1 e6bcace5 2021-06-22 stsp /*
   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>
   4 e6bcace5 2021-06-22 stsp  *
   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.
   8 e6bcace5 2021-06-22 stsp  *
   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.
  16 e6bcace5 2021-06-22 stsp  */
  17 e6bcace5 2021-06-22 stsp 
  18 08736cf9 2021-06-23 stsp #include <sys/types.h>
  19 08736cf9 2021-06-23 stsp #include <sys/uio.h>
  20 e6bcace5 2021-06-22 stsp #include <sys/stat.h>
  21 31ba2236 2022-01-05 thomas #include <sys/time.h>
  22 e6bcace5 2021-06-22 stsp 
  23 08736cf9 2021-06-23 stsp #include <stdint.h>
  24 e6bcace5 2021-06-22 stsp #include <stdio.h>
  25 e6bcace5 2021-06-22 stsp #include <stdlib.h>
  26 e6bcace5 2021-06-22 stsp #include <string.h>
  27 31ba2236 2022-01-05 thomas #include <time.h>
  28 e6bcace5 2021-06-22 stsp #include <limits.h>
  29 e6bcace5 2021-06-22 stsp #include <zlib.h>
  30 b347007e 2021-10-15 thomas 
  31 b347007e 2021-10-15 thomas #if defined(__FreeBSD__)
  32 b347007e 2021-10-15 thomas #include <unistd.h>
  33 b347007e 2021-10-15 thomas #endif
  34 e6bcace5 2021-06-22 stsp 
  35 e6bcace5 2021-06-22 stsp #include "got_error.h"
  36 e6bcace5 2021-06-22 stsp #include "got_cancel.h"
  37 e6bcace5 2021-06-22 stsp #include "got_object.h"
  38 0be8fa4c 2021-10-15 thomas #include "got_path.h"
  39 05118f5a 2021-06-22 stsp #include "got_reference.h"
  40 05118f5a 2021-06-22 stsp #include "got_repository_admin.h"
  41 e4d8ab47 2021-10-15 thomas #include "got_opentemp.h"
  42 e6bcace5 2021-06-22 stsp 
  43 e6bcace5 2021-06-22 stsp #include "got_lib_deltify.h"
  44 e6bcace5 2021-06-22 stsp #include "got_lib_delta.h"
  45 e6bcace5 2021-06-22 stsp #include "got_lib_object.h"
  46 e6bcace5 2021-06-22 stsp #include "got_lib_object_idset.h"
  47 05118f5a 2021-06-22 stsp #include "got_lib_object_cache.h"
  48 e6bcace5 2021-06-22 stsp #include "got_lib_deflate.h"
  49 e6bcace5 2021-06-22 stsp #include "got_lib_pack.h"
  50 05118f5a 2021-06-22 stsp #include "got_lib_privsep.h"
  51 05118f5a 2021-06-22 stsp #include "got_lib_repository.h"
  52 31ba2236 2022-01-05 thomas #include "got_lib_ratelimit.h"
  53 e6bcace5 2021-06-22 stsp 
  54 b79280dd 2021-10-15 thomas #ifndef MIN
  55 b79280dd 2021-10-15 thomas #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
  56 b79280dd 2021-10-15 thomas #endif
  57 b79280dd 2021-10-15 thomas 
  58 e6bcace5 2021-06-22 stsp #ifndef MAX
  59 e6bcace5 2021-06-22 stsp #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
  60 e6bcace5 2021-06-22 stsp #endif
  61 e6bcace5 2021-06-22 stsp 
  62 e6bcace5 2021-06-22 stsp struct got_pack_meta {
  63 e6bcace5 2021-06-22 stsp 	struct got_object_id id;
  64 e6bcace5 2021-06-22 stsp 	char	*path;
  65 e6bcace5 2021-06-22 stsp 	int	obj_type;
  66 ab6186ae 2021-10-15 thomas 	off_t	size;
  67 e6bcace5 2021-06-22 stsp 	time_t	mtime;
  68 e6bcace5 2021-06-22 stsp 
  69 e6bcace5 2021-06-22 stsp 	/* The best delta we picked */
  70 e6bcace5 2021-06-22 stsp 	struct got_pack_meta *head;
  71 e6bcace5 2021-06-22 stsp 	struct got_pack_meta *prev;
  72 3b6ceab7 2022-01-10 thomas 	unsigned char *delta_buf; /* if not encoded in delta cache file */
  73 b79280dd 2021-10-15 thomas 	off_t	delta_offset;	/* offset in delta cache file */
  74 3b6ceab7 2022-01-10 thomas 	off_t	delta_len;	/* encoded delta length */
  75 e6bcace5 2021-06-22 stsp 	int	nchain;
  76 e6bcace5 2021-06-22 stsp 
  77 f9c2e8e5 2022-02-13 thomas 	int	have_reused_delta;
  78 f9c2e8e5 2022-02-13 thomas 	off_t   reused_delta_offset; /* offset of delta in reused pack file */
  79 f9c2e8e5 2022-02-13 thomas 	struct got_object_id *base_obj_id;
  80 f9c2e8e5 2022-02-13 thomas 
  81 e6bcace5 2021-06-22 stsp 	/* Only used for delta window */
  82 e6bcace5 2021-06-22 stsp 	struct got_delta_table *dtab;
  83 e6bcace5 2021-06-22 stsp 
  84 e6bcace5 2021-06-22 stsp 	/* Only used for writing offset deltas */
  85 e6bcace5 2021-06-22 stsp 	off_t	off;
  86 e6bcace5 2021-06-22 stsp };
  87 e6bcace5 2021-06-22 stsp 
  88 e6bcace5 2021-06-22 stsp struct got_pack_metavec {
  89 e6bcace5 2021-06-22 stsp 	struct got_pack_meta **meta;
  90 e6bcace5 2021-06-22 stsp 	int nmeta;
  91 e6bcace5 2021-06-22 stsp 	int metasz;
  92 e6bcace5 2021-06-22 stsp };
  93 e6bcace5 2021-06-22 stsp 
  94 e6bcace5 2021-06-22 stsp static const struct got_error *
  95 e6bcace5 2021-06-22 stsp alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
  96 e6bcace5 2021-06-22 stsp     const char *path, int obj_type, time_t mtime)
  97 e6bcace5 2021-06-22 stsp {
  98 e6bcace5 2021-06-22 stsp 	const struct got_error *err = NULL;
  99 e6bcace5 2021-06-22 stsp 	struct got_pack_meta *m;
 100 e6bcace5 2021-06-22 stsp 
 101 e6bcace5 2021-06-22 stsp 	*new = NULL;
 102 e6bcace5 2021-06-22 stsp 
 103 e6bcace5 2021-06-22 stsp 	m = calloc(1, sizeof(*m));
 104 e6bcace5 2021-06-22 stsp 	if (m == NULL)
 105 b7e0a384 2021-10-15 thomas 		return got_error_from_errno("calloc");
 106 e6bcace5 2021-06-22 stsp 
 107 e6bcace5 2021-06-22 stsp 	memcpy(&m->id, id, sizeof(m->id));
 108 e6bcace5 2021-06-22 stsp 
 109 e6bcace5 2021-06-22 stsp 	m->path = strdup(path);
 110 e6bcace5 2021-06-22 stsp 	if (m->path == NULL) {
 111 e6bcace5 2021-06-22 stsp 		err = got_error_from_errno("strdup");
 112 e6bcace5 2021-06-22 stsp 		free(m);
 113 e6bcace5 2021-06-22 stsp 		return err;
 114 e6bcace5 2021-06-22 stsp 	}
 115 e6bcace5 2021-06-22 stsp 
 116 e6bcace5 2021-06-22 stsp 	m->obj_type = obj_type;
 117 e6bcace5 2021-06-22 stsp 	m->mtime = mtime;
 118 e6bcace5 2021-06-22 stsp 	*new = m;
 119 e6bcace5 2021-06-22 stsp 	return NULL;
 120 e6bcace5 2021-06-22 stsp }
 121 e6bcace5 2021-06-22 stsp 
 122 e6bcace5 2021-06-22 stsp static void
 123 e6bcace5 2021-06-22 stsp clear_meta(struct got_pack_meta *meta)
 124 e6bcace5 2021-06-22 stsp {
 125 e6bcace5 2021-06-22 stsp 	if (meta == NULL)
 126 e6bcace5 2021-06-22 stsp 		return;
 127 e6bcace5 2021-06-22 stsp 	free(meta->path);
 128 e6bcace5 2021-06-22 stsp 	meta->path = NULL;
 129 3b6ceab7 2022-01-10 thomas 	free(meta->delta_buf);
 130 3b6ceab7 2022-01-10 thomas 	meta->delta_buf = NULL;
 131 f9c2e8e5 2022-02-13 thomas 	free(meta->base_obj_id);
 132 f9c2e8e5 2022-02-13 thomas 	meta->base_obj_id = NULL;
 133 e6bcace5 2021-06-22 stsp }
 134 e6bcace5 2021-06-22 stsp 
 135 e6bcace5 2021-06-22 stsp static void
 136 e6bcace5 2021-06-22 stsp free_nmeta(struct got_pack_meta **meta, int nmeta)
 137 e6bcace5 2021-06-22 stsp {
 138 e6bcace5 2021-06-22 stsp 	int i;
 139 e6bcace5 2021-06-22 stsp 
 140 e6bcace5 2021-06-22 stsp 	for (i = 0; i < nmeta; i++)
 141 e6bcace5 2021-06-22 stsp 		clear_meta(meta[i]);
 142 e6bcace5 2021-06-22 stsp 	free(meta);
 143 e6bcace5 2021-06-22 stsp }
 144 e6bcace5 2021-06-22 stsp 
 145 e6bcace5 2021-06-22 stsp static int
 146 e6bcace5 2021-06-22 stsp delta_order_cmp(const void *pa, const void *pb)
 147 e6bcace5 2021-06-22 stsp {
 148 e6bcace5 2021-06-22 stsp 	struct got_pack_meta *a, *b;
 149 e6bcace5 2021-06-22 stsp 	int cmp;
 150 e6bcace5 2021-06-22 stsp 
 151 e6bcace5 2021-06-22 stsp 	a = *(struct got_pack_meta **)pa;
 152 e6bcace5 2021-06-22 stsp 	b = *(struct got_pack_meta **)pb;
 153 e6bcace5 2021-06-22 stsp 
 154 e6bcace5 2021-06-22 stsp 	if (a->obj_type != b->obj_type)
 155 e6bcace5 2021-06-22 stsp 		return a->obj_type - b->obj_type;
 156 e6bcace5 2021-06-22 stsp 	cmp = strcmp(a->path, b->path);
 157 e6bcace5 2021-06-22 stsp 	if (cmp != 0)
 158 e6bcace5 2021-06-22 stsp 		return cmp;
 159 e6bcace5 2021-06-22 stsp 	if (a->mtime != b->mtime)
 160 e6bcace5 2021-06-22 stsp 		return a->mtime - b->mtime;
 161 e6bcace5 2021-06-22 stsp 	return got_object_id_cmp(&a->id, &b->id);
 162 e6bcace5 2021-06-22 stsp }
 163 e6bcace5 2021-06-22 stsp 
 164 3b6ceab7 2022-01-10 thomas static off_t
 165 e6bcace5 2021-06-22 stsp delta_size(struct got_delta_instruction *deltas, int ndeltas)
 166 e6bcace5 2021-06-22 stsp {
 167 3b6ceab7 2022-01-10 thomas 	int i;
 168 3b6ceab7 2022-01-10 thomas 	off_t size = 32;
 169 e6bcace5 2021-06-22 stsp 	for (i = 0; i < ndeltas; i++) {
 170 e6bcace5 2021-06-22 stsp 		if (deltas[i].copy)
 171 e6bcace5 2021-06-22 stsp 			size += GOT_DELTA_SIZE_SHIFT;
 172 e6bcace5 2021-06-22 stsp 		else
 173 e6bcace5 2021-06-22 stsp 			size += deltas[i].len + 1;
 174 e6bcace5 2021-06-22 stsp 	}
 175 e6bcace5 2021-06-22 stsp 	return size;
 176 3b6ceab7 2022-01-10 thomas }
 177 3b6ceab7 2022-01-10 thomas 
 178 3b6ceab7 2022-01-10 thomas static const struct got_error *
 179 3b6ceab7 2022-01-10 thomas append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
 180 3b6ceab7 2022-01-10 thomas {
 181 3b6ceab7 2022-01-10 thomas 	char *n;
 182 3b6ceab7 2022-01-10 thomas 
 183 3b6ceab7 2022-01-10 thomas 	if (*len + nseg >= *sz) {
 184 3b6ceab7 2022-01-10 thomas 		while (*len + nseg >= *sz)
 185 3b6ceab7 2022-01-10 thomas 			*sz += *sz / 2;
 186 3b6ceab7 2022-01-10 thomas 		n = realloc(*p, *sz);
 187 3b6ceab7 2022-01-10 thomas 		if (n == NULL)
 188 3b6ceab7 2022-01-10 thomas 			return got_error_from_errno("realloc");
 189 3b6ceab7 2022-01-10 thomas 		*p = n;
 190 3b6ceab7 2022-01-10 thomas 	}
 191 3b6ceab7 2022-01-10 thomas 	memcpy(*p + *len, seg, nseg);
 192 3b6ceab7 2022-01-10 thomas 	*len += nseg;
 193 3b6ceab7 2022-01-10 thomas 	return NULL;
 194 3b6ceab7 2022-01-10 thomas }
 195 3b6ceab7 2022-01-10 thomas 
 196 3b6ceab7 2022-01-10 thomas static const struct got_error *
 197 3b6ceab7 2022-01-10 thomas encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
 198 3b6ceab7 2022-01-10 thomas     struct got_delta_instruction *deltas, int ndeltas,
 199 3b6ceab7 2022-01-10 thomas     off_t delta_size, off_t base_size)
 200 3b6ceab7 2022-01-10 thomas {
 201 3b6ceab7 2022-01-10 thomas 	const struct got_error *err;
 202 3b6ceab7 2022-01-10 thomas 	unsigned char buf[16], *bp;
 203 3b6ceab7 2022-01-10 thomas 	int i, j;
 204 3b6ceab7 2022-01-10 thomas 	size_t len = 0;
 205 3b6ceab7 2022-01-10 thomas 	off_t n;
 206 3b6ceab7 2022-01-10 thomas 	struct got_delta_instruction *d;
 207 3b6ceab7 2022-01-10 thomas 
 208 3b6ceab7 2022-01-10 thomas 	m->delta_buf = malloc(delta_size);
 209 3b6ceab7 2022-01-10 thomas 	if (m->delta_buf == NULL)
 210 3b6ceab7 2022-01-10 thomas 		return got_error_from_errno("calloc");
 211 3b6ceab7 2022-01-10 thomas 
 212 3b6ceab7 2022-01-10 thomas 	/* base object size */
 213 3b6ceab7 2022-01-10 thomas 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
 214 3b6ceab7 2022-01-10 thomas 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
 215 3b6ceab7 2022-01-10 thomas 	for (i = 1; n > 0; i++) {
 216 3b6ceab7 2022-01-10 thomas 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
 217 3b6ceab7 2022-01-10 thomas 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
 218 3b6ceab7 2022-01-10 thomas 		n >>= GOT_DELTA_SIZE_SHIFT;
 219 3b6ceab7 2022-01-10 thomas 	}
 220 3b6ceab7 2022-01-10 thomas 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
 221 3b6ceab7 2022-01-10 thomas 	if (err)
 222 3b6ceab7 2022-01-10 thomas 		return err;
 223 3b6ceab7 2022-01-10 thomas 
 224 3b6ceab7 2022-01-10 thomas 	/* target object size */
 225 3b6ceab7 2022-01-10 thomas 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
 226 3b6ceab7 2022-01-10 thomas 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
 227 3b6ceab7 2022-01-10 thomas 	for (i = 1; n > 0; i++) {
 228 3b6ceab7 2022-01-10 thomas 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
 229 3b6ceab7 2022-01-10 thomas 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
 230 3b6ceab7 2022-01-10 thomas 		n >>= GOT_DELTA_SIZE_SHIFT;
 231 3b6ceab7 2022-01-10 thomas 	}
 232 3b6ceab7 2022-01-10 thomas 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
 233 3b6ceab7 2022-01-10 thomas 	if (err)
 234 3b6ceab7 2022-01-10 thomas 		return err;
 235 3b6ceab7 2022-01-10 thomas 
 236 3b6ceab7 2022-01-10 thomas 	for (j = 0; j < ndeltas; j++) {
 237 3b6ceab7 2022-01-10 thomas 		d = &deltas[j];
 238 3b6ceab7 2022-01-10 thomas 		if (d->copy) {
 239 3b6ceab7 2022-01-10 thomas 			n = d->offset;
 240 3b6ceab7 2022-01-10 thomas 			bp = &buf[1];
 241 3b6ceab7 2022-01-10 thomas 			buf[0] = GOT_DELTA_BASE_COPY;
 242 3b6ceab7 2022-01-10 thomas 			for (i = 0; i < 4; i++) {
 243 3b6ceab7 2022-01-10 thomas 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
 244 3b6ceab7 2022-01-10 thomas 				buf[0] |= 1 << i;
 245 3b6ceab7 2022-01-10 thomas 				*bp++ = n & 0xff;
 246 3b6ceab7 2022-01-10 thomas 				n >>= 8;
 247 3b6ceab7 2022-01-10 thomas 				if (n == 0)
 248 3b6ceab7 2022-01-10 thomas 					break;
 249 3b6ceab7 2022-01-10 thomas 			}
 250 3b6ceab7 2022-01-10 thomas 
 251 3b6ceab7 2022-01-10 thomas 			n = d->len;
 252 3b6ceab7 2022-01-10 thomas 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
 253 3b6ceab7 2022-01-10 thomas 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
 254 3b6ceab7 2022-01-10 thomas 				for (i = 0; i < 3 && n > 0; i++) {
 255 3b6ceab7 2022-01-10 thomas 					buf[0] |= 1 << (i + 4);
 256 3b6ceab7 2022-01-10 thomas 					*bp++ = n & 0xff;
 257 3b6ceab7 2022-01-10 thomas 					n >>= 8;
 258 3b6ceab7 2022-01-10 thomas 				}
 259 3b6ceab7 2022-01-10 thomas 			}
 260 3b6ceab7 2022-01-10 thomas 			err = append(&m->delta_buf, &len, &delta_size,
 261 3b6ceab7 2022-01-10 thomas 			    buf, bp - buf);
 262 3b6ceab7 2022-01-10 thomas 			if (err)
 263 3b6ceab7 2022-01-10 thomas 				return err;
 264 3b6ceab7 2022-01-10 thomas 		} else if (o->f == NULL) {
 265 3b6ceab7 2022-01-10 thomas 			n = 0;
 266 3b6ceab7 2022-01-10 thomas 			while (n != d->len) {
 267 3b6ceab7 2022-01-10 thomas 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
 268 3b6ceab7 2022-01-10 thomas 				err = append(&m->delta_buf, &len, &delta_size,
 269 3b6ceab7 2022-01-10 thomas 				    buf, 1);
 270 3b6ceab7 2022-01-10 thomas 				if (err)
 271 3b6ceab7 2022-01-10 thomas 					return err;
 272 3b6ceab7 2022-01-10 thomas 				err = append(&m->delta_buf, &len, &delta_size,
 273 3b6ceab7 2022-01-10 thomas 				    o->data + o->hdrlen + d->offset + n,
 274 3b6ceab7 2022-01-10 thomas 				    buf[0]);
