltify(&deltas, &ndeltas,
769 64a8571e 2022-01-07 stsp raw->f, raw->hdrlen,
770 64a8571e 2022-01-07 stsp raw->size + raw->hdrlen,
771 64a8571e 2022-01-07 stsp base->dtab, base_raw->f, base_raw->hdrlen,
772 64a8571e 2022-01-07 stsp base_raw->size + base_raw->hdrlen);
774 e6bcace5 2021-06-22 stsp got_object_raw_close(base_raw);
775 e6bcace5 2021-06-22 stsp base_raw = NULL;
779 e6bcace5 2021-06-22 stsp size = delta_size(deltas, ndeltas);
780 74881701 2021-10-15 stsp if (size + 32 < best_size){
782 e6bcace5 2021-06-22 stsp * if we already picked a best delta,
783 e6bcace5 2021-06-22 stsp * replace it.
785 74881701 2021-10-15 stsp best_size = size;
786 74881701 2021-10-15 stsp free(best_deltas);
787 74881701 2021-10-15 stsp best_deltas = deltas;
788 74881701 2021-10-15 stsp best_ndeltas = ndeltas;
789 74881701 2021-10-15 stsp deltas = NULL;
790 e6bcace5 2021-06-22 stsp m->nchain = base->nchain + 1;
791 e6bcace5 2021-06-22 stsp m->prev = base;
792 e6bcace5 2021-06-22 stsp m->head = base->head;
793 e6bcace5 2021-06-22 stsp if (m->head == NULL)
794 e6bcace5 2021-06-22 stsp m->head = base;
796 e6bcace5 2021-06-22 stsp free(deltas);
797 e6bcace5 2021-06-22 stsp deltas = NULL;
798 e6bcace5 2021-06-22 stsp ndeltas = 0;
802 74881701 2021-10-15 stsp if (best_ndeltas > 0) {
803 402a5ec1 2022-01-10 stsp if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
804 402a5ec1 2022-01-10 stsp delta_memsize + best_size <= max_delta_memsize) {
805 402a5ec1 2022-01-10 stsp delta_memsize += best_size;
806 5060d5a1 2022-01-10 stsp err = encode_delta_in_mem(m, raw, best_deltas,
807 5060d5a1 2022-01-10 stsp best_ndeltas, best_size, m->prev->size);
809 5060d5a1 2022-01-10 stsp m->delta_offset = ftello(delta_cache);
810 5060d5a1 2022-01-10 stsp err = encode_delta(m, raw, best_deltas,
811 5060d5a1 2022-01-10 stsp best_ndeltas, m->prev->size, delta_cache);
813 74881701 2021-10-15 stsp free(best_deltas);
814 74881701 2021-10-15 stsp best_deltas = NULL;
815 74881701 2021-10-15 stsp best_ndeltas = 0;
820 e6bcace5 2021-06-22 stsp got_object_raw_close(raw);
821 e6bcace5 2021-06-22 stsp raw = NULL;
824 e6bcace5 2021-06-22 stsp for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
825 e6bcace5 2021-06-22 stsp got_deltify_free(meta[i]->dtab);
826 e6bcace5 2021-06-22 stsp meta[i]->dtab = NULL;
829 e6bcace5 2021-06-22 stsp got_object_raw_close(raw);
830 e6bcace5 2021-06-22 stsp if (base_raw)
831 e6bcace5 2021-06-22 stsp got_object_raw_close(base_raw);
832 cc7a354a 2021-10-15 stsp if (outfd != -1 && close(outfd) == -1 && err == NULL)
833 cc7a354a 2021-10-15 stsp err = got_error_from_errno("close");
834 74881701 2021-10-15 stsp free(deltas);
835 74881701 2021-10-15 stsp free(best_deltas);
836 e6bcace5 2021-06-22 stsp return err;
839 e6bcace5 2021-06-22 stsp static const struct got_error *
840 05118f5a 2021-06-22 stsp search_packidx(int *found, struct got_object_id *id,
841 05118f5a 2021-06-22 stsp struct got_repository *repo)
843 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
844 05118f5a 2021-06-22 stsp struct got_packidx *packidx = NULL;
847 05118f5a 2021-06-22 stsp *found = 0;
849 05118f5a 2021-06-22 stsp err = got_repo_search_packidx(&packidx, &idx, repo, id);
850 05118f5a 2021-06-22 stsp if (err == NULL)
851 05118f5a 2021-06-22 stsp *found = 1; /* object is already packed */
852 05118f5a 2021-06-22 stsp else if (err->code == GOT_ERR_NO_OBJ)
853 05118f5a 2021-06-22 stsp err = NULL;
854 05118f5a 2021-06-22 stsp return err;
857 05118f5a 2021-06-22 stsp static const struct got_error *
858 67fd6849 2022-02-13 stsp add_object(int want_meta, struct got_object_idset *idset,
859 e6bcace5 2021-06-22 stsp struct got_object_id *id, const char *path, int obj_type,
860 6863cbf9 2022-03-21 stsp time_t mtime, int loose_obj_only, struct got_repository *repo,
861 6863cbf9 2022-03-21 stsp int *ncolored, int *nfound, int *ntrees,
862 6863cbf9 2022-03-21 stsp got_pack_progress_cb progress_cb, void *progress_arg,
863 6863cbf9 2022-03-21 stsp struct got_ratelimit *rl)
865 e6bcace5 2021-06-22 stsp const struct got_error *err;
866 67fd6849 2022-02-13 stsp struct got_pack_meta *m = NULL;
868 05118f5a 2021-06-22 stsp if (loose_obj_only) {
869 05118f5a 2021-06-22 stsp int is_packed;
870 05118f5a 2021-06-22 stsp err = search_packidx(&is_packed, id, repo);
872 05118f5a 2021-06-22 stsp return err;
873 cdeb891a 2022-03-21 stsp if (is_packed && want_meta)
874 05118f5a 2021-06-22 stsp return NULL;
877 67fd6849 2022-02-13 stsp if (want_meta) {
878 67fd6849 2022-02-13 stsp err = alloc_meta(&m, id, path, obj_type, mtime);
880 6863cbf9 2022-03-21 stsp return err;
882 6863cbf9 2022-03-21 stsp (*nfound)++;
883 6863cbf9 2022-03-21 stsp err = report_progress(progress_cb, progress_arg, rl,
884 6863cbf9 2022-03-21 stsp *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
886 bb6672b6 2022-04-14 tb clear_meta(m);
888 67fd6849 2022-02-13 stsp return err;
892 bb6672b6 2022-04-14 tb err = got_object_idset_add(idset, id, m);
894 bb6672b6 2022-04-14 tb clear_meta(m);
900 e6bcace5 2021-06-22 stsp static const struct got_error *
901 67fd6849 2022-02-13 stsp load_tree_entries(struct got_object_id_queue *ids, int want_meta,
902 2f8438b0 2022-05-04 stsp struct got_object_idset *idset, struct got_object_idset *idset_exclude,
903 2f8438b0 2022-05-04 stsp struct got_object_id *tree_id,
904 e6bcace5 2021-06-22 stsp const char *dpath, time_t mtime, struct got_repository *repo,
905 b8af7c06 2022-03-15 stsp int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
906 b8af7c06 2022-03-15 stsp got_pack_progress_cb progress_cb, void *progress_arg,
907 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
909 e6bcace5 2021-06-22 stsp const struct got_error *err;
910 e6bcace5 2021-06-22 stsp struct got_tree_object *tree;
911 e6bcace5 2021-06-22 stsp char *p = NULL;
914 e6bcace5 2021-06-22 stsp err = got_object_open_as_tree(&tree, repo, tree_id);
916 b8af7c06 2022-03-15 stsp return err;
918 b8af7c06 2022-03-15 stsp (*ntrees)++;
919 b8af7c06 2022-03-15 stsp err = report_progress(progress_cb, progress_arg, rl,
920 b8af7c06 2022-03-15 stsp *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
922 e6bcace5 2021-06-22 stsp return err;
924 e6bcace5 2021-06-22 stsp for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
925 e6bcace5 2021-06-22 stsp struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
926 e6bcace5 2021-06-22 stsp struct got_object_id *id = got_tree_entry_get_id(e);
927 e6bcace5 2021-06-22 stsp mode_t mode = got_tree_entry_get_mode(e);
929 e6bcace5 2021-06-22 stsp if (cancel_cb) {
