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 8b925c6c 2022-07-16 thomas #include <sys/queue.h>
20 08736cf9 2021-06-23 stsp #include <sys/uio.h>
21 e6bcace5 2021-06-22 stsp #include <sys/stat.h>
22 31ba2236 2022-01-05 thomas #include <sys/time.h>
23 d93dda9a 2022-05-12 thomas #include <sys/mman.h>
25 d93dda9a 2022-05-12 thomas #include <errno.h>
26 08736cf9 2021-06-23 stsp #include <stdint.h>
27 60b94e7d 2022-07-19 thomas #include <inttypes.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 31ba2236 2022-01-05 thomas #include <time.h>
32 db847d2c 2022-03-22 thomas #include <unistd.h>
33 e6bcace5 2021-06-22 stsp #include <limits.h>
34 e6bcace5 2021-06-22 stsp #include <zlib.h>
36 b347007e 2021-10-15 thomas #if defined(__FreeBSD__)
37 b347007e 2021-10-15 thomas #include <unistd.h>
40 e6bcace5 2021-06-22 stsp #include "got_error.h"
41 e6bcace5 2021-06-22 stsp #include "got_cancel.h"
42 e6bcace5 2021-06-22 stsp #include "got_object.h"
43 0be8fa4c 2021-10-15 thomas #include "got_path.h"
44 05118f5a 2021-06-22 stsp #include "got_reference.h"
45 05118f5a 2021-06-22 stsp #include "got_repository_admin.h"
46 e4d8ab47 2021-10-15 thomas #include "got_opentemp.h"
48 e6bcace5 2021-06-22 stsp #include "got_lib_deltify.h"
49 e6bcace5 2021-06-22 stsp #include "got_lib_delta.h"
50 e6bcace5 2021-06-22 stsp #include "got_lib_object.h"
51 e6bcace5 2021-06-22 stsp #include "got_lib_object_idset.h"
52 05118f5a 2021-06-22 stsp #include "got_lib_object_cache.h"
53 e6bcace5 2021-06-22 stsp #include "got_lib_deflate.h"
54 e6bcace5 2021-06-22 stsp #include "got_lib_pack.h"
55 036966ba 2022-05-31 thomas #include "got_lib_pack_create.h"
56 05118f5a 2021-06-22 stsp #include "got_lib_privsep.h"
57 05118f5a 2021-06-22 stsp #include "got_lib_repository.h"
58 31ba2236 2022-01-05 thomas #include "got_lib_ratelimit.h"
59 9249e7e3 2022-05-12 thomas #include "got_lib_inflate.h"
61 d23a6c95 2022-05-31 thomas #include "murmurhash2.h"
63 b79280dd 2021-10-15 thomas #ifndef MIN
64 b79280dd 2021-10-15 thomas #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
68 e6bcace5 2021-06-22 stsp #define MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
71 5fe7b5c1 2022-04-16 thomas #ifndef nitems
72 5fe7b5c1 2022-04-16 thomas #define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
75 e6bcace5 2021-06-22 stsp struct got_pack_meta {
76 e6bcace5 2021-06-22 stsp struct got_object_id id;
77 d23a6c95 2022-05-31 thomas uint32_t path_hash;
78 e6bcace5 2021-06-22 stsp int obj_type;
79 ab6186ae 2021-10-15 thomas off_t size;
80 e6bcace5 2021-06-22 stsp time_t mtime;
82 e6bcace5 2021-06-22 stsp /* The best delta we picked */
83 e6bcace5 2021-06-22 stsp struct got_pack_meta *head;
84 e6bcace5 2021-06-22 stsp struct got_pack_meta *prev;
85 9249e7e3 2022-05-12 thomas unsigned char *delta_buf; /* if encoded in memory (compressed) */
86 9249e7e3 2022-05-12 thomas off_t delta_offset; /* offset in delta cache file (compressed) */
87 3b6ceab7 2022-01-10 thomas off_t delta_len; /* encoded delta length */
88 9249e7e3 2022-05-12 thomas off_t delta_compressed_len; /* encoded+compressed delta length */
91 f9c2e8e5 2022-02-13 thomas off_t reused_delta_offset; /* offset of delta in reused pack file */
92 f9c2e8e5 2022-02-13 thomas struct got_object_id *base_obj_id;
94 e6bcace5 2021-06-22 stsp /* Only used for delta window */
95 e6bcace5 2021-06-22 stsp struct got_delta_table *dtab;
97 e6bcace5 2021-06-22 stsp /* Only used for writing offset deltas */
101 e6bcace5 2021-06-22 stsp struct got_pack_metavec {
102 e6bcace5 2021-06-22 stsp struct got_pack_meta **meta;
104 e6bcace5 2021-06-22 stsp int metasz;
107 e6bcace5 2021-06-22 stsp static const struct got_error *
108 e6bcace5 2021-06-22 stsp alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
109 cc524d36 2022-05-31 thomas const char *path, int obj_type, time_t mtime, uint32_t seed)
111 e6bcace5 2021-06-22 stsp struct got_pack_meta *m;
113 e6bcace5 2021-06-22 stsp *new = NULL;
115 e6bcace5 2021-06-22 stsp m = calloc(1, sizeof(*m));
116 e6bcace5 2021-06-22 stsp if (m == NULL)
117 b7e0a384 2021-10-15 thomas return got_error_from_errno("calloc");
119 e6bcace5 2021-06-22 stsp memcpy(&m->id, id, sizeof(m->id));
121 cc524d36 2022-05-31 thomas m->path_hash = murmurhash2(path, strlen(path), seed);
122 e6bcace5 2021-06-22 stsp m->obj_type = obj_type;
123 e6bcace5 2021-06-22 stsp m->mtime = mtime;
125 e6bcace5 2021-06-22 stsp return NULL;
128 e6bcace5 2021-06-22 stsp static void
129 e6bcace5 2021-06-22 stsp clear_meta(struct got_pack_meta *meta)
131 e6bcace5 2021-06-22 stsp if (meta == NULL)
133 d23a6c95 2022-05-31 thomas meta->path_hash = 0;
134 3b6ceab7 2022-01-10 thomas free(meta->delta_buf);
135 3b6ceab7 2022-01-10 thomas meta->delta_buf = NULL;
136 f9c2e8e5 2022-02-13 thomas free(meta->base_obj_id);
137 f9c2e8e5 2022-02-13 thomas meta->base_obj_id = NULL;
138 6c0788af 2022-05-31 thomas meta->reused_delta_offset = 0;
141 e6bcace5 2021-06-22 stsp static void
142 e6bcace5 2021-06-22 stsp free_nmeta(struct got_pack_meta **meta, int nmeta)
146 e6bcace5 2021-06-22 stsp for (i = 0; i < nmeta; i++)
147 e6bcace5 2021-06-22 stsp clear_meta(meta[i]);
148 e6bcace5 2021-06-22 stsp free(meta);
152 e6bcace5 2021-06-22 stsp delta_order_cmp(const void *pa, const void *pb)
154 e6bcace5 2021-06-22 stsp struct got_pack_meta *a, *b;
156 e6bcace5 2021-06-22 stsp a = *(struct got_pack_meta **)pa;
157 e6bcace5 2021-06-22 stsp b = *(struct got_pack_meta **)pb;
159 e6bcace5 2021-06-22 stsp if (a->obj_type != b->obj_type)
160 e6bcace5 2021-06-22 stsp return a->obj_type - b->obj_type;
161 d23a6c95 2022-05-31 thomas if (a->path_hash < b->path_hash)
162 d23a6c95 2022-05-31 thomas return -1;
163 d23a6c95 2022-05-31 thomas if (a->path_hash > b->path_hash)
164 d23a6c95 2022-05-31 thomas return 1;
165 f6b43367 2022-05-01 thomas if (a->mtime < b->mtime)
166 f6b43367 2022-05-01 thomas return -1;
167 f6b43367 2022-05-01 thomas if (a->mtime > b->mtime)
168 f6b43367 2022-05-01 thomas return 1;
169 e6bcace5 2021-06-22 stsp return got_object_id_cmp(&a->id, &b->id);
172 3b6ceab7 2022-01-10 thomas static off_t
173 e6bcace5 2021-06-22 stsp delta_size(struct got_delta_instruction *deltas, int ndeltas)
176 3b6ceab7 2022-01-10 thomas off_t size = 32;
177 e6bcace5 2021-06-22 stsp for (i = 0; i < ndeltas; i++) {
178 e6bcace5 2021-06-22 stsp if (deltas[i].copy)
179 e6bcace5 2021-06-22 stsp size += GOT_DELTA_SIZE_SHIFT;
181 e6bcace5 2021-06-22 stsp size += deltas[i].len + 1;
183 e6bcace5 2021-06-22 stsp return size;
186 3b6ceab7 2022-01-10 thomas static const struct got_error *
187 3b6ceab7 2022-01-10 thomas append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
191 3b6ceab7 2022-01-10 thomas if (*len + nseg >= *sz) {
192 3b6ceab7 2022-01-10 thomas while (*len + nseg >= *sz)
193 3b6ceab7 2022-01-10 thomas *sz += *sz / 2;
194 3b6ceab7 2022-01-10 thomas n = realloc(*p, *sz);
195 3b6ceab7 2022-01-10 thomas if (n == NULL)
196 3b6ceab7 2022-01-10 thomas return got_error_from_errno("realloc");
199 3b6ceab7 2022-01-10 thomas memcpy(*p + *len, seg, nseg);
200 3b6ceab7 2022-01-10 thomas *len += nseg;
201 3b6ceab7 2022-01-10 thomas return NULL;
204 3b6ceab7 2022-01-10 thomas static const struct got_error *
205 3b6ceab7 2022-01-10 thomas encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
206 3b6ceab7 2022-01-10 thomas struct got_delta_instruction *deltas, int ndeltas,
207 3b6ceab7 2022-01-10 thomas off_t delta_size, off_t base_size)
209 3b6ceab7 2022-01-10 thomas const struct got_error *err;
210 3b6ceab7 2022-01-10 thomas unsigned char buf[16], *bp;
211 3b6ceab7 2022-01-10 thomas int i, j;
212 9249e7e3 2022-05-12 thomas size_t len = 0, compressed_len;
213 9249e7e3 2022-05-12 thomas off_t bufsize = delta_size;
215 3b6ceab7 2022-01-10 thomas struct got_delta_instruction *d;
216 9249e7e3 2022-05-12 thomas uint8_t *delta_buf;
218 9249e7e3 2022-05-12 thomas delta_buf = malloc(bufsize);
219 9249e7e3 2022-05-12 thomas if (delta_buf == NULL)
220 9249e7e3 2022-05-12 thomas return got_error_from_errno("malloc");
222 3b6ceab7 2022-01-10 thomas /* base object size */
223 3b6ceab7 2022-01-10 thomas buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
224 3b6ceab7 2022-01-10 thomas n = base_size >> GOT_DELTA_SIZE_SHIFT;
225 3b6ceab7 2022-01-10 thomas for (i = 1; n > 0; i++) {
226 3b6ceab7 2022-01-10 thomas buf[i - 1] |= GOT_DELTA_SIZE_MORE;
227 3b6ceab7 2022-01-10 thomas buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
228 3b6ceab7 2022-01-10 thomas n >>= GOT_DELTA_SIZE_SHIFT;
230 9249e7e3 2022-05-12 thomas err = append(&delta_buf, &len, &bufsize, buf, i);
232 9249e7e3 2022-05-12 thomas goto done;
234 3b6ceab7 2022-01-10 thomas /* target object size */
235 3b6ceab7 2022-01-10 thomas buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
236 3b6ceab7 2022-01-10 thomas n = o->size >> GOT_DELTA_SIZE_SHIFT;
237 3b6ceab7 2022-01-10 thomas for (i = 1; n > 0; i++) {
238 3b6ceab7 2022-01-10 thomas buf[i - 1] |= GOT_DELTA_SIZE_MORE;
239 3b6ceab7 2022-01-10 thomas buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
240 3b6ceab7 2022-01-10 thomas n >>= GOT_DELTA_SIZE_SHIFT;
242 9249e7e3 2022-05-12 thomas err = append(&delta_buf, &len, &bufsize, buf, i);
244 9249e7e3 2022-05-12 thomas goto done;
246 3b6ceab7 2022-01-10 thomas for (j = 0; j < ndeltas; j++) {
247 3b6ceab7 2022-01-10 thomas d = &deltas[j];
248 3b6ceab7 2022-01-10 thomas if (d->copy) {
249 3b6ceab7 2022-01-10 thomas n = d->offset;
250 3b6ceab7 2022-01-10 thomas bp = &buf[1];
251 3b6ceab7 2022-01-10 thomas buf[0] = GOT_DELTA_BASE_COPY;
252 3b6ceab7 2022-01-10 thomas for (i = 0; i < 4; i++) {
253 3b6ceab7 2022-01-10 thomas /* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
254 3b6ceab7 2022-01-10 thomas buf[0] |= 1 << i;
255 3b6ceab7 2022-01-10 thomas *bp++ = n & 0xff;
257 3b6ceab7 2022-01-10 thomas if (n == 0)
261 3b6ceab7 2022-01-10 thomas n = d->len;
262 3b6ceab7 2022-01-10 thomas if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
263 3b6ceab7 2022-01-10 thomas /* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
264 3b6ceab7 2022-01-10 thomas for (i = 0; i < 3 && n > 0; i++) {
265 3b6ceab7 2022-01-10 thomas buf[0] |= 1 << (i + 4);
266 3b6ceab7 2022-01-10 thomas *bp++ = n & 0xff;
270 9249e7e3 2022-05-12 thomas err = append(&delta_buf, &len, &bufsize,
271 3b6ceab7 2022-01-10 thomas buf, bp - buf);
273 9249e7e3 2022-05-12 thomas goto done;
274 3b6ceab7 2022-01-10 thomas } else if (o->f == NULL) {
276 3b6ceab7 2022-01-10 thomas while (n != d->len) {
277 3b6ceab7 2022-01-10 thomas buf[0] = (d->len - n < 127) ? d->len - n : 127;
278 9249e7e3 2022-05-12 thomas err = append(&delta_buf, &len, &bufsize,
281 9249e7e3 2022-05-12 thomas goto done;
282 9249e7e3 2022-05-12 thomas err = append(&delta_buf, &len, &bufsize,
283 3b6ceab7 2022-01-10 thomas o->data + o->hdrlen + d->offset + n,
286 9249e7e3 2022-05-12 thomas goto done;
287 3b6ceab7 2022-01-10 thomas n += buf[0];
290 3b6ceab7 2022-01-10 thomas char content[128];
291 3b6ceab7 2022-01-10 thomas size_t r;
292 9249e7e3 2022-05-12 thomas if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
293 9249e7e3 2022-05-12 thomas err = got_error_from_errno("fseeko");
294 9249e7e3 2022-05-12 thomas goto done;
297 3b6ceab7 2022-01-10 thomas while (n != d->len) {
298 3b6ceab7 2022-01-10 thomas buf[0] = (d->len - n < 127) ? d->len - n : 127;
299 9249e7e3 2022-05-12 thomas err = append(&delta_buf, &len, &bufsize,
302 9249e7e3 2022-05-12 thomas goto done;
303 3b6ceab7 2022-01-10 thomas r = fread(content, 1, buf[0], o->f);
304 9249e7e3 2022-05-12 thomas if (r != buf[0]) {
305 9249e7e3 2022-05-12 thomas err = got_ferror(o->f, GOT_ERR_IO);
306 9249e7e3 2022-05-12 thomas goto done;
308 9249e7e3 2022-05-12 thomas err = append(&delta_buf, &len, &bufsize,
309 3b6ceab7 2022-01-10 thomas content, buf[0]);
311 9249e7e3 2022-05-12 thomas goto done;
312 3b6ceab7 2022-01-10 thomas n += buf[0];
317 9249e7e3 2022-05-12 thomas err = got_deflate_to_mem_mmap(&m->delta_buf, &compressed_len,
318 9249e7e3 2022-05-12 thomas NULL, NULL, delta_buf, 0, len);
320 9249e7e3 2022-05-12 thomas goto done;
322 3b6ceab7 2022-01-10 thomas m->delta_len = len;
323 9249e7e3 2022-05-12 thomas m->delta_compressed_len = compressed_len;
325 9249e7e3 2022-05-12 thomas free(delta_buf);
326 9249e7e3 2022-05-12 thomas return err;
329 b79280dd 2021-10-15 thomas static const struct got_error *
330 b79280dd 2021-10-15 thomas encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
331 b79280dd 2021-10-15 thomas struct got_delta_instruction *deltas, int ndeltas,
332 d0772de9 2021-10-15 thomas off_t base_size, FILE *f)
334 9249e7e3 2022-05-12 thomas const struct got_error *err;
335 d0772de9 2021-10-15 thomas unsigned char buf[16], *bp;
336 d0772de9 2021-10-15 thomas int i, j;
338 9249e7e3 2022-05-12 thomas struct got_deflate_buf zb;
339 d0772de9 2021-10-15 thomas struct got_delta_instruction *d;
340 9249e7e3 2022-05-12 thomas off_t delta_len = 0, compressed_len = 0;
342 9249e7e3 2022-05-12 thomas err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
344 9249e7e3 2022-05-12 thomas return err;
346 d0772de9 2021-10-15 thomas /* base object size */
347 d0772de9 2021-10-15 thomas buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
348 d0772de9 2021-10-15 thomas n = base_size >> GOT_DELTA_SIZE_SHIFT;
349 d0772de9 2021-10-15 thomas for (i = 1; n > 0; i++) {
350 d0772de9 2021-10-15 thomas buf[i - 1] |= GOT_DELTA_SIZE_MORE;
351 d0772de9 2021-10-15 thomas buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
352 d0772de9 2021-10-15 thomas n >>= GOT_DELTA_SIZE_SHIFT;
355 9249e7e3 2022-05-12 thomas err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