 275 3b6ceab7 2022-01-10 thomas 				if (err)
 276 3b6ceab7 2022-01-10 thomas 					return err;
 277 3b6ceab7 2022-01-10 thomas 				n += buf[0];
 278 3b6ceab7 2022-01-10 thomas 			}
 279 3b6ceab7 2022-01-10 thomas 		} else {
 280 3b6ceab7 2022-01-10 thomas 			char content[128];
 281 3b6ceab7 2022-01-10 thomas 			size_t r;
 282 3b6ceab7 2022-01-10 thomas 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
 283 3b6ceab7 2022-01-10 thomas 				return got_error_from_errno("fseeko");
 284 3b6ceab7 2022-01-10 thomas 			n = 0;
 285 3b6ceab7 2022-01-10 thomas 			while (n != d->len) {
 286 3b6ceab7 2022-01-10 thomas 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
 287 3b6ceab7 2022-01-10 thomas 				err = append(&m->delta_buf, &len, &delta_size,
 288 3b6ceab7 2022-01-10 thomas 				    buf, 1);
 289 3b6ceab7 2022-01-10 thomas 				if (err)
 290 3b6ceab7 2022-01-10 thomas 					return err;
 291 3b6ceab7 2022-01-10 thomas 				r = fread(content, 1, buf[0], o->f);
 292 3b6ceab7 2022-01-10 thomas 				if (r != buf[0])
 293 3b6ceab7 2022-01-10 thomas 					return got_ferror(o->f, GOT_ERR_IO);
 294 3b6ceab7 2022-01-10 thomas 				err = append(&m->delta_buf, &len, &delta_size,
 295 3b6ceab7 2022-01-10 thomas 				    content, buf[0]);
 296 3b6ceab7 2022-01-10 thomas 				if (err)
 297 3b6ceab7 2022-01-10 thomas 					return err;
 298 3b6ceab7 2022-01-10 thomas 				n += buf[0];
 299 3b6ceab7 2022-01-10 thomas 			}
 300 3b6ceab7 2022-01-10 thomas 		}
 301 3b6ceab7 2022-01-10 thomas 	}
 302 3b6ceab7 2022-01-10 thomas 
 303 3b6ceab7 2022-01-10 thomas 	m->delta_len = len;
 304 3b6ceab7 2022-01-10 thomas 	return NULL;
 305 e6bcace5 2021-06-22 stsp }
 306 e6bcace5 2021-06-22 stsp 
 307 b79280dd 2021-10-15 thomas static const struct got_error *
 308 b79280dd 2021-10-15 thomas encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
 309 b79280dd 2021-10-15 thomas     struct got_delta_instruction *deltas, int ndeltas,
 310 d0772de9 2021-10-15 thomas     off_t base_size, FILE *f)
 311 d0772de9 2021-10-15 thomas {
 312 d0772de9 2021-10-15 thomas 	unsigned char buf[16], *bp;
 313 d0772de9 2021-10-15 thomas 	int i, j;
 314 d0772de9 2021-10-15 thomas 	off_t n;
 315 d0772de9 2021-10-15 thomas 	size_t w;
 316 d0772de9 2021-10-15 thomas 	struct got_delta_instruction *d;
 317 d0772de9 2021-10-15 thomas 
 318 d0772de9 2021-10-15 thomas 	/* base object size */
 319 d0772de9 2021-10-15 thomas 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
 320 d0772de9 2021-10-15 thomas 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
 321 d0772de9 2021-10-15 thomas 	for (i = 1; n > 0; i++) {
 322 d0772de9 2021-10-15 thomas 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
 323 d0772de9 2021-10-15 thomas 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
 324 d0772de9 2021-10-15 thomas 		n >>= GOT_DELTA_SIZE_SHIFT;
 325 d0772de9 2021-10-15 thomas 	}
 326 d0772de9 2021-10-15 thomas 	w = fwrite(buf, 1, i, f);
 327 d0772de9 2021-10-15 thomas 	if (w != i)
 328 d0772de9 2021-10-15 thomas 		return got_ferror(f, GOT_ERR_IO);
 329 d0772de9 2021-10-15 thomas 
 330 d0772de9 2021-10-15 thomas 	/* target object size */
 331 d0772de9 2021-10-15 thomas 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
 332 d0772de9 2021-10-15 thomas 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
 333 d0772de9 2021-10-15 thomas 	for (i = 1; n > 0; i++) {
 334 d0772de9 2021-10-15 thomas 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
 335 d0772de9 2021-10-15 thomas 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
 336 d0772de9 2021-10-15 thomas 		n >>= GOT_DELTA_SIZE_SHIFT;
 337 d0772de9 2021-10-15 thomas 	}
 338 d0772de9 2021-10-15 thomas 	w = fwrite(buf, 1, i, f);
 339 d0772de9 2021-10-15 thomas 	if (w != i)
 340 d0772de9 2021-10-15 thomas 		return got_ferror(f, GOT_ERR_IO);
 341 d0772de9 2021-10-15 thomas 
 342 d0772de9 2021-10-15 thomas 	for (j = 0; j < ndeltas; j++) {
 343 d0772de9 2021-10-15 thomas 		d = &deltas[j];
 344 d0772de9 2021-10-15 thomas 		if (d->copy) {
 345 d0772de9 2021-10-15 thomas 			n = d->offset;
 346 d0772de9 2021-10-15 thomas 			bp = &buf[1];
 347 d0772de9 2021-10-15 thomas 			buf[0] = GOT_DELTA_BASE_COPY;
 348 d0772de9 2021-10-15 thomas 			for (i = 0; i < 4; i++) {
 349 d0772de9 2021-10-15 thomas 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
 350 d0772de9 2021-10-15 thomas 				buf[0] |= 1 << i;
 351 d0772de9 2021-10-15 thomas 				*bp++ = n & 0xff;
 352 d0772de9 2021-10-15 thomas 				n >>= 8;
 353 d0772de9 2021-10-15 thomas 				if (n == 0)
 354 d0772de9 2021-10-15 thomas 					break;
 355 d0772de9 2021-10-15 thomas 			}
 356 d0772de9 2021-10-15 thomas 
 357 d0772de9 2021-10-15 thomas 			n = d->len;
 358 d0772de9 2021-10-15 thomas 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
 359 d0772de9 2021-10-15 thomas 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
 360 d0772de9 2021-10-15 thomas 				for (i = 0; i < 3 && n > 0; i++) {
 361 d0772de9 2021-10-15 thomas 					buf[0] |= 1 << (i + 4);
 362 d0772de9 2021-10-15 thomas 					*bp++ = n & 0xff;
 363 d0772de9 2021-10-15 thomas 					n >>= 8;
 364 d0772de9 2021-10-15 thomas 				}
 365 d0772de9 2021-10-15 thomas 			}
 366 d0772de9 2021-10-15 thomas 			w = fwrite(buf, 1, bp - buf, f);
 367 d0772de9 2021-10-15 thomas 			if (w != bp - buf)
 368 d0772de9 2021-10-15 thomas 				return got_ferror(f, GOT_ERR_IO);
 369 2b0ae357 2022-01-10 thomas 		} else if (o->f == NULL) {
 370 2b0ae357 2022-01-10 thomas 			n = 0;
 371 2b0ae357 2022-01-10 thomas 			while (n != d->len) {
 372 2b0ae357 2022-01-10 thomas 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
 373 2b0ae357 2022-01-10 thomas 				w = fwrite(buf, 1, 1, f);
 374 2b0ae357 2022-01-10 thomas 				if (w != 1)
 375 2b0ae357 2022-01-10 thomas 					return got_ferror(f, GOT_ERR_IO);
 376 2b0ae357 2022-01-10 thomas 				w = fwrite(o->data + o->hdrlen + d->offset + n,
 377 2b0ae357 2022-01-10 thomas 				    1, buf[0], f);
 378 2b0ae357 2022-01-10 thomas 				if (w != buf[0])
 379 2b0ae357 2022-01-10 thomas 					return got_ferror(f, GOT_ERR_IO);
 380 2b0ae357 2022-01-10 thomas 				n += buf[0];
 381 2b0ae357 2022-01-10 thomas 			}
 382 d0772de9 2021-10-15 thomas 		} else {
 383 d0772de9 2021-10-15 thomas 			char content[128];
 384 d0772de9 2021-10-15 thomas 			size_t r;
 385 d0772de9 2021-10-15 thomas 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
 386 d0772de9 2021-10-15 thomas 				return got_error_from_errno("fseeko");
 387 d0772de9 2021-10-15 thomas 			n = 0;
 388 d0772de9 2021-10-15 thomas 			while (n != d->len) {
 389 d0772de9 2021-10-15 thomas 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
 390 d0772de9 2021-10-15 thomas 				w = fwrite(buf, 1, 1, f);
 391 d0772de9 2021-10-15 thomas 				if (w != 1)
 392 d0772de9 2021-10-15 thomas 					return got_ferror(f, GOT_ERR_IO);
 393 d0772de9 2021-10-15 thomas 				r = fread(content, 1, buf[0], o->f);
 394 d0772de9 2021-10-15 thomas 				if (r != buf[0])
 395 d0772de9 2021-10-15 thomas 					return got_ferror(o->f, GOT_ERR_IO);
 396 d0772de9 2021-10-15 thomas 				w = fwrite(content, 1, buf[0], f);
 397 d0772de9 2021-10-15 thomas 				if (w != buf[0])
 398 d0772de9 2021-10-15 thomas 					return got_ferror(f, GOT_ERR_IO);
 399 d0772de9 2021-10-15 thomas 				n += buf[0];
 400 d0772de9 2021-10-15 thomas 			}
 401 d0772de9 2021-10-15 thomas 		}
 402 d0772de9 2021-10-15 thomas 	}
 403 e6bcace5 2021-06-22 stsp 
 404 3b6ceab7 2022-01-10 thomas 	m->delta_len = ftello(f) - m->delta_offset;
 405 d0772de9 2021-10-15 thomas 	return NULL;
 406 d0772de9 2021-10-15 thomas }
 407 31ba2236 2022-01-05 thomas 
 408 31ba2236 2022-01-05 thomas static const struct got_error *
 409 31ba2236 2022-01-05 thomas report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
 410 85220b0e 2022-03-22 thomas     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
 411 85220b0e 2022-03-22 thomas     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
 412 85220b0e 2022-03-22 thomas     int nobj_written)
 413 31ba2236 2022-01-05 thomas {
 414 31ba2236 2022-01-05 thomas 	const struct got_error *err;
 415 31ba2236 2022-01-05 thomas 	int elapsed;
 416 31ba2236 2022-01-05 thomas 
 417 31ba2236 2022-01-05 thomas 	if (progress_cb == NULL)
 418 31ba2236 2022-01-05 thomas 		return NULL;
 419 d0772de9 2021-10-15 thomas 
 420 31ba2236 2022-01-05 thomas 	err = got_ratelimit_check(&elapsed, rl);
 421 31ba2236 2022-01-05 thomas 	if (err || !elapsed)
 422 31ba2236 2022-01-05 thomas 		return err;
 423 d0772de9 2021-10-15 thomas 
 424 85220b0e 2022-03-22 thomas 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
 425 85220b0e 2022-03-22 thomas 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
 426 31ba2236 2022-01-05 thomas }
 427 31ba2236 2022-01-05 thomas 
 428 e6bcace5 2021-06-22 stsp static const struct got_error *
 429 f9c2e8e5 2022-02-13 thomas add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
 430 f9c2e8e5 2022-02-13 thomas {
 431 f9c2e8e5 2022-02-13 thomas 	if (v->nmeta == v->metasz){
 432 f9c2e8e5 2022-02-13 thomas 		size_t newsize = 2 * v->metasz;
 433 f9c2e8e5 2022-02-13 thomas 		struct got_pack_meta **new;
 434 f9c2e8e5 2022-02-13 thomas 		new = reallocarray(v->meta, newsize, sizeof(*new));
 435 f9c2e8e5 2022-02-13 thomas 		if (new == NULL)
 436 f9c2e8e5 2022-02-13 thomas 			return got_error_from_errno("reallocarray");
 437 f9c2e8e5 2022-02-13 thomas 		v->meta = new;
 438 f9c2e8e5 2022-02-13 thomas 		v->metasz = newsize; 
 439 f9c2e8e5 2022-02-13 thomas 	}
 440 f9c2e8e5 2022-02-13 thomas 
 441 f9c2e8e5 2022-02-13 thomas 	v->meta[v->nmeta++] = m;
 442 f9c2e8e5 2022-02-13 thomas 	return NULL;
 443 f9c2e8e5 2022-02-13 thomas }
 444 f9c2e8e5 2022-02-13 thomas 
 445 f9c2e8e5 2022-02-13 thomas static const struct got_error *
 446 f9c2e8e5 2022-02-13 thomas reuse_delta(int idx, struct got_pack_meta *m, struct got_pack_metavec *v,
 447 f9c2e8e5 2022-02-13 thomas     struct got_object_idset *idset, struct got_pack *pack,
 448 f9c2e8e5 2022-02-13 thomas     struct got_packidx *packidx, int delta_cache_fd,
 449 f9c2e8e5 2022-02-13 thomas     struct got_repository *repo)
 450 f9c2e8e5 2022-02-13 thomas {
 451 f9c2e8e5 2022-02-13 thomas 	const struct got_error *err = NULL;
 452 f9c2e8e5 2022-02-13 thomas 	struct got_pack_meta *base = NULL;
 453 f9c2e8e5 2022-02-13 thomas 	struct got_object_id *base_obj_id = NULL;
 454 f9c2e8e5 2022-02-13 thomas 	off_t delta_len = 0, delta_offset = 0, delta_cache_offset = 0;
 455 f9c2e8e5 2022-02-13 thomas 	uint64_t base_size, result_size;
 456 f9c2e8e5 2022-02-13 thomas 
 457 f9c2e8e5 2022-02-13 thomas 	if (m->have_reused_delta)
 458 f9c2e8e5 2022-02-13 thomas 		return NULL;
 459 f9c2e8e5 2022-02-13 thomas 
 460 f9c2e8e5 2022-02-13 thomas 	err = got_object_read_raw_delta(&base_size, &result_size, &delta_len,
 461 f9c2e8e5 2022-02-13 thomas 	    &delta_offset, &delta_cache_offset, &base_obj_id, delta_cache_fd,
 462 f9c2e8e5 2022-02-13 thomas 	    packidx, idx, &m->id, repo);
 463 f9c2e8e5 2022-02-13 thomas 	if (err)
 464 f9c2e8e5 2022-02-13 thomas 		return err;
 465 f9c2e8e5 2022-02-13 thomas 
 466 f9c2e8e5 2022-02-13 thomas 	if (delta_offset + delta_len < delta_offset)
 467 f9c2e8e5 2022-02-13 thomas 		return got_error(GOT_ERR_BAD_PACKFILE);
 468 f9c2e8e5 2022-02-13 thomas 
 469 f9c2e8e5 2022-02-13 thomas 	base = got_object_idset_get(idset, base_obj_id);
 470 f9c2e8e5 2022-02-13 thomas 	if (base == NULL)
 471 f9c2e8e5 2022-02-13 thomas 		goto done;
 472 f9c2e8e5 2022-02-13 thomas 
 473 f9c2e8e5 2022-02-13 thomas 	m->delta_len = delta_len;
 474 f9c2e8e5 2022-02-13 thomas 	m->delta_offset = delta_cache_offset;
 475 f9c2e8e5 2022-02-13 thomas 	m->prev = base;
 476 f9c2e8e5 2022-02-13 thomas 	m->size = result_size;
 477 f9c2e8e5 2022-02-13 thomas 	m->have_reused_delta = 1;
 478 f9c2e8e5 2022-02-13 thomas 	m->reused_delta_offset = delta_offset;
 479 f9c2e8e5 2022-02-13 thomas 	m->base_obj_id = base_obj_id;
 480 f9c2e8e5 2022-02-13 thomas 	base_obj_id = NULL;
 481 f9c2e8e5 2022-02-13 thomas 	err = add_meta(m, v);
 482 f9c2e8e5 2022-02-13 thomas done:
 483 f9c2e8e5 2022-02-13 thomas 	free(base_obj_id);
 484 f9c2e8e5 2022-02-13 thomas 	return err;
 485 f9c2e8e5 2022-02-13 thomas }
 486 f9c2e8e5 2022-02-13 thomas 
 487 f9c2e8e5 2022-02-13 thomas static const struct got_error *
 488 f9c2e8e5 2022-02-13 thomas find_pack_for_reuse(struct got_packidx **best_packidx,
 489 f9c2e8e5 2022-02-13 thomas     struct got_repository *repo)
 490 f9c2e8e5 2022-02-13 thomas {
 491 de47d040 2022-03-22 thomas 	const struct got_error *err = NULL;
 492 f9c2e8e5 2022-02-13 thomas 	struct got_pathlist_entry *pe;
 493 f9c2e8e5 2022-02-13 thomas 	const char *best_packidx_path = NULL;
 494 f9c2e8e5 2022-02-13 thomas 	int nobj_max = 0;
 495 f9c2e8e5 2022-02-13 thomas 
 496 f9c2e8e5 2022-02-13 thomas 	*best_packidx = NULL;
 497 f9c2e8e5 2022-02-13 thomas 
 498 de47d040 2022-03-22 thomas 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
 499 f9c2e8e5 2022-02-13 thomas 		const char *path_packidx = pe->path;
 500 f9c2e8e5 2022-02-13 thomas 		struct got_packidx *packidx;
 501 f9c2e8e5 2022-02-13 thomas 		int nobj;
 502 f9c2e8e5 2022-02-13 thomas 
 503 f9c2e8e5 2022-02-13 thomas 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
 504 f9c2e8e5 2022-02-13 thomas 		if (err)
 505 f9c2e8e5 2022-02-13 thomas 			break;
 506 f9c2e8e5 2022-02-13 thomas 
 507 f9c2e8e5 2022-02-13 thomas 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
 508 f9c2e8e5 2022-02-13 thomas 		if (nobj > nobj_max) {
 509 f9c2e8e5 2022-02-13 thomas 			best_packidx_path = path_packidx;
 510 f9c2e8e5 2022-02-13 thomas 			nobj_max = nobj;
 511 f9c2e8e5 2022-02-13 thomas 		}
 512 f9c2e8e5 2022-02-13 thomas 	}
 513 f9c2e8e5 2022-02-13 thomas 
 514 f9c2e8e5 2022-02-13 thomas 	if (best_packidx_path) {
 515 f9c2e8e5 2022-02-13 thomas 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
 516 f9c2e8e5 2022-02-13 thomas 		    repo);
 517 f9c2e8e5 2022-02-13 thomas 	}
 518 f9c2e8e5 2022-02-13 thomas 
 519 f9c2e8e5 2022-02-13 thomas 	return err;