930 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
935 3af9de88 2021-09-22 stsp if (got_object_tree_entry_is_submodule(e) ||
936 2f8438b0 2022-05-04 stsp got_object_idset_contains(idset, id) ||
937 2f8438b0 2022-05-04 stsp got_object_idset_contains(idset_exclude, id))
940 e6bcace5 2021-06-22 stsp if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
941 e6bcace5 2021-06-22 stsp got_tree_entry_get_name(e)) == -1) {
942 e6bcace5 2021-06-22 stsp err = got_error_from_errno("asprintf");
946 e6bcace5 2021-06-22 stsp if (S_ISDIR(mode)) {
947 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
948 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, id);
951 3e6ceea0 2022-05-20 stsp qid->data = p;
953 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(ids, qid, entry);
954 3af9de88 2021-09-22 stsp } else if (S_ISREG(mode) || S_ISLNK(mode)) {
955 2f8438b0 2022-05-04 stsp err = add_object(want_meta,
956 2f8438b0 2022-05-04 stsp want_meta ? idset : idset_exclude, id, p,
957 6863cbf9 2022-03-21 stsp GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo,
958 6863cbf9 2022-03-21 stsp ncolored, nfound, ntrees,
959 6863cbf9 2022-03-21 stsp progress_cb, progress_arg, rl);
970 e6bcace5 2021-06-22 stsp got_object_tree_close(tree);
972 e6bcace5 2021-06-22 stsp return err;
975 e6bcace5 2021-06-22 stsp static const struct got_error *
976 67fd6849 2022-02-13 stsp load_tree(int want_meta, struct got_object_idset *idset,
977 2f8438b0 2022-05-04 stsp struct got_object_idset *idset_exclude,
978 e6bcace5 2021-06-22 stsp struct got_object_id *tree_id, const char *dpath, time_t mtime,
979 b8af7c06 2022-03-15 stsp struct got_repository *repo, int loose_obj_only,
980 b8af7c06 2022-03-15 stsp int *ncolored, int *nfound, int *ntrees,
981 b8af7c06 2022-03-15 stsp got_pack_progress_cb progress_cb, void *progress_arg,
982 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
984 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
985 e6bcace5 2021-06-22 stsp struct got_object_id_queue tree_ids;
986 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
988 2f8438b0 2022-05-04 stsp if (got_object_idset_contains(idset, tree_id) ||
989 2f8438b0 2022-05-04 stsp got_object_idset_contains(idset_exclude, tree_id))
990 e6bcace5 2021-06-22 stsp return NULL;
992 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, tree_id);
994 3e6ceea0 2022-05-20 stsp return err;
995 3e6ceea0 2022-05-20 stsp qid->data = strdup(dpath);
996 3e6ceea0 2022-05-20 stsp if (qid->data == NULL) {
997 3e6ceea0 2022-05-20 stsp err = got_error_from_errno("strdup");
998 3e6ceea0 2022-05-20 stsp got_object_qid_free(qid);
999 e6bcace5 2021-06-22 stsp return err;
1002 dbdddfee 2021-06-23 naddy STAILQ_INIT(&tree_ids);
1003 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
1005 dbdddfee 2021-06-23 naddy while (!STAILQ_EMPTY(&tree_ids)) {
1006 3e6ceea0 2022-05-20 stsp const char *path;
1007 e6bcace5 2021-06-22 stsp if (cancel_cb) {
1008 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
1013 dbdddfee 2021-06-23 naddy qid = STAILQ_FIRST(&tree_ids);
1014 dbdddfee 2021-06-23 naddy STAILQ_REMOVE_HEAD(&tree_ids, entry);
1015 3e6ceea0 2022-05-20 stsp path = qid->data;
1017 2f8438b0 2022-05-04 stsp if (got_object_idset_contains(idset, &qid->id) ||
1018 2f8438b0 2022-05-04 stsp got_object_idset_contains(idset_exclude, &qid->id)) {
1019 3e6ceea0 2022-05-20 stsp free(qid->data);
1020 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1024 2f8438b0 2022-05-04 stsp err = add_object(want_meta, want_meta ? idset : idset_exclude,
1025 3e6ceea0 2022-05-20 stsp &qid->id, path, GOT_OBJ_TYPE_TREE,
1026 2f8438b0 2022-05-04 stsp mtime, loose_obj_only, repo,
1027 6863cbf9 2022-03-21 stsp ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1028 e6bcace5 2021-06-22 stsp if (err) {
1029 3e6ceea0 2022-05-20 stsp free(qid->data);
1030 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1034 2f8438b0 2022-05-04 stsp err = load_tree_entries(&tree_ids, want_meta, idset,
1035 2f8438b0 2022-05-04 stsp idset_exclude, &qid->id,
1036 3e6ceea0 2022-05-20 stsp path, mtime, repo, loose_obj_only, ncolored, nfound,
1037 b8af7c06 2022-03-15 stsp ntrees, progress_cb, progress_arg, rl,
1038 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1039 3e6ceea0 2022-05-20 stsp free(qid->data);
1040 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1045 3e6ceea0 2022-05-20 stsp STAILQ_FOREACH(qid, &tree_ids, entry)
1046 3e6ceea0 2022-05-20 stsp free(qid->data);
1047 e6bcace5 2021-06-22 stsp got_object_id_queue_free(&tree_ids);
1048 e6bcace5 2021-06-22 stsp return err;
1051 e6bcace5 2021-06-22 stsp static const struct got_error *
1052 67fd6849 2022-02-13 stsp load_commit(int want_meta, struct got_object_idset *idset,
1053 2f8438b0 2022-05-04 stsp struct got_object_idset *idset_exclude,
1054 05118f5a 2021-06-22 stsp struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1055 b8af7c06 2022-03-15 stsp int *ncolored, int *nfound, int *ntrees,
1056 b8af7c06 2022-03-15 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1057 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1059 e6bcace5 2021-06-22 stsp const struct got_error *err;
1060 e6bcace5 2021-06-22 stsp struct got_commit_object *commit;
1062 2f8438b0 2022-05-04 stsp if (got_object_idset_contains(idset, id) ||
1063 2f8438b0 2022-05-04 stsp got_object_idset_contains(idset_exclude, id))
1064 e6bcace5 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 cdeb891a 2022-03-21 stsp if (is_packed && want_meta)
1072 05118f5a 2021-06-22 stsp return NULL;
1075 e6bcace5 2021-06-22 stsp err = got_object_open_as_commit(&commit, repo, id);
1077 e6bcace5 2021-06-22 stsp return err;
1079 2f8438b0 2022-05-04 stsp err = add_object(want_meta, want_meta ? idset : idset_exclude,
1080 2f8438b0 2022-05-04 stsp id, "", GOT_OBJ_TYPE_COMMIT,
1081 05118f5a 2021-06-22 stsp got_object_commit_get_committer_time(commit),
1082 6863cbf9 2022-03-21 stsp loose_obj_only, repo,
1083 6863cbf9 2022-03-21 stsp ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1085 e6bcace5 2021-06-22 stsp goto done;
1087 2f8438b0 2022-05-04 stsp err = load_tree(want_meta, idset, idset_exclude,
1088 2f8438b0 2022-05-04 stsp got_object_commit_get_tree_id(commit),
1089 05118f5a 2021-06-22 stsp "", got_object_commit_get_committer_time(commit),
1090 b8af7c06 2022-03-15 stsp repo, loose_obj_only, ncolored, nfound, ntrees,
1091 b8af7c06 2022-03-15 stsp progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1093 e6bcace5 2021-06-22 stsp got_object_commit_close(commit);
1094 e6bcace5 2021-06-22 stsp return err;