356 9249e7e3 2022-05-12 thomas buf, 0, i, f, NULL);
358 9249e7e3 2022-05-12 thomas goto done;
359 9249e7e3 2022-05-12 thomas delta_len += i;
361 d0772de9 2021-10-15 thomas /* target object size */
362 d0772de9 2021-10-15 thomas buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
363 d0772de9 2021-10-15 thomas n = o->size >> GOT_DELTA_SIZE_SHIFT;
364 d0772de9 2021-10-15 thomas for (i = 1; n > 0; i++) {
365 d0772de9 2021-10-15 thomas buf[i - 1] |= GOT_DELTA_SIZE_MORE;
366 d0772de9 2021-10-15 thomas buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
367 d0772de9 2021-10-15 thomas n >>= GOT_DELTA_SIZE_SHIFT;
370 9249e7e3 2022-05-12 thomas err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
371 9249e7e3 2022-05-12 thomas buf, 0, i, f, NULL);
373 9249e7e3 2022-05-12 thomas goto done;
374 9249e7e3 2022-05-12 thomas delta_len += i;
376 d0772de9 2021-10-15 thomas for (j = 0; j < ndeltas; j++) {
377 d0772de9 2021-10-15 thomas d = &deltas[j];
378 d0772de9 2021-10-15 thomas if (d->copy) {
379 d0772de9 2021-10-15 thomas n = d->offset;
380 d0772de9 2021-10-15 thomas bp = &buf[1];
381 d0772de9 2021-10-15 thomas buf[0] = GOT_DELTA_BASE_COPY;
382 d0772de9 2021-10-15 thomas for (i = 0; i < 4; i++) {
383 d0772de9 2021-10-15 thomas /* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
384 d0772de9 2021-10-15 thomas buf[0] |= 1 << i;
385 d0772de9 2021-10-15 thomas *bp++ = n & 0xff;
387 d0772de9 2021-10-15 thomas if (n == 0)
390 d0772de9 2021-10-15 thomas n = d->len;
391 d0772de9 2021-10-15 thomas if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
392 d0772de9 2021-10-15 thomas /* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
393 d0772de9 2021-10-15 thomas for (i = 0; i < 3 && n > 0; i++) {
394 d0772de9 2021-10-15 thomas buf[0] |= 1 << (i + 4);
395 d0772de9 2021-10-15 thomas *bp++ = n & 0xff;
399 9249e7e3 2022-05-12 thomas err = got_deflate_append_to_file_mmap(&zb,
400 9249e7e3 2022-05-12 thomas &compressed_len, buf, 0, bp - buf, f, NULL);
402 9249e7e3 2022-05-12 thomas goto done;
403 9249e7e3 2022-05-12 thomas delta_len += (bp - buf);
404 2b0ae357 2022-01-10 thomas } else if (o->f == NULL) {
406 2b0ae357 2022-01-10 thomas while (n != d->len) {
407 2b0ae357 2022-01-10 thomas buf[0] = (d->len - n < 127) ? d->len - n : 127;
408 9249e7e3 2022-05-12 thomas err = got_deflate_append_to_file_mmap(&zb,
409 9249e7e3 2022-05-12 thomas &compressed_len, buf, 0, 1, f, NULL);
411 9249e7e3 2022-05-12 thomas goto done;
412 9249e7e3 2022-05-12 thomas delta_len++;
413 9249e7e3 2022-05-12 thomas err = got_deflate_append_to_file_mmap(&zb,
414 9249e7e3 2022-05-12 thomas &compressed_len,
415 9249e7e3 2022-05-12 thomas o->data + o->hdrlen + d->offset + n, 0,
416 9249e7e3 2022-05-12 thomas buf[0], f, NULL);
418 9249e7e3 2022-05-12 thomas goto done;
419 9249e7e3 2022-05-12 thomas delta_len += buf[0];
420 2b0ae357 2022-01-10 thomas n += buf[0];
423 d0772de9 2021-10-15 thomas char content[128];
424 d0772de9 2021-10-15 thomas size_t r;
425 9249e7e3 2022-05-12 thomas if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
426 9249e7e3 2022-05-12 thomas err = got_error_from_errno("fseeko");
427 9249e7e3 2022-05-12 thomas goto done;
430 d0772de9 2021-10-15 thomas while (n != d->len) {
431 d0772de9 2021-10-15 thomas buf[0] = (d->len - n < 127) ? d->len - n : 127;
432 9249e7e3 2022-05-12 thomas err = got_deflate_append_to_file_mmap(&zb,
433 9249e7e3 2022-05-12 thomas &compressed_len, buf, 0, 1, f, NULL);
435 9249e7e3 2022-05-12 thomas goto done;
436 9249e7e3 2022-05-12 thomas delta_len++;
437 d0772de9 2021-10-15 thomas r = fread(content, 1, buf[0], o->f);
438 9249e7e3 2022-05-12 thomas if (r != buf[0]) {
439 9249e7e3 2022-05-12 thomas err = got_ferror(o->f, GOT_ERR_IO);
440 9249e7e3 2022-05-12 thomas goto done;
442 9249e7e3 2022-05-12 thomas err = got_deflate_append_to_file_mmap(&zb,
443 9249e7e3 2022-05-12 thomas &compressed_len, content, 0, buf[0], f,
446 9249e7e3 2022-05-12 thomas goto done;
447 9249e7e3 2022-05-12 thomas delta_len += buf[0];
448 d0772de9 2021-10-15 thomas n += buf[0];
453 9249e7e3 2022-05-12 thomas err = got_deflate_flush(&zb, f, NULL, &compressed_len);
455 9249e7e3 2022-05-12 thomas goto done;
457 9249e7e3 2022-05-12 thomas /* sanity check */
458 9249e7e3 2022-05-12 thomas if (compressed_len != ftello(f) - m->delta_offset) {
459 9249e7e3 2022-05-12 thomas err = got_error(GOT_ERR_COMPRESSION);
460 9249e7e3 2022-05-12 thomas goto done;
463 9249e7e3 2022-05-12 thomas m->delta_len = delta_len;
464 9249e7e3 2022-05-12 thomas m->delta_compressed_len = compressed_len;
466 9249e7e3 2022-05-12 thomas got_deflate_end(&zb);
467 9249e7e3 2022-05-12 thomas return err;
470 31ba2236 2022-01-05 thomas static const struct got_error *
471 31ba2236 2022-01-05 thomas report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
472 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
473 85220b0e 2022-03-22 thomas off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
474 85220b0e 2022-03-22 thomas int nobj_written)
476 31ba2236 2022-01-05 thomas const struct got_error *err;
477 31ba2236 2022-01-05 thomas int elapsed;
479 31ba2236 2022-01-05 thomas if (progress_cb == NULL)
480 31ba2236 2022-01-05 thomas return NULL;
482 31ba2236 2022-01-05 thomas err = got_ratelimit_check(&elapsed, rl);
483 31ba2236 2022-01-05 thomas if (err || !elapsed)
484 31ba2236 2022-01-05 thomas return err;
486 85220b0e 2022-03-22 thomas return progress_cb(progress_arg, ncolored, nfound, ntrees,
487 85220b0e 2022-03-22 thomas packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
490 e6bcace5 2021-06-22 stsp static const struct got_error *
491 f9c2e8e5 2022-02-13 thomas add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
493 f9c2e8e5 2022-02-13 thomas if (v->nmeta == v->metasz){
494 f9c2e8e5 2022-02-13 thomas size_t newsize = 2 * v->metasz;
495 f9c2e8e5 2022-02-13 thomas struct got_pack_meta **new;
496 f9c2e8e5 2022-02-13 thomas new = reallocarray(v->meta, newsize, sizeof(*new));
497 f9c2e8e5 2022-02-13 thomas if (new == NULL)
498 f9c2e8e5 2022-02-13 thomas return got_error_from_errno("reallocarray");
499 f9c2e8e5 2022-02-13 thomas v->meta = new;
500 f9c2e8e5 2022-02-13 thomas v->metasz = newsize;
503 f9c2e8e5 2022-02-13 thomas v->meta[v->nmeta++] = m;
504 f9c2e8e5 2022-02-13 thomas return NULL;
507 f9c2e8e5 2022-02-13 thomas static const struct got_error *
508 f9c2e8e5 2022-02-13 thomas find_pack_for_reuse(struct got_packidx **best_packidx,
509 f9c2e8e5 2022-02-13 thomas struct got_repository *repo)
511 de47d040 2022-03-22 thomas const struct got_error *err = NULL;
512 f9c2e8e5 2022-02-13 thomas struct got_pathlist_entry *pe;
513 f9c2e8e5 2022-02-13 thomas const char *best_packidx_path = NULL;
514 f9c2e8e5 2022-02-13 thomas int nobj_max = 0;
516 f9c2e8e5 2022-02-13 thomas *best_packidx = NULL;
518 de47d040 2022-03-22 thomas TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
519 f9c2e8e5 2022-02-13 thomas const char *path_packidx = pe->path;
520 f9c2e8e5 2022-02-13 thomas struct got_packidx *packidx;
521 f9c2e8e5 2022-02-13 thomas int nobj;
523 f9c2e8e5 2022-02-13 thomas err = got_repo_get_packidx(&packidx, path_packidx, repo);
527 f9c2e8e5 2022-02-13 thomas nobj = be32toh(packidx->hdr.fanout_table[0xff]);
528 f9c2e8e5 2022-02-13 thomas if (nobj > nobj_max) {
529 f9c2e8e5 2022-02-13 thomas best_packidx_path = path_packidx;
530 f9c2e8e5 2022-02-13 thomas nobj_max = nobj;
534 f9c2e8e5 2022-02-13 thomas if (best_packidx_path) {
535 f9c2e8e5 2022-02-13 thomas err = got_repo_get_packidx(best_packidx, best_packidx_path,
539 f9c2e8e5 2022-02-13 thomas return err;
542 7d0d4920 2022-05-12 thomas struct send_id_arg {
543 7d0d4920 2022-05-12 thomas struct imsgbuf *ibuf;
544 7d0d4920 2022-05-12 thomas struct got_object_id *ids[GOT_IMSG_OBJ_ID_LIST_MAX_NIDS];
545 7d0d4920 2022-05-12 thomas size_t nids;
548 f9c2e8e5 2022-02-13 thomas static const struct got_error *
549 7d0d4920 2022-05-12 thomas send_id(struct got_object_id *id, void *data, void *arg)
551 7d0d4920 2022-05-12 thomas const struct got_error *err = NULL;
552 7d0d4920 2022-05-12 thomas struct send_id_arg *a = arg;
554 7d0d4920 2022-05-12 thomas a->ids[a->nids++] = id;
556 7d0d4920 2022-05-12 thomas if (a->nids >= GOT_IMSG_OBJ_ID_LIST_MAX_NIDS) {
557 7d0d4920 2022-05-12 thomas err = got_privsep_send_object_idlist(a->ibuf, a->ids, a->nids);
559 f9c2e8e5 2022-02-13 thomas return err;
560 7d0d4920 2022-05-12 thomas a->nids = 0;
563 7d0d4920 2022-05-12 thomas return NULL;
566 ec2b23c5 2022-07-01 thomas static const struct got_error *
567 ec2b23c5 2022-07-01 thomas send_idset(struct imsgbuf *ibuf, struct got_object_idset *idset)
569 ec2b23c5 2022-07-01 thomas const struct got_error *err;
570 ec2b23c5 2022-07-01 thomas struct send_id_arg sia;
572 ec2b23c5 2022-07-01 thomas memset(&sia, 0, sizeof(sia));
573 ec2b23c5 2022-07-01 thomas sia.ibuf = ibuf;
574 ec2b23c5 2022-07-01 thomas err = got_object_idset_for_each(idset, send_id, &sia);
576 ec2b23c5 2022-07-01 thomas return err;
578 ec2b23c5 2022-07-01 thomas if (sia.nids > 0) {
579 ec2b23c5 2022-07-01 thomas err = got_privsep_send_object_idlist(ibuf, sia.ids, sia.nids);
581 ec2b23c5 2022-07-01 thomas return err;
584 ec2b23c5 2022-07-01 thomas return got_privsep_send_object_idlist_done(ibuf);
588 7d0d4920 2022-05-12 thomas static const struct got_error *
589 7d0d4920 2022-05-12 thomas recv_reused_delta(struct got_imsg_reused_delta *delta,
590 7d0d4920 2022-05-12 thomas struct got_object_idset *idset, struct got_pack_metavec *v)
592 7d0d4920 2022-05-12 thomas struct got_pack_meta *m, *base;
594 7d0d4920 2022-05-12 thomas if (delta->delta_offset + delta->delta_size < delta->delta_offset ||
595 7d0d4920 2022-05-12 thomas delta->delta_offset +
596 7d0d4920 2022-05-12 thomas delta->delta_compressed_size < delta->delta_offset)
597 7d0d4920 2022-05-12 thomas return got_error(GOT_ERR_BAD_PACKFILE);
599 7d0d4920 2022-05-12 thomas m = got_object_idset_get(idset, &delta->id);
600 7d0d4920 2022-05-12 thomas if (m == NULL)
601 7d0d4920 2022-05-12 thomas return got_error(GOT_ERR_NO_OBJ);
603 7d0d4920 2022-05-12 thomas base = got_object_idset_get(idset, &delta->base_id);
604 7d0d4920 2022-05-12 thomas if (base == NULL)
605 7d0d4920 2022-05-12 thomas return got_error(GOT_ERR_NO_OBJ);
607 7d0d4920 2022-05-12 thomas m->delta_len = delta->delta_size;
608 7d0d4920 2022-05-12 thomas m->delta_compressed_len = delta->delta_compressed_size;
609 7d0d4920 2022-05-12 thomas m->delta_offset = delta->delta_out_offset;
610 7d0d4920 2022-05-12 thomas m->prev = base;
611 7d0d4920 2022-05-12 thomas m->size = delta->result_size;
612 7d0d4920 2022-05-12 thomas m->reused_delta_offset = delta->delta_offset;
613 7d0d4920 2022-05-12 thomas m->base_obj_id = got_object_id_dup(&delta->base_id);
614 7d0d4920 2022-05-12 thomas if (m->base_obj_id == NULL)
615 7d0d4920 2022-05-12 thomas return got_error_from_errno("got_object_id_dup");
617 7d0d4920 2022-05-12 thomas return add_meta(m, v);
620 68036464 2022-06-26 thomas static const struct got_error *
621 ec2b23c5 2022-07-01 thomas cache_pack_for_packidx(struct got_pack **pack, struct got_packidx *packidx,
622 ec2b23c5 2022-07-01 thomas struct got_repository *repo)
624 ec2b23c5 2022-07-01 thomas const struct got_error *err;
625 68036464 2022-06-26 thomas char *path_packfile = NULL;
627 68036464 2022-06-26 thomas err = got_packidx_get_packfile_path(&path_packfile,
628 68036464 2022-06-26 thomas packidx->path_packidx);
630 68036464 2022-06-26 thomas return err;
632 68036464 2022-06-26 thomas *pack = got_repo_get_cached_pack(repo, path_packfile);
633 68036464 2022-06-26 thomas if (*pack == NULL) {
634 68036464 2022-06-26 thomas err = got_repo_cache_pack(pack, repo, path_packfile, packidx);
636 68036464 2022-06-26 thomas goto done;
638 68036464 2022-06-26 thomas if ((*pack)->privsep_child == NULL) {
639 68036464 2022-06-26 thomas err = got_pack_start_privsep_child(*pack, packidx);
641 68036464 2022-06-26 thomas goto done;
644 ec2b23c5 2022-07-01 thomas free(path_packfile);
645 ec2b23c5 2022-07-01 thomas return err;
648 ec2b23c5 2022-07-01 thomas static const struct got_error *
649 ec2b23c5 2022-07-01 thomas prepare_delta_reuse(struct got_pack *pack, struct got_packidx *packidx,
650 ec2b23c5 2022-07-01 thomas int delta_outfd, struct got_repository *repo)
652 ec2b23c5 2022-07-01 thomas const struct got_error *err = NULL;
654 ec2b23c5 2022-07-01 thomas if (!pack->child_has_delta_outfd) {
655 68036464 2022-06-26 thomas int outfd_child;
656 68036464 2022-06-26 thomas outfd_child = dup(delta_outfd);
657 68036464 2022-06-26 thomas if (outfd_child == -1) {
658 68036464 2022-06-26 thomas err = got_error_from_errno("dup");
659 68036464 2022-06-26 thomas goto done;
661 68036464 2022-06-26 thomas err = got_privsep_send_raw_delta_outfd(
662 ec2b23c5 2022-07-01 thomas pack->privsep_child->ibuf, outfd_child);
664 68036464 2022-06-26 thomas goto done;
665 ec2b23c5 2022-07-01 thomas pack->child_has_delta_outfd = 1;
668 ec2b23c5 2022-07-01 thomas err = got_privsep_send_delta_reuse_req(pack->privsep_child->ibuf);
670 68036464 2022-06-26 thomas return err;
674 f9c2e8e5 2022-02-13 thomas static const struct got_error *
675 f9c2e8e5 2022-02-13 thomas search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
676 85220b0e 2022-03-22 thomas int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
677 85220b0e 2022-03-22 thomas struct got_repository *repo,
678 f9c2e8e5 2022-02-13 thomas got_pack_progress_cb progress_cb, void *progress_arg,
679 f9c2e8e5 2022-02-13 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
681 f9c2e8e5 2022-02-13 thomas const struct got_error *err = NULL;
682 f9c2e8e5 2022-02-13 thomas struct got_packidx *packidx;