 520 f9c2e8e5 2022-02-13 thomas }
 521 f9c2e8e5 2022-02-13 thomas 
 522 f9c2e8e5 2022-02-13 thomas struct search_deltas_arg {
 523 f9c2e8e5 2022-02-13 thomas 	struct got_packidx *packidx;
 524 f9c2e8e5 2022-02-13 thomas 	struct got_pack *pack;
 525 f9c2e8e5 2022-02-13 thomas 	struct got_object_idset *idset;
 526 f9c2e8e5 2022-02-13 thomas 	struct got_pack_metavec *v;
 527 f9c2e8e5 2022-02-13 thomas 	int delta_cache_fd;
 528 f9c2e8e5 2022-02-13 thomas 	struct got_repository *repo;
 529 f9c2e8e5 2022-02-13 thomas 	got_pack_progress_cb progress_cb;
 530 f9c2e8e5 2022-02-13 thomas 	void *progress_arg;
 531 f9c2e8e5 2022-02-13 thomas 	struct got_ratelimit *rl;
 532 f9c2e8e5 2022-02-13 thomas 	got_cancel_cb cancel_cb;
 533 f9c2e8e5 2022-02-13 thomas 	void *cancel_arg;
 534 85220b0e 2022-03-22 thomas 	int ncolored;
 535 85220b0e 2022-03-22 thomas 	int nfound;
 536 85220b0e 2022-03-22 thomas 	int ntrees;
 537 f9c2e8e5 2022-02-13 thomas 	int ncommits;
 538 f9c2e8e5 2022-02-13 thomas };
 539 f9c2e8e5 2022-02-13 thomas 
 540 f9c2e8e5 2022-02-13 thomas static const struct got_error *
 541 f9c2e8e5 2022-02-13 thomas search_delta_for_object(struct got_object_id *id, void *data, void *arg)
 542 f9c2e8e5 2022-02-13 thomas {
 543 f9c2e8e5 2022-02-13 thomas 	const struct got_error *err;
 544 f9c2e8e5 2022-02-13 thomas 	struct got_pack_meta *m = data;
 545 f9c2e8e5 2022-02-13 thomas 	struct search_deltas_arg *a = arg;
 546 f9c2e8e5 2022-02-13 thomas 	int obj_idx;
 547 f9c2e8e5 2022-02-13 thomas 	struct got_object *obj = NULL;
 548 f9c2e8e5 2022-02-13 thomas 
 549 f9c2e8e5 2022-02-13 thomas 	if (a->cancel_cb) {
 550 f9c2e8e5 2022-02-13 thomas 		err = (*a->cancel_cb)(a->cancel_arg);
 551 f9c2e8e5 2022-02-13 thomas 		if (err)
 552 f9c2e8e5 2022-02-13 thomas 			return err;
 553 f9c2e8e5 2022-02-13 thomas 	}
 554 f9c2e8e5 2022-02-13 thomas 
 555 f9c2e8e5 2022-02-13 thomas 	if (!got_repo_check_packidx_bloom_filter(a->repo,
 556 f9c2e8e5 2022-02-13 thomas 	    a->packidx->path_packidx, id))
 557 f9c2e8e5 2022-02-13 thomas 		return NULL;
 558 f9c2e8e5 2022-02-13 thomas 
 559 f9c2e8e5 2022-02-13 thomas 	obj_idx = got_packidx_get_object_idx(a->packidx, id);
 560 f9c2e8e5 2022-02-13 thomas 	if (obj_idx == -1)
 561 f9c2e8e5 2022-02-13 thomas 		return NULL;
 562 f9c2e8e5 2022-02-13 thomas 
 563 f9c2e8e5 2022-02-13 thomas 	/* TODO:
 564 f9c2e8e5 2022-02-13 thomas 	 * Opening and closing an object just to check its flags
 565 f9c2e8e5 2022-02-13 thomas 	 * is a bit expensive. We could have an imsg which requests
 566 f9c2e8e5 2022-02-13 thomas 	 * plain type/size information for an object without doing
 567 f9c2e8e5 2022-02-13 thomas 	 * work such as traversing the object's entire delta chain
 568 f9c2e8e5 2022-02-13 thomas 	 * to find the base object type, and other such info which
 569 f9c2e8e5 2022-02-13 thomas 	 * we don't really need here.
 570 f9c2e8e5 2022-02-13 thomas 	 */
 571 f9c2e8e5 2022-02-13 thomas 	err = got_object_open_from_packfile(&obj, &m->id, a->pack,
 572 f9c2e8e5 2022-02-13 thomas 	    a->packidx, obj_idx, a->repo);
 573 f9c2e8e5 2022-02-13 thomas 	if (err)
 574 f9c2e8e5 2022-02-13 thomas 		return err;
 575 f9c2e8e5 2022-02-13 thomas 
 576 f9c2e8e5 2022-02-13 thomas 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
 577 f9c2e8e5 2022-02-13 thomas 		reuse_delta(obj_idx, m, a->v, a->idset, a->pack, a->packidx,
 578 f9c2e8e5 2022-02-13 thomas 		    a->delta_cache_fd, a->repo);
 579 f9c2e8e5 2022-02-13 thomas 		if (err)
 580 f9c2e8e5 2022-02-13 thomas 			goto done;
 581 f9c2e8e5 2022-02-13 thomas 		err = report_progress(a->progress_cb, a->progress_arg, a->rl,
 582 85220b0e 2022-03-22 thomas 		    a->ncolored, a->nfound, a->ntrees, 0L, a->ncommits,
 583 85220b0e 2022-03-22 thomas 		    got_object_idset_num_elements(a->idset), a->v->nmeta, 0);
 584 f9c2e8e5 2022-02-13 thomas 	}
 585 f9c2e8e5 2022-02-13 thomas done:
 586 f9c2e8e5 2022-02-13 thomas 	got_object_close(obj);
 587 f9c2e8e5 2022-02-13 thomas 	return err;
 588 f9c2e8e5 2022-02-13 thomas }
 589 f9c2e8e5 2022-02-13 thomas 
 590 f9c2e8e5 2022-02-13 thomas static const struct got_error *
 591 f9c2e8e5 2022-02-13 thomas search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
 592 85220b0e 2022-03-22 thomas     int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
 593 85220b0e 2022-03-22 thomas     struct got_repository *repo,
 594 f9c2e8e5 2022-02-13 thomas     got_pack_progress_cb progress_cb, void *progress_arg,
 595 f9c2e8e5 2022-02-13 thomas     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
 596 f9c2e8e5 2022-02-13 thomas {
 597 f9c2e8e5 2022-02-13 thomas 	const struct got_error *err = NULL;
 598 f9c2e8e5 2022-02-13 thomas 	char *path_packfile = NULL;
 599 f9c2e8e5 2022-02-13 thomas 	struct got_packidx *packidx;
 600 f9c2e8e5 2022-02-13 thomas 	struct got_pack *pack;
 601 f9c2e8e5 2022-02-13 thomas 	struct search_deltas_arg sda;
 602 f9c2e8e5 2022-02-13 thomas 
 603 f9c2e8e5 2022-02-13 thomas 	err = find_pack_for_reuse(&packidx, repo);
 604 f9c2e8e5 2022-02-13 thomas 	if (err)
 605 f9c2e8e5 2022-02-13 thomas 		return err;
 606 f9c2e8e5 2022-02-13 thomas 
 607 f9c2e8e5 2022-02-13 thomas 	if (packidx == NULL)
 608 f9c2e8e5 2022-02-13 thomas 		return NULL;
 609 f9c2e8e5 2022-02-13 thomas 
 610 f9c2e8e5 2022-02-13 thomas 	err = got_packidx_get_packfile_path(&path_packfile,
 611 f9c2e8e5 2022-02-13 thomas 	    packidx->path_packidx);
 612 f9c2e8e5 2022-02-13 thomas 	if (err)
 613 f9c2e8e5 2022-02-13 thomas 		return err;
 614 f9c2e8e5 2022-02-13 thomas 
 615 f9c2e8e5 2022-02-13 thomas 	pack = got_repo_get_cached_pack(repo, path_packfile);
 616 f9c2e8e5 2022-02-13 thomas 	if (pack == NULL) {
 617 f9c2e8e5 2022-02-13 thomas 		err = got_repo_cache_pack(&pack, repo, path_packfile, packidx);
 618 f9c2e8e5 2022-02-13 thomas 		if (err)
 619 f9c2e8e5 2022-02-13 thomas 			goto done;
 620 f9c2e8e5 2022-02-13 thomas 	}
 621 f9c2e8e5 2022-02-13 thomas 
 622 f9c2e8e5 2022-02-13 thomas 	sda.packidx = packidx;
 623 f9c2e8e5 2022-02-13 thomas 	sda.pack = pack;
 624 f9c2e8e5 2022-02-13 thomas 	sda.idset = idset;
 625 f9c2e8e5 2022-02-13 thomas 	sda.v = v;
 626 f9c2e8e5 2022-02-13 thomas 	sda.delta_cache_fd = delta_cache_fd;
 627 f9c2e8e5 2022-02-13 thomas 	sda.repo = repo;
 628 f9c2e8e5 2022-02-13 thomas 	sda.progress_cb = progress_cb;
 629 f9c2e8e5 2022-02-13 thomas 	sda.progress_arg = progress_arg;
 630 f9c2e8e5 2022-02-13 thomas 	sda.rl = rl;
 631 f9c2e8e5 2022-02-13 thomas 	sda.cancel_cb = cancel_cb;
 632 f9c2e8e5 2022-02-13 thomas 	sda.cancel_arg = cancel_arg;
 633 85220b0e 2022-03-22 thomas 	sda.ncolored = ncolored;
 634 85220b0e 2022-03-22 thomas 	sda.nfound = nfound;
 635 85220b0e 2022-03-22 thomas 	sda.ntrees = ntrees;
 636 f9c2e8e5 2022-02-13 thomas 	sda.ncommits = ncommits;
 637 f9c2e8e5 2022-02-13 thomas 	err = got_object_idset_for_each(idset, search_delta_for_object, &sda);
 638 f9c2e8e5 2022-02-13 thomas done:
 639 f9c2e8e5 2022-02-13 thomas 	free(path_packfile);
 640 f9c2e8e5 2022-02-13 thomas 	return err;
 641 f9c2e8e5 2022-02-13 thomas }
 642 f9c2e8e5 2022-02-13 thomas 
 643 f9c2e8e5 2022-02-13 thomas static const struct got_error *
 644 85220b0e 2022-03-22 thomas pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
 645 85220b0e 2022-03-22 thomas     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
 646 85220b0e 2022-03-22 thomas     struct got_repository *repo,
 647 05118f5a 2021-06-22 stsp     got_pack_progress_cb progress_cb, void *progress_arg,
 648 31ba2236 2022-01-05 thomas     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
 649 e6bcace5 2021-06-22 stsp {
 650 e6bcace5 2021-06-22 stsp 	const struct got_error *err = NULL;
 651 e6bcace5 2021-06-22 stsp 	struct got_pack_meta *m = NULL, *base = NULL;
 652 e6bcace5 2021-06-22 stsp 	struct got_raw_object *raw = NULL, *base_raw = NULL;
 653 b79280dd 2021-10-15 thomas 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
 654 3b6ceab7 2022-01-10 thomas 	int i, j, ndeltas, best_ndeltas;
 655 3b6ceab7 2022-01-10 thomas 	off_t size, best_size;
 656 959daf23 2021-10-29 thomas 	const int max_base_candidates = 3;
 657 510885f7 2022-01-10 thomas 	size_t delta_memsize = 0;
 658 510885f7 2022-01-10 thomas 	const size_t max_delta_memsize = 25 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
 659 ecf9545f 2021-10-15 thomas 	int outfd = -1;
 660 e6bcace5 2021-06-22 stsp 
 661 e6bcace5 2021-06-22 stsp 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
 662 e6bcace5 2021-06-22 stsp 	for (i = 0; i < nmeta; i++) {
 663 e6bcace5 2021-06-22 stsp 		if (cancel_cb) {
 664 e6bcace5 2021-06-22 stsp 			err = (*cancel_cb)(cancel_arg);
 665 e6bcace5 2021-06-22 stsp 			if (err)
 666 e6bcace5 2021-06-22 stsp 				break;
 667 e6bcace5 2021-06-22 stsp 		}
 668 31ba2236 2022-01-05 thomas 		err = report_progress(progress_cb, progress_arg, rl,
 669 85220b0e 2022-03-22 thomas 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
 670 85220b0e 2022-03-22 thomas 		    nreused + i, 0);
 671 31ba2236 2022-01-05 thomas 		if (err)
 672 31ba2236 2022-01-05 thomas 			goto done;
 673 e6bcace5 2021-06-22 stsp 		m = meta[i];
 674 e6bcace5 2021-06-22 stsp 
 675 e6bcace5 2021-06-22 stsp 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
 676 e6bcace5 2021-06-22 stsp 		    m->obj_type == GOT_OBJ_TYPE_TAG)
 677 e6bcace5 2021-06-22 stsp 			continue;
 678 e6bcace5 2021-06-22 stsp 
 679 f8bb1d3e 2021-10-17 thomas 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
 680 e6bcace5 2021-06-22 stsp 		if (err)
 681 e6bcace5 2021-06-22 stsp 			goto done;
 682 ab6186ae 2021-10-15 thomas 		m->size = raw->size;
 683 e6bcace5 2021-06-22 stsp 
 684 2b0ae357 2022-01-10 thomas 		if (raw->f == NULL) {
 685 2b0ae357 2022-01-10 thomas 			err = got_deltify_init_mem(&m->dtab, raw->data,
 686 2b0ae357 2022-01-10 thomas 			    raw->hdrlen, raw->size + raw->hdrlen);
 687 2b0ae357 2022-01-10 thomas 		} else {
 688 2b0ae357 2022-01-10 thomas 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
 689 2b0ae357 2022-01-10 thomas 			    raw->size + raw->hdrlen);
 690 2b0ae357 2022-01-10 thomas 		}
 691 e6bcace5 2021-06-22 stsp 		if (err)
 692 e6bcace5 2021-06-22 stsp 			goto done;
 693 e6bcace5 2021-06-22 stsp 
 694 e6bcace5 2021-06-22 stsp 		if (i > max_base_candidates) {
 695 e6bcace5 2021-06-22 stsp 			struct got_pack_meta *n = NULL;
 696 e6bcace5 2021-06-22 stsp 			n = meta[i - (max_base_candidates + 1)];
 697 e6bcace5 2021-06-22 stsp 			got_deltify_free(n->dtab);
 698 e6bcace5 2021-06-22 stsp 			n->dtab = NULL;
 699 e6bcace5 2021-06-22 stsp 		}
 700 e6bcace5 2021-06-22 stsp 
 701 b79280dd 2021-10-15 thomas 		best_size = raw->size;
 702 b79280dd 2021-10-15 thomas 		best_ndeltas = 0;
 703 e6bcace5 2021-06-22 stsp 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
 704 e6bcace5 2021-06-22 stsp 			if (cancel_cb) {
 705 e6bcace5 2021-06-22 stsp 				err = (*cancel_cb)(cancel_arg);
 706 e6bcace5 2021-06-22 stsp 				if (err)
 707 e6bcace5 2021-06-22 stsp 					goto done;
 708 e6bcace5 2021-06-22 stsp 			}
 709 e6bcace5 2021-06-22 stsp 			base = meta[j];
 710 e6bcace5 2021-06-22 stsp 			/* long chains make unpacking slow, avoid such bases */
 711 62a05ae9 2021-10-29 thomas 			if (base->nchain >= 128 ||
 712 e6bcace5 2021-06-22 stsp 			    base->obj_type != m->obj_type)
 713 e6bcace5 2021-06-22 stsp 				continue;
 714 e6bcace5 2021-06-22 stsp 
 715 8ab9215c 2021-10-15 thomas 			err = got_object_raw_open(&base_raw, &outfd, repo,
 716 f8bb1d3e 2021-10-17 thomas 			    &base->id);
 717 e6bcace5 2021-06-22 stsp 			if (err)
 718 e6bcace5 2021-06-22 stsp 				goto done;
 719 f9c2e8e5 2022-02-13 thomas 
 720 2b0ae357 2022-01-10 thomas 			if (raw->f == NULL && base_raw->f == NULL) {
 721 2b0ae357 2022-01-10 thomas 				err = got_deltify_mem_mem(&deltas, &ndeltas,
 722 2b0ae357 2022-01-10 thomas 				    raw->data, raw->hdrlen,
 723 2b0ae357 2022-01-10 thomas 				    raw->size + raw->hdrlen,
 724 2b0ae357 2022-01-10 thomas 				    base->dtab, base_raw->data,
 725 2b0ae357 2022-01-10 thomas 				    base_raw->hdrlen,
 726 2b0ae357 2022-01-10 thomas 				    base_raw->size + base_raw->hdrlen);
 727 2b0ae357 2022-01-10 thomas 			} else if (raw->f == NULL) {
 728 2b0ae357 2022-01-10 thomas 				err = got_deltify_mem_file(&deltas, &ndeltas,
 729 2b0ae357 2022-01-10 thomas 				    raw->data, raw->hdrlen,
 730 2b0ae357 2022-01-10 thomas 				    raw->size + raw->hdrlen,
 731 2b0ae357 2022-01-10 thomas 				    base->dtab, base_raw->f,
 732 2b0ae357 2022-01-10 thomas 				    base_raw->hdrlen,
 733 2b0ae357 2022-01-10 thomas 				    base_raw->size + base_raw->hdrlen);
 734 2b0ae357 2022-01-10 thomas 			} else if (base_raw->f == NULL) {
 735 2b0ae357 2022-01-10 thomas 				err = got_deltify_file_mem(&deltas, &ndeltas,
 736 2b0ae357 2022-01-10 thomas 				    raw->f, raw->hdrlen,
 737 2b0ae357 2022-01-10 thomas 				    raw->size + raw->hdrlen,
 738 2b0ae357 2022-01-10 thomas 				    base->dtab, base_raw->data,
 739 2b0ae357 2022-01-10 thomas 				    base_raw->hdrlen,
 740 2b0ae357 2022-01-10 thomas 				    base_raw->size + base_raw->hdrlen);
 741 2b0ae357 2022-01-10 thomas 			} else {
 742 2b0ae357 2022-01-10 thomas 				err = got_deltify(&deltas, &ndeltas,
 743 2b0ae357 2022-01-10 thomas 				    raw->f, raw->hdrlen,
 744 2b0ae357 2022-01-10 thomas 				    raw->size + raw->hdrlen,
 745 2b0ae357 2022-01-10 thomas 				    base->dtab, base_raw->f, base_raw->hdrlen,
 746 2b0ae357 2022-01-10 thomas 				    base_raw->size + base_raw->hdrlen);
 747 2b0ae357 2022-01-10 thomas 			}
 748 e6bcace5 2021-06-22 stsp 			got_object_raw_close(base_raw);
 749 e6bcace5 2021-06-22 stsp 			base_raw = NULL;
 750 e6bcace5 2021-06-22 stsp 			if (err)
 751 e6bcace5 2021-06-22 stsp 				goto done;
 752 e6bcace5 2021-06-22 stsp 
 753 e6bcace5 2021-06-22 stsp 			size = delta_size(deltas, ndeltas);
 754 b79280dd 2021-10-15 thomas 			if (size + 32 < best_size){
 755 e6bcace5 2021-06-22 stsp 				/*
 756 e6bcace5 2021-06-22 stsp 				 * if we already picked a best delta,
 757 e6bcace5 2021-06-22 stsp 				 * replace it.