1097 05118f5a 2021-06-22 stsp static const struct got_error *
1098 67fd6849 2022-02-13 stsp load_tag(int want_meta, struct got_object_idset *idset,
1099 2f8438b0 2022-05-04 stsp struct got_object_idset *idset_exclude,
1100 05118f5a 2021-06-22 stsp struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1101 b8af7c06 2022-03-15 stsp int *ncolored, int *nfound, int *ntrees,
1102 b8af7c06 2022-03-15 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1103 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1105 05118f5a 2021-06-22 stsp const struct got_error *err;
1106 05118f5a 2021-06-22 stsp struct got_tag_object *tag = NULL;
1108 2f8438b0 2022-05-04 stsp if (got_object_idset_contains(idset, id) ||
1109 2f8438b0 2022-05-04 stsp got_object_idset_contains(idset_exclude, id))
1110 05118f5a 2021-06-22 stsp return NULL;
1112 05118f5a 2021-06-22 stsp if (loose_obj_only) {
1113 05118f5a 2021-06-22 stsp int is_packed;
1114 05118f5a 2021-06-22 stsp err = search_packidx(&is_packed, id, repo);
1116 05118f5a 2021-06-22 stsp return err;
1117 cdeb891a 2022-03-21 stsp if (is_packed && want_meta)
1118 05118f5a 2021-06-22 stsp return NULL;
1121 05118f5a 2021-06-22 stsp err = got_object_open_as_tag(&tag, repo, id);
1123 05118f5a 2021-06-22 stsp return err;
1125 2f8438b0 2022-05-04 stsp err = add_object(want_meta, want_meta ? idset : idset_exclude,
1126 2f8438b0 2022-05-04 stsp id, "", GOT_OBJ_TYPE_TAG,
1127 6863cbf9 2022-03-21 stsp got_object_tag_get_tagger_time(tag), loose_obj_only, repo,
1128 6863cbf9 2022-03-21 stsp ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1130 05118f5a 2021-06-22 stsp goto done;
1132 05118f5a 2021-06-22 stsp switch (got_object_tag_get_object_type(tag)) {
1133 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_COMMIT:
1134 2f8438b0 2022-05-04 stsp err = load_commit(want_meta, idset, idset_exclude,
1135 b8af7c06 2022-03-15 stsp got_object_tag_get_object_id(tag), repo, loose_obj_only,
1136 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1137 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1139 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_TREE:
1140 2f8438b0 2022-05-04 stsp err = load_tree(want_meta, idset, idset_exclude,
1141 67fd6849 2022-02-13 stsp got_object_tag_get_object_id(tag), "",
1142 b8af7c06 2022-03-15 stsp got_object_tag_get_tagger_time(tag), repo, loose_obj_only,
1143 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1144 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1151 05118f5a 2021-06-22 stsp got_object_tag_close(tag);
1152 05118f5a 2021-06-22 stsp return err;
1155 e6bcace5 2021-06-22 stsp enum findtwixt_color {
1156 e6bcace5 2021-06-22 stsp COLOR_KEEP = 0,
1157 e6bcace5 2021-06-22 stsp COLOR_DROP,
1158 e6bcace5 2021-06-22 stsp COLOR_BLANK,
1159 70f8f24d 2022-04-14 stsp COLOR_SKIP,
1162 e6bcace5 2021-06-22 stsp static const int findtwixt_colors[] = {
1163 e6bcace5 2021-06-22 stsp COLOR_KEEP,
1164 e6bcace5 2021-06-22 stsp COLOR_DROP,
1165 70f8f24d 2022-04-14 stsp COLOR_BLANK,
1166 70f8f24d 2022-04-14 stsp COLOR_SKIP,
1169 e6bcace5 2021-06-22 stsp static const struct got_error *
1170 70f8f24d 2022-04-14 stsp paint_commit(struct got_object_qid *qid, int color)
1172 70f8f24d 2022-04-14 stsp if (color < 0 || color >= nitems(findtwixt_colors))
1173 70f8f24d 2022-04-14 stsp return got_error(GOT_ERR_RANGE);
1175 e6bcace5 2021-06-22 stsp qid->data = (void *)&findtwixt_colors[color];
1176 e6bcace5 2021-06-22 stsp return NULL;
1179 e6bcace5 2021-06-22 stsp static const struct got_error *
1180 70f8f24d 2022-04-14 stsp queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1181 70f8f24d 2022-04-14 stsp int color, struct got_repository *repo)
1183 70f8f24d 2022-04-14 stsp const struct got_error *err;
1184 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
1186 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, id);
1188 e6bcace5 2021-06-22 stsp return err;
1190 70f8f24d 2022-04-14 stsp STAILQ_INSERT_TAIL(ids, qid, entry);
1191 70f8f24d 2022-04-14 stsp return paint_commit(qid, color);
1194 e6bcace5 2021-06-22 stsp struct append_id_arg {
1195 e6bcace5 2021-06-22 stsp struct got_object_id **array;
1197 70f8f24d 2022-04-14 stsp struct got_object_idset *drop;
1198 70f8f24d 2022-04-14 stsp struct got_object_idset *skip;
1201 e6bcace5 2021-06-22 stsp static const struct got_error *
1202 e6bcace5 2021-06-22 stsp append_id(struct got_object_id *id, void *data, void *arg)
1204 e6bcace5 2021-06-22 stsp struct append_id_arg *a = arg;
1206 70f8f24d 2022-04-14 stsp if (got_object_idset_contains(a->skip, id) ||
1207 70f8f24d 2022-04-14 stsp got_object_idset_contains(a->drop, id))
1208 70f8f24d 2022-04-14 stsp return NULL;
1210 70f8f24d 2022-04-14 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");
1214 e6bcace5 2021-06-22 stsp return NULL;
1217 29e0594f 2022-04-09 stsp static const struct got_error *
1218 29e0594f 2022-04-09 stsp queue_commit_or_tag_id(struct got_object_id *id, int color,
1219 29e0594f 2022-04-09 stsp struct got_object_id_queue *ids, struct got_repository *repo)
1221 29e0594f 2022-04-09 stsp const struct got_error *err;
1222 29e0594f 2022-04-09 stsp struct got_tag_object *tag = NULL;
1223 29e0594f 2022-04-09 stsp int obj_type;
1225 29e0594f 2022-04-09 stsp err = got_object_get_type(&obj_type, repo, id);
1227 29e0594f 2022-04-09 stsp return err;
1229 29e0594f 2022-04-09 stsp if (obj_type == GOT_OBJ_TYPE_TAG) {
1230 29e0594f 2022-04-09 stsp err = got_object_open_as_tag(&tag, repo, id);
1232 29e0594f 2022-04-09 stsp return err;
1233 29e0594f 2022-04-09 stsp obj_type = got_object_tag_get_object_type(tag);
1234 29e0594f 2022-04-09 stsp id = got_object_tag_get_object_id(tag);
1237 29e0594f 2022-04-09 stsp if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1238 29e0594f 2022-04-09 stsp err = queue_commit_id(ids, id, color, repo);
1240 29e0594f 2022-04-09 stsp goto done;
1244 29e0594f 2022-04-09 stsp got_object_tag_close(tag);
1245 29e0594f 2022-04-09 stsp return err;
1248 e6bcace5 2021-06-22 stsp static const struct got_error *
1249 70f8f24d 2022-04-14 stsp paint_commits(int *ncolored, struct got_object_id_queue *ids, int nids,
1250 14dbbf48 2022-04-10 stsp struct got_object_idset *keep, struct got_object_idset *drop,
1251 70f8f24d 2022-04-14 stsp struct got_object_idset *skip, struct got_repository *repo,
1252 b8af7c06 2022-03-15 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1253 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1255 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1256 14dbbf48 2022-04-10 stsp struct got_commit_object *commit = NULL;
