2 5aa81393 2020-01-06 stsp * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
4 718b3ab0 2018-03-17 stsp * Permission to use, copy, modify, and distribute this software for any
5 718b3ab0 2018-03-17 stsp * purpose with or without fee is hereby granted, provided that the above
6 718b3ab0 2018-03-17 stsp * copyright notice and this permission notice appear in all copies.
8 718b3ab0 2018-03-17 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 718b3ab0 2018-03-17 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 718b3ab0 2018-03-17 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 718b3ab0 2018-03-17 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 718b3ab0 2018-03-17 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 718b3ab0 2018-03-17 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 718b3ab0 2018-03-17 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 b9de8018 2022-10-13 stsp struct got_pack_privsep_child {
18 b9de8018 2022-10-13 stsp int imsg_fd;
20 b9de8018 2022-10-13 stsp struct imsgbuf *ibuf;
23 718b3ab0 2018-03-17 stsp /* An open pack file. */
24 718b3ab0 2018-03-17 stsp struct got_pack {
25 718b3ab0 2018-03-17 stsp char *path_packfile;
27 d7464085 2018-07-09 stsp uint8_t *map;
28 ad4cc361 2022-10-27 op off_t filesize;
29 b9de8018 2022-10-13 stsp struct got_pack_privsep_child *privsep_child;
31 57160834 2022-05-31 stsp int accumfd;
32 db696021 2022-01-04 stsp int child_has_tempfiles;
33 67fd6849 2022-02-13 stsp int child_has_delta_outfd;
34 ab2f42e7 2019-11-10 stsp struct got_delta_cache *delta_cache;
37 136ec6c9 2021-06-22 stsp struct got_packidx;
39 3d589bee 2022-06-25 stsp const struct got_error *got_pack_start_privsep_child(struct got_pack *,
40 3d589bee 2022-06-25 stsp struct got_packidx *);
41 d7464085 2018-07-09 stsp const struct got_error *got_pack_close(struct got_pack *);
43 668a20f6 2020-03-18 stsp const struct got_error *got_pack_parse_offset_delta(off_t *, size_t *,
44 24d916d2 2022-10-24 op struct got_pack *, off_t, size_t);
45 c4330eff 2021-06-22 stsp const struct got_error *got_pack_parse_ref_delta(struct got_object_id *,
46 c4330eff 2021-06-22 stsp struct got_pack *, off_t, int);
47 668a20f6 2020-03-18 stsp const struct got_error *got_pack_resolve_delta_chain(struct got_delta_chain *,
48 668a20f6 2020-03-18 stsp struct got_packidx *, struct got_pack *, off_t, size_t, int, size_t,
49 668a20f6 2020-03-18 stsp unsigned int);
50 668a20f6 2020-03-18 stsp const struct got_error *got_pack_parse_object_type_and_size(uint8_t *,
51 668a20f6 2020-03-18 stsp uint64_t *, size_t *, struct got_pack *, off_t);
53 1510f469 2018-09-09 stsp #define GOT_PACK_PREFIX "pack-"
54 1510f469 2018-09-09 stsp #define GOT_PACKFILE_SUFFIX ".pack"
55 1510f469 2018-09-09 stsp #define GOT_PACKIDX_SUFFIX ".idx"
56 1510f469 2018-09-09 stsp #define GOT_PACKFILE_NAMELEN (strlen(GOT_PACK_PREFIX) + \
57 1510f469 2018-09-09 stsp SHA1_DIGEST_STRING_LENGTH - 1 + \
58 1510f469 2018-09-09 stsp strlen(GOT_PACKFILE_SUFFIX))
59 1510f469 2018-09-09 stsp #define GOT_PACKIDX_NAMELEN (strlen(GOT_PACK_PREFIX) + \
60 1510f469 2018-09-09 stsp SHA1_DIGEST_STRING_LENGTH - 1 + \
61 1510f469 2018-09-09 stsp strlen(GOT_PACKIDX_SUFFIX))
63 718b3ab0 2018-03-17 stsp /* See Documentation/technical/pack-format.txt in Git. */
65 718b3ab0 2018-03-17 stsp struct got_packidx_trailer {
66 718b3ab0 2018-03-17 stsp u_int8_t packfile_sha1[SHA1_DIGEST_LENGTH];
67 718b3ab0 2018-03-17 stsp u_int8_t packidx_sha1[SHA1_DIGEST_LENGTH];
68 718b3ab0 2018-03-17 stsp } __attribute__((__packed__));
70 57b35b75 2018-06-22 stsp struct got_packidx_object_id {
71 57b35b75 2018-06-22 stsp u_int8_t sha1[SHA1_DIGEST_LENGTH];
72 57b35b75 2018-06-22 stsp } __attribute__((__packed__));
74 718b3ab0 2018-03-17 stsp /* Ignore pack index version 1 which is no longer written by Git. */
75 718b3ab0 2018-03-17 stsp #define GOT_PACKIDX_VERSION 2
77 718b3ab0 2018-03-17 stsp struct got_packidx_v2_hdr {
78 57b35b75 2018-06-22 stsp uint32_t *magic; /* big endian */
79 718b3ab0 2018-03-17 stsp #define GOT_PACKIDX_V2_MAGIC 0xff744f63 /* "\377t0c" */
80 57b35b75 2018-06-22 stsp uint32_t *version;
83 718b3ab0 2018-03-17 stsp * Each entry N in the fanout table contains the number of objects in
84 718b3ab0 2018-03-17 stsp * the packfile whose SHA1 begins with a byte less than or equal to N.