 758 e6bcace5 2021-06-22 stsp 				 */
 759 b79280dd 2021-10-15 thomas 				best_size = size;
 760 b79280dd 2021-10-15 thomas 				free(best_deltas);
 761 b79280dd 2021-10-15 thomas 				best_deltas = deltas;
 762 b79280dd 2021-10-15 thomas 				best_ndeltas = ndeltas;
 763 b79280dd 2021-10-15 thomas 				deltas = NULL;
 764 e6bcace5 2021-06-22 stsp 				m->nchain = base->nchain + 1;
 765 e6bcace5 2021-06-22 stsp 				m->prev = base;
 766 e6bcace5 2021-06-22 stsp 				m->head = base->head;
 767 e6bcace5 2021-06-22 stsp 				if (m->head == NULL)
 768 e6bcace5 2021-06-22 stsp 					m->head = base;
 769 e6bcace5 2021-06-22 stsp 			} else {
 770 e6bcace5 2021-06-22 stsp 				free(deltas);
 771 e6bcace5 2021-06-22 stsp 				deltas = NULL;
 772 e6bcace5 2021-06-22 stsp 				ndeltas = 0;
 773 e6bcace5 2021-06-22 stsp 			}
 774 e6bcace5 2021-06-22 stsp 		}
 775 e6bcace5 2021-06-22 stsp 
 776 b79280dd 2021-10-15 thomas 		if (best_ndeltas > 0) {
 777 510885f7 2022-01-10 thomas 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
 778 510885f7 2022-01-10 thomas 			    delta_memsize + best_size <= max_delta_memsize) {
 779 510885f7 2022-01-10 thomas 				delta_memsize += best_size;
 780 3b6ceab7 2022-01-10 thomas 				err = encode_delta_in_mem(m, raw, best_deltas,
 781 3b6ceab7 2022-01-10 thomas 				    best_ndeltas, best_size, m->prev->size);
 782 3b6ceab7 2022-01-10 thomas 			} else {
 783 3b6ceab7 2022-01-10 thomas 				m->delta_offset = ftello(delta_cache);
 784 f9c2e8e5 2022-02-13 thomas 				/*
 785 f9c2e8e5 2022-02-13 thomas 				 * TODO:
 786 f9c2e8e5 2022-02-13 thomas 				 * Storing compressed delta data in the delta
 787 f9c2e8e5 2022-02-13 thomas 				 * cache file would probably be more efficient
 788 f9c2e8e5 2022-02-13 thomas 				 * than writing uncompressed delta data here
 789 f9c2e8e5 2022-02-13 thomas 				 * and compressing it while writing the pack
 790 f9c2e8e5 2022-02-13 thomas 				 * file. This would also allow for reusing
 791 f9c2e8e5 2022-02-13 thomas 				 * deltas in their compressed form.
 792 f9c2e8e5 2022-02-13 thomas 				 */
 793 3b6ceab7 2022-01-10 thomas 				err = encode_delta(m, raw, best_deltas,
 794 3b6ceab7 2022-01-10 thomas 				    best_ndeltas, m->prev->size, delta_cache);
 795 3b6ceab7 2022-01-10 thomas 			}
 796 b79280dd 2021-10-15 thomas 			free(best_deltas);
 797 b79280dd 2021-10-15 thomas 			best_deltas = NULL;
 798 b79280dd 2021-10-15 thomas 			best_ndeltas = 0;
 799 b79280dd 2021-10-15 thomas 			if (err)
 800 b79280dd 2021-10-15 thomas 				goto done;
 801 b79280dd 2021-10-15 thomas 		}
 802 b79280dd 2021-10-15 thomas 
 803 e6bcace5 2021-06-22 stsp 		got_object_raw_close(raw);
 804 e6bcace5 2021-06-22 stsp 		raw = NULL;
 805 e6bcace5 2021-06-22 stsp 	}
 806 e6bcace5 2021-06-22 stsp done:
 807 e6bcace5 2021-06-22 stsp 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
 808 e6bcace5 2021-06-22 stsp 		got_deltify_free(meta[i]->dtab);
 809 e6bcace5 2021-06-22 stsp 		meta[i]->dtab = NULL;
 810 e6bcace5 2021-06-22 stsp 	}
 811 e6bcace5 2021-06-22 stsp 	if (raw)
 812 e6bcace5 2021-06-22 stsp 		got_object_raw_close(raw);
 813 e6bcace5 2021-06-22 stsp 	if (base_raw)
 814 e6bcace5 2021-06-22 stsp 		got_object_raw_close(base_raw);
 815 ecf9545f 2021-10-15 thomas 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
 816 ecf9545f 2021-10-15 thomas 		err = got_error_from_errno("close");
 817 b79280dd 2021-10-15 thomas 	free(deltas);
 818 b79280dd 2021-10-15 thomas 	free(best_deltas);
 819 e6bcace5 2021-06-22 stsp 	return err;
 820 e6bcace5 2021-06-22 stsp }
 821 e6bcace5 2021-06-22 stsp 
 822 e6bcace5 2021-06-22 stsp static const struct got_error *
 823 05118f5a 2021-06-22 stsp search_packidx(int *found, struct got_object_id *id,
 824 05118f5a 2021-06-22 stsp     struct got_repository *repo)
 825 05118f5a 2021-06-22 stsp {
 826 05118f5a 2021-06-22 stsp 	const struct got_error *err = NULL;
 827 05118f5a 2021-06-22 stsp 	struct got_packidx *packidx = NULL;
 828 05118f5a 2021-06-22 stsp 	int idx;
 829 05118f5a 2021-06-22 stsp 
 830 05118f5a 2021-06-22 stsp 	*found = 0;
 831 05118f5a 2021-06-22 stsp 
 832 05118f5a 2021-06-22 stsp 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
 833 05118f5a 2021-06-22 stsp 	if (err == NULL)
 834 05118f5a 2021-06-22 stsp 		*found = 1; /* object is already packed */
 835 05118f5a 2021-06-22 stsp 	else if (err->code == GOT_ERR_NO_OBJ)
 836 05118f5a 2021-06-22 stsp 		err = NULL;
 837 05118f5a 2021-06-22 stsp 	return err;
 838 05118f5a 2021-06-22 stsp }
 839 07165b17 2021-07-01 stsp 
 840 07165b17 2021-07-01 stsp static const int obj_types[] = {
 841 07165b17 2021-07-01 stsp 	GOT_OBJ_TYPE_ANY,
 842 07165b17 2021-07-01 stsp 	GOT_OBJ_TYPE_COMMIT,
 843 07165b17 2021-07-01 stsp 	GOT_OBJ_TYPE_TREE,
 844 07165b17 2021-07-01 stsp 	GOT_OBJ_TYPE_BLOB,
 845 07165b17 2021-07-01 stsp 	GOT_OBJ_TYPE_TAG,
 846 07165b17 2021-07-01 stsp 	GOT_OBJ_TYPE_OFFSET_DELTA,
 847 07165b17 2021-07-01 stsp 	GOT_OBJ_TYPE_REF_DELTA
 848 07165b17 2021-07-01 stsp };
 849 05118f5a 2021-06-22 stsp 
 850 05118f5a 2021-06-22 stsp static const struct got_error *
 851 f9c2e8e5 2022-02-13 thomas add_object(int want_meta, struct got_object_idset *idset,
 852 e6bcace5 2021-06-22 stsp     struct got_object_id *id, const char *path, int obj_type,
 853 05118f5a 2021-06-22 stsp     time_t mtime, int loose_obj_only, struct got_repository *repo)
 854 e6bcace5 2021-06-22 stsp {
 855 e6bcace5 2021-06-22 stsp 	const struct got_error *err;
 856 f9c2e8e5 2022-02-13 thomas 	struct got_pack_meta *m = NULL;
 857 e6bcace5 2021-06-22 stsp 
 858 05118f5a 2021-06-22 stsp 	if (loose_obj_only) {
 859 05118f5a 2021-06-22 stsp 		int is_packed;
 860 05118f5a 2021-06-22 stsp 		err = search_packidx(&is_packed, id, repo);
 861 05118f5a 2021-06-22 stsp 		if (err)
 862 05118f5a 2021-06-22 stsp 			return err;
 863 05118f5a 2021-06-22 stsp 		if (is_packed)
 864 05118f5a 2021-06-22 stsp 			return NULL;
 865 05118f5a 2021-06-22 stsp 	}
 866 05118f5a 2021-06-22 stsp 
 867 f9c2e8e5 2022-02-13 thomas 	if (want_meta) {
 868 f9c2e8e5 2022-02-13 thomas 		err = alloc_meta(&m, id, path, obj_type, mtime);
 869 f9c2e8e5 2022-02-13 thomas 		if (err)
 870 f9c2e8e5 2022-02-13 thomas 			return err;
 871 e6bcace5 2021-06-22 stsp 	}
 872 e6bcace5 2021-06-22 stsp 
 873 f9c2e8e5 2022-02-13 thomas 	return got_object_idset_add(idset, id, m);
 874 e6bcace5 2021-06-22 stsp }
 875 e6bcace5 2021-06-22 stsp 
 876 e6bcace5 2021-06-22 stsp static const struct got_error *
 877 f9c2e8e5 2022-02-13 thomas load_tree_entries(struct got_object_id_queue *ids, int want_meta,
 878 e6bcace5 2021-06-22 stsp     struct got_object_idset *idset, struct got_object_id *tree_id,
 879 e6bcace5 2021-06-22 stsp     const char *dpath, time_t mtime, struct got_repository *repo,
 880 85220b0e 2022-03-22 thomas     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
 881 85220b0e 2022-03-22 thomas     got_pack_progress_cb progress_cb, void *progress_arg,
 882 85220b0e 2022-03-22 thomas     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
 883 e6bcace5 2021-06-22 stsp {
 884 e6bcace5 2021-06-22 stsp 	const struct got_error *err;
 885 e6bcace5 2021-06-22 stsp 	struct got_tree_object *tree;
 886 e6bcace5 2021-06-22 stsp 	char *p = NULL;
 887 e6bcace5 2021-06-22 stsp 	int i;
 888 e6bcace5 2021-06-22 stsp 
 889 e6bcace5 2021-06-22 stsp 	err = got_object_open_as_tree(&tree, repo, tree_id);
 890 85220b0e 2022-03-22 thomas 	if (err)
 891 85220b0e 2022-03-22 thomas 		return err;
 892 85220b0e 2022-03-22 thomas 
 893 85220b0e 2022-03-22 thomas 	(*ntrees)++;
 894 85220b0e 2022-03-22 thomas 	err = report_progress(progress_cb, progress_arg, rl,
 895 85220b0e 2022-03-22 thomas 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
 896 e6bcace5 2021-06-22 stsp 	if (err)
 897 e6bcace5 2021-06-22 stsp 		return err;
 898 e6bcace5 2021-06-22 stsp 
 899 e6bcace5 2021-06-22 stsp 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
 900 e6bcace5 2021-06-22 stsp 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
 901 e6bcace5 2021-06-22 stsp 		struct got_object_id *id = got_tree_entry_get_id(e);
 902 e6bcace5 2021-06-22 stsp 		mode_t mode = got_tree_entry_get_mode(e);
 903 e6bcace5 2021-06-22 stsp 
 904 e6bcace5 2021-06-22 stsp 		if (cancel_cb) {
 905 e6bcace5 2021-06-22 stsp 			err = (*cancel_cb)(cancel_arg);
 906 e6bcace5 2021-06-22 stsp 			if (err)
 907 e6bcace5 2021-06-22 stsp 				break;
 908 e6bcace5 2021-06-22 stsp 		}
 909 e6bcace5 2021-06-22 stsp 
 910 08511b5e 2021-09-24 thomas 		if (got_object_tree_entry_is_submodule(e) ||
 911 e6bcace5 2021-06-22 stsp 		    got_object_idset_contains(idset, id))
 912 e6bcace5 2021-06-22 stsp 			continue;
 913 e6bcace5 2021-06-22 stsp 		
 914 e6bcace5 2021-06-22 stsp 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
 915 e6bcace5 2021-06-22 stsp 		    got_tree_entry_get_name(e)) == -1) {
 916 e6bcace5 2021-06-22 stsp 			err = got_error_from_errno("asprintf");
 917 e6bcace5 2021-06-22 stsp 			break;
 918 e6bcace5 2021-06-22 stsp 		}
 919 e6bcace5 2021-06-22 stsp 
 920 e6bcace5 2021-06-22 stsp 		if (S_ISDIR(mode)) {
 921 e6bcace5 2021-06-22 stsp 			struct got_object_qid *qid;
 922 e6bcace5 2021-06-22 stsp 			err = got_object_qid_alloc(&qid, id);
 923 e6bcace5 2021-06-22 stsp 			if (err)
 924 e6bcace5 2021-06-22 stsp 				break;
 925 dbdddfee 2021-06-23 naddy 			STAILQ_INSERT_TAIL(ids, qid, entry);
 926 08511b5e 2021-09-24 thomas 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
 927 f9c2e8e5 2022-02-13 thomas 			err = add_object(want_meta, idset, id, p,
 928 f9c2e8e5 2022-02-13 thomas 			    GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo);
 929 e6bcace5 2021-06-22 stsp 			if (err)
 930 e6bcace5 2021-06-22 stsp 				break;
 931 85220b0e 2022-03-22 thomas 			(*nfound)++;
 932 85220b0e 2022-03-22 thomas 			err = report_progress(progress_cb, progress_arg, rl,
 933 85220b0e 2022-03-22 thomas 			    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
 934 85220b0e 2022-03-22 thomas 			if (err)
 935 85220b0e 2022-03-22 thomas 				break;
 936 e6bcace5 2021-06-22 stsp 		}
 937 e6bcace5 2021-06-22 stsp 		free(p);
 938 e6bcace5 2021-06-22 stsp 		p = NULL;
 939 e6bcace5 2021-06-22 stsp 	}
 940 e6bcace5 2021-06-22 stsp 
 941 e6bcace5 2021-06-22 stsp 	got_object_tree_close(tree);
 942 e6bcace5 2021-06-22 stsp 	free(p);
 943 e6bcace5 2021-06-22 stsp 	return err;
 944 e6bcace5 2021-06-22 stsp }
 945 e6bcace5 2021-06-22 stsp 
 946 e6bcace5 2021-06-22 stsp static const struct got_error *
 947 f9c2e8e5 2022-02-13 thomas load_tree(int want_meta, struct got_object_idset *idset,
 948 e6bcace5 2021-06-22 stsp     struct got_object_id *tree_id, const char *dpath, time_t mtime,
 949 85220b0e 2022-03-22 thomas     struct got_repository *repo, int loose_obj_only,
 950 85220b0e 2022-03-22 thomas     int *ncolored, int *nfound, int *ntrees,
 951 85220b0e 2022-03-22 thomas     got_pack_progress_cb progress_cb, void *progress_arg,
 952 85220b0e 2022-03-22 thomas     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
 953 e6bcace5 2021-06-22 stsp {
 954 e6bcace5 2021-06-22 stsp 	const struct got_error *err = NULL;
 955 e6bcace5 2021-06-22 stsp 	struct got_object_id_queue tree_ids;
 956 e6bcace5 2021-06-22 stsp 	struct got_object_qid *qid;
 957 e6bcace5 2021-06-22 stsp 
 958 e6bcace5 2021-06-22 stsp 	if (got_object_idset_contains(idset, tree_id))
 959 e6bcace5 2021-06-22 stsp 		return NULL;
 960 e6bcace5 2021-06-22 stsp 
 961 e6bcace5 2021-06-22 stsp 	err = got_object_qid_alloc(&qid, tree_id);
 962 e6bcace5 2021-06-22 stsp 	if (err)
 963 e6bcace5 2021-06-22 stsp 		return err;
 964 e6bcace5 2021-06-22 stsp 
 965 dbdddfee 2021-06-23 naddy 	STAILQ_INIT(&tree_ids);
 966 dbdddfee 2021-06-23 naddy 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
 967 e6bcace5 2021-06-22 stsp 
 968 dbdddfee 2021-06-23 naddy 	while (!STAILQ_EMPTY(&tree_ids)) {
 969 e6bcace5 2021-06-22 stsp 		if (cancel_cb) {
 970 e6bcace5 2021-06-22 stsp 			err = (*cancel_cb)(cancel_arg);
 971 e6bcace5 2021-06-22 stsp 			if (err)
 972 e6bcace5 2021-06-22 stsp 				break;
 973 e6bcace5 2021-06-22 stsp 		}
 974 e6bcace5 2021-06-22 stsp 
 975 dbdddfee 2021-06-23 naddy 		qid = STAILQ_FIRST(&tree_ids);
 976 dbdddfee 2021-06-23 naddy 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
 977 e6bcace5 2021-06-22 stsp 
 978 e6bcace5 2021-06-22 stsp 		if (got_object_idset_contains(idset, qid->id)) {
 979 e6bcace5 2021-06-22 stsp 			got_object_qid_free(qid);
 980 e6bcace5 2021-06-22 stsp 			continue;
 981 e6bcace5 2021-06-22 stsp 		}
 982 e6bcace5 2021-06-22 stsp 
 983 f9c2e8e5 2022-02-13 thomas 		err = add_object(want_meta, idset, qid->id, dpath,
 984 f9c2e8e5 2022-02-13 thomas 		    GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo);
 985 e6bcace5 2021-06-22 stsp 		if (err) {
 986 e6bcace5 2021-06-22 stsp 			got_object_qid_free(qid);
 987 e6bcace5 2021-06-22 stsp 			break;
 988 e6bcace5 2021-06-22 stsp 		}
 989 e6bcace5 2021-06-22 stsp 
 990 85220b0e 2022-03-22 thomas 		(*nfound)++;
 991 85220b0e 2022-03-22 thomas 		err = report_progress(progress_cb, progress_arg, rl,