1257 70f8f24d 2022-04-14 stsp const struct got_object_id_queue *parents;
1258 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
1259 70f8f24d 2022-04-14 stsp int nqueued = nids, nskip = 0;
1261 70f8f24d 2022-04-14 stsp while (!STAILQ_EMPTY(ids) && nskip != nqueued) {
1262 70f8f24d 2022-04-14 stsp int color;
1264 57bc7b6d 2022-04-10 stsp if (cancel_cb) {
1265 57bc7b6d 2022-04-10 stsp err = cancel_cb(cancel_arg);
1270 14dbbf48 2022-04-10 stsp qid = STAILQ_FIRST(ids);
1271 70f8f24d 2022-04-14 stsp STAILQ_REMOVE_HEAD(ids, entry);
1272 70f8f24d 2022-04-14 stsp nqueued--;
1273 70f8f24d 2022-04-14 stsp color = *((int *)qid->data);
1274 70f8f24d 2022-04-14 stsp if (color == COLOR_SKIP)
1277 d7b5a0e8 2022-04-20 stsp if (got_object_idset_contains(skip, &qid->id)) {
1278 70f8f24d 2022-04-14 stsp got_object_qid_free(qid);
1282 70f8f24d 2022-04-14 stsp switch (color) {
1283 70f8f24d 2022-04-14 stsp case COLOR_KEEP:
1284 d7b5a0e8 2022-04-20 stsp if (got_object_idset_contains(keep, &qid->id)) {
1285 70f8f24d 2022-04-14 stsp got_object_qid_free(qid);
1288 d7b5a0e8 2022-04-20 stsp if (got_object_idset_contains(drop, &qid->id)) {
1289 70f8f24d 2022-04-14 stsp err = paint_commit(qid, COLOR_SKIP);
1291 70f8f24d 2022-04-14 stsp goto done;
1294 70f8f24d 2022-04-14 stsp (*ncolored)++;
1295 d7b5a0e8 2022-04-20 stsp err = got_object_idset_add(keep, &qid->id, NULL);
1297 70f8f24d 2022-04-14 stsp goto done;
1299 70f8f24d 2022-04-14 stsp case COLOR_DROP:
1300 d7b5a0e8 2022-04-20 stsp if (got_object_idset_contains(drop, &qid->id)) {
1301 70f8f24d 2022-04-14 stsp got_object_qid_free(qid);
1304 d7b5a0e8 2022-04-20 stsp if (got_object_idset_contains(keep, &qid->id)) {
1305 70f8f24d 2022-04-14 stsp err = paint_commit(qid, COLOR_SKIP);
1307 70f8f24d 2022-04-14 stsp goto done;
1310 70f8f24d 2022-04-14 stsp (*ncolored)++;
1311 d7b5a0e8 2022-04-20 stsp err = got_object_idset_add(drop, &qid->id, NULL);
1313 70f8f24d 2022-04-14 stsp goto done;
1315 70f8f24d 2022-04-14 stsp case COLOR_SKIP:
1316 d7b5a0e8 2022-04-20 stsp if (!got_object_idset_contains(skip, &qid->id)) {
1317 d7b5a0e8 2022-04-20 stsp err = got_object_idset_add(skip, &qid->id,
1320 70f8f24d 2022-04-14 stsp goto done;
1324 70f8f24d 2022-04-14 stsp /* should not happen */
1325 70f8f24d 2022-04-14 stsp err = got_error_fmt(GOT_ERR_NOT_IMPL,
1326 70f8f24d 2022-04-14 stsp "%s invalid commit color %d", __func__, color);
1327 70f8f24d 2022-04-14 stsp goto done;
1330 b8af7c06 2022-03-15 stsp err = report_progress(progress_cb, progress_arg, rl,
1331 b8af7c06 2022-03-15 stsp *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1336 d7b5a0e8 2022-04-20 stsp err = got_object_open_as_commit(&commit, repo, &qid->id);
1340 70f8f24d 2022-04-14 stsp parents = got_object_commit_get_parent_ids(commit);
1341 70f8f24d 2022-04-14 stsp if (parents) {
1342 70f8f24d 2022-04-14 stsp struct got_object_qid *pid;
1343 70f8f24d 2022-04-14 stsp color = *((int *)qid->data);
1344 70f8f24d 2022-04-14 stsp STAILQ_FOREACH(pid, parents, entry) {
1345 d7b5a0e8 2022-04-20 stsp err = queue_commit_id(ids, &pid->id, color,
1349 70f8f24d 2022-04-14 stsp nqueued++;
1350 70f8f24d 2022-04-14 stsp if (color == COLOR_SKIP)
1355 70f8f24d 2022-04-14 stsp got_object_commit_close(commit);
1356 70f8f24d 2022-04-14 stsp commit = NULL;
1357 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1360 14dbbf48 2022-04-10 stsp if (commit)
1361 14dbbf48 2022-04-10 stsp got_object_commit_close(commit);
1362 14dbbf48 2022-04-10 stsp return err;
1365 14dbbf48 2022-04-10 stsp static const struct got_error *
1366 14dbbf48 2022-04-10 stsp findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1367 14dbbf48 2022-04-10 stsp struct got_object_id **head, int nhead,
1368 14dbbf48 2022-04-10 stsp struct got_object_id **tail, int ntail,
1369 14dbbf48 2022-04-10 stsp struct got_repository *repo,
1370 14dbbf48 2022-04-10 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1371 14dbbf48 2022-04-10 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1373 14dbbf48 2022-04-10 stsp const struct got_error *err = NULL;
1374 14dbbf48 2022-04-10 stsp struct got_object_id_queue ids;
1375 70f8f24d 2022-04-14 stsp struct got_object_idset *keep, *drop, *skip = NULL;
1376 14dbbf48 2022-04-10 stsp int i, nkeep;
1378 14dbbf48 2022-04-10 stsp STAILQ_INIT(&ids);
1379 14dbbf48 2022-04-10 stsp *res = NULL;
1380 14dbbf48 2022-04-10 stsp *nres = 0;
1381 14dbbf48 2022-04-10 stsp *ncolored = 0;
1383 14dbbf48 2022-04-10 stsp keep = got_object_idset_alloc();
1384 14dbbf48 2022-04-10 stsp if (keep == NULL)
1385 14dbbf48 2022-04-10 stsp return got_error_from_errno("got_object_idset_alloc");
1387 14dbbf48 2022-04-10 stsp drop = got_object_idset_alloc();
1388 14dbbf48 2022-04-10 stsp if (drop == NULL) {
1389 14dbbf48 2022-04-10 stsp err = got_error_from_errno("got_object_idset_alloc");
1390 14dbbf48 2022-04-10 stsp goto done;
1393 70f8f24d 2022-04-14 stsp skip = got_object_idset_alloc();
1394 70f8f24d 2022-04-14 stsp if (skip == NULL) {
1395 70f8f24d 2022-04-14 stsp err = got_error_from_errno("got_object_idset_alloc");
1396 70f8f24d 2022-04-14 stsp goto done;
1399 14dbbf48 2022-04-10 stsp for (i = 0; i < nhead; i++) {
1400 14dbbf48 2022-04-10 stsp struct got_object_id *id = head[i];
1401 14dbbf48 2022-04-10 stsp if (id == NULL)
1403 fbafdecf 2022-04-10 stsp err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1405 14dbbf48 2022-04-10 stsp goto done;
1408 14dbbf48 2022-04-10 stsp for (i = 0; i < ntail; i++) {
1409 14dbbf48 2022-04-10 stsp struct got_object_id *id = tail[i];
1410 14dbbf48 2022-04-10 stsp if (id == NULL)
1412 14dbbf48 2022-04-10 stsp err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1414 14dbbf48 2022-04-10 stsp goto done;
1417 70f8f24d 2022-04-14 stsp err = paint_commits(ncolored, &ids, nhead + ntail,
1418 70f8f24d 2022-04-14 stsp keep, drop, skip, repo, progress_cb, progress_arg, rl,
1419 70f8f24d 2022-04-14 stsp cancel_cb, cancel_arg);
1421 14dbbf48 2022-04-10 stsp goto done;
1423 e6bcace5 2021-06-22 stsp nkeep = got_object_idset_num_elements(keep);
1424 e6bcace5 2021-06-22 stsp if (nkeep > 0) {
1425 e6bcace5 2021-06-22 stsp struct append_id_arg arg;
1426 e6bcace5 2021-06-22 stsp arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1427 e6bcace5 2021-06-22 stsp if (arg.array == NULL) {
1428 e6bcace5 2021-06-22 stsp err = got_error_from_errno("calloc");
1429 e6bcace5 2021-06-22 stsp goto done;
1431 70f8f24d 2022-04-14 stsp arg.idx = -1;
1432 70f8f24d 2022-04-14 stsp arg.skip = skip;
1433 70f8f24d 2022-04-14 stsp arg.drop = drop;