683 f9c2e8e5 2022-02-13 thomas struct got_pack *pack;
684 7d0d4920 2022-05-12 thomas struct got_imsg_reused_delta deltas[GOT_IMSG_REUSED_DELTAS_MAX_NDELTAS];
685 7d0d4920 2022-05-12 thomas size_t ndeltas, i;
687 f9c2e8e5 2022-02-13 thomas err = find_pack_for_reuse(&packidx, repo);
689 f9c2e8e5 2022-02-13 thomas return err;
691 f9c2e8e5 2022-02-13 thomas if (packidx == NULL)
692 f9c2e8e5 2022-02-13 thomas return NULL;
694 ec2b23c5 2022-07-01 thomas err = cache_pack_for_packidx(&pack, packidx, repo);
696 f9c2e8e5 2022-02-13 thomas return err;
698 ec2b23c5 2022-07-01 thomas err = prepare_delta_reuse(pack, packidx, delta_cache_fd, repo);
700 7d0d4920 2022-05-12 thomas return err;
702 ec2b23c5 2022-07-01 thomas err = send_idset(pack->privsep_child->ibuf, idset);
704 7d0d4920 2022-05-12 thomas return err;
706 7d0d4920 2022-05-12 thomas for (;;) {
707 7d0d4920 2022-05-12 thomas int done = 0;
709 7d0d4920 2022-05-12 thomas if (cancel_cb) {
710 7d0d4920 2022-05-12 thomas err = (*cancel_cb)(cancel_arg);
715 7d0d4920 2022-05-12 thomas err = got_privsep_recv_reused_deltas(&done, deltas, &ndeltas,
716 7d0d4920 2022-05-12 thomas pack->privsep_child->ibuf);
717 7d0d4920 2022-05-12 thomas if (err || done)
720 7d0d4920 2022-05-12 thomas for (i = 0; i < ndeltas; i++) {
721 7d0d4920 2022-05-12 thomas struct got_imsg_reused_delta *delta = &deltas[i];
722 7d0d4920 2022-05-12 thomas err = recv_reused_delta(delta, idset, v);
724 7d0d4920 2022-05-12 thomas goto done;
727 7d0d4920 2022-05-12 thomas err = report_progress(progress_cb, progress_arg, rl,
728 7d0d4920 2022-05-12 thomas ncolored, nfound, ntrees, 0L, ncommits,
729 7d0d4920 2022-05-12 thomas got_object_idset_num_elements(idset), v->nmeta, 0);
734 f9c2e8e5 2022-02-13 thomas return err;
737 f9c2e8e5 2022-02-13 thomas static const struct got_error *
738 85220b0e 2022-03-22 thomas pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
739 85220b0e 2022-03-22 thomas int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
740 85220b0e 2022-03-22 thomas struct got_repository *repo,
741 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
742 31ba2236 2022-01-05 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
744 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
745 e6bcace5 2021-06-22 stsp struct got_pack_meta *m = NULL, *base = NULL;
746 e6bcace5 2021-06-22 stsp struct got_raw_object *raw = NULL, *base_raw = NULL;
747 b79280dd 2021-10-15 thomas struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
748 3b6ceab7 2022-01-10 thomas int i, j, ndeltas, best_ndeltas;
749 3b6ceab7 2022-01-10 thomas off_t size, best_size;
750 959daf23 2021-10-29 thomas const int max_base_candidates = 3;
751 510885f7 2022-01-10 thomas size_t delta_memsize = 0;
752 1569c56d 2022-05-31 thomas const size_t max_delta_memsize = 4 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
753 ecf9545f 2021-10-15 thomas int outfd = -1;
754 cc524d36 2022-05-31 thomas uint32_t delta_seed;
756 cc524d36 2022-05-31 thomas delta_seed = arc4random();
758 e6bcace5 2021-06-22 stsp qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
759 e6bcace5 2021-06-22 stsp for (i = 0; i < nmeta; i++) {
760 e6bcace5 2021-06-22 stsp if (cancel_cb) {
761 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
765 31ba2236 2022-01-05 thomas err = report_progress(progress_cb, progress_arg, rl,
766 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
767 85220b0e 2022-03-22 thomas nreused + i, 0);
769 31ba2236 2022-01-05 thomas goto done;
770 e6bcace5 2021-06-22 stsp m = meta[i];
772 e6bcace5 2021-06-22 stsp if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
773 e6bcace5 2021-06-22 stsp m->obj_type == GOT_OBJ_TYPE_TAG)
776 f8bb1d3e 2021-10-17 thomas err = got_object_raw_open(&raw, &outfd, repo, &m->id);
779 ab6186ae 2021-10-15 thomas m->size = raw->size;
781 2b0ae357 2022-01-10 thomas if (raw->f == NULL) {
782 2b0ae357 2022-01-10 thomas err = got_deltify_init_mem(&m->dtab, raw->data,
783 cc524d36 2022-05-31 thomas raw->hdrlen, raw->size + raw->hdrlen, delta_seed);
785 2b0ae357 2022-01-10 thomas err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
786 cc524d36 2022-05-31 thomas raw->size + raw->hdrlen, delta_seed);
791 e6bcace5 2021-06-22 stsp if (i > max_base_candidates) {
792 e6bcace5 2021-06-22 stsp struct got_pack_meta *n = NULL;
793 e6bcace5 2021-06-22 stsp n = meta[i - (max_base_candidates + 1)];
794 e6bcace5 2021-06-22 stsp got_deltify_free(n->dtab);
795 e6bcace5 2021-06-22 stsp n->dtab = NULL;
798 b79280dd 2021-10-15 thomas best_size = raw->size;
799 b79280dd 2021-10-15 thomas best_ndeltas = 0;
800 e6bcace5 2021-06-22 stsp for (j = MAX(0, i - max_base_candidates); j < i; j++) {
801 e6bcace5 2021-06-22 stsp if (cancel_cb) {
802 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
806 e6bcace5 2021-06-22 stsp base = meta[j];
807 e6bcace5 2021-06-22 stsp /* long chains make unpacking slow, avoid such bases */
808 62a05ae9 2021-10-29 thomas if (base->nchain >= 128 ||
809 e6bcace5 2021-06-22 stsp base->obj_type != m->obj_type)
812 8ab9215c 2021-10-15 thomas err = got_object_raw_open(&base_raw, &outfd, repo,
813 f8bb1d3e 2021-10-17 thomas &base->id);
817 2b0ae357 2022-01-10 thomas if (raw->f == NULL && base_raw->f == NULL) {
818 2b0ae357 2022-01-10 thomas err = got_deltify_mem_mem(&deltas, &ndeltas,
819 2b0ae357 2022-01-10 thomas raw->data, raw->hdrlen,
820 cc524d36 2022-05-31 thomas raw->size + raw->hdrlen, delta_seed,
821 2b0ae357 2022-01-10 thomas base->dtab, base_raw->data,
822 2b0ae357 2022-01-10 thomas base_raw->hdrlen,
823 2b0ae357 2022-01-10 thomas base_raw->size + base_raw->hdrlen);
824 2b0ae357 2022-01-10 thomas } else if (raw->f == NULL) {
825 2b0ae357 2022-01-10 thomas err = got_deltify_mem_file(&deltas, &ndeltas,
826 2b0ae357 2022-01-10 thomas raw->data, raw->hdrlen,
827 cc524d36 2022-05-31 thomas raw->size + raw->hdrlen, delta_seed,
828 2b0ae357 2022-01-10 thomas base->dtab, base_raw->f,
829 2b0ae357 2022-01-10 thomas base_raw->hdrlen,
830 2b0ae357 2022-01-10 thomas base_raw->size + base_raw->hdrlen);
831 2b0ae357 2022-01-10 thomas } else if (base_raw->f == NULL) {
832 2b0ae357 2022-01-10 thomas err = got_deltify_file_mem(&deltas, &ndeltas,
833 2b0ae357 2022-01-10 thomas raw->f, raw->hdrlen,
834 cc524d36 2022-05-31 thomas raw->size + raw->hdrlen, delta_seed,
835 2b0ae357 2022-01-10 thomas base->dtab, base_raw->data,
836 2b0ae357 2022-01-10 thomas base_raw->hdrlen,
837 2b0ae357 2022-01-10 thomas base_raw->size + base_raw->hdrlen);
839 2b0ae357 2022-01-10 thomas err = got_deltify(&deltas, &ndeltas,
840 2b0ae357 2022-01-10 thomas raw->f, raw->hdrlen,
841 cc524d36 2022-05-31 thomas raw->size + raw->hdrlen, delta_seed,
842 2b0ae357 2022-01-10 thomas base->dtab, base_raw->f, base_raw->hdrlen,
843 2b0ae357 2022-01-10 thomas base_raw->size + base_raw->hdrlen);
845 e6bcace5 2021-06-22 stsp got_object_raw_close(base_raw);
846 e6bcace5 2021-06-22 stsp base_raw = NULL;
850 e6bcace5 2021-06-22 stsp size = delta_size(deltas, ndeltas);
851 b79280dd 2021-10-15 thomas if (size + 32 < best_size){
853 e6bcace5 2021-06-22 stsp * if we already picked a best delta,
854 e6bcace5 2021-06-22 stsp * replace it.
856 b79280dd 2021-10-15 thomas best_size = size;
857 b79280dd 2021-10-15 thomas free(best_deltas);
858 b79280dd 2021-10-15 thomas best_deltas = deltas;
859 b79280dd 2021-10-15 thomas best_ndeltas = ndeltas;
860 b79280dd 2021-10-15 thomas deltas = NULL;
861 e6bcace5 2021-06-22 stsp m->nchain = base->nchain + 1;
862 e6bcace5 2021-06-22 stsp m->prev = base;
863 e6bcace5 2021-06-22 stsp m->head = base->head;
864 e6bcace5 2021-06-22 stsp if (m->head == NULL)
865 e6bcace5 2021-06-22 stsp m->head = base;
867 e6bcace5 2021-06-22 stsp free(deltas);
868 e6bcace5 2021-06-22 stsp deltas = NULL;
869 e6bcace5 2021-06-22 stsp ndeltas = 0;
873 b79280dd 2021-10-15 thomas if (best_ndeltas > 0) {
874 510885f7 2022-01-10 thomas if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
875 510885f7 2022-01-10 thomas delta_memsize + best_size <= max_delta_memsize) {
876 510885f7 2022-01-10 thomas delta_memsize += best_size;
877 3b6ceab7 2022-01-10 thomas err = encode_delta_in_mem(m, raw, best_deltas,
878 3b6ceab7 2022-01-10 thomas best_ndeltas, best_size, m->prev->size);
880 3b6ceab7 2022-01-10 thomas m->delta_offset = ftello(delta_cache);
881 3b6ceab7 2022-01-10 thomas err = encode_delta(m, raw, best_deltas,
882 3b6ceab7 2022-01-10 thomas best_ndeltas, m->prev->size, delta_cache);
884 b79280dd 2021-10-15 thomas free(best_deltas);
885 b79280dd 2021-10-15 thomas best_deltas = NULL;
886 b79280dd 2021-10-15 thomas best_ndeltas = 0;
888 b79280dd 2021-10-15 thomas goto done;
891 e6bcace5 2021-06-22 stsp got_object_raw_close(raw);
892 e6bcace5 2021-06-22 stsp raw = NULL;
895 e6bcace5 2021-06-22 stsp for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
896 e6bcace5 2021-06-22 stsp got_deltify_free(meta[i]->dtab);
897 e6bcace5 2021-06-22 stsp meta[i]->dtab = NULL;
900 e6bcace5 2021-06-22 stsp got_object_raw_close(raw);
901 e6bcace5 2021-06-22 stsp if (base_raw)
902 e6bcace5 2021-06-22 stsp got_object_raw_close(base_raw);
903 ecf9545f 2021-10-15 thomas if (outfd != -1 && close(outfd) == -1 && err == NULL)
904 ecf9545f 2021-10-15 thomas err = got_error_from_errno("close");
905 b79280dd 2021-10-15 thomas free(deltas);
906 b79280dd 2021-10-15 thomas free(best_deltas);
907 e6bcace5 2021-06-22 stsp return err;
910 e6bcace5 2021-06-22 stsp static const struct got_error *
911 05118f5a 2021-06-22 stsp search_packidx(int *found, struct got_object_id *id,
912 05118f5a 2021-06-22 stsp struct got_repository *repo)
914 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
915 05118f5a 2021-06-22 stsp struct got_packidx *packidx = NULL;
918 05118f5a 2021-06-22 stsp *found = 0;
920 05118f5a 2021-06-22 stsp err = got_repo_search_packidx(&packidx, &idx, repo, id);
921 05118f5a 2021-06-22 stsp if (err == NULL)
922 05118f5a 2021-06-22 stsp *found = 1; /* object is already packed */
923 05118f5a 2021-06-22 stsp else if (err->code == GOT_ERR_NO_OBJ)
924 05118f5a 2021-06-22 stsp err = NULL;
925 05118f5a 2021-06-22 stsp return err;
928 05118f5a 2021-06-22 stsp static const struct got_error *
929 f9c2e8e5 2022-02-13 thomas add_object(int want_meta, struct got_object_idset *idset,
930 e6bcace5 2021-06-22 stsp struct got_object_id *id, const char *path, int obj_type,
931 cc524d36 2022-05-31 thomas time_t mtime, uint32_t seed, int loose_obj_only,
932 cc524d36 2022-05-31 thomas struct got_repository *repo, int *ncolored, int *nfound, int *ntrees,
933 78a00876 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
934 78a00876 2022-03-22 thomas struct got_ratelimit *rl)
936 e6bcace5 2021-06-22 stsp const struct got_error *err;
937 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *m = NULL;
939 05118f5a 2021-06-22 stsp if (loose_obj_only) {
940 05118f5a 2021-06-22 stsp int is_packed;
941 05118f5a 2021-06-22 stsp err = search_packidx(&is_packed, id, repo);
943 05118f5a 2021-06-22 stsp return err;
944 a605f678 2022-03-22 thomas if (is_packed && want_meta)
945 05118f5a 2021-06-22 stsp return NULL;
948 f9c2e8e5 2022-02-13 thomas if (want_meta) {
949 cc524d36 2022-05-31 thomas err = alloc_meta(&m, id, path, obj_type, mtime, seed);
951 78a00876 2022-03-22 thomas return err;
953 78a00876 2022-03-22 thomas (*nfound)++;
954 78a00876 2022-03-22 thomas err = report_progress(progress_cb, progress_arg, rl,
955 78a00876 2022-03-22 thomas *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
956 ee7b409c 2022-04-16 thomas if (err) {
957 ee7b409c 2022-04-16 thomas clear_meta(m);
959 f9c2e8e5 2022-02-13 thomas return err;
963 ee7b409c 2022-04-16 thomas err = got_object_idset_add(idset, id, m);
964 ee7b409c 2022-04-16 thomas if (err) {
965 ee7b409c 2022-04-16 thomas clear_meta(m);
968 ee7b409c 2022-04-16 thomas return err;
971 e6bcace5 2021-06-22 stsp static const struct got_error *
972 f9c2e8e5 2022-02-13 thomas load_tree_entries(struct got_object_id_queue *ids, int want_meta,
973 7afbdbad 2022-05-12 thomas struct got_object_idset *idset, struct got_object_idset *idset_exclude,
974 63915ee5 2022-06-23 thomas struct got_tree_object *tree,
975 cc524d36 2022-05-31 thomas const char *dpath, time_t mtime, uint32_t seed, struct got_repository *repo,
976 85220b0e 2022-03-22 thomas int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
977 85220b0e 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
978 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
980 e6bcace5 2021-06-22 stsp const struct got_error *err;
981 e6bcace5 2021-06-22 stsp char *p = NULL;
984 85220b0e 2022-03-22 thomas (*ntrees)++;
985 85220b0e 2022-03-22 thomas err = report_progress(progress_cb, progress_arg, rl,
986 85220b0e 2022-03-22 thomas *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
988 e6bcace5 2021-06-22 stsp return err;
990 e6bcace5 2021-06-22 stsp for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
991 e6bcace5 2021-06-22 stsp struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
992 e6bcace5 2021-06-22 stsp struct got_object_id *id = got_tree_entry_get_id(e);
993 e6bcace5 2021-06-22 stsp mode_t mode = got_tree_entry_get_mode(e);
995 e6bcace5 2021-06-22 stsp if (cancel_cb) {
996 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
1001 08511b5e 2021-09-24 thomas if (got_object_tree_entry_is_submodule(e) ||
1002 7afbdbad 2022-05-12 thomas got_object_idset_contains(idset, id) ||
1003 7afbdbad 2022-05-12 thomas got_object_idset_contains(idset_exclude, id))
1004 13280750 2022-06-13 thomas continue;
1007 63915ee5 2022-06-23 thomas * If got-read-pack is crawling trees for us then
1008 63915ee5 2022-06-23 thomas * we are only here to collect blob IDs.