85 718b3ab0 2018-03-17 stsp * The last entry (index 255) contains the number of objects in the
86 718b3ab0 2018-03-17 stsp * pack file whose first SHA1 byte is <= 0xff, and thus records the
87 718b3ab0 2018-03-17 stsp * total number of objects in the pack file. All pointer variables
88 718b3ab0 2018-03-17 stsp * below point to tables with a corresponding number of entries.
90 57b35b75 2018-06-22 stsp uint32_t *fanout_table; /* values are big endian */
91 57b35b75 2018-06-22 stsp #define GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS (0xff + 1)
93 718b3ab0 2018-03-17 stsp /* Sorted SHA1 checksums for each object in the pack file. */
94 57b35b75 2018-06-22 stsp struct got_packidx_object_id *sorted_ids;
96 718b3ab0 2018-03-17 stsp /* CRC32 of the packed representation of each object. */
97 718b3ab0 2018-03-17 stsp uint32_t *crc32;
99 718b3ab0 2018-03-17 stsp /* Offset into the pack file for each object. */
100 718b3ab0 2018-03-17 stsp uint32_t *offsets; /* values are big endian */
101 718b3ab0 2018-03-17 stsp #define GOT_PACKIDX_OFFSET_VAL_MASK 0x7fffffff
102 718b3ab0 2018-03-17 stsp #define GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX 0x80000000
104 718b3ab0 2018-03-17 stsp /* Large offsets table is empty for pack files < 2 GB. */
105 718b3ab0 2018-03-17 stsp uint64_t *large_offsets; /* values are big endian */
107 57b35b75 2018-06-22 stsp struct got_packidx_trailer *trailer;
110 67fd6849 2022-02-13 stsp struct got_pack_offset_index {
111 67fd6849 2022-02-13 stsp uint32_t offset;
112 67fd6849 2022-02-13 stsp uint32_t idx;
115 67fd6849 2022-02-13 stsp struct got_pack_large_offset_index {
116 67fd6849 2022-02-13 stsp uint64_t offset;
117 67fd6849 2022-02-13 stsp uint32_t idx;
120 6fd11751 2018-06-04 stsp /* An open pack index file. */
121 6fd11751 2018-06-04 stsp struct got_packidx {
122 6fd11751 2018-06-04 stsp char *path_packidx; /* actual on-disk path */
124 57b35b75 2018-06-22 stsp uint8_t *map;
125 57b35b75 2018-06-22 stsp size_t len;
126 5e6be232 2019-11-08 stsp size_t nlargeobj;
127 57b35b75 2018-06-22 stsp struct got_packidx_v2_hdr hdr; /* convenient pointers into map */
128 67fd6849 2022-02-13 stsp struct got_pack_offset_index *sorted_offsets;
129 67fd6849 2022-02-13 stsp struct got_pack_large_offset_index *sorted_large_offsets;
132 718b3ab0 2018-03-17 stsp struct got_packfile_hdr {
133 718b3ab0 2018-03-17 stsp uint32_t signature;
134 718b3ab0 2018-03-17 stsp #define GOT_PACKFILE_SIGNATURE 0x5041434b /* 'P' 'A' 'C' 'K' */
135 718b3ab0 2018-03-17 stsp uint32_t version; /* big endian */
136 718b3ab0 2018-03-17 stsp #define GOT_PACKFILE_VERSION 2
137 718b3ab0 2018-03-17 stsp uint32_t nobjects; /* big endian */
140 718b3ab0 2018-03-17 stsp struct got_packfile_obj_hdr {
142 718b3ab0 2018-03-17 stsp * The object size field uses a variable length encoding:
143 718b3ab0 2018-03-17 stsp * size0...sizeN form a 4+7+7+...+7 bit integer, where size0 is the
144 718b3ab0 2018-03-17 stsp * least significant part and sizeN is the most significant part.