1434 e6bcace5 2021-06-22 stsp err = got_object_idset_for_each(keep, append_id, &arg);
1435 e6bcace5 2021-06-22 stsp if (err) {
1436 e6bcace5 2021-06-22 stsp free(arg.array);
1437 e6bcace5 2021-06-22 stsp goto done;
1439 e6bcace5 2021-06-22 stsp *res = arg.array;
1440 70f8f24d 2022-04-14 stsp *nres = arg.idx + 1;
1443 e6bcace5 2021-06-22 stsp got_object_idset_free(keep);
1444 e6bcace5 2021-06-22 stsp got_object_idset_free(drop);
1446 70f8f24d 2022-04-14 stsp got_object_idset_free(skip);
1447 e6bcace5 2021-06-22 stsp got_object_id_queue_free(&ids);
1448 e6bcace5 2021-06-22 stsp return err;
1451 e6bcace5 2021-06-22 stsp static const struct got_error *
1452 b8af7c06 2022-03-15 stsp load_object_ids(int *ncolored, int *nfound, int *ntrees,
1453 b8af7c06 2022-03-15 stsp struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1454 e6bcace5 2021-06-22 stsp struct got_object_id **ours, int nours, struct got_repository *repo,
1455 b8af7c06 2022-03-15 stsp int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
1456 b8af7c06 2022-03-15 stsp struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1458 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1459 e6bcace5 2021-06-22 stsp struct got_object_id **ids = NULL;
1460 05118f5a 2021-06-22 stsp int i, nobj = 0, obj_type;
1461 2f8438b0 2022-05-04 stsp struct got_object_idset *idset_exclude;
1463 2f8438b0 2022-05-04 stsp idset_exclude = got_object_idset_alloc();
1464 2f8438b0 2022-05-04 stsp if (idset_exclude == NULL)
1465 2f8438b0 2022-05-04 stsp return got_error_from_errno("got_object_idset_alloc");
1467 b8af7c06 2022-03-15 stsp *ncolored = 0;
1468 b8af7c06 2022-03-15 stsp *nfound = 0;
1469 b8af7c06 2022-03-15 stsp *ntrees = 0;
1471 b8af7c06 2022-03-15 stsp err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1472 b8af7c06 2022-03-15 stsp repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1474 e6bcace5 2021-06-22 stsp goto done;
1476 e6bcace5 2021-06-22 stsp for (i = 0; i < ntheirs; i++) {
1477 05118f5a 2021-06-22 stsp struct got_object_id *id = theirs[i];
1478 05118f5a 2021-06-22 stsp if (id == NULL)
1480 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
1482 05118f5a 2021-06-22 stsp return err;
1483 9d34261e 2022-04-07 stsp if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1484 2f8438b0 2022-05-04 stsp err = load_commit(0, idset, idset_exclude, id, repo,
1485 9d34261e 2022-04-07 stsp loose_obj_only, ncolored, nfound, ntrees,
1486 9d34261e 2022-04-07 stsp progress_cb, progress_arg, rl,
1487 9d34261e 2022-04-07 stsp cancel_cb, cancel_arg);
1489 07165b17 2021-07-01 stsp goto done;
1490 9d34261e 2022-04-07 stsp } else if (obj_type == GOT_OBJ_TYPE_TAG) {
1491 2f8438b0 2022-05-04 stsp err = load_tag(0, idset, idset_exclude, id, repo,
1492 9d34261e 2022-04-07 stsp loose_obj_only, ncolored, nfound, ntrees,
1493 9d34261e 2022-04-07 stsp progress_cb, progress_arg, rl,
1494 9d34261e 2022-04-07 stsp cancel_cb, cancel_arg);
1496 9d34261e 2022-04-07 stsp goto done;
1500 eca70f98 2021-09-03 stsp for (i = 0; i < nobj; i++) {
1501 2f8438b0 2022-05-04 stsp err = load_commit(1, idset, idset_exclude,
1502 2f8438b0 2022-05-04 stsp ids[i], repo, loose_obj_only, ncolored, nfound, ntrees,
1503 2f8438b0 2022-05-04 stsp progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1505 eca70f98 2021-09-03 stsp goto done;
1508 05118f5a 2021-06-22 stsp for (i = 0; i < nours; i++) {
1509 05118f5a 2021-06-22 stsp struct got_object_id *id = ours[i];
1510 67fd6849 2022-02-13 stsp struct got_pack_meta *m;
1511 05118f5a 2021-06-22 stsp if (id == NULL)
1513 67fd6849 2022-02-13 stsp m = got_object_idset_get(idset, id);
1514 67fd6849 2022-02-13 stsp if (m == NULL) {
1515 07165b17 2021-07-01 stsp err = got_object_get_type(&obj_type, repo, id);
1517 07165b17 2021-07-01 stsp goto done;
1519 67fd6849 2022-02-13 stsp obj_type = m->obj_type;
1520 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_TAG)
1522 2f8438b0 2022-05-04 stsp err = load_tag(1, idset, idset_exclude, id, repo,
1523 2f8438b0 2022-05-04 stsp loose_obj_only, ncolored, nfound, ntrees,
1524 2f8438b0 2022-05-04 stsp progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1526 e6bcace5 2021-06-22 stsp goto done;
1529 e6bcace5 2021-06-22 stsp for (i = 0; i < nobj; i++) {
1530 e6bcace5 2021-06-22 stsp free(ids[i]);
1532 e6bcace5 2021-06-22 stsp free(ids);
1533 2f8438b0 2022-05-04 stsp got_object_idset_free(idset_exclude);
1534 e6bcace5 2021-06-22 stsp return err;
1537 e6bcace5 2021-06-22 stsp const struct got_error *
1538 2d9e6abf 2022-05-04 stsp hwrite(FILE *f, void *buf, off_t len, SHA1_CTX *ctx)
1542 e6bcace5 2021-06-22 stsp SHA1Update(ctx, buf, len);
1543 e6bcace5 2021-06-22 stsp n = fwrite(buf, 1, len, f);
1544 e6bcace5 2021-06-22 stsp if (n != len)
1545 e6bcace5 2021-06-22 stsp return got_ferror(f, GOT_ERR_IO);
1546 2d9e6abf 2022-05-04 stsp return NULL;
1549 2d9e6abf 2022-05-04 stsp const struct got_error *
1550 2d9e6abf 2022-05-04 stsp hcopy(FILE *fsrc, FILE *fdst, off_t len, SHA1_CTX *ctx)
1552 2d9e6abf 2022-05-04 stsp unsigned char buf[65536];
1553 2d9e6abf 2022-05-04 stsp off_t remain = len;
1556 2d9e6abf 2022-05-04 stsp while (remain > 0) {
1557 2d9e6abf 2022-05-04 stsp size_t copylen = MIN(sizeof(buf), remain);
1558 2d9e6abf 2022-05-04 stsp n = fread(buf, 1, copylen, fsrc);
1559 2d9e6abf 2022-05-04 stsp if (n != copylen)
1560 2d9e6abf 2022-05-04 stsp return got_ferror(fsrc, GOT_ERR_IO);
1561 2d9e6abf 2022-05-04 stsp SHA1Update(ctx, buf, copylen);
1562 2d9e6abf 2022-05-04 stsp n = fwrite(buf, 1, copylen, fdst);
1563 2d9e6abf 2022-05-04 stsp if (n != copylen)
1564 2d9e6abf 2022-05-04 stsp return got_ferror(fdst, GOT_ERR_IO);
1565 2d9e6abf 2022-05-04 stsp remain -= copylen;
1568 e6bcace5 2021-06-22 stsp return NULL;
1571 e93fb944 2022-05-10 stsp const struct got_error *
1572 e93fb944 2022-05-10 stsp hcopy_mmap(uint8_t *src, off_t src_offset, size_t src_size,
1573 e93fb944 2022-05-10 stsp FILE *fdst, off_t len, SHA1_CTX *ctx)
1577 e93fb944 2022-05-10 stsp if (src_offset + len > src_size)
1578 e93fb944 2022-05-10 stsp return got_error(GOT_ERR_RANGE);
1580 e93fb944 2022-05-10 stsp SHA1Update(ctx, src + src_offset, len);
1581 e93fb944 2022-05-10 stsp n = fwrite(src + src_offset, 1, len, fdst);
1582 e93fb944 2022-05-10 stsp if (n != len)
1583 e93fb944 2022-05-10 stsp return got_ferror(fdst, GOT_ERR_IO);
1585 e93fb944 2022-05-10 stsp return NULL;
1588 e6bcace5 2021-06-22 stsp static void
1589 e6bcace5 2021-06-22 stsp putbe32(char *b, uint32_t n)
1591 e6bcace5 2021-06-22 stsp b[0] = n >> 24;
1592 e6bcace5 2021-06-22 stsp b[1] = n >> 16;