1010 63915ee5 2022-06-23 thomas if (ids == NULL && S_ISDIR(mode))
1011 63915ee5 2022-06-23 thomas continue;
1013 63915ee5 2022-06-23 thomas if (asprintf(&p, "%s%s%s", dpath,
1014 63915ee5 2022-06-23 thomas got_path_is_root_dir(dpath) ? "" : "/",
1015 e6bcace5 2021-06-22 stsp got_tree_entry_get_name(e)) == -1) {
1016 e6bcace5 2021-06-22 stsp err = got_error_from_errno("asprintf");
1020 e6bcace5 2021-06-22 stsp if (S_ISDIR(mode)) {
1021 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
1022 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, id);
1025 eee114b4 2022-05-31 thomas qid->data = p;
1026 eee114b4 2022-05-31 thomas p = NULL;
1027 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(ids, qid, entry);
1028 08511b5e 2021-09-24 thomas } else if (S_ISREG(mode) || S_ISLNK(mode)) {
1029 7afbdbad 2022-05-12 thomas err = add_object(want_meta,
1030 7afbdbad 2022-05-12 thomas want_meta ? idset : idset_exclude, id, p,
1031 cc524d36 2022-05-31 thomas GOT_OBJ_TYPE_BLOB, mtime, seed, loose_obj_only,
1032 cc524d36 2022-05-31 thomas repo, ncolored, nfound, ntrees,
1033 78a00876 2022-03-22 thomas progress_cb, progress_arg, rl);
1034 85220b0e 2022-03-22 thomas if (err)
1036 eee114b4 2022-05-31 thomas free(p);
1037 eee114b4 2022-05-31 thomas p = NULL;
1038 eee114b4 2022-05-31 thomas } else {
1039 eee114b4 2022-05-31 thomas free(p);
1040 eee114b4 2022-05-31 thomas p = NULL;
1045 e6bcace5 2021-06-22 stsp return err;
1048 e6bcace5 2021-06-22 stsp static const struct got_error *
1049 f9c2e8e5 2022-02-13 thomas load_tree(int want_meta, struct got_object_idset *idset,
1050 7afbdbad 2022-05-12 thomas struct got_object_idset *idset_exclude,
1051 e6bcace5 2021-06-22 stsp struct got_object_id *tree_id, const char *dpath, time_t mtime,
1052 cc524d36 2022-05-31 thomas uint32_t seed, struct got_repository *repo, int loose_obj_only,
1053 85220b0e 2022-03-22 thomas int *ncolored, int *nfound, int *ntrees,
1054 85220b0e 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
1055 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1057 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1058 e6bcace5 2021-06-22 stsp struct got_object_id_queue tree_ids;
1059 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
1060 63915ee5 2022-06-23 thomas struct got_tree_object *tree = NULL;
1062 7afbdbad 2022-05-12 thomas if (got_object_idset_contains(idset, tree_id) ||
1063 7afbdbad 2022-05-12 thomas got_object_idset_contains(idset_exclude, tree_id))
1064 e6bcace5 2021-06-22 stsp return NULL;
1066 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, tree_id);
1068 eee114b4 2022-05-31 thomas return err;
1069 eee114b4 2022-05-31 thomas qid->data = strdup(dpath);
1070 eee114b4 2022-05-31 thomas if (qid->data == NULL) {
1071 eee114b4 2022-05-31 thomas err = got_error_from_errno("strdup");
1072 eee114b4 2022-05-31 thomas got_object_qid_free(qid);
1073 e6bcace5 2021-06-22 stsp return err;
1076 dbdddfee 2021-06-23 naddy STAILQ_INIT(&tree_ids);
1077 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
1079 dbdddfee 2021-06-23 naddy while (!STAILQ_EMPTY(&tree_ids)) {
1080 eee114b4 2022-05-31 thomas const char *path;
1081 e6bcace5 2021-06-22 stsp if (cancel_cb) {
1082 e6bcace5 2021-06-22 stsp err = (*cancel_cb)(cancel_arg);
1087 dbdddfee 2021-06-23 naddy qid = STAILQ_FIRST(&tree_ids);
1088 dbdddfee 2021-06-23 naddy STAILQ_REMOVE_HEAD(&tree_ids, entry);
1089 eee114b4 2022-05-31 thomas path = qid->data;
1091 7afbdbad 2022-05-12 thomas if (got_object_idset_contains(idset, &qid->id) ||
1092 7afbdbad 2022-05-12 thomas got_object_idset_contains(idset_exclude, &qid->id)) {
1093 eee114b4 2022-05-31 thomas free(qid->data);
1094 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1098 7afbdbad 2022-05-12 thomas err = add_object(want_meta, want_meta ? idset : idset_exclude,
1099 eee114b4 2022-05-31 thomas &qid->id, path, GOT_OBJ_TYPE_TREE,
1100 cc524d36 2022-05-31 thomas mtime, seed, loose_obj_only, repo,
1101 78a00876 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1102 e6bcace5 2021-06-22 stsp if (err) {
1103 eee114b4 2022-05-31 thomas free(qid->data);
1104 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1108 63915ee5 2022-06-23 thomas err = got_object_open_as_tree(&tree, repo, &qid->id);
1109 63915ee5 2022-06-23 thomas if (err) {
1110 63915ee5 2022-06-23 thomas free(qid->data);
1111 63915ee5 2022-06-23 thomas got_object_qid_free(qid);
1115 7afbdbad 2022-05-12 thomas err = load_tree_entries(&tree_ids, want_meta, idset,
1116 63915ee5 2022-06-23 thomas idset_exclude, tree, path, mtime, seed, repo,
1117 63915ee5 2022-06-23 thomas loose_obj_only, ncolored, nfound, ntrees,
1118 63915ee5 2022-06-23 thomas progress_cb, progress_arg, rl,
1119 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
1120 eee114b4 2022-05-31 thomas free(qid->data);
1121 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1125 63915ee5 2022-06-23 thomas got_object_tree_close(tree);
1126 63915ee5 2022-06-23 thomas tree = NULL;
1129 eee114b4 2022-05-31 thomas STAILQ_FOREACH(qid, &tree_ids, entry)
1130 eee114b4 2022-05-31 thomas free(qid->data);
1131 e6bcace5 2021-06-22 stsp got_object_id_queue_free(&tree_ids);
1132 63915ee5 2022-06-23 thomas if (tree)
1133 63915ee5 2022-06-23 thomas got_object_tree_close(tree);
1134 e6bcace5 2021-06-22 stsp return err;
1137 e6bcace5 2021-06-22 stsp static const struct got_error *
1138 f9c2e8e5 2022-02-13 thomas load_commit(int want_meta, struct got_object_idset *idset,
1139 7afbdbad 2022-05-12 thomas struct got_object_idset *idset_exclude,
1140 cc524d36 2022-05-31 thomas struct got_object_id *id, struct got_repository *repo, uint32_t seed,
1141 cc524d36 2022-05-31 thomas int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
1142 85220b0e 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
1143 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1145 e6bcace5 2021-06-22 stsp const struct got_error *err;
1146 e6bcace5 2021-06-22 stsp struct got_commit_object *commit;
1148 7afbdbad 2022-05-12 thomas if (got_object_idset_contains(idset, id) ||
1149 7afbdbad 2022-05-12 thomas got_object_idset_contains(idset_exclude, id))
1150 e6bcace5 2021-06-22 stsp return NULL;
1152 05118f5a 2021-06-22 stsp if (loose_obj_only) {
1153 05118f5a 2021-06-22 stsp int is_packed;
1154 05118f5a 2021-06-22 stsp err = search_packidx(&is_packed, id, repo);
1156 05118f5a 2021-06-22 stsp return err;
1157 a605f678 2022-03-22 thomas if (is_packed && want_meta)
1158 05118f5a 2021-06-22 stsp return NULL;
1161 e6bcace5 2021-06-22 stsp err = got_object_open_as_commit(&commit, repo, id);
1163 e6bcace5 2021-06-22 stsp return err;
1165 7afbdbad 2022-05-12 thomas err = add_object(want_meta, want_meta ? idset : idset_exclude,
1166 7afbdbad 2022-05-12 thomas id, "", GOT_OBJ_TYPE_COMMIT,
1167 cc524d36 2022-05-31 thomas got_object_commit_get_committer_time(commit), seed,
1168 78a00876 2022-03-22 thomas loose_obj_only, repo,
1169 78a00876 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1171 e6bcace5 2021-06-22 stsp goto done;
1173 7afbdbad 2022-05-12 thomas err = load_tree(want_meta, idset, idset_exclude,
1174 7afbdbad 2022-05-12 thomas got_object_commit_get_tree_id(commit),
1175 cc524d36 2022-05-31 thomas "", got_object_commit_get_committer_time(commit), seed,
1176 85220b0e 2022-03-22 thomas repo, loose_obj_only, ncolored, nfound, ntrees,
1177 85220b0e 2022-03-22 thomas progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1179 e6bcace5 2021-06-22 stsp got_object_commit_close(commit);
1180 e6bcace5 2021-06-22 stsp return err;
1183 05118f5a 2021-06-22 stsp static const struct got_error *
1184 f9c2e8e5 2022-02-13 thomas load_tag(int want_meta, struct got_object_idset *idset,
1185 7afbdbad 2022-05-12 thomas struct got_object_idset *idset_exclude,
1186 cc524d36 2022-05-31 thomas struct got_object_id *id, struct got_repository *repo, uint32_t seed,
1187 cc524d36 2022-05-31 thomas int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
1188 85220b0e 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
1189 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1191 05118f5a 2021-06-22 stsp const struct got_error *err;
1192 05118f5a 2021-06-22 stsp struct got_tag_object *tag = NULL;
1194 7afbdbad 2022-05-12 thomas if (got_object_idset_contains(idset, id) ||
1195 7afbdbad 2022-05-12 thomas got_object_idset_contains(idset_exclude, id))
1196 05118f5a 2021-06-22 stsp return NULL;
1198 05118f5a 2021-06-22 stsp if (loose_obj_only) {
1199 05118f5a 2021-06-22 stsp int is_packed;
1200 05118f5a 2021-06-22 stsp err = search_packidx(&is_packed, id, repo);
1202 05118f5a 2021-06-22 stsp return err;
1203 a605f678 2022-03-22 thomas if (is_packed && want_meta)
1204 05118f5a 2021-06-22 stsp return NULL;
1207 05118f5a 2021-06-22 stsp err = got_object_open_as_tag(&tag, repo, id);
1209 05118f5a 2021-06-22 stsp return err;
1211 7afbdbad 2022-05-12 thomas err = add_object(want_meta, want_meta ? idset : idset_exclude,
1212 7afbdbad 2022-05-12 thomas id, "", GOT_OBJ_TYPE_TAG,
1213 cc524d36 2022-05-31 thomas got_object_tag_get_tagger_time(tag), seed, loose_obj_only, repo,
1214 78a00876 2022-03-22 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1216 05118f5a 2021-06-22 stsp goto done;
1218 05118f5a 2021-06-22 stsp switch (got_object_tag_get_object_type(tag)) {
1219 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_COMMIT:
1220 7afbdbad 2022-05-12 thomas err = load_commit(want_meta, idset, idset_exclude,
1221 cc524d36 2022-05-31 thomas got_object_tag_get_object_id(tag), repo, seed,
1222 cc524d36 2022-05-31 thomas loose_obj_only, ncolored, nfound, ntrees,
1223 cc524d36 2022-05-31 thomas progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1225 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_TREE:
1226 7afbdbad 2022-05-12 thomas err = load_tree(want_meta, idset, idset_exclude,
1227 f9c2e8e5 2022-02-13 thomas got_object_tag_get_object_id(tag), "",
1228 cc524d36 2022-05-31 thomas got_object_tag_get_tagger_time(tag), seed, repo,
1229 cc524d36 2022-05-31 thomas loose_obj_only, ncolored, nfound, ntrees,
1230 cc524d36 2022-05-31 thomas progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1237 05118f5a 2021-06-22 stsp got_object_tag_close(tag);
1238 05118f5a 2021-06-22 stsp return err;
1241 e6bcace5 2021-06-22 stsp enum findtwixt_color {
1242 e6bcace5 2021-06-22 stsp COLOR_KEEP = 0,
1243 e6bcace5 2021-06-22 stsp COLOR_DROP,
1244 5fe7b5c1 2022-04-16 thomas COLOR_SKIP,
1245 ec2b23c5 2022-07-01 thomas COLOR_MAX,
1248 e6bcace5 2021-06-22 stsp static const struct got_error *
1249 ec2b23c5 2022-07-01 thomas paint_commit(struct got_object_qid *qid, intptr_t color)
1251 ec2b23c5 2022-07-01 thomas if (color < 0 || color >= COLOR_MAX)
1252 5fe7b5c1 2022-04-16 thomas return got_error(GOT_ERR_RANGE);
1254 ec2b23c5 2022-07-01 thomas qid->data = (void *)color;
1255 e6bcace5 2021-06-22 stsp return NULL;
1258 e6bcace5 2021-06-22 stsp static const struct got_error *
1259 5fe7b5c1 2022-04-16 thomas queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1260 ec2b23c5 2022-07-01 thomas intptr_t color, struct got_repository *repo)
1262 5fe7b5c1 2022-04-16 thomas const struct got_error *err;
1263 e6bcace5 2021-06-22 stsp struct got_object_qid *qid;
1265 e6bcace5 2021-06-22 stsp err = got_object_qid_alloc(&qid, id);
1267 e6bcace5 2021-06-22 stsp return err;
1269 5fe7b5c1 2022-04-16 thomas STAILQ_INSERT_TAIL(ids, qid, entry);
1270 5fe7b5c1 2022-04-16 thomas return paint_commit(qid, color);
1273 e6bcace5 2021-06-22 stsp struct append_id_arg {
1274 e6bcace5 2021-06-22 stsp struct got_object_id **array;
1276 5fe7b5c1 2022-04-16 thomas struct got_object_idset *drop;
1277 5fe7b5c1 2022-04-16 thomas struct got_object_idset *skip;
1280 e6bcace5 2021-06-22 stsp static const struct got_error *
1281 e6bcace5 2021-06-22 stsp append_id(struct got_object_id *id, void *data, void *arg)
1283 e6bcace5 2021-06-22 stsp struct append_id_arg *a = arg;
1285 5fe7b5c1 2022-04-16 thomas if (got_object_idset_contains(a->skip, id) ||
1286 5fe7b5c1 2022-04-16 thomas got_object_idset_contains(a->drop, id))
1287 5fe7b5c1 2022-04-16 thomas return NULL;
1289 5fe7b5c1 2022-04-16 thomas a->array[++a->idx] = got_object_id_dup(id);
1290 e6bcace5 2021-06-22 stsp if (a->array[a->idx] == NULL)
1291 e6bcace5 2021-06-22 stsp return got_error_from_errno("got_object_id_dup");
1293 e6bcace5 2021-06-22 stsp return NULL;
1296 a3a0e34e 2022-04-16 thomas static const struct got_error *
1297 ec2b23c5 2022-07-01 thomas queue_commit_or_tag_id(struct got_object_id *id, intptr_t color,
1298 a3a0e34e 2022-04-16 thomas struct got_object_id_queue *ids, struct got_repository *repo)
1300 a3a0e34e 2022-04-16 thomas const struct got_error *err;
1301 a3a0e34e 2022-04-16 thomas struct got_tag_object *tag = NULL;
1302 a3a0e34e 2022-04-16 thomas int obj_type;
1304 a3a0e34e 2022-04-16 thomas err = got_object_get_type(&obj_type, repo, id);
1305 a3a0e34e 2022-04-16 thomas if (err)
1306 a3a0e34e 2022-04-16 thomas return err;
1308 a3a0e34e 2022-04-16 thomas if (obj_type == GOT_OBJ_TYPE_TAG) {
1309 a3a0e34e 2022-04-16 thomas err = got_object_open_as_tag(&tag, repo, id);
1310 a3a0e34e 2022-04-16 thomas if (err)
1311 a3a0e34e 2022-04-16 thomas return err;
1312 a3a0e34e 2022-04-16 thomas obj_type = got_object_tag_get_object_type(tag);
1313 a3a0e34e 2022-04-16 thomas id = got_object_tag_get_object_id(tag);
1316 a3a0e34e 2022-04-16 thomas if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1317 a3a0e34e 2022-04-16 thomas err = queue_commit_id(ids, id, color, repo);
1318 a3a0e34e 2022-04-16 thomas if (err)
1319 a3a0e34e 2022-04-16 thomas goto done;
1322 a3a0e34e 2022-04-16 thomas if (tag)
1323 a3a0e34e 2022-04-16 thomas got_object_tag_close(tag);
1324 a3a0e34e 2022-04-16 thomas return err;
1327 ec2b23c5 2022-07-01 thomas struct recv_painted_commit_arg {
1328 ec2b23c5 2022-07-01 thomas int *ncolored;
1329 ec2b23c5 2022-07-01 thomas int *nqueued;
1330 ec2b23c5 2022-07-01 thomas int *nskip;
1331 ec2b23c5 2022-07-01 thomas struct got_object_id_queue *ids;
1332 ec2b23c5 2022-07-01 thomas struct got_object_idset *keep;
1333 ec2b23c5 2022-07-01 thomas struct got_object_idset *drop;
1334 ec2b23c5 2022-07-01 thomas struct got_object_idset *skip;
1335 ec2b23c5 2022-07-01 thomas got_pack_progress_cb progress_cb;
1336 ec2b23c5 2022-07-01 thomas void *progress_arg;
1337 ec2b23c5 2022-07-01 thomas struct got_ratelimit *rl;
1338 ec2b23c5 2022-07-01 thomas got_cancel_cb cancel_cb;
1339 ec2b23c5 2022-07-01 thomas void *cancel_arg;
1342 e6bcace5 2021-06-22 stsp static const struct got_error *
1343 ec2b23c5 2022-07-01 thomas recv_painted_commit(void *arg, struct got_object_id *id, intptr_t color)
1345 ec2b23c5 2022-07-01 thomas const struct got_error *err = NULL;
1346 ec2b23c5 2022-07-01 thomas struct recv_painted_commit_arg *a = arg;
1347 ec2b23c5 2022-07-01 thomas struct got_object_qid *qid, *tmp;
1349 ec2b23c5 2022-07-01 thomas if (a->cancel_cb) {
1350 ec2b23c5 2022-07-01 thomas err = a->cancel_cb(a->cancel_arg);
1351 ec2b23c5 2022-07-01 thomas if (err)
1352 ec2b23c5 2022-07-01 thomas return err;
1355 ec2b23c5 2022-07-01 thomas switch (color) {
1356 ec2b23c5 2022-07-01 thomas case COLOR_KEEP:
1357 ec2b23c5 2022-07-01 thomas err = got_object_idset_add(a->keep, id, NULL);
1358 ec2b23c5 2022-07-01 thomas if (err)
1359 ec2b23c5 2022-07-01 thomas return err;
1360 ec2b23c5 2022-07-01 thomas (*a->ncolored)++;
1362 ec2b23c5 2022-07-01 thomas case COLOR_DROP:
1363 ec2b23c5 2022-07-01 thomas err = got_object_idset_add(a->drop, id, NULL);
1364 ec2b23c5 2022-07-01 thomas if (err)
1365 ec2b23c5 2022-07-01 thomas return err;
1366 ec2b23c5 2022-07-01 thomas (*a->ncolored)++;
1368 ec2b23c5 2022-07-01 thomas case COLOR_SKIP:
1369 ec2b23c5 2022-07-01 thomas err = got_object_idset_add(a->skip, id, NULL);
1370 ec2b23c5 2022-07-01 thomas if (err)
1371 ec2b23c5 2022-07-01 thomas return err;
1373 ec2b23c5 2022-07-01 thomas default:
1374 ec2b23c5 2022-07-01 thomas /* should not happen */
1375 ec2b23c5 2022-07-01 thomas return got_error_fmt(GOT_ERR_NOT_IMPL,
1376 60b94e7d 2022-07-19 thomas "%s invalid commit color %"PRIdPTR, __func__, color);
1379 ec2b23c5 2022-07-01 thomas STAILQ_FOREACH_SAFE(qid, a->ids, entry, tmp) {
1380 ec2b23c5 2022-07-01 thomas if (got_object_id_cmp(&qid->id, id) != 0)
1381 ec2b23c5 2022-07-01 thomas continue;
1382 ec2b23c5 2022-07-01 thomas STAILQ_REMOVE(a->ids, qid, got_object_qid, entry);
1383 ec2b23c5 2022-07-01 thomas color = (intptr_t)qid->data;
1384 ec2b23c5 2022-07-01 thomas got_object_qid_free(qid);
1385 ec2b23c5 2022-07-01 thomas (*a->nqueued)--;
1386 ec2b23c5 2022-07-01 thomas if (color == COLOR_SKIP)
1387 ec2b23c5 2022-07-01 thomas (*a->nskip)--;
1391 ec2b23c5 2022-07-01 thomas return report_progress(a->progress_cb, a->progress_arg, a->rl,
1392 ec2b23c5 2022-07-01 thomas *a->ncolored, 0, 0, 0L, 0, 0, 0, 0);
1395 ec2b23c5 2022-07-01 thomas static const struct got_error *
1396 ec2b23c5 2022-07-01 thomas paint_packed_commits(struct got_pack *pack, struct got_object_id *id,
1397 ec2b23c5 2022-07-01 thomas int idx, intptr_t color, int *ncolored, int *nqueued, int *nskip,
1398 ec2b23c5 2022-07-01 thomas struct got_object_id_queue *ids,
1399 ec2b23c5 2022-07-01 thomas struct got_object_idset *keep, struct got_object_idset *drop,
1400 ec2b23c5 2022-07-01 thomas struct got_object_idset *skip, struct got_repository *repo,
1401 ec2b23c5 2022-07-01 thomas got_pack_progress_cb progress_cb, void *progress_arg,
1402 ec2b23c5 2022-07-01 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1404 ec2b23c5 2022-07-01 thomas const struct got_error *err = NULL;
1405 ec2b23c5 2022-07-01 thomas struct got_object_id_queue next_ids;
1406 ec2b23c5 2022-07-01 thomas struct got_object_qid *qid, *tmp;
1407 ec2b23c5 2022-07-01 thomas struct recv_painted_commit_arg arg;
1409 ec2b23c5 2022-07-01 thomas STAILQ_INIT(&next_ids);
1411 ec2b23c5 2022-07-01 thomas err = got_privsep_send_painting_request(pack->privsep_child->ibuf,
1412 ec2b23c5 2022-07-01 thomas idx, id, color);
1413 ec2b23c5 2022-07-01 thomas if (err)
1414 ec2b23c5 2022-07-01 thomas return err;
1416 ec2b23c5 2022-07-01 thomas arg.ncolored = ncolored;
1417 ec2b23c5 2022-07-01 thomas arg.nqueued = nqueued;
1418 ec2b23c5 2022-07-01 thomas arg.nskip = nskip;
1419 ec2b23c5 2022-07-01 thomas arg.ids = ids;
1420 ec2b23c5 2022-07-01 thomas arg.keep = keep;
1421 ec2b23c5 2022-07-01 thomas arg.drop = drop;
1422 ec2b23c5 2022-07-01 thomas arg.skip = skip;
1423 ec2b23c5 2022-07-01 thomas arg.progress_cb = progress_cb;
1424 ec2b23c5 2022-07-01 thomas arg.progress_arg = progress_arg;
1425 ec2b23c5 2022-07-01 thomas arg.rl = rl;
1426 ec2b23c5 2022-07-01 thomas arg.cancel_cb = cancel_cb;
1427 ec2b23c5 2022-07-01 thomas arg.cancel_arg = cancel_arg;
1428 ec2b23c5 2022-07-01 thomas err = got_privsep_recv_painted_commits(&next_ids,
1429 ec2b23c5 2022-07-01 thomas recv_painted_commit, &arg, pack->privsep_child->ibuf);
1430 ec2b23c5 2022-07-01 thomas if (err)
1431 ec2b23c5 2022-07-01 thomas return err;
1433 ec2b23c5 2022-07-01 thomas STAILQ_FOREACH_SAFE(qid, &next_ids, entry, tmp) {
1434 ec2b23c5 2022-07-01 thomas struct got_object_qid *old_id;
1435 ec2b23c5 2022-07-01 thomas intptr_t qcolor, ocolor;
1436 ec2b23c5 2022-07-01 thomas STAILQ_FOREACH(old_id, ids, entry) {
1437 ec2b23c5 2022-07-01 thomas if (got_object_id_cmp(&qid->id, &old_id->id))
1438 ec2b23c5 2022-07-01 thomas continue;
1439 ec2b23c5 2022-07-01 thomas qcolor = (intptr_t)qid->data;
1440 ec2b23c5 2022-07-01 thomas ocolor = (intptr_t)old_id->data;
1441 ec2b23c5 2022-07-01 thomas STAILQ_REMOVE(&next_ids, qid, got_object_qid, entry);
1442 ec2b23c5 2022-07-01 thomas got_object_qid_free(qid);
1443 ec2b23c5 2022-07-01 thomas qid = NULL;
1444 ec2b23c5 2022-07-01 thomas if (qcolor != ocolor) {
1445 ec2b23c5 2022-07-01 thomas paint_commit(old_id, qcolor);
1446 ec2b23c5 2022-07-01 thomas if (ocolor == COLOR_SKIP)
1447 ec2b23c5 2022-07-01 thomas (*nskip)--;
1448 ec2b23c5 2022-07-01 thomas else if (qcolor == COLOR_SKIP)
1449 ec2b23c5 2022-07-01 thomas (*nskip)++;
1454 ec2b23c5 2022-07-01 thomas while (!STAILQ_EMPTY(&next_ids)) {
1455 ec2b23c5 2022-07-01 thomas qid = STAILQ_FIRST(&next_ids);
1456 ec2b23c5 2022-07-01 thomas STAILQ_REMOVE_HEAD(&next_ids, entry);
1457 ec2b23c5 2022-07-01 thomas paint_commit(qid, color);
1458 ec2b23c5 2022-07-01 thomas STAILQ_INSERT_TAIL(ids, qid, entry);
1459 ec2b23c5 2022-07-01 thomas (*nqueued)++;
1460 ec2b23c5 2022-07-01 thomas if (color == COLOR_SKIP)
1461 ec2b23c5 2022-07-01 thomas (*nskip)++;
1464 ec2b23c5 2022-07-01 thomas return err;
1467 ec2b23c5 2022-07-01 thomas static const struct got_error *
1468 ec2b23c5 2022-07-01 thomas find_pack_for_commit_painting(struct got_packidx **best_packidx,
1469 ec2b23c5 2022-07-01 thomas struct got_object_id_queue *ids, int nids, struct got_repository *repo)
1471 ec2b23c5 2022-07-01 thomas const struct got_error *err = NULL;
1472 ec2b23c5 2022-07-01 thomas struct got_pathlist_entry *pe;
1473 ec2b23c5 2022-07-01 thomas const char *best_packidx_path = NULL;
1474 ec2b23c5 2022-07-01 thomas int nobj_max = 0;
1475 ec2b23c5 2022-07-01 thomas int ncommits_max = 0;
1477 ec2b23c5 2022-07-01 thomas *best_packidx = NULL;
1480 ec2b23c5 2022-07-01 thomas * Find the largest pack which contains at least some of the
1481 ec2b23c5 2022-07-01 thomas * commits we are interested in.