 992 85220b0e 2022-03-22 thomas 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
 993 85220b0e 2022-03-22 thomas 		if (err)
 994 85220b0e 2022-03-22 thomas 			break;
 995 85220b0e 2022-03-22 thomas 
 996 f9c2e8e5 2022-02-13 thomas 		err = load_tree_entries(&tree_ids, want_meta, idset, qid->id,
 997 85220b0e 2022-03-22 thomas 		    dpath, mtime, repo, loose_obj_only, ncolored, nfound,
 998 85220b0e 2022-03-22 thomas 		    ntrees, progress_cb, progress_arg, rl,
 999 85220b0e 2022-03-22 thomas 		    cancel_cb, cancel_arg);
1000 e6bcace5 2021-06-22 stsp 		got_object_qid_free(qid);
1001 e6bcace5 2021-06-22 stsp 		if (err)
1002 e6bcace5 2021-06-22 stsp 			break;
1003 e6bcace5 2021-06-22 stsp 	}
1004 e6bcace5 2021-06-22 stsp 
1005 e6bcace5 2021-06-22 stsp 	got_object_id_queue_free(&tree_ids);
1006 e6bcace5 2021-06-22 stsp 	return err;
1007 e6bcace5 2021-06-22 stsp }
1008 e6bcace5 2021-06-22 stsp 
1009 e6bcace5 2021-06-22 stsp static const struct got_error *
1010 f9c2e8e5 2022-02-13 thomas load_commit(int want_meta, struct got_object_idset *idset,
1011 05118f5a 2021-06-22 stsp     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1012 85220b0e 2022-03-22 thomas     int *ncolored, int *nfound, int *ntrees,
1013 85220b0e 2022-03-22 thomas     got_pack_progress_cb progress_cb, void *progress_arg,
1014 85220b0e 2022-03-22 thomas     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1015 e6bcace5 2021-06-22 stsp {
1016 e6bcace5 2021-06-22 stsp 	const struct got_error *err;
1017 e6bcace5 2021-06-22 stsp 	struct got_commit_object *commit;
1018 e6bcace5 2021-06-22 stsp 
1019 e6bcace5 2021-06-22 stsp 	if (got_object_idset_contains(idset, id))
1020 e6bcace5 2021-06-22 stsp 		return NULL;
1021 05118f5a 2021-06-22 stsp 
1022 05118f5a 2021-06-22 stsp 	if (loose_obj_only) {
1023 05118f5a 2021-06-22 stsp 		int is_packed;
1024 05118f5a 2021-06-22 stsp 		err = search_packidx(&is_packed, id, repo);
1025 05118f5a 2021-06-22 stsp 		if (err)
1026 05118f5a 2021-06-22 stsp 			return err;
1027 05118f5a 2021-06-22 stsp 		if (is_packed)
1028 05118f5a 2021-06-22 stsp 			return NULL;
1029 05118f5a 2021-06-22 stsp 	}
1030 e6bcace5 2021-06-22 stsp 
1031 e6bcace5 2021-06-22 stsp 	err = got_object_open_as_commit(&commit, repo, id);
1032 e6bcace5 2021-06-22 stsp 	if (err)
1033 e6bcace5 2021-06-22 stsp 		return err;
1034 e6bcace5 2021-06-22 stsp 
1035 f9c2e8e5 2022-02-13 thomas 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_COMMIT,
1036 05118f5a 2021-06-22 stsp 	    got_object_commit_get_committer_time(commit),
1037 05118f5a 2021-06-22 stsp 	    loose_obj_only, repo);
1038 e6bcace5 2021-06-22 stsp 	if (err)
1039 e6bcace5 2021-06-22 stsp 		goto done;
1040 e6bcace5 2021-06-22 stsp 
1041 85220b0e 2022-03-22 thomas 	(*nfound)++;
1042 85220b0e 2022-03-22 thomas 	err = report_progress(progress_cb, progress_arg, rl,
1043 85220b0e 2022-03-22 thomas 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
1044 85220b0e 2022-03-22 thomas 	if (err)
1045 85220b0e 2022-03-22 thomas 		goto done;
1046 85220b0e 2022-03-22 thomas 
1047 f9c2e8e5 2022-02-13 thomas 	err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1048 05118f5a 2021-06-22 stsp 	    "", got_object_commit_get_committer_time(commit),
1049 85220b0e 2022-03-22 thomas 	    repo, loose_obj_only, ncolored, nfound, ntrees,
1050 85220b0e 2022-03-22 thomas 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1051 e6bcace5 2021-06-22 stsp done:
1052 e6bcace5 2021-06-22 stsp 	got_object_commit_close(commit);
1053 e6bcace5 2021-06-22 stsp 	return err;
1054 e6bcace5 2021-06-22 stsp }
1055 e6bcace5 2021-06-22 stsp 
1056 05118f5a 2021-06-22 stsp static const struct got_error *
1057 f9c2e8e5 2022-02-13 thomas load_tag(int want_meta, struct got_object_idset *idset,
1058 05118f5a 2021-06-22 stsp     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1059 85220b0e 2022-03-22 thomas     int *ncolored, int *nfound, int *ntrees,
1060 85220b0e 2022-03-22 thomas     got_pack_progress_cb progress_cb, void *progress_arg,
1061 85220b0e 2022-03-22 thomas     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1062 05118f5a 2021-06-22 stsp {
1063 05118f5a 2021-06-22 stsp 	const struct got_error *err;
1064 05118f5a 2021-06-22 stsp 	struct got_tag_object *tag = NULL;
1065 05118f5a 2021-06-22 stsp 
1066 05118f5a 2021-06-22 stsp 	if (got_object_idset_contains(idset, id))
1067 05118f5a 2021-06-22 stsp 		return NULL;
1068 05118f5a 2021-06-22 stsp 
1069 05118f5a 2021-06-22 stsp 	if (loose_obj_only) {
1070 05118f5a 2021-06-22 stsp 		int is_packed;
1071 05118f5a 2021-06-22 stsp 		err = search_packidx(&is_packed, id, repo);
1072 05118f5a 2021-06-22 stsp 		if (err)
1073 05118f5a 2021-06-22 stsp 			return err;
1074 05118f5a 2021-06-22 stsp 		if (is_packed)
1075 05118f5a 2021-06-22 stsp 			return NULL;
1076 05118f5a 2021-06-22 stsp 	}
1077 05118f5a 2021-06-22 stsp 
1078 05118f5a 2021-06-22 stsp 	err = got_object_open_as_tag(&tag, repo, id);
1079 05118f5a 2021-06-22 stsp 	if (err)
1080 05118f5a 2021-06-22 stsp 		return err;
1081 05118f5a 2021-06-22 stsp 
1082 f9c2e8e5 2022-02-13 thomas 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1083 05118f5a 2021-06-22 stsp 	    got_object_tag_get_tagger_time(tag),
1084 05118f5a 2021-06-22 stsp 	    loose_obj_only, repo);
1085 05118f5a 2021-06-22 stsp 	if (err)
1086 05118f5a 2021-06-22 stsp 		goto done;
1087 05118f5a 2021-06-22 stsp 
1088 85220b0e 2022-03-22 thomas 	(*nfound)++;
1089 85220b0e 2022-03-22 thomas 	err = report_progress(progress_cb, progress_arg, rl,
1090 85220b0e 2022-03-22 thomas 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
1091 85220b0e 2022-03-22 thomas 	if (err)
1092 85220b0e 2022-03-22 thomas 		goto done;
1093 85220b0e 2022-03-22 thomas 
1094 05118f5a 2021-06-22 stsp 	switch (got_object_tag_get_object_type(tag)) {
1095 05118f5a 2021-06-22 stsp 	case GOT_OBJ_TYPE_COMMIT:
1096 f9c2e8e5 2022-02-13 thomas 		err = load_commit(want_meta, idset,
1097 85220b0e 2022-03-22 thomas 		    got_object_tag_get_object_id(tag), repo, loose_obj_only,
1098 85220b0e 2022-03-22 thomas 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1099 85220b0e 2022-03-22 thomas 		    cancel_cb, cancel_arg);
1100 05118f5a 2021-06-22 stsp 		break;
1101 05118f5a 2021-06-22 stsp 	case GOT_OBJ_TYPE_TREE:
1102 f9c2e8e5 2022-02-13 thomas 		err = load_tree(want_meta, idset,
1103 f9c2e8e5 2022-02-13 thomas 		    got_object_tag_get_object_id(tag), "",
1104 85220b0e 2022-03-22 thomas 		    got_object_tag_get_tagger_time(tag), repo, loose_obj_only,
1105 85220b0e 2022-03-22 thomas 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1106 85220b0e 2022-03-22 thomas 		    cancel_cb, cancel_arg);
1107 05118f5a 2021-06-22 stsp 		break;
1108 05118f5a 2021-06-22 stsp 	default:
1109 05118f5a 2021-06-22 stsp 		break;
1110 05118f5a 2021-06-22 stsp 	}
1111 05118f5a 2021-06-22 stsp 
1112 05118f5a 2021-06-22 stsp done:
1113 05118f5a 2021-06-22 stsp 	got_object_tag_close(tag);
1114 05118f5a 2021-06-22 stsp 	return err;
1115 05118f5a 2021-06-22 stsp }
1116 05118f5a 2021-06-22 stsp 
1117 e6bcace5 2021-06-22 stsp enum findtwixt_color {
1118 e6bcace5 2021-06-22 stsp 	COLOR_KEEP = 0,
1119 e6bcace5 2021-06-22 stsp 	COLOR_DROP,
1120 e6bcace5 2021-06-22 stsp 	COLOR_BLANK,
1121 e6bcace5 2021-06-22 stsp };
1122 e6bcace5 2021-06-22 stsp static const int findtwixt_colors[] = {
1123 e6bcace5 2021-06-22 stsp 	COLOR_KEEP,
1124 e6bcace5 2021-06-22 stsp 	COLOR_DROP,
1125 e6bcace5 2021-06-22 stsp 	COLOR_BLANK
1126 e6bcace5 2021-06-22 stsp };
1127 e6bcace5 2021-06-22 stsp 
1128 e6bcace5 2021-06-22 stsp static const struct got_error *
1129 e6bcace5 2021-06-22 stsp queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1130 e6bcace5 2021-06-22 stsp     int color, struct got_repository *repo)
1131 e6bcace5 2021-06-22 stsp {
1132 e6bcace5 2021-06-22 stsp 	const struct got_error *err;
1133 e6bcace5 2021-06-22 stsp 	struct got_object_qid *qid;
1134 e6bcace5 2021-06-22 stsp 
1135 e6bcace5 2021-06-22 stsp 	err = got_object_qid_alloc(&qid, id);
1136 e6bcace5 2021-06-22 stsp 	if (err)
1137 e6bcace5 2021-06-22 stsp 		return err;
1138 e6bcace5 2021-06-22 stsp 
1139 dbdddfee 2021-06-23 naddy 	STAILQ_INSERT_TAIL(ids, qid, entry);
1140 e6bcace5 2021-06-22 stsp 	qid->data = (void *)&findtwixt_colors[color];
1141 e6bcace5 2021-06-22 stsp 	return NULL;
1142 e6bcace5 2021-06-22 stsp }
1143 e6bcace5 2021-06-22 stsp 
1144 e6bcace5 2021-06-22 stsp static const struct got_error *
1145 e6bcace5 2021-06-22 stsp drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
1146 e6bcace5 2021-06-22 stsp     struct got_object_id *id, struct got_repository *repo,
1147 e6bcace5 2021-06-22 stsp     got_cancel_cb cancel_cb, void *cancel_arg)
1148 e6bcace5 2021-06-22 stsp {
1149 e6bcace5 2021-06-22 stsp 	const struct got_error *err = NULL;
1150 e6bcace5 2021-06-22 stsp 	struct got_commit_object *commit;
1151 e6bcace5 2021-06-22 stsp 	const struct got_object_id_queue *parents;
1152 e6bcace5 2021-06-22 stsp 	struct got_object_id_queue ids;
1153 e6bcace5 2021-06-22 stsp 	struct got_object_qid *qid;
1154 e6bcace5 2021-06-22 stsp 
1155 dbdddfee 2021-06-23 naddy 	STAILQ_INIT(&ids);
1156 e6bcace5 2021-06-22 stsp 
1157 e6bcace5 2021-06-22 stsp 	err = got_object_qid_alloc(&qid, id);
1158 e6bcace5 2021-06-22 stsp 	if (err)
1159 e6bcace5 2021-06-22 stsp 		return err;
1160 dbdddfee 2021-06-23 naddy 	STAILQ_INSERT_HEAD(&ids, qid, entry);
1161 e6bcace5 2021-06-22 stsp 
1162 dbdddfee 2021-06-23 naddy 	while (!STAILQ_EMPTY(&ids)) {
1163 e6bcace5 2021-06-22 stsp 		if (cancel_cb) {
1164 e6bcace5 2021-06-22 stsp 			err = (*cancel_cb)(cancel_arg);
1165 e6bcace5 2021-06-22 stsp 			if (err)
1166 e6bcace5 2021-06-22 stsp 				break;
1167 e6bcace5 2021-06-22 stsp 		}
1168 e6bcace5 2021-06-22 stsp 
1169 dbdddfee 2021-06-23 naddy 		qid = STAILQ_FIRST(&ids);
1170 dbdddfee 2021-06-23 naddy 		STAILQ_REMOVE_HEAD(&ids, entry);
1171 e6bcace5 2021-06-22 stsp 
1172 e6bcace5 2021-06-22 stsp 		if (got_object_idset_contains(drop, qid->id)) {
1173 e6bcace5 2021-06-22 stsp 			got_object_qid_free(qid);
1174 e6bcace5 2021-06-22 stsp 			continue;
1175 e6bcace5 2021-06-22 stsp 		}
1176 e6bcace5 2021-06-22 stsp 
1177 07165b17 2021-07-01 stsp 		err = got_object_idset_add(drop, qid->id,
1178 07165b17 2021-07-01 stsp 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1179 e6bcace5 2021-06-22 stsp 		if (err) {
1180 e6bcace5 2021-06-22 stsp 			got_object_qid_free(qid);
1181 e6bcace5 2021-06-22 stsp 			break;
1182 e6bcace5 2021-06-22 stsp 		}
1183 e6bcace5 2021-06-22 stsp 
1184 e6bcace5 2021-06-22 stsp 		if (!got_object_idset_contains(keep, qid->id)) {
1185 e6bcace5 2021-06-22 stsp 			got_object_qid_free(qid);
1186 e6bcace5 2021-06-22 stsp 			continue;
1187 e6bcace5 2021-06-22 stsp 		}
1188 e6bcace5 2021-06-22 stsp 
1189 e6bcace5 2021-06-22 stsp 		err = got_object_open_as_commit(&commit, repo, qid->id);
1190 e6bcace5 2021-06-22 stsp 		got_object_qid_free(qid);
1191 e6bcace5 2021-06-22 stsp 		if (err)
1192 e6bcace5 2021-06-22 stsp 			break;
1193 e6bcace5 2021-06-22 stsp 
1194 e6bcace5 2021-06-22 stsp 		parents = got_object_commit_get_parent_ids(commit);
1195 e6bcace5 2021-06-22 stsp 		if (parents) {
1196 e6bcace5 2021-06-22 stsp 			err = got_object_id_queue_copy(parents, &ids);
1197 e6bcace5 2021-06-22 stsp 			if (err) {
1198 e6bcace5 2021-06-22 stsp 				got_object_commit_close(commit);
1199 e6bcace5 2021-06-22 stsp 				break;
1200 e6bcace5 2021-06-22 stsp 			}
1201 e6bcace5 2021-06-22 stsp 		}
1202 e6bcace5 2021-06-22 stsp 		got_object_commit_close(commit);
1203 e6bcace5 2021-06-22 stsp 	}
1204 e6bcace5 2021-06-22 stsp 
1205 e6bcace5 2021-06-22 stsp 	got_object_id_queue_free(&ids);
1206 e6bcace5 2021-06-22 stsp 	return err;
1207 e6bcace5 2021-06-22 stsp }
1208 e6bcace5 2021-06-22 stsp 
1209 e6bcace5 2021-06-22 stsp struct append_id_arg {
1210 e6bcace5 2021-06-22 stsp 	struct got_object_id **array;
1211 e6bcace5 2021-06-22 stsp 	int idx;
1212 e6bcace5 2021-06-22 stsp };
1213 e6bcace5 2021-06-22 stsp 
1214 e6bcace5 2021-06-22 stsp static const struct got_error *
1215 e6bcace5 2021-06-22 stsp append_id(struct got_object_id *id, void *data, void *arg)
1216 e6bcace5 2021-06-22 stsp {
1217 e6bcace5 2021-06-22 stsp 	struct append_id_arg *a = arg;
1218 e6bcace5 2021-06-22 stsp 
1219 e6bcace5 2021-06-22 stsp 	a->array[a->idx] = got_object_id_dup(id);
1220 e6bcace5 2021-06-22 stsp 	if (a->array[a->idx] == NULL)
1221 e6bcace5 2021-06-22 stsp 		return got_error_from_errno("got_object_id_dup");
1222 e6bcace5 2021-06-22 stsp 
1223 e6bcace5 2021-06-22 stsp 	a->idx++;
1224 e6bcace5 2021-06-22 stsp 	return NULL;
1225 e6bcace5 2021-06-22 stsp }
1226 e6bcace5 2021-06-22 stsp 
1227 e6bcace5 2021-06-22 stsp static const struct got_error *
1228 85220b0e 2022-03-22 thomas findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1229 e6bcace5 2021-06-22 stsp     struct got_object_id **head, int nhead,
1230 e6bcace5 2021-06-22 stsp     struct got_object_id **tail, int ntail,
1231 e6bcace5 2021-06-22 stsp     struct got_repository *repo,