1593 e6bcace5 2021-06-22 stsp b[2] = n >> 8;
1594 e6bcace5 2021-06-22 stsp b[3] = n >> 0;
1597 e6bcace5 2021-06-22 stsp static int
1598 e6bcace5 2021-06-22 stsp write_order_cmp(const void *pa, const void *pb)
1600 e6bcace5 2021-06-22 stsp struct got_pack_meta *a, *b, *ahd, *bhd;
1602 e6bcace5 2021-06-22 stsp a = *(struct got_pack_meta **)pa;
1603 e6bcace5 2021-06-22 stsp b = *(struct got_pack_meta **)pb;
1604 e6bcace5 2021-06-22 stsp ahd = (a->head == NULL) ? a : a->head;
1605 e6bcace5 2021-06-22 stsp bhd = (b->head == NULL) ? b : b->head;
1606 611e8e31 2022-05-01 stsp if (bhd->mtime < ahd->mtime)
1607 611e8e31 2022-05-01 stsp return -1;
1608 611e8e31 2022-05-01 stsp if (bhd->mtime > ahd->mtime)
1610 611e8e31 2022-05-01 stsp if (bhd < ahd)
1611 611e8e31 2022-05-01 stsp return -1;
1612 611e8e31 2022-05-01 stsp if (bhd > ahd)
1614 e6bcace5 2021-06-22 stsp if (a->nchain != b->nchain)
1615 e6bcace5 2021-06-22 stsp return a->nchain - b->nchain;
1616 611e8e31 2022-05-01 stsp if (a->mtime < b->mtime)
1617 611e8e31 2022-05-01 stsp return -1;
1618 611e8e31 2022-05-01 stsp if (a->mtime > b->mtime)
1620 611e8e31 2022-05-01 stsp return got_object_id_cmp(&a->id, &b->id);
1623 67fd6849 2022-02-13 stsp static int
1624 67fd6849 2022-02-13 stsp reuse_write_order_cmp(const void *pa, const void *pb)
1626 67fd6849 2022-02-13 stsp struct got_pack_meta *a, *b;
1628 67fd6849 2022-02-13 stsp a = *(struct got_pack_meta **)pa;
1629 67fd6849 2022-02-13 stsp b = *(struct got_pack_meta **)pb;
1631 67fd6849 2022-02-13 stsp if (a->reused_delta_offset < b->reused_delta_offset)
1632 67fd6849 2022-02-13 stsp return -1;
1633 67fd6849 2022-02-13 stsp if (a->reused_delta_offset > b->reused_delta_offset)
1638 e6bcace5 2021-06-22 stsp static const struct got_error *
1639 e6bcace5 2021-06-22 stsp packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1643 e6bcace5 2021-06-22 stsp *hdrlen = 0;
1645 e6bcace5 2021-06-22 stsp hdr[0] = obj_type << 4;
1646 e6bcace5 2021-06-22 stsp hdr[0] |= len & 0xf;
1647 e6bcace5 2021-06-22 stsp len >>= 4;
1648 e6bcace5 2021-06-22 stsp for (i = 1; len != 0; i++){
1649 e6bcace5 2021-06-22 stsp if (i >= bufsize)
1650 e6bcace5 2021-06-22 stsp return got_error(GOT_ERR_NO_SPACE);
1651 e6bcace5 2021-06-22 stsp hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1652 e6bcace5 2021-06-22 stsp hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1653 e6bcace5 2021-06-22 stsp len >>= GOT_DELTA_SIZE_SHIFT;
1656 e6bcace5 2021-06-22 stsp *hdrlen = i;
1657 e6bcace5 2021-06-22 stsp return NULL;
1660 e6bcace5 2021-06-22 stsp static int
1661 e6bcace5 2021-06-22 stsp packoff(char *hdr, off_t off)
1664 e6bcace5 2021-06-22 stsp char rbuf[8];
1666 e6bcace5 2021-06-22 stsp rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1667 e6bcace5 2021-06-22 stsp for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1668 e6bcace5 2021-06-22 stsp rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1669 e6bcace5 2021-06-22 stsp GOT_DELTA_SIZE_MORE;
1673 e6bcace5 2021-06-22 stsp while (i > 0)
1674 e6bcace5 2021-06-22 stsp hdr[j++] = rbuf[--i];
1678 e6bcace5 2021-06-22 stsp static const struct got_error *
1679 5060d5a1 2022-01-10 stsp deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1680 67fd6849 2022-02-13 stsp struct got_pack_meta *m)
1682 5060d5a1 2022-01-10 stsp const struct got_error *err;
1683 5060d5a1 2022-01-10 stsp char buf[32];
1686 67fd6849 2022-02-13 stsp if (m->prev->off != 0) {
1687 5060d5a1 2022-01-10 stsp err = packhdr(&nh, buf, sizeof(buf),
1688 5060d5a1 2022-01-10 stsp GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1690 5060d5a1 2022-01-10 stsp return err;
1691 5060d5a1 2022-01-10 stsp nh += packoff(buf + nh, m->off - m->prev->off);
1692 5060d5a1 2022-01-10 stsp err = hwrite(packfile, buf, nh, ctx);
1694 5060d5a1 2022-01-10 stsp return err;
1695 5060d5a1 2022-01-10 stsp *packfile_size += nh;
1697 5060d5a1 2022-01-10 stsp err = packhdr(&nh, buf, sizeof(buf),
1698 5060d5a1 2022-01-10 stsp GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1700 5060d5a1 2022-01-10 stsp return err;
1701 5060d5a1 2022-01-10 stsp err = hwrite(packfile, buf, nh, ctx);
1703 5060d5a1 2022-01-10 stsp return err;
1704 5060d5a1 2022-01-10 stsp *packfile_size += nh;
1705 5060d5a1 2022-01-10 stsp err = hwrite(packfile, m->prev->id.sha1,
1706 5060d5a1 2022-01-10 stsp sizeof(m->prev->id.sha1), ctx);
1708 5060d5a1 2022-01-10 stsp return err;
1709 5060d5a1 2022-01-10 stsp *packfile_size += sizeof(m->prev->id.sha1);
1712 5060d5a1 2022-01-10 stsp return NULL;
1715 5060d5a1 2022-01-10 stsp static const struct got_error *
1716 67fd6849 2022-02-13 stsp write_packed_object(off_t *packfile_size, FILE *packfile,
1717 e93fb944 2022-05-10 stsp FILE *delta_cache, uint8_t *delta_cache_map, size_t delta_cache_size,
1718 e93fb944 2022-05-10 stsp struct got_pack_meta *m, int *outfd, SHA1_CTX *ctx,
1719 e93fb944 2022-05-10 stsp struct got_repository *repo)
1721 67fd6849 2022-02-13 stsp const struct got_error *err = NULL;
1722 67fd6849 2022-02-13 stsp struct got_deflate_checksum csum;
1723 67fd6849 2022-02-13 stsp char buf[32];
1725 67fd6849 2022-02-13 stsp struct got_raw_object *raw = NULL;
1726 67fd6849 2022-02-13 stsp off_t outlen;
1728 67fd6849 2022-02-13 stsp csum.output_sha1 = ctx;
1729 67fd6849 2022-02-13 stsp csum.output_crc = NULL;
1731 67fd6849 2022-02-13 stsp m->off = ftello(packfile);
1732 67fd6849 2022-02-13 stsp if (m->delta_len == 0) {
1733 67fd6849 2022-02-13 stsp err = got_object_raw_open(&raw, outfd, repo, &m->id);
1735 67fd6849 2022-02-13 stsp goto done;
1736 67fd6849 2022-02-13 stsp err = packhdr(&nh, buf, sizeof(buf),
1737 67fd6849 2022-02-13 stsp m->obj_type, raw->size);
1739 67fd6849 2022-02-13 stsp goto done;
1740 67fd6849 2022-02-13 stsp err = hwrite(packfile, buf, nh, ctx);
1742 67fd6849 2022-02-13 stsp goto done;
1743 67fd6849 2022-02-13 stsp *packfile_size += nh;
1744 67fd6849 2022-02-13 stsp if (raw->f == NULL) {
1745 67fd6849 2022-02-13 stsp err = got_deflate_to_file_mmap(&outlen,
1746 67fd6849 2022-02-13 stsp raw->data + raw->hdrlen, 0, raw->size,
1747 67fd6849 2022-02-13 stsp packfile, &csum);
1749 67fd6849 2022-02-13 stsp goto done;
1751 67fd6849 2022-02-13 stsp if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1753 67fd6849 2022-02-13 stsp err = got_error_from_errno("fseeko");
1754 67fd6849 2022-02-13 stsp goto done;
1756 67fd6849 2022-02-13 stsp err = got_deflate_to_file(&outlen, raw->f,
1757 67fd6849 2022-02-13 stsp raw->size, packfile, &csum);
1759 67fd6849 2022-02-13 stsp goto done;
1761 67fd6849 2022-02-13 stsp *packfile_size += outlen;