1483 ec2b23c5 2022-07-01 thomas TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1484 ec2b23c5 2022-07-01 thomas const char *path_packidx = pe->path;
1485 ec2b23c5 2022-07-01 thomas struct got_packidx *packidx;
1486 ec2b23c5 2022-07-01 thomas int nobj, idx, ncommits = 0;
1487 ec2b23c5 2022-07-01 thomas struct got_object_qid *qid;
1489 ec2b23c5 2022-07-01 thomas err = got_repo_get_packidx(&packidx, path_packidx, repo);
1490 ec2b23c5 2022-07-01 thomas if (err)
1493 ec2b23c5 2022-07-01 thomas nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1494 ec2b23c5 2022-07-01 thomas if (nobj <= nobj_max)
1495 ec2b23c5 2022-07-01 thomas continue;
1497 ec2b23c5 2022-07-01 thomas STAILQ_FOREACH(qid, ids, entry) {
1498 ec2b23c5 2022-07-01 thomas idx = got_packidx_get_object_idx(packidx, &qid->id);
1499 ec2b23c5 2022-07-01 thomas if (idx != -1)
1500 ec2b23c5 2022-07-01 thomas ncommits++;
1502 ec2b23c5 2022-07-01 thomas if (ncommits > ncommits_max) {
1503 ec2b23c5 2022-07-01 thomas best_packidx_path = path_packidx;
1504 ec2b23c5 2022-07-01 thomas nobj_max = nobj;
1505 ec2b23c5 2022-07-01 thomas ncommits_max = ncommits;
1509 ec2b23c5 2022-07-01 thomas if (best_packidx_path && err == NULL) {
1510 ec2b23c5 2022-07-01 thomas err = got_repo_get_packidx(best_packidx, best_packidx_path,
1514 ec2b23c5 2022-07-01 thomas return err;
1517 ec2b23c5 2022-07-01 thomas static const struct got_error *
1518 5fe7b5c1 2022-04-16 thomas paint_commits(int *ncolored, struct got_object_id_queue *ids, int nids,
1519 e9371be6 2022-04-16 thomas struct got_object_idset *keep, struct got_object_idset *drop,
1520 5fe7b5c1 2022-04-16 thomas struct got_object_idset *skip, struct got_repository *repo,
1521 85220b0e 2022-03-22 thomas got_pack_progress_cb progress_cb, void *progress_arg,
1522 85220b0e 2022-03-22 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1524 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1525 e9371be6 2022-04-16 thomas struct got_commit_object *commit = NULL;
1526 ec2b23c5 2022-07-01 thomas struct got_packidx *packidx = NULL;
1527 ec2b23c5 2022-07-01 thomas struct got_pack *pack = NULL;
1528 5fe7b5c1 2022-04-16 thomas const struct got_object_id_queue *parents;
1529 ec2b23c5 2022-07-01 thomas struct got_object_qid *qid = NULL;
1530 5fe7b5c1 2022-04-16 thomas int nqueued = nids, nskip = 0;
1531 ec2b23c5 2022-07-01 thomas int idx;
1533 5fe7b5c1 2022-04-16 thomas while (!STAILQ_EMPTY(ids) && nskip != nqueued) {
1534 ec2b23c5 2022-07-01 thomas intptr_t color;
1536 21a0da85 2022-04-16 thomas if (cancel_cb) {
1537 21a0da85 2022-04-16 thomas err = cancel_cb(cancel_arg);
1538 21a0da85 2022-04-16 thomas if (err)
1542 e9371be6 2022-04-16 thomas qid = STAILQ_FIRST(ids);
1543 5fe7b5c1 2022-04-16 thomas STAILQ_REMOVE_HEAD(ids, entry);
1544 5fe7b5c1 2022-04-16 thomas nqueued--;
1545 ec2b23c5 2022-07-01 thomas color = (intptr_t)qid->data;
1546 5fe7b5c1 2022-04-16 thomas if (color == COLOR_SKIP)
1547 5fe7b5c1 2022-04-16 thomas nskip--;
1549 ec242592 2022-04-22 thomas if (got_object_idset_contains(skip, &qid->id)) {
1550 5fe7b5c1 2022-04-16 thomas got_object_qid_free(qid);
1551 ec2b23c5 2022-07-01 thomas qid = NULL;
1552 5fe7b5c1 2022-04-16 thomas continue;
1554 ec2b23c5 2022-07-01 thomas if (color == COLOR_KEEP &&
1555 ec2b23c5 2022-07-01 thomas got_object_idset_contains(keep, &qid->id)) {
1556 ec2b23c5 2022-07-01 thomas got_object_qid_free(qid);
1557 ec2b23c5 2022-07-01 thomas qid = NULL;
1558 ec2b23c5 2022-07-01 thomas continue;
1560 ec2b23c5 2022-07-01 thomas if (color == COLOR_DROP &&
1561 ec2b23c5 2022-07-01 thomas got_object_idset_contains(drop, &qid->id)) {
1562 ec2b23c5 2022-07-01 thomas got_object_qid_free(qid);
1563 ec2b23c5 2022-07-01 thomas qid = NULL;
1564 ec2b23c5 2022-07-01 thomas continue;
1567 ec2b23c5 2022-07-01 thomas /* Pinned pack may have moved to different cache slot. */
1568 ec2b23c5 2022-07-01 thomas pack = got_repo_get_pinned_pack(repo);
1570 ec2b23c5 2022-07-01 thomas if (packidx && pack) {
1571 ec2b23c5 2022-07-01 thomas idx = got_packidx_get_object_idx(packidx, &qid->id);
1572 ec2b23c5 2022-07-01 thomas if (idx != -1) {
1573 ec2b23c5 2022-07-01 thomas err = paint_packed_commits(pack, &qid->id,
1574 ec2b23c5 2022-07-01 thomas idx, color, ncolored, &nqueued, &nskip,
1575 ec2b23c5 2022-07-01 thomas ids, keep, drop, skip, repo,
1576 ec2b23c5 2022-07-01 thomas progress_cb, progress_arg, rl,
1577 ec2b23c5 2022-07-01 thomas cancel_cb, cancel_arg);
1578 ec2b23c5 2022-07-01 thomas if (err)
1580 5fe7b5c1 2022-04-16 thomas got_object_qid_free(qid);
1581 ec2b23c5 2022-07-01 thomas qid = NULL;
1582 5fe7b5c1 2022-04-16 thomas continue;
1586 ec2b23c5 2022-07-01 thomas switch (color) {
1587 ec2b23c5 2022-07-01 thomas case COLOR_KEEP:
1588 ec242592 2022-04-22 thomas if (got_object_idset_contains(drop, &qid->id)) {
1589 5fe7b5c1 2022-04-16 thomas err = paint_commit(qid, COLOR_SKIP);
1590 5fe7b5c1 2022-04-16 thomas if (err)
1591 5fe7b5c1 2022-04-16 thomas goto done;
1593 5fe7b5c1 2022-04-16 thomas (*ncolored)++;
1594 ec242592 2022-04-22 thomas err = got_object_idset_add(keep, &qid->id, NULL);
1595 5fe7b5c1 2022-04-16 thomas if (err)
1596 5fe7b5c1 2022-04-16 thomas goto done;
1598 5fe7b5c1 2022-04-16 thomas case COLOR_DROP:
1599 ec242592 2022-04-22 thomas if (got_object_idset_contains(keep, &qid->id)) {
1600 5fe7b5c1 2022-04-16 thomas err = paint_commit(qid, COLOR_SKIP);
1601 5fe7b5c1 2022-04-16 thomas if (err)
1602 5fe7b5c1 2022-04-16 thomas goto done;
1604 5fe7b5c1 2022-04-16 thomas (*ncolored)++;
1605 ec242592 2022-04-22 thomas err = got_object_idset_add(drop, &qid->id, NULL);
1606 5fe7b5c1 2022-04-16 thomas if (err)
1607 5fe7b5c1 2022-04-16 thomas goto done;
1609 5fe7b5c1 2022-04-16 thomas case COLOR_SKIP:
1610 ec242592 2022-04-22 thomas if (!got_object_idset_contains(skip, &qid->id)) {
1611 ec242592 2022-04-22 thomas err = got_object_idset_add(skip, &qid->id,
1613 5fe7b5c1 2022-04-16 thomas if (err)
1614 5fe7b5c1 2022-04-16 thomas goto done;
1617 5fe7b5c1 2022-04-16 thomas default:
1618 5fe7b5c1 2022-04-16 thomas /* should not happen */
1619 5fe7b5c1 2022-04-16 thomas err = got_error_fmt(GOT_ERR_NOT_IMPL,
1620 60b94e7d 2022-07-19 thomas "%s invalid commit color %"PRIdPTR, __func__,
1622 5fe7b5c1 2022-04-16 thomas goto done;
1625 85220b0e 2022-03-22 thomas err = report_progress(progress_cb, progress_arg, rl,
1626 85220b0e 2022-03-22 thomas *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1627 85220b0e 2022-03-22 thomas if (err)
1630 ec242592 2022-04-22 thomas err = got_object_open_as_commit(&commit, repo, &qid->id);
1631 5fe7b5c1 2022-04-16 thomas if (err)
1634 5fe7b5c1 2022-04-16 thomas parents = got_object_commit_get_parent_ids(commit);
1635 5fe7b5c1 2022-04-16 thomas if (parents) {
1636 5fe7b5c1 2022-04-16 thomas struct got_object_qid *pid;
1637 ec2b23c5 2022-07-01 thomas color = (intptr_t)qid->data;
1638 5fe7b5c1 2022-04-16 thomas STAILQ_FOREACH(pid, parents, entry) {
1639 ec2b23c5 2022-07-01 thomas err = queue_commit_id(ids, &pid->id,
1640 ec2b23c5 2022-07-01 thomas color, repo);
1641 5fe7b5c1 2022-04-16 thomas if (err)
1643 5fe7b5c1 2022-04-16 thomas nqueued++;
1644 5fe7b5c1 2022-04-16 thomas if (color == COLOR_SKIP)
1645 5fe7b5c1 2022-04-16 thomas nskip++;
1649 ec2b23c5 2022-07-01 thomas if (pack == NULL && (commit->flags & GOT_COMMIT_FLAG_PACKED)) {
1650 ec2b23c5 2022-07-01 thomas if (packidx == NULL) {
1651 ec2b23c5 2022-07-01 thomas err = find_pack_for_commit_painting(&packidx,
1652 ec2b23c5 2022-07-01 thomas ids, nqueued, repo);
1653 ec2b23c5 2022-07-01 thomas if (err)
1654 ec2b23c5 2022-07-01 thomas goto done;
1656 ec2b23c5 2022-07-01 thomas if (packidx != NULL) {
1657 ec2b23c5 2022-07-01 thomas err = cache_pack_for_packidx(&pack, packidx,
1659 ec2b23c5 2022-07-01 thomas if (err)
1660 ec2b23c5 2022-07-01 thomas goto done;
1661 ec2b23c5 2022-07-01 thomas err = got_privsep_init_commit_painting(
1662 ec2b23c5 2022-07-01 thomas pack->privsep_child->ibuf);
1663 ec2b23c5 2022-07-01 thomas if (err)
1664 ec2b23c5 2022-07-01 thomas goto done;
1665 ec2b23c5 2022-07-01 thomas err = send_idset(pack->privsep_child->ibuf,
1667 ec2b23c5 2022-07-01 thomas if (err)
1668 ec2b23c5 2022-07-01 thomas goto done;
1669 ec2b23c5 2022-07-01 thomas err = send_idset(pack->privsep_child->ibuf, drop);
1670 ec2b23c5 2022-07-01 thomas if (err)
1671 ec2b23c5 2022-07-01 thomas goto done;
1672 ec2b23c5 2022-07-01 thomas err = send_idset(pack->privsep_child->ibuf, skip);
1673 ec2b23c5 2022-07-01 thomas if (err)
1674 ec2b23c5 2022-07-01 thomas goto done;
1675 ec2b23c5 2022-07-01 thomas err = got_repo_pin_pack(repo, packidx, pack);
1676 ec2b23c5 2022-07-01 thomas if (err)
1677 ec2b23c5 2022-07-01 thomas goto done;
1681 5fe7b5c1 2022-04-16 thomas got_object_commit_close(commit);
1682 5fe7b5c1 2022-04-16 thomas commit = NULL;
1684 e6bcace5 2021-06-22 stsp got_object_qid_free(qid);
1685 ec2b23c5 2022-07-01 thomas qid = NULL;
1688 ec2b23c5 2022-07-01 thomas if (pack) {
1689 ec2b23c5 2022-07-01 thomas const struct got_error *pack_err;
1690 ec2b23c5 2022-07-01 thomas pack_err = got_privsep_send_painting_commits_done(
1691 ec2b23c5 2022-07-01 thomas pack->privsep_child->ibuf);
1692 ec2b23c5 2022-07-01 thomas if (err == NULL)
1693 ec2b23c5 2022-07-01 thomas err = pack_err;
1695 e9371be6 2022-04-16 thomas if (commit)
1696 e9371be6 2022-04-16 thomas got_object_commit_close(commit);
1697 ec2b23c5 2022-07-01 thomas got_object_qid_free(qid);
1698 ec2b23c5 2022-07-01 thomas got_repo_unpin_pack(repo);
1699 e9371be6 2022-04-16 thomas return err;
1702 e9371be6 2022-04-16 thomas static const struct got_error *
1703 e9371be6 2022-04-16 thomas findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1704 e9371be6 2022-04-16 thomas struct got_object_id **head, int nhead,
1705 e9371be6 2022-04-16 thomas struct got_object_id **tail, int ntail,
1706 e9371be6 2022-04-16 thomas struct got_repository *repo,
1707 e9371be6 2022-04-16 thomas got_pack_progress_cb progress_cb, void *progress_arg,
1708 e9371be6 2022-04-16 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1710 e9371be6 2022-04-16 thomas const struct got_error *err = NULL;
1711 e9371be6 2022-04-16 thomas struct got_object_id_queue ids;
1712 5fe7b5c1 2022-04-16 thomas struct got_object_idset *keep, *drop, *skip = NULL;
1713 e9371be6 2022-04-16 thomas int i, nkeep;
1715 e9371be6 2022-04-16 thomas STAILQ_INIT(&ids);
1716 e9371be6 2022-04-16 thomas *res = NULL;
1717 e9371be6 2022-04-16 thomas *nres = 0;
1718 e9371be6 2022-04-16 thomas *ncolored = 0;
1720 e9371be6 2022-04-16 thomas keep = got_object_idset_alloc();
1721 e9371be6 2022-04-16 thomas if (keep == NULL)
1722 e9371be6 2022-04-16 thomas return got_error_from_errno("got_object_idset_alloc");
1724 e9371be6 2022-04-16 thomas drop = got_object_idset_alloc();
1725 e9371be6 2022-04-16 thomas if (drop == NULL) {
1726 e9371be6 2022-04-16 thomas err = got_error_from_errno("got_object_idset_alloc");
1727 e9371be6 2022-04-16 thomas goto done;
1730 5fe7b5c1 2022-04-16 thomas skip = got_object_idset_alloc();
1731 5fe7b5c1 2022-04-16 thomas if (skip == NULL) {