1232 85220b0e 2022-03-22 thomas     got_pack_progress_cb progress_cb, void *progress_arg,
1233 85220b0e 2022-03-22 thomas     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1234 e6bcace5 2021-06-22 stsp {
1235 e6bcace5 2021-06-22 stsp 	const struct got_error *err = NULL;
1236 e6bcace5 2021-06-22 stsp 	struct got_object_id_queue ids;
1237 e6bcace5 2021-06-22 stsp 	struct got_object_idset *keep, *drop;
1238 e6bcace5 2021-06-22 stsp 	struct got_object_qid *qid;
1239 05118f5a 2021-06-22 stsp 	int i, ncolor, nkeep, obj_type;
1240 e6bcace5 2021-06-22 stsp 
1241 dbdddfee 2021-06-23 naddy 	STAILQ_INIT(&ids);
1242 e6bcace5 2021-06-22 stsp 	*res = NULL;
1243 e6bcace5 2021-06-22 stsp 	*nres = 0;
1244 85220b0e 2022-03-22 thomas 	*ncolored = 0;
1245 e6bcace5 2021-06-22 stsp 
1246 e6bcace5 2021-06-22 stsp 	keep = got_object_idset_alloc();
1247 e6bcace5 2021-06-22 stsp 	if (keep == NULL)
1248 e6bcace5 2021-06-22 stsp 		return got_error_from_errno("got_object_idset_alloc");
1249 e6bcace5 2021-06-22 stsp 
1250 e6bcace5 2021-06-22 stsp 	drop = got_object_idset_alloc();
1251 e6bcace5 2021-06-22 stsp 	if (drop == NULL) {
1252 e6bcace5 2021-06-22 stsp 		err = got_error_from_errno("got_object_idset_alloc");
1253 e6bcace5 2021-06-22 stsp 		goto done;
1254 e6bcace5 2021-06-22 stsp 	}
1255 e6bcace5 2021-06-22 stsp 
1256 05118f5a 2021-06-22 stsp 	for (i = 0; i < nhead; i++) {
1257 05118f5a 2021-06-22 stsp 		struct got_object_id *id = head[i];
1258 05118f5a 2021-06-22 stsp 		if (id == NULL)
1259 05118f5a 2021-06-22 stsp 			continue;
1260 05118f5a 2021-06-22 stsp 		err = got_object_get_type(&obj_type, repo, id);
1261 05118f5a 2021-06-22 stsp 		if (err)
1262 05118f5a 2021-06-22 stsp 			return err;
1263 05118f5a 2021-06-22 stsp 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1264 05118f5a 2021-06-22 stsp 			continue;
1265 05118f5a 2021-06-22 stsp 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
1266 05118f5a 2021-06-22 stsp 		if (err)
1267 05118f5a 2021-06-22 stsp 			goto done;
1268 e6bcace5 2021-06-22 stsp 	}		
1269 05118f5a 2021-06-22 stsp 	for (i = 0; i < ntail; i++) {
1270 05118f5a 2021-06-22 stsp 		struct got_object_id *id = tail[i];
1271 05118f5a 2021-06-22 stsp 		if (id == NULL)
1272 05118f5a 2021-06-22 stsp 			continue;
1273 05118f5a 2021-06-22 stsp 		err = got_object_get_type(&obj_type, repo, id);
1274 05118f5a 2021-06-22 stsp 		if (err)
1275 05118f5a 2021-06-22 stsp 			return err;
1276 05118f5a 2021-06-22 stsp 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1277 05118f5a 2021-06-22 stsp 			continue;
1278 05118f5a 2021-06-22 stsp 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
1279 05118f5a 2021-06-22 stsp 		if (err)
1280 05118f5a 2021-06-22 stsp 			goto done;
1281 e6bcace5 2021-06-22 stsp 	}
1282 e6bcace5 2021-06-22 stsp 
1283 dbdddfee 2021-06-23 naddy 	while (!STAILQ_EMPTY(&ids)) {
1284 e6bcace5 2021-06-22 stsp 		int qcolor;
1285 dbdddfee 2021-06-23 naddy 		qid = STAILQ_FIRST(&ids);
1286 e6bcace5 2021-06-22 stsp 		qcolor = *((int *)qid->data);
1287 e6bcace5 2021-06-22 stsp 
1288 e6bcace5 2021-06-22 stsp 		if (got_object_idset_contains(drop, qid->id))
1289 e6bcace5 2021-06-22 stsp 			ncolor = COLOR_DROP;
1290 e6bcace5 2021-06-22 stsp 		else if (got_object_idset_contains(keep, qid->id))
1291 e6bcace5 2021-06-22 stsp 			ncolor = COLOR_KEEP;
1292 e6bcace5 2021-06-22 stsp 		else
1293 e6bcace5 2021-06-22 stsp 			ncolor = COLOR_BLANK;
1294 85220b0e 2022-03-22 thomas 
1295 85220b0e 2022-03-22 thomas 		(*ncolored)++;
1296 85220b0e 2022-03-22 thomas 		err = report_progress(progress_cb, progress_arg, rl,
1297 85220b0e 2022-03-22 thomas 		    *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1298 85220b0e 2022-03-22 thomas 		if (err)
1299 85220b0e 2022-03-22 thomas 			goto done;
1300 e6bcace5 2021-06-22 stsp 
1301 e6bcace5 2021-06-22 stsp 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
1302 e6bcace5 2021-06-22 stsp 		    qcolor == COLOR_KEEP)) {
1303 dbdddfee 2021-06-23 naddy 			STAILQ_REMOVE_HEAD(&ids, entry);
1304 e6bcace5 2021-06-22 stsp 			got_object_qid_free(qid);
1305 e6bcace5 2021-06-22 stsp 			continue;
1306 e6bcace5 2021-06-22 stsp 		}
1307 e6bcace5 2021-06-22 stsp 
1308 e6bcace5 2021-06-22 stsp 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
1309 e6bcace5 2021-06-22 stsp 			err = drop_commit(keep, drop, qid->id, repo,
1310 e6bcace5 2021-06-22 stsp 			    cancel_cb, cancel_arg);
1311 e6bcace5 2021-06-22 stsp 			if (err)
1312 e6bcace5 2021-06-22 stsp 				goto done;
1313 e6bcace5 2021-06-22 stsp 		} else if (ncolor == COLOR_BLANK) {
1314 e6bcace5 2021-06-22 stsp 			struct got_commit_object *commit;
1315 e6bcace5 2021-06-22 stsp 			struct got_object_id *id;
1316 e6bcace5 2021-06-22 stsp 			const struct got_object_id_queue *parents;
1317 e6bcace5 2021-06-22 stsp 			struct got_object_qid *pid;
1318 e6bcace5 2021-06-22 stsp 
1319 e6bcace5 2021-06-22 stsp 			id = got_object_id_dup(qid->id);
1320 e6bcace5 2021-06-22 stsp 			if (id == NULL) {
1321 e6bcace5 2021-06-22 stsp 				err = got_error_from_errno("got_object_id_dup");
1322 e6bcace5 2021-06-22 stsp 				goto done;
1323 e6bcace5 2021-06-22 stsp 			}
1324 e6bcace5 2021-06-22 stsp 			if (qcolor == COLOR_KEEP)
1325 07165b17 2021-07-01 stsp 				err = got_object_idset_add(keep, id,
1326 07165b17 2021-07-01 stsp 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1327 e6bcace5 2021-06-22 stsp 			else
1328 07165b17 2021-07-01 stsp 				err = got_object_idset_add(drop, id,
1329 07165b17 2021-07-01 stsp 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1330 e6bcace5 2021-06-22 stsp 			if (err) {
1331 e6bcace5 2021-06-22 stsp 				free(id);
1332 e6bcace5 2021-06-22 stsp 				goto done;
1333 e6bcace5 2021-06-22 stsp 			}
1334 e6bcace5 2021-06-22 stsp 
1335 e6bcace5 2021-06-22 stsp 			err = got_object_open_as_commit(&commit, repo, id);
1336 e6bcace5 2021-06-22 stsp 			if (err) {
1337 e6bcace5 2021-06-22 stsp 				free(id);
1338 e6bcace5 2021-06-22 stsp 				goto done;
1339 e6bcace5 2021-06-22 stsp 			}
1340 e6bcace5 2021-06-22 stsp 			parents = got_object_commit_get_parent_ids(commit);
1341 e6bcace5 2021-06-22 stsp 			if (parents) {
1342 dbdddfee 2021-06-23 naddy 				STAILQ_FOREACH(pid, parents, entry) {
1343 e6bcace5 2021-06-22 stsp 					err = queue_commit_id(&ids, pid->id,
1344 e6bcace5 2021-06-22 stsp 					    qcolor, repo);
1345 e6bcace5 2021-06-22 stsp 					if (err) {
1346 e6bcace5 2021-06-22 stsp 						free(id);
1347 e6bcace5 2021-06-22 stsp 						goto done;
1348 e6bcace5 2021-06-22 stsp 					}
1349 e6bcace5 2021-06-22 stsp 				}
1350 e6bcace5 2021-06-22 stsp 			}
1351 e6bcace5 2021-06-22 stsp 			got_object_commit_close(commit);
1352 e6bcace5 2021-06-22 stsp 			commit = NULL;
1353 e6bcace5 2021-06-22 stsp 		} else {
1354 e6bcace5 2021-06-22 stsp 			/* should not happen */
1355 e6bcace5 2021-06-22 stsp 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1356 e6bcace5 2021-06-22 stsp 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
1357 e6bcace5 2021-06-22 stsp 			goto done;
1358 e6bcace5 2021-06-22 stsp 		}
1359 e6bcace5 2021-06-22 stsp 
1360 dbdddfee 2021-06-23 naddy 		STAILQ_REMOVE_HEAD(&ids, entry);
1361 e6bcace5 2021-06-22 stsp 		got_object_qid_free(qid);
1362 e6bcace5 2021-06-22 stsp 	}
1363 e6bcace5 2021-06-22 stsp 
1364 e6bcace5 2021-06-22 stsp 	nkeep = got_object_idset_num_elements(keep);
1365 e6bcace5 2021-06-22 stsp 	if (nkeep > 0) {
1366 e6bcace5 2021-06-22 stsp 		struct append_id_arg arg;
1367 e6bcace5 2021-06-22 stsp 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1368 e6bcace5 2021-06-22 stsp 		if (arg.array == NULL) {
1369 e6bcace5 2021-06-22 stsp 			err = got_error_from_errno("calloc");
1370 e6bcace5 2021-06-22 stsp 			goto done;
1371 e6bcace5 2021-06-22 stsp 		}
1372 e6bcace5 2021-06-22 stsp 		arg.idx = 0;
1373 e6bcace5 2021-06-22 stsp 		err = got_object_idset_for_each(keep, append_id, &arg);
1374 e6bcace5 2021-06-22 stsp 		if (err) {
1375 e6bcace5 2021-06-22 stsp 			free(arg.array);
1376 e6bcace5 2021-06-22 stsp 			goto done;
1377 e6bcace5 2021-06-22 stsp 		}
1378 e6bcace5 2021-06-22 stsp 		*res = arg.array;
1379 e6bcace5 2021-06-22 stsp 		*nres = nkeep;
1380 e6bcace5 2021-06-22 stsp 	}
1381 e6bcace5 2021-06-22 stsp done:
1382 e6bcace5 2021-06-22 stsp 	got_object_idset_free(keep);
1383 e6bcace5 2021-06-22 stsp 	got_object_idset_free(drop);
1384 e6bcace5 2021-06-22 stsp 	got_object_id_queue_free(&ids);
1385 e6bcace5 2021-06-22 stsp 	return err;
1386 e6bcace5 2021-06-22 stsp }
1387 e6bcace5 2021-06-22 stsp 
1388 e6bcace5 2021-06-22 stsp static const struct got_error *
1389 85220b0e 2022-03-22 thomas load_object_ids(int *ncolored, int *nfound, int *ntrees,
1390 85220b0e 2022-03-22 thomas     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1391 e6bcace5 2021-06-22 stsp     struct got_object_id **ours, int nours, struct got_repository *repo,
1392 85220b0e 2022-03-22 thomas     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
1393 85220b0e 2022-03-22 thomas     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1394 e6bcace5 2021-06-22 stsp {
1395 e6bcace5 2021-06-22 stsp 	const struct got_error *err = NULL;
1396 e6bcace5 2021-06-22 stsp 	struct got_object_id **ids = NULL;
1397 05118f5a 2021-06-22 stsp 	int i, nobj = 0, obj_type;
1398 e6bcace5 2021-06-22 stsp 
1399 85220b0e 2022-03-22 thomas 	*ncolored = 0;
1400 85220b0e 2022-03-22 thomas 	*nfound = 0;
1401 85220b0e 2022-03-22 thomas 	*ntrees = 0;
1402 85220b0e 2022-03-22 thomas 
1403 85220b0e 2022-03-22 thomas 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1404 85220b0e 2022-03-22 thomas 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1405 e6bcace5 2021-06-22 stsp 	if (err || nobj == 0)
1406 e6bcace5 2021-06-22 stsp 		goto done;
1407 e6bcace5 2021-06-22 stsp 
1408 e6bcace5 2021-06-22 stsp 	for (i = 0; i < ntheirs; i++) {
1409 05118f5a 2021-06-22 stsp 		struct got_object_id *id = theirs[i];
1410 05118f5a 2021-06-22 stsp 		if (id == NULL)
1411 05118f5a 2021-06-22 stsp 			continue;
1412 05118f5a 2021-06-22 stsp 		err = got_object_get_type(&obj_type, repo, id);
1413 05118f5a 2021-06-22 stsp 		if (err)
1414 05118f5a 2021-06-22 stsp 			return err;
1415 05118f5a 2021-06-22 stsp 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1416 05118f5a 2021-06-22 stsp 			continue;
1417 85220b0e 2022-03-22 thomas 		err = load_commit(0, idset, id, repo, loose_obj_only,
1418 85220b0e 2022-03-22 thomas 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1419 85220b0e 2022-03-22 thomas 		    cancel_cb, cancel_arg);
1420 05118f5a 2021-06-22 stsp 		if (err)
1421 05118f5a 2021-06-22 stsp 			goto done;
1422 05118f5a 2021-06-22 stsp 	}
1423 05118f5a 2021-06-22 stsp 
1424 05118f5a 2021-06-22 stsp 	for (i = 0; i < ntheirs; i++) {
1425 f4a2ff2d 2021-07-01 stsp 		struct got_object_id *id = theirs[i];
1426 f9c2e8e5 2022-02-13 thomas 		struct got_pack_meta *m;
1427 05118f5a 2021-06-22 stsp 		if (id == NULL)
1428 05118f5a 2021-06-22 stsp 			continue;
1429 f9c2e8e5 2022-02-13 thomas 		m = got_object_idset_get(idset, id);
1430 f9c2e8e5 2022-02-13 thomas 		if (m == NULL) {
1431 07165b17 2021-07-01 stsp 			err = got_object_get_type(&obj_type, repo, id);
1432 07165b17 2021-07-01 stsp 			if (err)
1433 07165b17 2021-07-01 stsp 				goto done;
1434 07165b17 2021-07-01 stsp 		} else
1435 f9c2e8e5 2022-02-13 thomas 			obj_type = m->obj_type;
1436 05118f5a 2021-06-22 stsp 		if (obj_type != GOT_OBJ_TYPE_TAG)
1437 05118f5a 2021-06-22 stsp 			continue;
1438 85220b0e 2022-03-22 thomas 		err = load_tag(0, idset, id, repo, loose_obj_only,
1439 85220b0e 2022-03-22 thomas 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1440 85220b0e 2022-03-22 thomas 		    cancel_cb, cancel_arg);
1441 05118f5a 2021-06-22 stsp 		if (err)
1442 05118f5a 2021-06-22 stsp 			goto done;
1443 05118f5a 2021-06-22 stsp 	}
1444 05118f5a 2021-06-22 stsp 
1445 eca70f98 2021-09-03 stsp 	for (i = 0; i < nobj; i++) {
1446 85220b0e 2022-03-22 thomas 		err = load_commit(1, idset, ids[i], repo, loose_obj_only,
1447 85220b0e 2022-03-22 thomas 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1448 85220b0e 2022-03-22 thomas 		    cancel_cb, cancel_arg);
1449 eca70f98 2021-09-03 stsp 		if (err)
1450 eca70f98 2021-09-03 stsp 			goto done;
1451 eca70f98 2021-09-03 stsp 	}
1452 eca70f98 2021-09-03 stsp 
1453 05118f5a 2021-06-22 stsp 	for (i = 0; i < nours; i++) {
1454 05118f5a 2021-06-22 stsp 		struct got_object_id *id = ours[i];
1455 f9c2e8e5 2022-02-13 thomas 		struct got_pack_meta *m;
1456 05118f5a 2021-06-22 stsp 		if (id == NULL)
1457 05118f5a 2021-06-22 stsp 			continue;
1458 f9c2e8e5 2022-02-13 thomas 		m = got_object_idset_get(idset, id);
1459 f9c2e8e5 2022-02-13 thomas 		if (m == NULL) {
1460 07165b17 2021-07-01 stsp 			err = got_object_get_type(&obj_type, repo, id);
1461 07165b17 2021-07-01 stsp 			if (err)
1462 07165b17 2021-07-01 stsp 				goto done;
1463 07165b17 2021-07-01 stsp 		} else
1464 f9c2e8e5 2022-02-13 thomas 			obj_type = m->obj_type;
1465 05118f5a 2021-06-22 stsp 		if (obj_type != GOT_OBJ_TYPE_TAG)
1466 05118f5a 2021-06-22 stsp 			continue;