1762 67fd6849 2022-02-13 stsp got_object_raw_close(raw);
1763 67fd6849 2022-02-13 stsp raw = NULL;
1764 67fd6849 2022-02-13 stsp } else if (m->delta_buf) {
1765 67fd6849 2022-02-13 stsp err = deltahdr(packfile_size, ctx, packfile, m);
1767 67fd6849 2022-02-13 stsp goto done;
1768 2d9e6abf 2022-05-04 stsp err = hwrite(packfile, m->delta_buf,
1769 2d9e6abf 2022-05-04 stsp m->delta_compressed_len, ctx);
1771 67fd6849 2022-02-13 stsp goto done;
1772 2d9e6abf 2022-05-04 stsp *packfile_size += m->delta_compressed_len;
1773 67fd6849 2022-02-13 stsp free(m->delta_buf);
1774 67fd6849 2022-02-13 stsp m->delta_buf = NULL;
1775 e93fb944 2022-05-10 stsp } else if (delta_cache_map) {
1776 e93fb944 2022-05-10 stsp err = deltahdr(packfile_size, ctx, packfile, m);
1778 e93fb944 2022-05-10 stsp goto done;
1779 e93fb944 2022-05-10 stsp err = hcopy_mmap(delta_cache_map, m->delta_offset,
1780 e93fb944 2022-05-10 stsp delta_cache_size, packfile, m->delta_compressed_len,
1783 e93fb944 2022-05-10 stsp goto done;
1784 e93fb944 2022-05-10 stsp *packfile_size += m->delta_compressed_len;
1786 67fd6849 2022-02-13 stsp if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1788 67fd6849 2022-02-13 stsp err = got_error_from_errno("fseeko");
1789 67fd6849 2022-02-13 stsp goto done;
1791 67fd6849 2022-02-13 stsp err = deltahdr(packfile_size, ctx, packfile, m);
1793 67fd6849 2022-02-13 stsp goto done;
1794 2d9e6abf 2022-05-04 stsp err = hcopy(delta_cache, packfile,
1795 2d9e6abf 2022-05-04 stsp m->delta_compressed_len, ctx);
1797 67fd6849 2022-02-13 stsp goto done;
1798 2d9e6abf 2022-05-04 stsp *packfile_size += m->delta_compressed_len;
1802 67fd6849 2022-02-13 stsp got_object_raw_close(raw);
1803 67fd6849 2022-02-13 stsp return err;
1806 67fd6849 2022-02-13 stsp static const struct got_error *
1807 74881701 2021-10-15 stsp genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1808 67fd6849 2022-02-13 stsp struct got_pack_meta **deltify, int ndeltify,
1809 67fd6849 2022-02-13 stsp struct got_pack_meta **reuse, int nreuse,
1810 b8af7c06 2022-03-15 stsp int ncolored, int nfound, int ntrees, int nours,
1811 b8af7c06 2022-03-15 stsp struct got_repository *repo,
1812 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1813 211cfef0 2022-01-05 stsp struct got_ratelimit *rl,
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 = NULL;
1818 e6bcace5 2021-06-22 stsp SHA1_CTX ctx;
1819 e6bcace5 2021-06-22 stsp struct got_pack_meta *m;
1820 08347b73 2021-10-14 stsp char buf[32];
1822 67fd6849 2022-02-13 stsp off_t packfile_size = 0;
1823 cc7a354a 2021-10-15 stsp int outfd = -1;
1824 e93fb944 2022-05-10 stsp int delta_cache_fd = -1;
1825 e93fb944 2022-05-10 stsp uint8_t *delta_cache_map = NULL;
1826 e93fb944 2022-05-10 stsp size_t delta_cache_size = 0;
1828 e6bcace5 2021-06-22 stsp SHA1Init(&ctx);
1830 e93fb944 2022-05-10 stsp #ifndef GOT_PACK_NO_MMAP
1831 e93fb944 2022-05-10 stsp delta_cache_fd = dup(fileno(delta_cache));
1832 e93fb944 2022-05-10 stsp if (delta_cache_fd != -1) {
1833 e93fb944 2022-05-10 stsp struct stat sb;
1834 e93fb944 2022-05-10 stsp if (fstat(delta_cache_fd, &sb) == -1) {
1835 e93fb944 2022-05-10 stsp err = got_error_from_errno("fstat");
1836 e93fb944 2022-05-10 stsp goto done;
1838 e93fb944 2022-05-10 stsp if (sb.st_size > 0 && sb.st_size <= SIZE_MAX) {
1839 e93fb944 2022-05-10 stsp delta_cache_map = mmap(NULL, sb.st_size,
1840 e93fb944 2022-05-10 stsp PROT_READ, MAP_PRIVATE, delta_cache_fd, 0);
1841 e93fb944 2022-05-10 stsp if (delta_cache_map == MAP_FAILED) {
1842 e93fb944 2022-05-10 stsp if (errno != ENOMEM) {
1843 e93fb944 2022-05-10 stsp err = got_error_from_errno("mmap");
1844 e93fb944 2022-05-10 stsp goto done;
1846 e93fb944 2022-05-10 stsp delta_cache_map = NULL; /* fallback on stdio */
1848 e93fb944 2022-05-10 stsp delta_cache_size = (size_t)sb.st_size;
1852 e6bcace5 2021-06-22 stsp err = hwrite(packfile, "PACK", 4, &ctx);
1854 e93fb944 2022-05-10 stsp goto done;
1855 e6bcace5 2021-06-22 stsp putbe32(buf, GOT_PACKFILE_VERSION);
1856 e6bcace5 2021-06-22 stsp err = hwrite(packfile, buf, 4, &ctx);
1858 e6bcace5 2021-06-22 stsp goto done;
1859 67fd6849 2022-02-13 stsp putbe32(buf, ndeltify + nreuse);
1860 e6bcace5 2021-06-22 stsp err = hwrite(packfile, buf, 4, &ctx);
1862 e6bcace5 2021-06-22 stsp goto done;
1864 67fd6849 2022-02-13 stsp qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1865 67fd6849 2022-02-13 stsp write_order_cmp);
1866 67fd6849 2022-02-13 stsp for (i = 0; i < ndeltify; i++) {
1867 211cfef0 2022-01-05 stsp err = report_progress(progress_cb, progress_arg, rl,
1868 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, packfile_size, nours,
1869 b8af7c06 2022-03-15 stsp ndeltify + nreuse, ndeltify + nreuse, i);
1871 211cfef0 2022-01-05 stsp goto done;
1872 67fd6849 2022-02-13 stsp m = deltify[i];
1873 67fd6849 2022-02-13 stsp err = write_packed_object(&packfile_size, packfile,
1874 e93fb944 2022-05-10 stsp delta_cache, delta_cache_map, delta_cache_size,
1875 e93fb944 2022-05-10 stsp m, &outfd, &ctx, repo);
1877 67fd6849 2022-02-13 stsp goto done;
1880 67fd6849 2022-02-13 stsp qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1881 67fd6849 2022-02-13 stsp reuse_write_order_cmp);
1882 67fd6849 2022-02-13 stsp for (i = 0; i < nreuse; i++) {
1883 67fd6849 2022-02-13 stsp err = report_progress(progress_cb, progress_arg, rl,
1884 b8af7c06 2022-03-15 stsp ncolored, nfound, ntrees, packfile_size, nours,
1885 b8af7c06 2022-03-15 stsp ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1887 67fd6849 2022-02-13 stsp goto done;
1888 67fd6849 2022-02-13 stsp m = reuse[i];
1889 67fd6849 2022-02-13 stsp err = write_packed_object(&packfile_size, packfile,
1890 e93fb944 2022-05-10 stsp delta_cache, delta_cache_map, delta_cache_size,
1891 e93fb944 2022-05-10 stsp m, &outfd, &ctx, repo);
1893 67fd6849 2022-02-13 stsp goto done;
1896 e6bcace5 2021-06-22 stsp SHA1Final(pack_sha1, &ctx);
1897 e6bcace5 2021-06-22 stsp n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1898 e6bcace5 2021-06-22 stsp if (n != SHA1_DIGEST_LENGTH)
1899 e6bcace5 2021-06-22 stsp err = got_ferror(packfile, GOT_ERR_IO);
1900 05118f5a 2021-06-22 stsp packfile_size += SHA1_DIGEST_LENGTH;
1901 dc7edd42 2021-08-22 stsp packfile_size += sizeof(struct got_packfile_hdr);
1902 211cfef0 2022-01-05 stsp if (progress_cb) {
1903 b8af7c06 2022-03-15 stsp err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1904 b8af7c06 2022-03-15 stsp packfile_size, nours, ndeltify + nreuse,
1905 b8af7c06 2022-03-15 stsp ndeltify + nreuse, ndeltify + nreuse);