1732 5fe7b5c1 2022-04-16 thomas err = got_error_from_errno("got_object_idset_alloc");
1733 5fe7b5c1 2022-04-16 thomas goto done;
1736 e9371be6 2022-04-16 thomas for (i = 0; i < nhead; i++) {
1737 e9371be6 2022-04-16 thomas struct got_object_id *id = head[i];
1738 e9371be6 2022-04-16 thomas if (id == NULL)
1739 e9371be6 2022-04-16 thomas continue;
1740 fc457d24 2022-04-16 thomas err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1741 e9371be6 2022-04-16 thomas if (err)
1742 e9371be6 2022-04-16 thomas goto done;
1745 e9371be6 2022-04-16 thomas for (i = 0; i < ntail; i++) {
1746 e9371be6 2022-04-16 thomas struct got_object_id *id = tail[i];
1747 e9371be6 2022-04-16 thomas if (id == NULL)
1748 e9371be6 2022-04-16 thomas continue;
1749 e9371be6 2022-04-16 thomas err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1750 e9371be6 2022-04-16 thomas if (err)
1751 e9371be6 2022-04-16 thomas goto done;
1754 5fe7b5c1 2022-04-16 thomas err = paint_commits(ncolored, &ids, nhead + ntail,
1755 5fe7b5c1 2022-04-16 thomas keep, drop, skip, repo, progress_cb, progress_arg, rl,
1756 5fe7b5c1 2022-04-16 thomas cancel_cb, cancel_arg);
1757 e9371be6 2022-04-16 thomas if (err)
1758 e9371be6 2022-04-16 thomas goto done;
1760 e6bcace5 2021-06-22 stsp nkeep = got_object_idset_num_elements(keep);
1761 e6bcace5 2021-06-22 stsp if (nkeep > 0) {
1762 e6bcace5 2021-06-22 stsp struct append_id_arg arg;
1763 e6bcace5 2021-06-22 stsp arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1764 e6bcace5 2021-06-22 stsp if (arg.array == NULL) {
1765 e6bcace5 2021-06-22 stsp err = got_error_from_errno("calloc");
1766 e6bcace5 2021-06-22 stsp goto done;
1768 5fe7b5c1 2022-04-16 thomas arg.idx = -1;
1769 5fe7b5c1 2022-04-16 thomas arg.skip = skip;
1770 5fe7b5c1 2022-04-16 thomas arg.drop = drop;
1771 e6bcace5 2021-06-22 stsp err = got_object_idset_for_each(keep, append_id, &arg);
1772 e6bcace5 2021-06-22 stsp if (err) {
1773 e6bcace5 2021-06-22 stsp free(arg.array);
1774 e6bcace5 2021-06-22 stsp goto done;
1776 e6bcace5 2021-06-22 stsp *res = arg.array;
1777 5fe7b5c1 2022-04-16 thomas *nres = arg.idx + 1;
1780 e6bcace5 2021-06-22 stsp got_object_idset_free(keep);
1781 e6bcace5 2021-06-22 stsp got_object_idset_free(drop);
1782 5fe7b5c1 2022-04-16 thomas if (skip)
1783 5fe7b5c1 2022-04-16 thomas got_object_idset_free(skip);
1784 e6bcace5 2021-06-22 stsp got_object_id_queue_free(&ids);
1785 63915ee5 2022-06-23 thomas return err;
1788 63915ee5 2022-06-23 thomas struct load_packed_obj_arg {
1789 63915ee5 2022-06-23 thomas /* output parameters: */
1790 63915ee5 2022-06-23 thomas struct got_object_id *id;
1791 63915ee5 2022-06-23 thomas char *dpath;
1792 63915ee5 2022-06-23 thomas time_t mtime;
1794 63915ee5 2022-06-23 thomas /* input parameters: */
1795 63915ee5 2022-06-23 thomas uint32_t seed;
1796 63915ee5 2022-06-23 thomas int want_meta;
1797 63915ee5 2022-06-23 thomas struct got_object_idset *idset;
1798 63915ee5 2022-06-23 thomas struct got_object_idset *idset_exclude;
1799 63915ee5 2022-06-23 thomas int loose_obj_only;
1800 63915ee5 2022-06-23 thomas int *ncolored;
1801 63915ee5 2022-06-23 thomas int *nfound;
1802 63915ee5 2022-06-23 thomas int *ntrees;
1803 63915ee5 2022-06-23 thomas got_pack_progress_cb progress_cb;
1804 63915ee5 2022-06-23 thomas void *progress_arg;
1805 63915ee5 2022-06-23 thomas struct got_ratelimit *rl;
1806 63915ee5 2022-06-23 thomas got_cancel_cb cancel_cb;
1807 63915ee5 2022-06-23 thomas void *cancel_arg;
1810 63915ee5 2022-06-23 thomas static const struct got_error *
1811 63915ee5 2022-06-23 thomas load_packed_commit_id(void *arg, time_t mtime, struct got_object_id *id,
1812 63915ee5 2022-06-23 thomas struct got_repository *repo)
1814 63915ee5 2022-06-23 thomas struct load_packed_obj_arg *a = arg;
1816 63915ee5 2022-06-23 thomas if (got_object_idset_contains(a->idset, id) ||
1817 63915ee5 2022-06-23 thomas got_object_idset_contains(a->idset_exclude, id))
1818 63915ee5 2022-06-23 thomas return NULL;
1820 63915ee5 2022-06-23 thomas return add_object(a->want_meta,
1821 63915ee5 2022-06-23 thomas a->want_meta ? a->idset : a->idset_exclude,
1822 63915ee5 2022-06-23 thomas id, "", GOT_OBJ_TYPE_COMMIT, mtime, a->seed, a->loose_obj_only,
1823 63915ee5 2022-06-23 thomas repo, a->ncolored, a->nfound, a->ntrees,
1824 63915ee5 2022-06-23 thomas a->progress_cb, a->progress_arg, a->rl);
1827 63915ee5 2022-06-23 thomas static const struct got_error *
1828 63915ee5 2022-06-23 thomas load_packed_tree_ids(void *arg, struct got_tree_object *tree, time_t mtime,
1829 63915ee5 2022-06-23 thomas struct got_object_id *id, const char *dpath, struct got_repository *repo)
1831 63915ee5 2022-06-23 thomas const struct got_error *err;
1832 63915ee5 2022-06-23 thomas struct load_packed_obj_arg *a = arg;
1833 63915ee5 2022-06-23 thomas const char *relpath;
1836 63915ee5 2022-06-23 thomas * When we receive a tree's ID and path but not the tree itself,
1837 63915ee5 2022-06-23 thomas * this tree object was not found in the pack file. This is the
1838 63915ee5 2022-06-23 thomas * last time we are being called for this optimized traversal.
1839 63915ee5 2022-06-23 thomas * Return from here and switch to loading objects the slow way.
1841 63915ee5 2022-06-23 thomas if (tree == NULL) {
1842 63915ee5 2022-06-23 thomas free(a->id);
1843 63915ee5 2022-06-23 thomas a->id = got_object_id_dup(id);
1844 63915ee5 2022-06-23 thomas if (a->id == NULL) {
1845 63915ee5 2022-06-23 thomas err = got_error_from_errno("got_object_id_dup");
1846 63915ee5 2022-06-23 thomas free(a->dpath);
1847 63915ee5 2022-06-23 thomas a->dpath = NULL;
1848 63915ee5 2022-06-23 thomas return err;
1851 63915ee5 2022-06-23 thomas free(a->dpath);
1852 63915ee5 2022-06-23 thomas a->dpath = strdup(dpath);
1853 63915ee5 2022-06-23 thomas if (a->dpath == NULL) {
1854 63915ee5 2022-06-23 thomas err = got_error_from_errno("strdup");
1855 63915ee5 2022-06-23 thomas free(a->id);
1856 63915ee5 2022-06-23 thomas a->id = NULL;
1857 63915ee5 2022-06-23 thomas return err;
1860 63915ee5 2022-06-23 thomas a->mtime = mtime;
1861 63915ee5 2022-06-23 thomas return NULL;
1864 63915ee5 2022-06-23 thomas if (got_object_idset_contains(a->idset, id) ||
1865 63915ee5 2022-06-23 thomas got_object_idset_contains(a->idset_exclude, id))
1866 63915ee5 2022-06-23 thomas return NULL;
1868 63915ee5 2022-06-23 thomas relpath = dpath;
1869 63915ee5 2022-06-23 thomas while (relpath[0] == '/')
1870 63915ee5 2022-06-23 thomas relpath++;
1872 63915ee5 2022-06-23 thomas err = add_object(a->want_meta,
1873 63915ee5 2022-06-23 thomas a->want_meta ? a->idset : a->idset_exclude,
1874 63915ee5 2022-06-23 thomas id, relpath, GOT_OBJ_TYPE_TREE, mtime, a->seed,
1875 63915ee5 2022-06-23 thomas a->loose_obj_only, repo, a->ncolored, a->nfound, a->ntrees,
1876 63915ee5 2022-06-23 thomas a->progress_cb, a->progress_arg, a->rl);
1877 63915ee5 2022-06-23 thomas if (err)
1878 63915ee5 2022-06-23 thomas return err;
1880 63915ee5 2022-06-23 thomas return load_tree_entries(NULL, a->want_meta, a->idset,
1881 63915ee5 2022-06-23 thomas a->idset_exclude, tree, dpath, mtime, a->seed, repo,
1882 63915ee5 2022-06-23 thomas a->loose_obj_only, a->ncolored, a->nfound, a->ntrees,
1883 63915ee5 2022-06-23 thomas a->progress_cb, a->progress_arg, a->rl,
1884 63915ee5 2022-06-23 thomas a->cancel_cb, a->cancel_arg);
1887 63915ee5 2022-06-23 thomas static const struct got_error *
1888 e71f1e62 2022-06-23 thomas load_packed_object_ids(int *found_all_objects,
1889 e71f1e62 2022-06-23 thomas struct got_object_id **ours, int nours,
1890 63915ee5 2022-06-23 thomas struct got_object_id **theirs, int ntheirs,
1891 63915ee5 2022-06-23 thomas int want_meta, uint32_t seed, struct got_object_idset *idset,
1892 63915ee5 2022-06-23 thomas struct got_object_idset *idset_exclude, int loose_obj_only,
1893 63915ee5 2022-06-23 thomas struct got_repository *repo, struct got_packidx *packidx,
1894 63915ee5 2022-06-23 thomas int *ncolored, int *nfound, int *ntrees,
1895 63915ee5 2022-06-23 thomas got_pack_progress_cb progress_cb, void *progress_arg,
1896 63915ee5 2022-06-23 thomas struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1898 63915ee5 2022-06-23 thomas const struct got_error *err = NULL;
1899 63915ee5 2022-06-23 thomas struct load_packed_obj_arg lpa;
1901 63915ee5 2022-06-23 thomas memset(&lpa, 0, sizeof(lpa));
1902 63915ee5 2022-06-23 thomas lpa.seed = seed;
1903 63915ee5 2022-06-23 thomas lpa.want_meta = want_meta;
1904 63915ee5 2022-06-23 thomas lpa.idset = idset;
1905 63915ee5 2022-06-23 thomas lpa.idset_exclude = idset_exclude;
1906 63915ee5 2022-06-23 thomas lpa.loose_obj_only = loose_obj_only;
1907 63915ee5 2022-06-23 thomas lpa.ncolored = ncolored;
1908 63915ee5 2022-06-23 thomas lpa.nfound = nfound;
1909 63915ee5 2022-06-23 thomas lpa.ntrees = ntrees;
1910 63915ee5 2022-06-23 thomas lpa.progress_cb = progress_cb;
1911 63915ee5 2022-06-23 thomas lpa.progress_arg = progress_arg;
1912 63915ee5 2022-06-23 thomas lpa.rl = rl;
1913 63915ee5 2022-06-23 thomas lpa.cancel_cb = cancel_cb;
1914 63915ee5 2022-06-23 thomas lpa.cancel_arg = cancel_arg;
1916 63915ee5 2022-06-23 thomas /* Attempt to load objects via got-read-pack, as far as possible. */
1917 e71f1e62 2022-06-23 thomas err = got_object_enumerate(found_all_objects, load_packed_commit_id,
1918 63915ee5 2022-06-23 thomas load_packed_tree_ids, &lpa, ours, nours, theirs, ntheirs,
1919 63915ee5 2022-06-23 thomas packidx, repo);
1920 63915ee5 2022-06-23 thomas if (err)
1921 63915ee5 2022-06-23 thomas return err;
1923 63915ee5 2022-06-23 thomas if (lpa.id == NULL)
1924 63915ee5 2022-06-23 thomas return NULL;
1927 63915ee5 2022-06-23 thomas * An incomplete tree hierarchy was present in the pack file
1928 e71f1e62 2022-06-23 thomas * and caused loading to be aborted.
1929 63915ee5 2022-06-23 thomas * Continue loading trees the slow way.
1931 63915ee5 2022-06-23 thomas err = load_tree(want_meta, idset, idset_exclude,
1932 63915ee5 2022-06-23 thomas lpa.id, lpa.dpath, lpa.mtime, seed, repo, loose_obj_only,
1933 63915ee5 2022-06-23 thomas ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1934 63915ee5 2022-06-23 thomas cancel_cb, cancel_arg);
1935 63915ee5 2022-06-23 thomas free(lpa.id);
1936 63915ee5 2022-06-23 thomas free(lpa.dpath);
1937 e6bcace5 2021-06-22 stsp return err;
1940 e6bcace5 2021-06-22 stsp static const struct got_error *
1941 63915ee5 2022-06-23 thomas find_pack_for_enumeration(struct got_packidx **best_packidx,
1942 63915ee5 2022-06-23 thomas struct got_object_id **ids, int nids, struct got_repository *repo)
1944 63915ee5 2022-06-23 thomas const struct got_error *err = NULL;
1945 63915ee5 2022-06-23 thomas struct got_pathlist_entry *pe;
1946 63915ee5 2022-06-23 thomas const char *best_packidx_path = NULL;
1947 63915ee5 2022-06-23 thomas int nobj_max = 0;
1948 63915ee5 2022-06-23 thomas int ncommits_max = 0;
1950 63915ee5 2022-06-23 thomas *best_packidx = NULL;
1953 63915ee5 2022-06-23 thomas * Find the largest pack which contains at least some of the
1954 63915ee5 2022-06-23 thomas * commits and tags we are interested in.