1467 85220b0e 2022-03-22 thomas 		err = load_tag(1, idset, id, repo, loose_obj_only,
1468 85220b0e 2022-03-22 thomas 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1469 85220b0e 2022-03-22 thomas 		    cancel_cb, cancel_arg);
1470 05118f5a 2021-06-22 stsp 		if (err)
1471 e6bcace5 2021-06-22 stsp 			goto done;
1472 e6bcace5 2021-06-22 stsp 	}
1473 e6bcace5 2021-06-22 stsp done:
1474 e6bcace5 2021-06-22 stsp 	for (i = 0; i < nobj; i++) {
1475 e6bcace5 2021-06-22 stsp 		free(ids[i]);
1476 e6bcace5 2021-06-22 stsp 	}
1477 e6bcace5 2021-06-22 stsp 	free(ids);
1478 e6bcace5 2021-06-22 stsp 	return err;
1479 e6bcace5 2021-06-22 stsp }
1480 e6bcace5 2021-06-22 stsp 
1481 e6bcace5 2021-06-22 stsp const struct got_error *
1482 e6bcace5 2021-06-22 stsp hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1483 e6bcace5 2021-06-22 stsp {
1484 e6bcace5 2021-06-22 stsp 	size_t n;
1485 e6bcace5 2021-06-22 stsp 
1486 e6bcace5 2021-06-22 stsp 	SHA1Update(ctx, buf, len);
1487 e6bcace5 2021-06-22 stsp 	n = fwrite(buf, 1, len, f);
1488 e6bcace5 2021-06-22 stsp 	if (n != len)
1489 e6bcace5 2021-06-22 stsp 		return got_ferror(f, GOT_ERR_IO);
1490 e6bcace5 2021-06-22 stsp 	return NULL;
1491 e6bcace5 2021-06-22 stsp }
1492 e6bcace5 2021-06-22 stsp 
1493 e6bcace5 2021-06-22 stsp static void
1494 e6bcace5 2021-06-22 stsp putbe32(char *b, uint32_t n)
1495 e6bcace5 2021-06-22 stsp {
1496 e6bcace5 2021-06-22 stsp 	b[0] = n >> 24;
1497 e6bcace5 2021-06-22 stsp 	b[1] = n >> 16;
1498 e6bcace5 2021-06-22 stsp 	b[2] = n >> 8;
1499 e6bcace5 2021-06-22 stsp 	b[3] = n >> 0;
1500 e6bcace5 2021-06-22 stsp }
1501 e6bcace5 2021-06-22 stsp 
1502 e6bcace5 2021-06-22 stsp static int
1503 e6bcace5 2021-06-22 stsp write_order_cmp(const void *pa, const void *pb)
1504 e6bcace5 2021-06-22 stsp {
1505 e6bcace5 2021-06-22 stsp 	struct got_pack_meta *a, *b, *ahd, *bhd;
1506 e6bcace5 2021-06-22 stsp 
1507 e6bcace5 2021-06-22 stsp 	a = *(struct got_pack_meta **)pa;
1508 e6bcace5 2021-06-22 stsp 	b = *(struct got_pack_meta **)pb;
1509 e6bcace5 2021-06-22 stsp 	ahd = (a->head == NULL) ? a : a->head;
1510 e6bcace5 2021-06-22 stsp 	bhd = (b->head == NULL) ? b : b->head;
1511 e6bcace5 2021-06-22 stsp 	if (ahd->mtime != bhd->mtime)
1512 e6bcace5 2021-06-22 stsp 		return bhd->mtime - ahd->mtime;
1513 e6bcace5 2021-06-22 stsp 	if (ahd != bhd)
1514 e6bcace5 2021-06-22 stsp 		return (uintptr_t)bhd - (uintptr_t)ahd;
1515 e6bcace5 2021-06-22 stsp 	if (a->nchain != b->nchain)
1516 e6bcace5 2021-06-22 stsp 		return a->nchain - b->nchain;
1517 e6bcace5 2021-06-22 stsp 	return a->mtime - b->mtime;
1518 e6bcace5 2021-06-22 stsp }
1519 e6bcace5 2021-06-22 stsp 
1520 f9c2e8e5 2022-02-13 thomas static int
1521 f9c2e8e5 2022-02-13 thomas reuse_write_order_cmp(const void *pa, const void *pb)
1522 f9c2e8e5 2022-02-13 thomas {
1523 f9c2e8e5 2022-02-13 thomas 	struct got_pack_meta *a, *b;
1524 f9c2e8e5 2022-02-13 thomas 
1525 f9c2e8e5 2022-02-13 thomas 	a = *(struct got_pack_meta **)pa;
1526 f9c2e8e5 2022-02-13 thomas 	b = *(struct got_pack_meta **)pb;
1527 f9c2e8e5 2022-02-13 thomas 
1528 f9c2e8e5 2022-02-13 thomas 	if (a->reused_delta_offset < b->reused_delta_offset)
1529 f9c2e8e5 2022-02-13 thomas 		return -1;
1530 f9c2e8e5 2022-02-13 thomas 	if (a->reused_delta_offset > b->reused_delta_offset)
1531 f9c2e8e5 2022-02-13 thomas 		return 1;
1532 f9c2e8e5 2022-02-13 thomas 	return 0;
1533 f9c2e8e5 2022-02-13 thomas }
1534 f9c2e8e5 2022-02-13 thomas 
1535 e6bcace5 2021-06-22 stsp static const struct got_error *
1536 e6bcace5 2021-06-22 stsp packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1537 e6bcace5 2021-06-22 stsp {
1538 e6bcace5 2021-06-22 stsp 	size_t i;
1539 e6bcace5 2021-06-22 stsp 
1540 e6bcace5 2021-06-22 stsp 	*hdrlen = 0;
1541 e6bcace5 2021-06-22 stsp 
1542 e6bcace5 2021-06-22 stsp 	hdr[0] = obj_type << 4;
1543 e6bcace5 2021-06-22 stsp 	hdr[0] |= len & 0xf;
1544 e6bcace5 2021-06-22 stsp 	len >>= 4;
1545 e6bcace5 2021-06-22 stsp 	for (i = 1; len != 0; i++){
1546 e6bcace5 2021-06-22 stsp 		if (i >= bufsize)
1547 e6bcace5 2021-06-22 stsp 			return got_error(GOT_ERR_NO_SPACE);
1548 e6bcace5 2021-06-22 stsp 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1549 e6bcace5 2021-06-22 stsp 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1550 e6bcace5 2021-06-22 stsp 		len >>= GOT_DELTA_SIZE_SHIFT;
1551 e6bcace5 2021-06-22 stsp 	}
1552 e6bcace5 2021-06-22 stsp 
1553 e6bcace5 2021-06-22 stsp 	*hdrlen = i;
1554 e6bcace5 2021-06-22 stsp 	return NULL;
1555 e6bcace5 2021-06-22 stsp }
1556 e6bcace5 2021-06-22 stsp 
1557 e6bcace5 2021-06-22 stsp static int
1558 e6bcace5 2021-06-22 stsp packoff(char *hdr, off_t off)
1559 e6bcace5 2021-06-22 stsp {
1560 e6bcace5 2021-06-22 stsp 	int i, j;
1561 e6bcace5 2021-06-22 stsp 	char rbuf[8];
1562 e6bcace5 2021-06-22 stsp 
1563 e6bcace5 2021-06-22 stsp 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1564 e6bcace5 2021-06-22 stsp 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1565 e6bcace5 2021-06-22 stsp 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1566 e6bcace5 2021-06-22 stsp 		    GOT_DELTA_SIZE_MORE;
1567 e6bcace5 2021-06-22 stsp 	}
1568 e6bcace5 2021-06-22 stsp 
1569 e6bcace5 2021-06-22 stsp 	j = 0;
1570 e6bcace5 2021-06-22 stsp 	while (i > 0)
1571 e6bcace5 2021-06-22 stsp 		hdr[j++] = rbuf[--i];
1572 e6bcace5 2021-06-22 stsp 	return j;
1573 e6bcace5 2021-06-22 stsp }
1574 e6bcace5 2021-06-22 stsp 
1575 e6bcace5 2021-06-22 stsp static const struct got_error *
1576 3b6ceab7 2022-01-10 thomas deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1577 f9c2e8e5 2022-02-13 thomas     struct got_pack_meta *m)
1578 3b6ceab7 2022-01-10 thomas {
1579 3b6ceab7 2022-01-10 thomas 	const struct got_error *err;
1580 3b6ceab7 2022-01-10 thomas 	char buf[32];
1581 3b6ceab7 2022-01-10 thomas 	int nh;
1582 3b6ceab7 2022-01-10 thomas 
1583 f9c2e8e5 2022-02-13 thomas 	if (m->prev->off != 0) {
1584 3b6ceab7 2022-01-10 thomas 		err = packhdr(&nh, buf, sizeof(buf),
1585 3b6ceab7 2022-01-10 thomas 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1586 3b6ceab7 2022-01-10 thomas 		if (err)
1587 3b6ceab7 2022-01-10 thomas 			return err;
1588 3b6ceab7 2022-01-10 thomas 		nh += packoff(buf + nh, m->off - m->prev->off);
1589 3b6ceab7 2022-01-10 thomas 		err = hwrite(packfile, buf, nh, ctx);
1590 3b6ceab7 2022-01-10 thomas 		if (err)
1591 3b6ceab7 2022-01-10 thomas 			return err;
1592 3b6ceab7 2022-01-10 thomas 		*packfile_size += nh;
1593 3b6ceab7 2022-01-10 thomas 	} else {
1594 3b6ceab7 2022-01-10 thomas 		err = packhdr(&nh, buf, sizeof(buf),
1595 3b6ceab7 2022-01-10 thomas 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1596 3b6ceab7 2022-01-10 thomas 		if (err)
1597 3b6ceab7 2022-01-10 thomas 			return err;
1598 3b6ceab7 2022-01-10 thomas 		err = hwrite(packfile, buf, nh, ctx);
1599 3b6ceab7 2022-01-10 thomas 		if (err)
1600 3b6ceab7 2022-01-10 thomas 			return err;
1601 3b6ceab7 2022-01-10 thomas 		*packfile_size += nh;
1602 3b6ceab7 2022-01-10 thomas 		err = hwrite(packfile, m->prev->id.sha1,
1603 3b6ceab7 2022-01-10 thomas 		    sizeof(m->prev->id.sha1), ctx);
1604 3b6ceab7 2022-01-10 thomas 		if (err)
1605 3b6ceab7 2022-01-10 thomas 			return err;
1606 3b6ceab7 2022-01-10 thomas 		*packfile_size += sizeof(m->prev->id.sha1);
1607 3b6ceab7 2022-01-10 thomas 	}
1608 3b6ceab7 2022-01-10 thomas 
1609 3b6ceab7 2022-01-10 thomas 	return NULL;
1610 3b6ceab7 2022-01-10 thomas }
1611 3b6ceab7 2022-01-10 thomas 
1612 3b6ceab7 2022-01-10 thomas static const struct got_error *
1613 f9c2e8e5 2022-02-13 thomas write_packed_object(off_t *packfile_size, FILE *packfile,
1614 f9c2e8e5 2022-02-13 thomas     FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1615 f9c2e8e5 2022-02-13 thomas     SHA1_CTX *ctx, struct got_repository *repo)
1616 f9c2e8e5 2022-02-13 thomas {
1617 f9c2e8e5 2022-02-13 thomas 	const struct got_error *err = NULL;
1618 f9c2e8e5 2022-02-13 thomas 	struct got_deflate_checksum csum;
1619 f9c2e8e5 2022-02-13 thomas 	char buf[32];
1620 f9c2e8e5 2022-02-13 thomas 	int nh;
1621 f9c2e8e5 2022-02-13 thomas 	struct got_raw_object *raw = NULL;
1622 f9c2e8e5 2022-02-13 thomas 	off_t outlen;
1623 f9c2e8e5 2022-02-13 thomas 
1624 f9c2e8e5 2022-02-13 thomas 	csum.output_sha1 = ctx;
1625 f9c2e8e5 2022-02-13 thomas 	csum.output_crc = NULL;
1626 f9c2e8e5 2022-02-13 thomas 
1627 f9c2e8e5 2022-02-13 thomas 	m->off = ftello(packfile);
1628 f9c2e8e5 2022-02-13 thomas 	if (m->delta_len == 0) {
1629 f9c2e8e5 2022-02-13 thomas 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1630 f9c2e8e5 2022-02-13 thomas 		if (err)
1631 f9c2e8e5 2022-02-13 thomas 			goto done;
1632 f9c2e8e5 2022-02-13 thomas 		err = packhdr(&nh, buf, sizeof(buf),
1633 f9c2e8e5 2022-02-13 thomas 		    m->obj_type, raw->size);
1634 f9c2e8e5 2022-02-13 thomas 		if (err)
1635 f9c2e8e5 2022-02-13 thomas 			goto done;
1636 f9c2e8e5 2022-02-13 thomas 		err = hwrite(packfile, buf, nh, ctx);
1637 f9c2e8e5 2022-02-13 thomas 		if (err)
1638 f9c2e8e5 2022-02-13 thomas 			goto done;
1639 f9c2e8e5 2022-02-13 thomas 		*packfile_size += nh;
1640 f9c2e8e5 2022-02-13 thomas 		if (raw->f == NULL) {
1641 f9c2e8e5 2022-02-13 thomas 			err = got_deflate_to_file_mmap(&outlen,
1642 f9c2e8e5 2022-02-13 thomas 			    raw->data + raw->hdrlen, 0, raw->size,
1643 f9c2e8e5 2022-02-13 thomas 			    packfile, &csum);
1644 f9c2e8e5 2022-02-13 thomas 			if (err)
1645 f9c2e8e5 2022-02-13 thomas 				goto done;
1646 f9c2e8e5 2022-02-13 thomas 		} else {
1647 f9c2e8e5 2022-02-13 thomas 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1648 f9c2e8e5 2022-02-13 thomas 			    == -1) {
1649 f9c2e8e5 2022-02-13 thomas 				err = got_error_from_errno("fseeko");
1650 f9c2e8e5 2022-02-13 thomas 				goto done;
1651 f9c2e8e5 2022-02-13 thomas 			}
1652 f9c2e8e5 2022-02-13 thomas 			err = got_deflate_to_file(&outlen, raw->f,
1653 f9c2e8e5 2022-02-13 thomas 			    raw->size, packfile, &csum);
1654 f9c2e8e5 2022-02-13 thomas 			if (err)
1655 f9c2e8e5 2022-02-13 thomas 				goto done;
1656 f9c2e8e5 2022-02-13 thomas 		}
1657 f9c2e8e5 2022-02-13 thomas 		*packfile_size += outlen;
1658 f9c2e8e5 2022-02-13 thomas 		got_object_raw_close(raw);
1659 f9c2e8e5 2022-02-13 thomas 		raw = NULL;
1660 f9c2e8e5 2022-02-13 thomas 	} else if (m->delta_buf) {
1661 f9c2e8e5 2022-02-13 thomas 		err = deltahdr(packfile_size, ctx, packfile, m);
1662 f9c2e8e5 2022-02-13 thomas 		if (err)
1663 f9c2e8e5 2022-02-13 thomas 			goto done;
1664 f9c2e8e5 2022-02-13 thomas 		err = got_deflate_to_file_mmap(&outlen,
1665 f9c2e8e5 2022-02-13 thomas 		    m->delta_buf, 0, m->delta_len, packfile, &csum);
1666 f9c2e8e5 2022-02-13 thomas 		if (err)
1667 f9c2e8e5 2022-02-13 thomas 			goto done;
1668 f9c2e8e5 2022-02-13 thomas 		*packfile_size += outlen;
1669 f9c2e8e5 2022-02-13 thomas 		free(m->delta_buf);
1670 f9c2e8e5 2022-02-13 thomas 		m->delta_buf = NULL;
1671 f9c2e8e5 2022-02-13 thomas 	} else {
1672 f9c2e8e5 2022-02-13 thomas 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1673 f9c2e8e5 2022-02-13 thomas 		    == -1) {
1674 f9c2e8e5 2022-02-13 thomas 			err = got_error_from_errno("fseeko");
1675 f9c2e8e5 2022-02-13 thomas 			goto done;
1676 f9c2e8e5 2022-02-13 thomas 		}
1677 f9c2e8e5 2022-02-13 thomas 		err = deltahdr(packfile_size, ctx, packfile, m);
1678 f9c2e8e5 2022-02-13 thomas 		if (err)
1679 f9c2e8e5 2022-02-13 thomas 			goto done;
1680 f9c2e8e5 2022-02-13 thomas 		err = got_deflate_to_file(&outlen, delta_cache,
1681 f9c2e8e5 2022-02-13 thomas 		    m->delta_len, packfile, &csum);
1682 f9c2e8e5 2022-02-13 thomas 		if (err)
1683 f9c2e8e5 2022-02-13 thomas 			goto done;
1684 f9c2e8e5 2022-02-13 thomas 		*packfile_size += outlen;
1685 f9c2e8e5 2022-02-13 thomas 	}
1686 f9c2e8e5 2022-02-13 thomas done:
1687 f9c2e8e5 2022-02-13 thomas 	if (raw)
1688 f9c2e8e5 2022-02-13 thomas 		got_object_raw_close(raw);
1689 f9c2e8e5 2022-02-13 thomas 	return err;
1690 f9c2e8e5 2022-02-13 thomas }
1691 f9c2e8e5 2022-02-13 thomas 
1692 f9c2e8e5 2022-02-13 thomas static const struct got_error *
1693 b79280dd 2021-10-15 thomas genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1694 f9c2e8e5 2022-02-13 thomas     struct got_pack_meta **deltify, int ndeltify,
1695 f9c2e8e5 2022-02-13 thomas     struct got_pack_meta **reuse, int nreuse,
1696 85220b0e 2022-03-22 thomas     int ncolored, int nfound, int ntrees, int nours,
1697 85220b0e 2022-03-22 thomas     struct got_repository *repo,
1698 05118f5a 2021-06-22 stsp     got_pack_progress_cb progress_cb, void *progress_arg,
1699 31ba2236 2022-01-05 thomas     struct got_ratelimit *rl,
1700 05118f5a 2021-06-22 stsp     got_cancel_cb cancel_cb, void *cancel_arg)
1701 e6bcace5 2021-06-22 stsp {
1702 e6bcace5 2021-06-22 stsp 	const struct got_error *err = NULL;
1703 f9c2e8e5 2022-02-13 thomas 	int i;
1704 e6bcace5 2021-06-22 stsp 	SHA1_CTX ctx;
1705 e6bcace5 2021-06-22 stsp 	struct got_pack_meta *m;
1706 e4d8ab47 2021-10-15 thomas 	char buf[32];
1707 e8f02263 2022-01-23 thomas 	size_t n;
1708 f9c2e8e5 2022-02-13 thomas 	off_t packfile_size = 0;