1907 211cfef0 2022-01-05 stsp goto done;
1910 cc7a354a 2021-10-15 stsp if (outfd != -1 && close(outfd) == -1 && err == NULL)
1911 cc7a354a 2021-10-15 stsp err = got_error_from_errno("close");
1912 e93fb944 2022-05-10 stsp if (delta_cache_map && munmap(delta_cache_map, delta_cache_size) == -1)
1913 e93fb944 2022-05-10 stsp err = got_error_from_errno("munmap");
1914 e93fb944 2022-05-10 stsp if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1915 e93fb944 2022-05-10 stsp err = got_error_from_errno("close");
1916 e6bcace5 2021-06-22 stsp return err;
1919 67fd6849 2022-02-13 stsp static const struct got_error *
1920 67fd6849 2022-02-13 stsp add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1922 67fd6849 2022-02-13 stsp struct got_pack_meta *m = data;
1923 67fd6849 2022-02-13 stsp struct got_pack_metavec *v = arg;
1925 f5e78e05 2022-05-04 stsp if (m->have_reused_delta)
1926 f5e78e05 2022-05-04 stsp return NULL;
1928 67fd6849 2022-02-13 stsp return add_meta(m, v);
1931 e6bcace5 2021-06-22 stsp const struct got_error *
1932 e6bcace5 2021-06-22 stsp got_pack_create(uint8_t *packsha1, FILE *packfile,
1933 e6bcace5 2021-06-22 stsp struct got_object_id **theirs, int ntheirs,
1934 e6bcace5 2021-06-22 stsp struct got_object_id **ours, int nours,
1935 f8a36e22 2021-08-26 stsp struct got_repository *repo, int loose_obj_only, int allow_empty,
1936 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
1937 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1939 e6bcace5 2021-06-22 stsp const struct got_error *err;
1940 67fd6849 2022-02-13 stsp int delta_cache_fd = -1;
1941 74881701 2021-10-15 stsp FILE *delta_cache = NULL;
1942 67fd6849 2022-02-13 stsp struct got_object_idset *idset;
1943 211cfef0 2022-01-05 stsp struct got_ratelimit rl;
1944 67fd6849 2022-02-13 stsp struct got_pack_metavec deltify, reuse;
1945 b8af7c06 2022-03-15 stsp int ncolored = 0, nfound = 0, ntrees = 0;
1946 f5e78e05 2022-05-04 stsp size_t ndeltify;
1948 67fd6849 2022-02-13 stsp memset(&deltify, 0, sizeof(deltify));
1949 67fd6849 2022-02-13 stsp memset(&reuse, 0, sizeof(reuse));
1951 211cfef0 2022-01-05 stsp got_ratelimit_init(&rl, 0, 500);
1953 67fd6849 2022-02-13 stsp idset = got_object_idset_alloc();
1954 67fd6849 2022-02-13 stsp if (idset == NULL)
1955 67fd6849 2022-02-13 stsp return got_error_from_errno("got_object_idset_alloc");
1957 b8af7c06 2022-03-15 stsp err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1958 b8af7c06 2022-03-15 stsp ntheirs, ours, nours, repo, loose_obj_only,
1959 b8af7c06 2022-03-15 stsp progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1961 17259bfa 2022-05-19 stsp goto done;
1963 28526235 2022-02-13 stsp if (progress_cb) {
1964 b8af7c06 2022-03-15 stsp err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1965 b8af7c06 2022-03-15 stsp 0L, nours, got_object_idset_num_elements(idset), 0, 0);
1967 28526235 2022-02-13 stsp goto done;
1970 67fd6849 2022-02-13 stsp if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1971 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_CANNOT_PACK);
1972 05118f5a 2021-06-22 stsp goto done;
1975 67fd6849 2022-02-13 stsp delta_cache_fd = got_opentempfd();
1976 67fd6849 2022-02-13 stsp if (delta_cache_fd == -1) {
1977 74881701 2021-10-15 stsp err = got_error_from_errno("got_opentemp");
1978 74881701 2021-10-15 stsp goto done;
1981 67fd6849 2022-02-13 stsp reuse.metasz = 64;
1982 67fd6849 2022-02-13 stsp reuse.meta = calloc(reuse.metasz,
1983 67fd6849 2022-02-13 stsp sizeof(struct got_pack_meta *));
1984 67fd6849 2022-02-13 stsp if (reuse.meta == NULL) {
1985 67fd6849 2022-02-13 stsp err = got_error_from_errno("calloc");
1986 67fd6849 2022-02-13 stsp goto done;
1989 b8af7c06 2022-03-15 stsp err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
1990 b8af7c06 2022-03-15 stsp ntrees, nours, repo, progress_cb, progress_arg, &rl,
1991 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
1993 67fd6849 2022-02-13 stsp goto done;
1995 67fd6849 2022-02-13 stsp delta_cache = fdopen(delta_cache_fd, "a+");
1996 67fd6849 2022-02-13 stsp if (delta_cache == NULL) {
1997 67fd6849 2022-02-13 stsp err = got_error_from_errno("fdopen");
1998 67fd6849 2022-02-13 stsp goto done;
2000 67fd6849 2022-02-13 stsp delta_cache_fd = -1;
2002 67fd6849 2022-02-13 stsp if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
2003 67fd6849 2022-02-13 stsp err = got_error_from_errno("fseeko");
2004 67fd6849 2022-02-13 stsp goto done;
2007 f5e78e05 2022-05-04 stsp ndeltify = got_object_idset_num_elements(idset) - reuse.nmeta;
2008 f5e78e05 2022-05-04 stsp if (ndeltify > 0) {
2009 f5e78e05 2022-05-04 stsp deltify.meta = calloc(ndeltify, sizeof(struct got_pack_meta *));
2010 f5e78e05 2022-05-04 stsp if (deltify.meta == NULL) {
2011 f5e78e05 2022-05-04 stsp err = got_error_from_errno("calloc");
2012 f5e78e05 2022-05-04 stsp goto done;
2014 f5e78e05 2022-05-04 stsp deltify.metasz = ndeltify;
2016 f5e78e05 2022-05-04 stsp err = got_object_idset_for_each(idset, add_meta_idset_cb,
2017 f5e78e05 2022-05-04 stsp &deltify);
2019 f8a36e22 2021-08-26 stsp goto done;
2020 f5e78e05 2022-05-04 stsp if (deltify.nmeta > 0) {
2021 f5e78e05 2022-05-04 stsp err = pick_deltas(deltify.meta, deltify.nmeta,
2022 f5e78e05 2022-05-04 stsp ncolored, nfound, ntrees, nours, reuse.nmeta,
2023 f5e78e05 2022-05-04 stsp delta_cache, repo, progress_cb, progress_arg, &rl,
2024 f5e78e05 2022-05-04 stsp cancel_cb, cancel_arg);
2026 f5e78e05 2022-05-04 stsp goto done;
2030 2d9e6abf 2022-05-04 stsp if (fflush(delta_cache) == EOF) {
2031 2d9e6abf 2022-05-04 stsp err = got_error_from_errno("fflush");
2032 2d9e6abf 2022-05-04 stsp goto done;
2034 67fd6849 2022-02-13 stsp err = genpack(packsha1, packfile, delta_cache, deltify.meta,
2035 b8af7c06 2022-03-15 stsp deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
2036 b8af7c06 2022-03-15 stsp nours, repo, progress_cb, progress_arg, &rl,
2037 b8af7c06 2022-03-15 stsp cancel_cb, cancel_arg);
2039 e6bcace5 2021-06-22 stsp goto done;
2041 67fd6849 2022-02-13 stsp free_nmeta(deltify.meta, deltify.nmeta);
2042 67fd6849 2022-02-13 stsp free_nmeta(reuse.meta, reuse.nmeta);
2043 67fd6849 2022-02-13 stsp got_object_idset_free(idset);
2044 67fd6849 2022-02-13 stsp if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
2045 67fd6849 2022-02-13 stsp err = got_error_from_errno("close");
2046 74881701 2021-10-15 stsp if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
2047 74881701 2021-10-15 stsp err = got_error_from_errno("fclose");
2048 e6bcace5 2021-06-22 stsp return err;