1956 63915ee5 2022-06-23 thomas TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1957 63915ee5 2022-06-23 thomas const char *path_packidx = pe->path;
1958 63915ee5 2022-06-23 thomas struct got_packidx *packidx;
1959 63915ee5 2022-06-23 thomas int nobj, i, idx, ncommits = 0;
1961 63915ee5 2022-06-23 thomas err = got_repo_get_packidx(&packidx, path_packidx, repo);
1962 63915ee5 2022-06-23 thomas if (err)
1965 63915ee5 2022-06-23 thomas nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1966 63915ee5 2022-06-23 thomas if (nobj <= nobj_max)
1967 63915ee5 2022-06-23 thomas continue;
1969 63915ee5 2022-06-23 thomas for (i = 0; i < nids; i++) {
1970 63915ee5 2022-06-23 thomas idx = got_packidx_get_object_idx(packidx, ids[i]);
1971 63915ee5 2022-06-23 thomas if (idx != -1)
1972 63915ee5 2022-06-23 thomas ncommits++;
1974 63915ee5 2022-06-23 thomas if (ncommits > ncommits_max) {
1975 63915ee5 2022-06-23 thomas best_packidx_path = path_packidx;
1976 63915ee5 2022-06-23 thomas nobj_max = nobj;
1977 63915ee5 2022-06-23 thomas ncommits_max = ncommits;
1981 eee80a61 2022-06-23 thomas if (best_packidx_path && err == NULL) {
1982 63915ee5 2022-06-23 thomas err = got_repo_get_packidx(best_packidx, best_packidx_path,
1986 63915ee5 2022-06-23 thomas return err;
1989 63915ee5 2022-06-23 thomas static const struct got_error *
1990 85220b0e 2022-03-22 thomas load_object_ids(int *ncolored, int *nfound, int *ntrees,
1991 85220b0e 2022-03-22 thomas struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1992 e6bcace5 2021-06-22 stsp struct got_object_id **ours, int nours, struct got_repository *repo,
1993 cc524d36 2022-05-31 thomas uint32_t seed, int loose_obj_only, got_pack_progress_cb progress_cb,
1994 cc524d36 2022-05-31 thomas void *progress_arg, struct got_ratelimit *rl, got_cancel_cb cancel_cb,
1995 cc524d36 2022-05-31 thomas void *cancel_arg)
1997 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
1998 e6bcace5 2021-06-22 stsp struct got_object_id **ids = NULL;
1999 63915ee5 2022-06-23 thomas struct got_packidx *packidx = NULL;
2000 e71f1e62 2022-06-23 thomas int i, nobj = 0, obj_type, found_all_objects = 0;
2001 7afbdbad 2022-05-12 thomas struct got_object_idset *idset_exclude;
2003 7afbdbad 2022-05-12 thomas idset_exclude = got_object_idset_alloc();
2004 7afbdbad 2022-05-12 thomas if (idset_exclude == NULL)
2005 7afbdbad 2022-05-12 thomas return got_error_from_errno("got_object_idset_alloc");
2007 85220b0e 2022-03-22 thomas *ncolored = 0;
2008 85220b0e 2022-03-22 thomas *nfound = 0;
2009 85220b0e 2022-03-22 thomas *ntrees = 0;
2011 85220b0e 2022-03-22 thomas err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
2012 85220b0e 2022-03-22 thomas repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
2013 ae2b0251 2022-05-12 thomas if (err)
2014 e6bcace5 2021-06-22 stsp goto done;
2016 63915ee5 2022-06-23 thomas err = find_pack_for_enumeration(&packidx, theirs, ntheirs, repo);
2017 63915ee5 2022-06-23 thomas if (err)
2018 63915ee5 2022-06-23 thomas goto done;
2019 63915ee5 2022-06-23 thomas if (packidx) {
2020 e71f1e62 2022-06-23 thomas err = load_packed_object_ids(&found_all_objects,
2021 e71f1e62 2022-06-23 thomas theirs, ntheirs, NULL, 0, 0, seed, idset, idset_exclude,
2022 e71f1e62 2022-06-23 thomas loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
2023 e71f1e62 2022-06-23 thomas progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
2024 63915ee5 2022-06-23 thomas if (err)
2025 63915ee5 2022-06-23 thomas goto done;
2028 e6bcace5 2021-06-22 stsp for (i = 0; i < ntheirs; i++) {
2029 05118f5a 2021-06-22 stsp struct got_object_id *id = theirs[i];
2030 05118f5a 2021-06-22 stsp if (id == NULL)
2032 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
2034 05118f5a 2021-06-22 stsp return err;
2035 8eef4276 2022-04-16 thomas if (obj_type == GOT_OBJ_TYPE_COMMIT) {
2036 e71f1e62 2022-06-23 thomas if (!found_all_objects) {
2037 e71f1e62 2022-06-23 thomas err = load_commit(0, idset, idset_exclude,
2038 e71f1e62 2022-06-23 thomas id, repo, seed, loose_obj_only,
2039 e71f1e62 2022-06-23 thomas ncolored, nfound, ntrees,
2040 e71f1e62 2022-06-23 thomas progress_cb, progress_arg, rl,
2041 e71f1e62 2022-06-23 thomas cancel_cb, cancel_arg);
2042 e71f1e62 2022-06-23 thomas if (err)
2043 e71f1e62 2022-06-23 thomas goto done;
2045 8eef4276 2022-04-16 thomas } else if (obj_type == GOT_OBJ_TYPE_TAG) {
2046 7afbdbad 2022-05-12 thomas err = load_tag(0, idset, idset_exclude, id, repo,
2047 cc524d36 2022-05-31 thomas seed, loose_obj_only, ncolored, nfound, ntrees,
2048 8eef4276 2022-04-16 thomas progress_cb, progress_arg, rl,
2049 8eef4276 2022-04-16 thomas cancel_cb, cancel_arg);
2050 8eef4276 2022-04-16 thomas if (err)
2051 8eef4276 2022-04-16 thomas goto done;
2055 e71f1e62 2022-06-23 thomas found_all_objects = 0;
2056 63915ee5 2022-06-23 thomas err = find_pack_for_enumeration(&packidx, ids, nobj, repo);
2057 63915ee5 2022-06-23 thomas if (err)
2058 63915ee5 2022-06-23 thomas goto done;
2059 63915ee5 2022-06-23 thomas if (packidx) {
2060 e71f1e62 2022-06-23 thomas err = load_packed_object_ids(&found_all_objects, ids,
2061 e71f1e62 2022-06-23 thomas nobj, theirs, ntheirs, 1, seed, idset, idset_exclude,
2062 e71f1e62 2022-06-23 thomas loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
2063 63915ee5 2022-06-23 thomas progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
2064 63915ee5 2022-06-23 thomas if (err)
2065 63915ee5 2022-06-23 thomas goto done;
2068 e71f1e62 2022-06-23 thomas if (!found_all_objects) {
2069 e71f1e62 2022-06-23 thomas for (i = 0; i < nobj; i++) {
2070 e71f1e62 2022-06-23 thomas err = load_commit(1, idset, idset_exclude, ids[i],
2071 e71f1e62 2022-06-23 thomas repo, seed, loose_obj_only, ncolored, nfound,
2072 e71f1e62 2022-06-23 thomas ntrees, progress_cb, progress_arg, rl,
2073 e71f1e62 2022-06-23 thomas cancel_cb, cancel_arg);
2074 e71f1e62 2022-06-23 thomas if (err)
2075 e71f1e62 2022-06-23 thomas goto done;
2079 05118f5a 2021-06-22 stsp for (i = 0; i < nours; i++) {
2080 05118f5a 2021-06-22 stsp struct got_object_id *id = ours[i];
2081 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *m;
2082 05118f5a 2021-06-22 stsp if (id == NULL)
2084 f9c2e8e5 2022-02-13 thomas m = got_object_idset_get(idset, id);
2085 f9c2e8e5 2022-02-13 thomas if (m == NULL) {
2086 07165b17 2021-07-01 stsp err = got_object_get_type(&obj_type, repo, id);
2088 07165b17 2021-07-01 stsp goto done;
2090 f9c2e8e5 2022-02-13 thomas obj_type = m->obj_type;
2091 05118f5a 2021-06-22 stsp if (obj_type != GOT_OBJ_TYPE_TAG)
2093 7afbdbad 2022-05-12 thomas err = load_tag(1, idset, idset_exclude, id, repo,
2094 cc524d36 2022-05-31 thomas seed, loose_obj_only, ncolored, nfound, ntrees,
2095 7afbdbad 2022-05-12 thomas progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
2097 e6bcace5 2021-06-22 stsp goto done;
2100 e6bcace5 2021-06-22 stsp for (i = 0; i < nobj; i++) {
2101 e6bcace5 2021-06-22 stsp free(ids[i]);
2103 e6bcace5 2021-06-22 stsp free(ids);
2104 7afbdbad 2022-05-12 thomas got_object_idset_free(idset_exclude);
2105 e6bcace5 2021-06-22 stsp return err;
2108 ef20f542 2022-06-26 thomas static const struct got_error *
2109 0c6f49ba 2022-07-01 thomas hwrite(FILE *f, const void *buf, off_t len, SHA1_CTX *ctx)
2113 e6bcace5 2021-06-22 stsp SHA1Update(ctx, buf, len);
2114 e6bcace5 2021-06-22 stsp n = fwrite(buf, 1, len, f);
2115 e6bcace5 2021-06-22 stsp if (n != len)
2116 e6bcace5 2021-06-22 stsp return got_ferror(f, GOT_ERR_IO);
2117 9249e7e3 2022-05-12 thomas return NULL;
2120 ef20f542 2022-06-26 thomas static const struct got_error *
2121 9249e7e3 2022-05-12 thomas hcopy(FILE *fsrc, FILE *fdst, off_t len, SHA1_CTX *ctx)
2123 9249e7e3 2022-05-12 thomas unsigned char buf[65536];
2124 9249e7e3 2022-05-12 thomas off_t remain = len;
2125 9249e7e3 2022-05-12 thomas size_t n;
2127 9249e7e3 2022-05-12 thomas while (remain > 0) {
2128 9249e7e3 2022-05-12 thomas size_t copylen = MIN(sizeof(buf), remain);
2129 9249e7e3 2022-05-12 thomas n = fread(buf, 1, copylen, fsrc);
2130 9249e7e3 2022-05-12 thomas if (n != copylen)
2131 9249e7e3 2022-05-12 thomas return got_ferror(fsrc, GOT_ERR_IO);
2132 9249e7e3 2022-05-12 thomas SHA1Update(ctx, buf, copylen);
2133 9249e7e3 2022-05-12 thomas n = fwrite(buf, 1, copylen, fdst);
2134 9249e7e3 2022-05-12 thomas if (n != copylen)
2135 9249e7e3 2022-05-12 thomas return got_ferror(fdst, GOT_ERR_IO);
2136 9249e7e3 2022-05-12 thomas remain -= copylen;
2139 e6bcace5 2021-06-22 stsp return NULL;
2142 ef20f542 2022-06-26 thomas static const struct got_error *
2143 d93dda9a 2022-05-12 thomas hcopy_mmap(uint8_t *src, off_t src_offset, size_t src_size,
2144 d93dda9a 2022-05-12 thomas FILE *fdst, off_t len, SHA1_CTX *ctx)
2146 d93dda9a 2022-05-12 thomas size_t n;
2148 d93dda9a 2022-05-12 thomas if (src_offset + len > src_size)
2149 d93dda9a 2022-05-12 thomas return got_error(GOT_ERR_RANGE);
2151 d93dda9a 2022-05-12 thomas SHA1Update(ctx, src + src_offset, len);
2152 d93dda9a 2022-05-12 thomas n = fwrite(src + src_offset, 1, len, fdst);
2153 d93dda9a 2022-05-12 thomas if (n != len)
2154 d93dda9a 2022-05-12 thomas return got_ferror(fdst, GOT_ERR_IO);
2156 d93dda9a 2022-05-12 thomas return NULL;
2159 e6bcace5 2021-06-22 stsp static void
2160 e6bcace5 2021-06-22 stsp putbe32(char *b, uint32_t n)
2162 e6bcace5 2021-06-22 stsp b[0] = n >> 24;
2163 e6bcace5 2021-06-22 stsp b[1] = n >> 16;
2164 e6bcace5 2021-06-22 stsp b[2] = n >> 8;
2165 e6bcace5 2021-06-22 stsp b[3] = n >> 0;
2168 e6bcace5 2021-06-22 stsp static int
2169 e6bcace5 2021-06-22 stsp write_order_cmp(const void *pa, const void *pb)
2171 e6bcace5 2021-06-22 stsp struct got_pack_meta *a, *b, *ahd, *bhd;
2173 e6bcace5 2021-06-22 stsp a = *(struct got_pack_meta **)pa;
2174 e6bcace5 2021-06-22 stsp b = *(struct got_pack_meta **)pb;
2175 e6bcace5 2021-06-22 stsp ahd = (a->head == NULL) ? a : a->head;
2176 e6bcace5 2021-06-22 stsp bhd = (b->head == NULL) ? b : b->head;
2177 f6b43367 2022-05-01 thomas if (bhd->mtime < ahd->mtime)
2178 f6b43367 2022-05-01 thomas return -1;
2179 f6b43367 2022-05-01 thomas if (bhd->mtime > ahd->mtime)
2180 f6b43367 2022-05-01 thomas return 1;
2181 f6b43367 2022-05-01 thomas if (bhd < ahd)
2182 f6b43367 2022-05-01 thomas return -1;
2183 f6b43367 2022-05-01 thomas if (bhd > ahd)
2184 f6b43367 2022-05-01 thomas return 1;
2185 e6bcace5 2021-06-22 stsp if (a->nchain != b->nchain)
2186 e6bcace5 2021-06-22 stsp return a->nchain - b->nchain;
2187 f6b43367 2022-05-01 thomas if (a->mtime < b->mtime)
2188 f6b43367 2022-05-01 thomas return -1;
2189 f6b43367 2022-05-01 thomas if (a->mtime > b->mtime)
2190 f6b43367 2022-05-01 thomas return 1;
2191 f6b43367 2022-05-01 thomas return got_object_id_cmp(&a->id, &b->id);
2194 f9c2e8e5 2022-02-13 thomas static int
2195 f9c2e8e5 2022-02-13 thomas reuse_write_order_cmp(const void *pa, const void *pb)
2197 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *a, *b;
2199 f9c2e8e5 2022-02-13 thomas a = *(struct got_pack_meta **)pa;
2200 f9c2e8e5 2022-02-13 thomas b = *(struct got_pack_meta **)pb;
2202 f9c2e8e5 2022-02-13 thomas if (a->reused_delta_offset < b->reused_delta_offset)
2203 f9c2e8e5 2022-02-13 thomas return -1;
2204 f9c2e8e5 2022-02-13 thomas if (a->reused_delta_offset > b->reused_delta_offset)
2205 f9c2e8e5 2022-02-13 thomas return 1;
2206 f9c2e8e5 2022-02-13 thomas return 0;
2209 e6bcace5 2021-06-22 stsp static const struct got_error *
2210 e6bcace5 2021-06-22 stsp packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
2214 e6bcace5 2021-06-22 stsp *hdrlen = 0;
2216 e6bcace5 2021-06-22 stsp hdr[0] = obj_type << 4;
2217 e6bcace5 2021-06-22 stsp hdr[0] |= len & 0xf;
2218 e6bcace5 2021-06-22 stsp len >>= 4;
2219 e6bcace5 2021-06-22 stsp for (i = 1; len != 0; i++){
2220 e6bcace5 2021-06-22 stsp if (i >= bufsize)
2221 e6bcace5 2021-06-22 stsp return got_error(GOT_ERR_NO_SPACE);
2222 e6bcace5 2021-06-22 stsp hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
2223 e6bcace5 2021-06-22 stsp hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
2224 e6bcace5 2021-06-22 stsp len >>= GOT_DELTA_SIZE_SHIFT;
2227 e6bcace5 2021-06-22 stsp *hdrlen = i;
2228 e6bcace5 2021-06-22 stsp return NULL;
2231 e6bcace5 2021-06-22 stsp static int
2232 e6bcace5 2021-06-22 stsp packoff(char *hdr, off_t off)
2235 e6bcace5 2021-06-22 stsp char rbuf[8];
2237 e6bcace5 2021-06-22 stsp rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
2238 e6bcace5 2021-06-22 stsp for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
2239 e6bcace5 2021-06-22 stsp rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
2240 e6bcace5 2021-06-22 stsp GOT_DELTA_SIZE_MORE;
2244 e6bcace5 2021-06-22 stsp while (i > 0)
2245 e6bcace5 2021-06-22 stsp hdr[j++] = rbuf[--i];
2249 e6bcace5 2021-06-22 stsp static const struct got_error *
2250 3b6ceab7 2022-01-10 thomas deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
2251 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *m)
2253 3b6ceab7 2022-01-10 thomas const struct got_error *err;
2254 3b6ceab7 2022-01-10 thomas char buf[32];
2257 f9c2e8e5 2022-02-13 thomas if (m->prev->off != 0) {
2258 3b6ceab7 2022-01-10 thomas err = packhdr(&nh, buf, sizeof(buf),
2259 3b6ceab7 2022-01-10 thomas GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
2260 3b6ceab7 2022-01-10 thomas if (err)
2261 3b6ceab7 2022-01-10 thomas return err;
2262 3b6ceab7 2022-01-10 thomas nh += packoff(buf + nh, m->off - m->prev->off);
2263 3b6ceab7 2022-01-10 thomas err = hwrite(packfile, buf, nh, ctx);
2264 3b6ceab7 2022-01-10 thomas if (err)
2265 3b6ceab7 2022-01-10 thomas return err;
2266 3b6ceab7 2022-01-10 thomas *packfile_size += nh;
2267 3b6ceab7 2022-01-10 thomas } else {
2268 3b6ceab7 2022-01-10 thomas err = packhdr(&nh, buf, sizeof(buf),
2269 3b6ceab7 2022-01-10 thomas GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
2270 3b6ceab7 2022-01-10 thomas if (err)
2271 3b6ceab7 2022-01-10 thomas return err;
2272 3b6ceab7 2022-01-10 thomas err = hwrite(packfile, buf, nh, ctx);
2273 3b6ceab7 2022-01-10 thomas if (err)
2274 3b6ceab7 2022-01-10 thomas return err;
2275 3b6ceab7 2022-01-10 thomas *packfile_size += nh;
2276 3b6ceab7 2022-01-10 thomas err = hwrite(packfile, m->prev->id.sha1,
2277 3b6ceab7 2022-01-10 thomas sizeof(m->prev->id.sha1), ctx);
2278 3b6ceab7 2022-01-10 thomas if (err)
2279 3b6ceab7 2022-01-10 thomas return err;
2280 3b6ceab7 2022-01-10 thomas *packfile_size += sizeof(m->prev->id.sha1);
2283 3b6ceab7 2022-01-10 thomas return NULL;
2286 3b6ceab7 2022-01-10 thomas static const struct got_error *
2287 f9c2e8e5 2022-02-13 thomas write_packed_object(off_t *packfile_size, FILE *packfile,
2288 d93dda9a 2022-05-12 thomas FILE *delta_cache, uint8_t *delta_cache_map, size_t delta_cache_size,
2289 d93dda9a 2022-05-12 thomas struct got_pack_meta *m, int *outfd, SHA1_CTX *ctx,
2290 d93dda9a 2022-05-12 thomas struct got_repository *repo)
2292 f9c2e8e5 2022-02-13 thomas const struct got_error *err = NULL;
2293 f9c2e8e5 2022-02-13 thomas struct got_deflate_checksum csum;
2294 f9c2e8e5 2022-02-13 thomas char buf[32];
2296 f9c2e8e5 2022-02-13 thomas struct got_raw_object *raw = NULL;
2297 f9c2e8e5 2022-02-13 thomas off_t outlen;
2299 f9c2e8e5 2022-02-13 thomas csum.output_sha1 = ctx;
2300 f9c2e8e5 2022-02-13 thomas csum.output_crc = NULL;
2302 f9c2e8e5 2022-02-13 thomas m->off = ftello(packfile);
2303 f9c2e8e5 2022-02-13 thomas if (m->delta_len == 0) {
2304 f9c2e8e5 2022-02-13 thomas err = got_object_raw_open(&raw, outfd, repo, &m->id);
2305 f9c2e8e5 2022-02-13 thomas if (err)
2306 f9c2e8e5 2022-02-13 thomas goto done;
2307 f9c2e8e5 2022-02-13 thomas err = packhdr(&nh, buf, sizeof(buf),