1709 ecf9545f 2021-10-15 thomas 	int outfd = -1;
1710 e6bcace5 2021-06-22 stsp 
1711 e6bcace5 2021-06-22 stsp 	SHA1Init(&ctx);
1712 e6bcace5 2021-06-22 stsp 
1713 e6bcace5 2021-06-22 stsp 	err = hwrite(packfile, "PACK", 4, &ctx);
1714 e6bcace5 2021-06-22 stsp 	if (err)
1715 e6bcace5 2021-06-22 stsp 		return err;
1716 e6bcace5 2021-06-22 stsp 	putbe32(buf, GOT_PACKFILE_VERSION);
1717 e6bcace5 2021-06-22 stsp 	err = hwrite(packfile, buf, 4, &ctx);
1718 e6bcace5 2021-06-22 stsp 	if (err)
1719 e6bcace5 2021-06-22 stsp 		goto done;
1720 f9c2e8e5 2022-02-13 thomas 	putbe32(buf, ndeltify + nreuse);
1721 e6bcace5 2021-06-22 stsp 	err = hwrite(packfile, buf, 4, &ctx);
1722 e6bcace5 2021-06-22 stsp 	if (err)
1723 e6bcace5 2021-06-22 stsp 		goto done;
1724 f9c2e8e5 2022-02-13 thomas 
1725 f9c2e8e5 2022-02-13 thomas 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1726 f9c2e8e5 2022-02-13 thomas 	    write_order_cmp);
1727 f9c2e8e5 2022-02-13 thomas 	for (i = 0; i < ndeltify; i++) {
1728 31ba2236 2022-01-05 thomas 		err = report_progress(progress_cb, progress_arg, rl,
1729 85220b0e 2022-03-22 thomas 		    ncolored, nfound, ntrees, packfile_size, nours,
1730 85220b0e 2022-03-22 thomas 		    ndeltify + nreuse, ndeltify + nreuse, i);
1731 31ba2236 2022-01-05 thomas 		if (err)
1732 31ba2236 2022-01-05 thomas 			goto done;
1733 f9c2e8e5 2022-02-13 thomas 		m = deltify[i];
1734 f9c2e8e5 2022-02-13 thomas 		err = write_packed_object(&packfile_size, packfile,
1735 f9c2e8e5 2022-02-13 thomas 		    delta_cache, m, &outfd, &ctx, repo);
1736 f9c2e8e5 2022-02-13 thomas 		if (err)
1737 f9c2e8e5 2022-02-13 thomas 			goto done;
1738 e6bcace5 2021-06-22 stsp 	}
1739 f9c2e8e5 2022-02-13 thomas 
1740 f9c2e8e5 2022-02-13 thomas 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1741 f9c2e8e5 2022-02-13 thomas 	    reuse_write_order_cmp);
1742 f9c2e8e5 2022-02-13 thomas 	for (i = 0; i < nreuse; i++) {
1743 f9c2e8e5 2022-02-13 thomas 		err = report_progress(progress_cb, progress_arg, rl,
1744 85220b0e 2022-03-22 thomas 		    ncolored, nfound, ntrees, packfile_size, nours,
1745 85220b0e 2022-03-22 thomas 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1746 f9c2e8e5 2022-02-13 thomas 		if (err)
1747 f9c2e8e5 2022-02-13 thomas 			goto done;
1748 f9c2e8e5 2022-02-13 thomas 		m = reuse[i];
1749 f9c2e8e5 2022-02-13 thomas 		err = write_packed_object(&packfile_size, packfile,
1750 f9c2e8e5 2022-02-13 thomas 		    delta_cache, m, &outfd, &ctx, repo);
1751 f9c2e8e5 2022-02-13 thomas 		if (err)
1752 f9c2e8e5 2022-02-13 thomas 			goto done;
1753 f9c2e8e5 2022-02-13 thomas 	}
1754 f9c2e8e5 2022-02-13 thomas 
1755 e6bcace5 2021-06-22 stsp 	SHA1Final(pack_sha1, &ctx);
1756 e6bcace5 2021-06-22 stsp 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1757 e6bcace5 2021-06-22 stsp 	if (n != SHA1_DIGEST_LENGTH)
1758 e6bcace5 2021-06-22 stsp 		err = got_ferror(packfile, GOT_ERR_IO);
1759 05118f5a 2021-06-22 stsp 	packfile_size += SHA1_DIGEST_LENGTH;
1760 dc7edd42 2021-08-22 stsp 	packfile_size += sizeof(struct got_packfile_hdr);
1761 31ba2236 2022-01-05 thomas 	if (progress_cb) {
1762 85220b0e 2022-03-22 thomas 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1763 85220b0e 2022-03-22 thomas 		    packfile_size, nours, ndeltify + nreuse,
1764 85220b0e 2022-03-22 thomas 		    ndeltify + nreuse, ndeltify + nreuse);
1765 31ba2236 2022-01-05 thomas 		if (err)
1766 31ba2236 2022-01-05 thomas 			goto done;
1767 31ba2236 2022-01-05 thomas 	}
1768 e6bcace5 2021-06-22 stsp done:
1769 ecf9545f 2021-10-15 thomas 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1770 ecf9545f 2021-10-15 thomas 		err = got_error_from_errno("close");
1771 e6bcace5 2021-06-22 stsp 	return err;
1772 e6bcace5 2021-06-22 stsp }
1773 e6bcace5 2021-06-22 stsp 
1774 f9c2e8e5 2022-02-13 thomas static const struct got_error *
1775 f9c2e8e5 2022-02-13 thomas remove_unused_object(struct got_object_idset_element *entry, void *arg)
1776 f9c2e8e5 2022-02-13 thomas {
1777 f9c2e8e5 2022-02-13 thomas 	struct got_object_idset *idset = arg;
1778 f9c2e8e5 2022-02-13 thomas 
1779 f9c2e8e5 2022-02-13 thomas 	if (got_object_idset_get_element_data(entry) == NULL)
1780 f9c2e8e5 2022-02-13 thomas 		got_object_idset_remove_element(idset, entry);
1781 f9c2e8e5 2022-02-13 thomas 
1782 f9c2e8e5 2022-02-13 thomas 	return NULL;
1783 f9c2e8e5 2022-02-13 thomas }
1784 f9c2e8e5 2022-02-13 thomas 
1785 f9c2e8e5 2022-02-13 thomas static const struct got_error *
1786 f9c2e8e5 2022-02-13 thomas remove_reused_object(struct got_object_idset_element *entry, void *arg)
1787 f9c2e8e5 2022-02-13 thomas {
1788 f9c2e8e5 2022-02-13 thomas 	struct got_object_idset *idset = arg;
1789 f9c2e8e5 2022-02-13 thomas 	struct got_pack_meta *m;
1790 f9c2e8e5 2022-02-13 thomas 
1791 f9c2e8e5 2022-02-13 thomas 	m = got_object_idset_get_element_data(entry);
1792 f9c2e8e5 2022-02-13 thomas 	if (m->have_reused_delta)
1793 f9c2e8e5 2022-02-13 thomas 		got_object_idset_remove_element(idset, entry);
1794 f9c2e8e5 2022-02-13 thomas 
1795 f9c2e8e5 2022-02-13 thomas 	return NULL;
1796 f9c2e8e5 2022-02-13 thomas }
1797 f9c2e8e5 2022-02-13 thomas 
1798 f9c2e8e5 2022-02-13 thomas static const struct got_error *
1799 f9c2e8e5 2022-02-13 thomas add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1800 f9c2e8e5 2022-02-13 thomas {
1801 f9c2e8e5 2022-02-13 thomas 	struct got_pack_meta *m = data;
1802 f9c2e8e5 2022-02-13 thomas 	struct got_pack_metavec *v = arg;
1803 f9c2e8e5 2022-02-13 thomas 
1804 f9c2e8e5 2022-02-13 thomas 	return add_meta(m, v);
1805 f9c2e8e5 2022-02-13 thomas }
1806 f9c2e8e5 2022-02-13 thomas 
1807 e6bcace5 2021-06-22 stsp const struct got_error *
1808 e6bcace5 2021-06-22 stsp got_pack_create(uint8_t *packsha1, FILE *packfile,
1809 e6bcace5 2021-06-22 stsp     struct got_object_id **theirs, int ntheirs,
1810 e6bcace5 2021-06-22 stsp     struct got_object_id **ours, int nours,
1811 f8a36e22 2021-08-26 stsp     struct got_repository *repo, int loose_obj_only, int allow_empty,
1812 05118f5a 2021-06-22 stsp     got_pack_progress_cb progress_cb, void *progress_arg,
1813 05118f5a 2021-06-22 stsp     got_cancel_cb cancel_cb, void *cancel_arg)
1814 e6bcace5 2021-06-22 stsp {
1815 e6bcace5 2021-06-22 stsp 	const struct got_error *err;
1816 f9c2e8e5 2022-02-13 thomas 	int delta_cache_fd = -1;
1817 b79280dd 2021-10-15 thomas 	FILE *delta_cache = NULL;
1818 f9c2e8e5 2022-02-13 thomas 	struct got_object_idset *idset;
1819 31ba2236 2022-01-05 thomas 	struct got_ratelimit rl;
1820 f9c2e8e5 2022-02-13 thomas 	struct got_pack_metavec deltify, reuse;
1821 85220b0e 2022-03-22 thomas 	int ncolored = 0, nfound = 0, ntrees = 0;
1822 e6bcace5 2021-06-22 stsp 
1823 f9c2e8e5 2022-02-13 thomas 	memset(&deltify, 0, sizeof(deltify));
1824 f9c2e8e5 2022-02-13 thomas 	memset(&reuse, 0, sizeof(reuse));
1825 f9c2e8e5 2022-02-13 thomas 
1826 31ba2236 2022-01-05 thomas 	got_ratelimit_init(&rl, 0, 500);
1827 31ba2236 2022-01-05 thomas 
1828 f9c2e8e5 2022-02-13 thomas 	idset = got_object_idset_alloc();
1829 f9c2e8e5 2022-02-13 thomas 	if (idset == NULL)
1830 f9c2e8e5 2022-02-13 thomas 		return got_error_from_errno("got_object_idset_alloc");
1831 f9c2e8e5 2022-02-13 thomas 
1832 85220b0e 2022-03-22 thomas 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1833 85220b0e 2022-03-22 thomas 	    ntheirs, ours, nours, repo, loose_obj_only,
1834 85220b0e 2022-03-22 thomas 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1835 e6bcace5 2021-06-22 stsp 	if (err)
1836 e6bcace5 2021-06-22 stsp 		return err;
1837 e6bcace5 2021-06-22 stsp 
1838 f9c2e8e5 2022-02-13 thomas 	err = got_object_idset_for_each_element(idset,
1839 f9c2e8e5 2022-02-13 thomas 	    remove_unused_object, idset);
1840 f9c2e8e5 2022-02-13 thomas 	if (err)
1841 f9c2e8e5 2022-02-13 thomas 		goto done;
1842 f9c2e8e5 2022-02-13 thomas 
1843 f9a643e8 2022-02-13 thomas 	if (progress_cb) {
1844 85220b0e 2022-03-22 thomas 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1845 85220b0e 2022-03-22 thomas 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
1846 f9a643e8 2022-02-13 thomas 		if (err)
1847 f9a643e8 2022-02-13 thomas 			goto done;
1848 f9a643e8 2022-02-13 thomas 	}
1849 f9a643e8 2022-02-13 thomas 
1850 f9c2e8e5 2022-02-13 thomas 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1851 05118f5a 2021-06-22 stsp 		err = got_error(GOT_ERR_CANNOT_PACK);
1852 05118f5a 2021-06-22 stsp 		goto done;
1853 05118f5a 2021-06-22 stsp 	}
1854 b79280dd 2021-10-15 thomas 
1855 f9c2e8e5 2022-02-13 thomas 	delta_cache_fd = got_opentempfd();
1856 f9c2e8e5 2022-02-13 thomas 	if (delta_cache_fd == -1) {
1857 b79280dd 2021-10-15 thomas 		err = got_error_from_errno("got_opentemp");
1858 b79280dd 2021-10-15 thomas 		goto done;
1859 b79280dd 2021-10-15 thomas 	}
1860 b79280dd 2021-10-15 thomas 
1861 f9c2e8e5 2022-02-13 thomas 	reuse.metasz = 64;
1862 f9c2e8e5 2022-02-13 thomas 	reuse.meta = calloc(reuse.metasz,
1863 f9c2e8e5 2022-02-13 thomas 	    sizeof(struct got_pack_meta *));
1864 f9c2e8e5 2022-02-13 thomas 	if (reuse.meta == NULL) {
1865 f9c2e8e5 2022-02-13 thomas 		err = got_error_from_errno("calloc");
1866 f9c2e8e5 2022-02-13 thomas 		goto done;
1867 f9c2e8e5 2022-02-13 thomas 	}
1868 f9c2e8e5 2022-02-13 thomas 
1869 85220b0e 2022-03-22 thomas 	err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
1870 85220b0e 2022-03-22 thomas 	    ntrees, nours, repo, progress_cb, progress_arg, &rl,
1871 85220b0e 2022-03-22 thomas 	    cancel_cb, cancel_arg);
1872 f9c2e8e5 2022-02-13 thomas 	if (err)
1873 f9c2e8e5 2022-02-13 thomas 		goto done;
1874 f9c2e8e5 2022-02-13 thomas 	if (reuse.nmeta > 0) {
1875 f9c2e8e5 2022-02-13 thomas 		err = got_object_idset_for_each_element(idset,
1876 f9c2e8e5 2022-02-13 thomas 		    remove_reused_object, idset);
1877 f9c2e8e5 2022-02-13 thomas 		if (err)
1878 f9c2e8e5 2022-02-13 thomas 			goto done;
1879 f9c2e8e5 2022-02-13 thomas 	}
1880 f9c2e8e5 2022-02-13 thomas 
1881 f9c2e8e5 2022-02-13 thomas 	delta_cache = fdopen(delta_cache_fd, "a+");
1882 f9c2e8e5 2022-02-13 thomas 	if (delta_cache == NULL) {
1883 f9c2e8e5 2022-02-13 thomas 		err = got_error_from_errno("fdopen");
1884 f9c2e8e5 2022-02-13 thomas 		goto done;
1885 f9c2e8e5 2022-02-13 thomas 	}
1886 f9c2e8e5 2022-02-13 thomas 	delta_cache_fd = -1;
1887 f9c2e8e5 2022-02-13 thomas 
1888 f9c2e8e5 2022-02-13 thomas 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1889 f9c2e8e5 2022-02-13 thomas 		err = got_error_from_errno("fseeko");
1890 f9c2e8e5 2022-02-13 thomas 		goto done;
1891 f9c2e8e5 2022-02-13 thomas 	}
1892 f9c2e8e5 2022-02-13 thomas 
1893 f9c2e8e5 2022-02-13 thomas 	deltify.meta = calloc(got_object_idset_num_elements(idset),
1894 f9c2e8e5 2022-02-13 thomas 	    sizeof(struct got_pack_meta *));
1895 f9c2e8e5 2022-02-13 thomas 	if (deltify.meta == NULL) {
1896 f9c2e8e5 2022-02-13 thomas 		err = got_error_from_errno("calloc");
1897 f9c2e8e5 2022-02-13 thomas 		goto done;
1898 f9c2e8e5 2022-02-13 thomas 	}
1899 f9c2e8e5 2022-02-13 thomas 	deltify.metasz = got_object_idset_num_elements(idset);
1900 f9c2e8e5 2022-02-13 thomas 
1901 f9c2e8e5 2022-02-13 thomas 	err = got_object_idset_for_each(idset, add_meta_idset_cb, &deltify);
1902 f9c2e8e5 2022-02-13 thomas 	if (err)
1903 f9c2e8e5 2022-02-13 thomas 		goto done;
1904 f9c2e8e5 2022-02-13 thomas 	if (deltify.nmeta > 0) {
1905 85220b0e 2022-03-22 thomas 		err = pick_deltas(deltify.meta, deltify.nmeta, ncolored,
1906 85220b0e 2022-03-22 thomas 		    nfound, ntrees, nours, reuse.nmeta, delta_cache, repo,
1907 31ba2236 2022-01-05 thomas 		    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1908 f8a36e22 2021-08-26 stsp 		if (err)
1909 f8a36e22 2021-08-26 stsp 			goto done;
1910 b79280dd 2021-10-15 thomas 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1911 b79280dd 2021-10-15 thomas 			err = got_error_from_errno("fseeko");
1912 b79280dd 2021-10-15 thomas 			goto done;
1913 b79280dd 2021-10-15 thomas 		}
1914 f8a36e22 2021-08-26 stsp 	}
1915 e6bcace5 2021-06-22 stsp 
1916 f9c2e8e5 2022-02-13 thomas 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
1917 85220b0e 2022-03-22 thomas 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1918 85220b0e 2022-03-22 thomas 	    nours, repo, progress_cb, progress_arg, &rl,
1919 85220b0e 2022-03-22 thomas 	    cancel_cb, cancel_arg);
1920 e6bcace5 2021-06-22 stsp 	if (err)
1921 e6bcace5 2021-06-22 stsp 		goto done;
1922 e6bcace5 2021-06-22 stsp done:
1923 f9c2e8e5 2022-02-13 thomas 	free_nmeta(deltify.meta, deltify.nmeta);
1924 f9c2e8e5 2022-02-13 thomas 	free_nmeta(reuse.meta, reuse.nmeta);
1925 f9c2e8e5 2022-02-13 thomas 	got_object_idset_free(idset);
1926 f9c2e8e5 2022-02-13 thomas 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1927 f9c2e8e5 2022-02-13 thomas 		err = got_error_from_errno("close");
1928 b79280dd 2021-10-15 thomas 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1929 b79280dd 2021-10-15 thomas 		err = got_error_from_errno("fclose");
1930 e6bcace5 2021-06-22 stsp 	return err;
1931 e6bcace5 2021-06-22 stsp }