2308 f9c2e8e5 2022-02-13 thomas m->obj_type, raw->size);
2309 f9c2e8e5 2022-02-13 thomas if (err)
2310 f9c2e8e5 2022-02-13 thomas goto done;
2311 f9c2e8e5 2022-02-13 thomas err = hwrite(packfile, buf, nh, ctx);
2312 f9c2e8e5 2022-02-13 thomas if (err)
2313 f9c2e8e5 2022-02-13 thomas goto done;
2314 f9c2e8e5 2022-02-13 thomas *packfile_size += nh;
2315 f9c2e8e5 2022-02-13 thomas if (raw->f == NULL) {
2316 f9c2e8e5 2022-02-13 thomas err = got_deflate_to_file_mmap(&outlen,
2317 f9c2e8e5 2022-02-13 thomas raw->data + raw->hdrlen, 0, raw->size,
2318 f9c2e8e5 2022-02-13 thomas packfile, &csum);
2319 f9c2e8e5 2022-02-13 thomas if (err)
2320 f9c2e8e5 2022-02-13 thomas goto done;
2321 f9c2e8e5 2022-02-13 thomas } else {
2322 f9c2e8e5 2022-02-13 thomas if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
2323 f9c2e8e5 2022-02-13 thomas == -1) {
2324 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("fseeko");
2325 f9c2e8e5 2022-02-13 thomas goto done;
2327 f9c2e8e5 2022-02-13 thomas err = got_deflate_to_file(&outlen, raw->f,
2328 f9c2e8e5 2022-02-13 thomas raw->size, packfile, &csum);
2329 f9c2e8e5 2022-02-13 thomas if (err)
2330 f9c2e8e5 2022-02-13 thomas goto done;
2332 f9c2e8e5 2022-02-13 thomas *packfile_size += outlen;
2333 f9c2e8e5 2022-02-13 thomas got_object_raw_close(raw);
2334 f9c2e8e5 2022-02-13 thomas raw = NULL;
2335 f9c2e8e5 2022-02-13 thomas } else if (m->delta_buf) {
2336 f9c2e8e5 2022-02-13 thomas err = deltahdr(packfile_size, ctx, packfile, m);
2337 f9c2e8e5 2022-02-13 thomas if (err)
2338 f9c2e8e5 2022-02-13 thomas goto done;
2339 9249e7e3 2022-05-12 thomas err = hwrite(packfile, m->delta_buf,
2340 9249e7e3 2022-05-12 thomas m->delta_compressed_len, ctx);
2341 f9c2e8e5 2022-02-13 thomas if (err)
2342 f9c2e8e5 2022-02-13 thomas goto done;
2343 9249e7e3 2022-05-12 thomas *packfile_size += m->delta_compressed_len;
2344 f9c2e8e5 2022-02-13 thomas free(m->delta_buf);
2345 f9c2e8e5 2022-02-13 thomas m->delta_buf = NULL;
2346 d93dda9a 2022-05-12 thomas } else if (delta_cache_map) {
2347 d93dda9a 2022-05-12 thomas err = deltahdr(packfile_size, ctx, packfile, m);
2348 d93dda9a 2022-05-12 thomas if (err)
2349 d93dda9a 2022-05-12 thomas goto done;
2350 d93dda9a 2022-05-12 thomas err = hcopy_mmap(delta_cache_map, m->delta_offset,
2351 d93dda9a 2022-05-12 thomas delta_cache_size, packfile, m->delta_compressed_len,
2353 d93dda9a 2022-05-12 thomas if (err)
2354 d93dda9a 2022-05-12 thomas goto done;
2355 d93dda9a 2022-05-12 thomas *packfile_size += m->delta_compressed_len;
2356 f9c2e8e5 2022-02-13 thomas } else {
2357 f9c2e8e5 2022-02-13 thomas if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
2358 f9c2e8e5 2022-02-13 thomas == -1) {
2359 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("fseeko");
2360 f9c2e8e5 2022-02-13 thomas goto done;
2362 f9c2e8e5 2022-02-13 thomas err = deltahdr(packfile_size, ctx, packfile, m);
2363 f9c2e8e5 2022-02-13 thomas if (err)
2364 f9c2e8e5 2022-02-13 thomas goto done;
2365 9249e7e3 2022-05-12 thomas err = hcopy(delta_cache, packfile,
2366 9249e7e3 2022-05-12 thomas m->delta_compressed_len, ctx);
2367 f9c2e8e5 2022-02-13 thomas if (err)
2368 f9c2e8e5 2022-02-13 thomas goto done;
2369 9249e7e3 2022-05-12 thomas *packfile_size += m->delta_compressed_len;
2372 f9c2e8e5 2022-02-13 thomas if (raw)
2373 f9c2e8e5 2022-02-13 thomas got_object_raw_close(raw);
2374 f9c2e8e5 2022-02-13 thomas return err;
2377 f9c2e8e5 2022-02-13 thomas static const struct got_error *
2378 b79280dd 2021-10-15 thomas genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
2379 f9c2e8e5 2022-02-13 thomas struct got_pack_meta **deltify, int ndeltify,
2380 f9c2e8e5 2022-02-13 thomas struct got_pack_meta **reuse, int nreuse,
2381 85220b0e 2022-03-22 thomas int ncolored, int nfound, int ntrees, int nours,
2382 85220b0e 2022-03-22 thomas struct got_repository *repo,
2383 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
2384 31ba2236 2022-01-05 thomas struct got_ratelimit *rl,
2385 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
2387 e6bcace5 2021-06-22 stsp const struct got_error *err = NULL;
2389 e6bcace5 2021-06-22 stsp SHA1_CTX ctx;
2390 e6bcace5 2021-06-22 stsp struct got_pack_meta *m;
2391 e4d8ab47 2021-10-15 thomas char buf[32];
2392 e8f02263 2022-01-23 thomas size_t n;
2393 f9c2e8e5 2022-02-13 thomas off_t packfile_size = 0;
2394 ecf9545f 2021-10-15 thomas int outfd = -1;
2395 d93dda9a 2022-05-12 thomas int delta_cache_fd = -1;
2396 d93dda9a 2022-05-12 thomas uint8_t *delta_cache_map = NULL;
2397 d93dda9a 2022-05-12 thomas size_t delta_cache_size = 0;
2399 e6bcace5 2021-06-22 stsp SHA1Init(&ctx);
2401 d93dda9a 2022-05-12 thomas #ifndef GOT_PACK_NO_MMAP
2402 d93dda9a 2022-05-12 thomas delta_cache_fd = dup(fileno(delta_cache));
2403 d93dda9a 2022-05-12 thomas if (delta_cache_fd != -1) {
2404 d93dda9a 2022-05-12 thomas struct stat sb;
2405 d93dda9a 2022-05-12 thomas if (fstat(delta_cache_fd, &sb) == -1) {
2406 d93dda9a 2022-05-12 thomas err = got_error_from_errno("fstat");
2407 d93dda9a 2022-05-12 thomas goto done;
2409 d93dda9a 2022-05-12 thomas if (sb.st_size > 0 && sb.st_size <= SIZE_MAX) {
2410 d93dda9a 2022-05-12 thomas delta_cache_map = mmap(NULL, sb.st_size,
2411 d93dda9a 2022-05-12 thomas PROT_READ, MAP_PRIVATE, delta_cache_fd, 0);
2412 d93dda9a 2022-05-12 thomas if (delta_cache_map == MAP_FAILED) {
2413 d93dda9a 2022-05-12 thomas if (errno != ENOMEM) {
2414 d93dda9a 2022-05-12 thomas err = got_error_from_errno("mmap");
2415 d93dda9a 2022-05-12 thomas goto done;
2417 d93dda9a 2022-05-12 thomas delta_cache_map = NULL; /* fallback on stdio */
2419 d93dda9a 2022-05-12 thomas delta_cache_size = (size_t)sb.st_size;
2423 e6bcace5 2021-06-22 stsp err = hwrite(packfile, "PACK", 4, &ctx);
2425 d93dda9a 2022-05-12 thomas goto done;
2426 e6bcace5 2021-06-22 stsp putbe32(buf, GOT_PACKFILE_VERSION);
2427 e6bcace5 2021-06-22 stsp err = hwrite(packfile, buf, 4, &ctx);
2429 e6bcace5 2021-06-22 stsp goto done;
2430 f9c2e8e5 2022-02-13 thomas putbe32(buf, ndeltify + nreuse);
2431 e6bcace5 2021-06-22 stsp err = hwrite(packfile, buf, 4, &ctx);
2433 e6bcace5 2021-06-22 stsp goto done;
2435 f9c2e8e5 2022-02-13 thomas qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
2436 f9c2e8e5 2022-02-13 thomas write_order_cmp);
2437 f9c2e8e5 2022-02-13 thomas for (i = 0; i < ndeltify; i++) {
2438 31ba2236 2022-01-05 thomas err = report_progress(progress_cb, progress_arg, rl,
2439 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, packfile_size, nours,
2440 85220b0e 2022-03-22 thomas ndeltify + nreuse, ndeltify + nreuse, i);
2441 31ba2236 2022-01-05 thomas if (err)
2442 31ba2236 2022-01-05 thomas goto done;
2443 f9c2e8e5 2022-02-13 thomas m = deltify[i];
2444 f9c2e8e5 2022-02-13 thomas err = write_packed_object(&packfile_size, packfile,
2445 d93dda9a 2022-05-12 thomas delta_cache, delta_cache_map, delta_cache_size,
2446 d93dda9a 2022-05-12 thomas m, &outfd, &ctx, repo);
2447 f9c2e8e5 2022-02-13 thomas if (err)
2448 f9c2e8e5 2022-02-13 thomas goto done;
2451 f9c2e8e5 2022-02-13 thomas qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
2452 f9c2e8e5 2022-02-13 thomas reuse_write_order_cmp);
2453 f9c2e8e5 2022-02-13 thomas for (i = 0; i < nreuse; i++) {
2454 f9c2e8e5 2022-02-13 thomas err = report_progress(progress_cb, progress_arg, rl,
2455 85220b0e 2022-03-22 thomas ncolored, nfound, ntrees, packfile_size, nours,
2456 85220b0e 2022-03-22 thomas ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
2457 f9c2e8e5 2022-02-13 thomas if (err)
2458 f9c2e8e5 2022-02-13 thomas goto done;
2459 f9c2e8e5 2022-02-13 thomas m = reuse[i];
2460 f9c2e8e5 2022-02-13 thomas err = write_packed_object(&packfile_size, packfile,
2461 d93dda9a 2022-05-12 thomas delta_cache, delta_cache_map, delta_cache_size,
2462 d93dda9a 2022-05-12 thomas m, &outfd, &ctx, repo);
2463 f9c2e8e5 2022-02-13 thomas if (err)
2464 f9c2e8e5 2022-02-13 thomas goto done;
2467 e6bcace5 2021-06-22 stsp SHA1Final(pack_sha1, &ctx);
2468 e6bcace5 2021-06-22 stsp n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
2469 e6bcace5 2021-06-22 stsp if (n != SHA1_DIGEST_LENGTH)
2470 e6bcace5 2021-06-22 stsp err = got_ferror(packfile, GOT_ERR_IO);
2471 05118f5a 2021-06-22 stsp packfile_size += SHA1_DIGEST_LENGTH;
2472 dc7edd42 2021-08-22 stsp packfile_size += sizeof(struct got_packfile_hdr);
2473 31ba2236 2022-01-05 thomas if (progress_cb) {
2474 85220b0e 2022-03-22 thomas err = progress_cb(progress_arg, ncolored, nfound, ntrees,
2475 85220b0e 2022-03-22 thomas packfile_size, nours, ndeltify + nreuse,
2476 85220b0e 2022-03-22 thomas ndeltify + nreuse, ndeltify + nreuse);
2477 31ba2236 2022-01-05 thomas if (err)
2478 31ba2236 2022-01-05 thomas goto done;
2481 ecf9545f 2021-10-15 thomas if (outfd != -1 && close(outfd) == -1 && err == NULL)
2482 ecf9545f 2021-10-15 thomas err = got_error_from_errno("close");
2483 d93dda9a 2022-05-12 thomas if (delta_cache_map && munmap(delta_cache_map, delta_cache_size) == -1)
2484 d93dda9a 2022-05-12 thomas err = got_error_from_errno("munmap");
2485 d93dda9a 2022-05-12 thomas if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
2486 d93dda9a 2022-05-12 thomas err = got_error_from_errno("close");
2487 e6bcace5 2021-06-22 stsp return err;
2490 f9c2e8e5 2022-02-13 thomas static const struct got_error *
2491 f9c2e8e5 2022-02-13 thomas add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
2493 f9c2e8e5 2022-02-13 thomas struct got_pack_meta *m = data;
2494 f9c2e8e5 2022-02-13 thomas struct got_pack_metavec *v = arg;
2496 6c0788af 2022-05-31 thomas if (m->reused_delta_offset != 0)
2497 f291ef1f 2022-05-12 thomas return NULL;
2499 f9c2e8e5 2022-02-13 thomas return add_meta(m, v);
2502 e6bcace5 2021-06-22 stsp const struct got_error *
2503 e6bcace5 2021-06-22 stsp got_pack_create(uint8_t *packsha1, FILE *packfile,
2504 e6bcace5 2021-06-22 stsp struct got_object_id **theirs, int ntheirs,
2505 e6bcace5 2021-06-22 stsp struct got_object_id **ours, int nours,
2506 f8a36e22 2021-08-26 stsp struct got_repository *repo, int loose_obj_only, int allow_empty,
2507 05118f5a 2021-06-22 stsp got_pack_progress_cb progress_cb, void *progress_arg,
2508 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
2510 e6bcace5 2021-06-22 stsp const struct got_error *err;
2511 f9c2e8e5 2022-02-13 thomas int delta_cache_fd = -1;
2512 b79280dd 2021-10-15 thomas FILE *delta_cache = NULL;
2513 f9c2e8e5 2022-02-13 thomas struct got_object_idset *idset;
2514 31ba2236 2022-01-05 thomas struct got_ratelimit rl;
2515 f9c2e8e5 2022-02-13 thomas struct got_pack_metavec deltify, reuse;
2516 85220b0e 2022-03-22 thomas int ncolored = 0, nfound = 0, ntrees = 0;
2517 f291ef1f 2022-05-12 thomas size_t ndeltify;
2518 cc524d36 2022-05-31 thomas uint32_t seed;
2520 cc524d36 2022-05-31 thomas seed = arc4random();
2522 f9c2e8e5 2022-02-13 thomas memset(&deltify, 0, sizeof(deltify));
2523 f9c2e8e5 2022-02-13 thomas memset(&reuse, 0, sizeof(reuse));
2525 31ba2236 2022-01-05 thomas got_ratelimit_init(&rl, 0, 500);
2527 f9c2e8e5 2022-02-13 thomas idset = got_object_idset_alloc();
2528 f9c2e8e5 2022-02-13 thomas if (idset == NULL)
2529 f9c2e8e5 2022-02-13 thomas return got_error_from_errno("got_object_idset_alloc");
2531 85220b0e 2022-03-22 thomas err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
2532 cc524d36 2022-05-31 thomas ntheirs, ours, nours, repo, seed, loose_obj_only,
2533 85220b0e 2022-03-22 thomas progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
2535 20c6bdb6 2022-05-19 thomas goto done;
2537 f9a643e8 2022-02-13 thomas if (progress_cb) {
2538 85220b0e 2022-03-22 thomas err = progress_cb(progress_arg, ncolored, nfound, ntrees,
2539 85220b0e 2022-03-22 thomas 0L, nours, got_object_idset_num_elements(idset), 0, 0);
2540 f9a643e8 2022-02-13 thomas if (err)
2541 f9a643e8 2022-02-13 thomas goto done;
2544 f9c2e8e5 2022-02-13 thomas if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
2545 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_CANNOT_PACK);
2546 05118f5a 2021-06-22 stsp goto done;
2549 f9c2e8e5 2022-02-13 thomas delta_cache_fd = got_opentempfd();
2550 f9c2e8e5 2022-02-13 thomas if (delta_cache_fd == -1) {
2551 b79280dd 2021-10-15 thomas err = got_error_from_errno("got_opentemp");
2552 b79280dd 2021-10-15 thomas goto done;
2555 f9c2e8e5 2022-02-13 thomas reuse.metasz = 64;
2556 f9c2e8e5 2022-02-13 thomas reuse.meta = calloc(reuse.metasz,
2557 f9c2e8e5 2022-02-13 thomas sizeof(struct got_pack_meta *));
2558 f9c2e8e5 2022-02-13 thomas if (reuse.meta == NULL) {
2559 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("calloc");
2560 f9c2e8e5 2022-02-13 thomas goto done;
2563 85220b0e 2022-03-22 thomas err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
2564 85220b0e 2022-03-22 thomas ntrees, nours, repo, progress_cb, progress_arg, &rl,
2565 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
2566 f9c2e8e5 2022-02-13 thomas if (err)
2567 f9c2e8e5 2022-02-13 thomas goto done;
2569 f9c2e8e5 2022-02-13 thomas delta_cache = fdopen(delta_cache_fd, "a+");
2570 f9c2e8e5 2022-02-13 thomas if (delta_cache == NULL) {
2571 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("fdopen");
2572 f9c2e8e5 2022-02-13 thomas goto done;
2574 f9c2e8e5 2022-02-13 thomas delta_cache_fd = -1;
2576 f9c2e8e5 2022-02-13 thomas if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
2577 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("fseeko");
2578 f9c2e8e5 2022-02-13 thomas goto done;
2581 f291ef1f 2022-05-12 thomas ndeltify = got_object_idset_num_elements(idset) - reuse.nmeta;
2582 f291ef1f 2022-05-12 thomas if (ndeltify > 0) {
2583 f291ef1f 2022-05-12 thomas deltify.meta = calloc(ndeltify, sizeof(struct got_pack_meta *));
2584 f291ef1f 2022-05-12 thomas if (deltify.meta == NULL) {
2585 f291ef1f 2022-05-12 thomas err = got_error_from_errno("calloc");
2586 f291ef1f 2022-05-12 thomas goto done;
2588 f291ef1f 2022-05-12 thomas deltify.metasz = ndeltify;
2590 f291ef1f 2022-05-12 thomas err = got_object_idset_for_each(idset, add_meta_idset_cb,
2591 f291ef1f 2022-05-12 thomas &deltify);
2593 f8a36e22 2021-08-26 stsp goto done;
2594 f291ef1f 2022-05-12 thomas if (deltify.nmeta > 0) {
2595 f291ef1f 2022-05-12 thomas err = pick_deltas(deltify.meta, deltify.nmeta,
2596 f291ef1f 2022-05-12 thomas ncolored, nfound, ntrees, nours, reuse.nmeta,
2597 f291ef1f 2022-05-12 thomas delta_cache, repo, progress_cb, progress_arg, &rl,
2598 f291ef1f 2022-05-12 thomas cancel_cb, cancel_arg);
2599 f291ef1f 2022-05-12 thomas if (err)
2600 f291ef1f 2022-05-12 thomas goto done;
2604 9249e7e3 2022-05-12 thomas if (fflush(delta_cache) == EOF) {
2605 9249e7e3 2022-05-12 thomas err = got_error_from_errno("fflush");
2606 9249e7e3 2022-05-12 thomas goto done;
2608 f9c2e8e5 2022-02-13 thomas err = genpack(packsha1, packfile, delta_cache, deltify.meta,
2609 85220b0e 2022-03-22 thomas deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
2610 85220b0e 2022-03-22 thomas nours, repo, progress_cb, progress_arg, &rl,
2611 85220b0e 2022-03-22 thomas cancel_cb, cancel_arg);
2613 e6bcace5 2021-06-22 stsp goto done;
2615 f9c2e8e5 2022-02-13 thomas free_nmeta(deltify.meta, deltify.nmeta);
2616 f9c2e8e5 2022-02-13 thomas free_nmeta(reuse.meta, reuse.nmeta);
2617 f9c2e8e5 2022-02-13 thomas got_object_idset_free(idset);
2618 f9c2e8e5 2022-02-13 thomas if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
2619 f9c2e8e5 2022-02-13 thomas err = got_error_from_errno("close");
2620 b79280dd 2021-10-15 thomas if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
2621 b79280dd 2021-10-15 thomas err = got_error_from_errno("fclose");
2622 e6bcace5 2021-06-22 stsp return err;