145 718b3ab0 2018-03-17 stsp * If the MSB of a size byte is set, an additional size byte follows.
146 718b3ab0 2018-03-17 stsp * Of the 7 remaining bits of size0, the first 3 bits indicate the
147 718b3ab0 2018-03-17 stsp * object's type, and the remaining 4 bits contribute to the size.
149 718b3ab0 2018-03-17 stsp uint8_t *size; /* variable length */
150 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_SIZE_MORE 0x80
151 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_SIZE0_TYPE_MASK 0x70 /* See struct got_object->type */
152 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT 4
153 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_SIZE0_VAL_MASK 0x0f
154 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_SIZE_VAL_MASK 0x7f
157 718b3ab0 2018-03-17 stsp /* If object is not a DELTA type. */
158 718b3ab0 2018-03-17 stsp struct got_packfile_object_data {
159 718b3ab0 2018-03-17 stsp uint8_t *data; /* compressed */
162 718b3ab0 2018-03-17 stsp /* If object is of type GOT_OBJ_TYPE_REF_DELTA. */
163 718b3ab0 2018-03-17 stsp struct got_packfile_object_data_ref_delta {
164 718b3ab0 2018-03-17 stsp uint8_t sha1[SHA1_DIGEST_LENGTH];
165 718b3ab0 2018-03-17 stsp uint8_t *delta_data; /* compressed */
168 718b3ab0 2018-03-17 stsp /* If object is of type GOT_OBJ_TYPE_OFFSET_DELTA. */
169 718b3ab0 2018-03-17 stsp struct got_packfile_object_data_offset_delta {
171 718b3ab0 2018-03-17 stsp * This offset is interpreted as a negative offset from
172 718b3ab0 2018-03-17 stsp * the got_packfile_obj_hdr corresponding to this object.
173 718b3ab0 2018-03-17 stsp * The size provided in the header specifies the amount
174 718b3ab0 2018-03-17 stsp * of compressed delta data that follows.
176 718b3ab0 2018-03-17 stsp * This field uses a variable length encoding of N bytes,
177 718b3ab0 2018-03-17 stsp * where the MSB is always set except for the last byte.
178 718b3ab0 2018-03-17 stsp * The value is encoded as a series of N 7 bit integers,
179 718b3ab0 2018-03-17 stsp * which are concatenated, and if N > 1 the value 2^7 +
180 718b3ab0 2018-03-17 stsp * 2^14 + ... + 2^(7 * (n-1)) is added to the result.
182 718b3ab0 2018-03-17 stsp uint8_t *offset; /* variable length */
183 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_DELTA_OFF_MORE 0x80
184 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_DELTA_OFF_VAL_MASK 0x7f
185 718b3ab0 2018-03-17 stsp uint8_t *delta_data; /* compressed */
188 718b3ab0 2018-03-17 stsp struct got_packfile_obj_data {
190 718b3ab0 2018-03-17 stsp struct got_packfile_object_data data;
191 718b3ab0 2018-03-17 stsp struct got_packfile_object_data_ref_delta ref_delta;
192 718b3ab0 2018-03-17 stsp struct got_packfile_object_data_offset_delta offset_delta;
193 718b3ab0 2018-03-17 stsp } __attribute__((__packed__));
194 718b3ab0 2018-03-17 stsp } __attribute__((__packed__));
196 c3564dfa 2021-07-15 stsp const struct got_error *got_packidx_init_hdr(struct got_packidx *, int, off_t);
197 6fd11751 2018-06-04 stsp const struct got_error *got_packidx_open(struct got_packidx **,
198 6d5a9006 2020-12-16 yzhong int, const char *, int);
199 4277420a 2019-06-29 stsp const struct got_error *got_packidx_close(struct got_packidx *);
200 aea75d87 2021-07-06 stsp const struct got_error *got_packidx_get_packfile_path(char **, const char *);
201 02828bfd 2021-06-22 stsp off_t got_packidx_get_object_offset(struct got_packidx *, int idx);
202 1510f469 2018-09-09 stsp int got_packidx_get_object_idx(struct got_packidx *, struct got_object_id *);
203 67fd6849 2022-02-13 stsp const struct got_error *got_packidx_get_offset_idx(int *, struct got_packidx *,
205 67fd6849 2022-02-13 stsp const struct got_error *got_packidx_get_object_id(struct got_object_id *,
206 67fd6849 2022-02-13 stsp struct got_packidx *, int);
207 dd88155e 2019-06-29 stsp const struct got_error *got_packidx_match_id_str_prefix(
208 dd88155e 2019-06-29 stsp struct got_object_id_queue *, struct got_packidx *, const char *);
210 718b3ab0 2018-03-17 stsp const struct got_error *got_packfile_open_object(struct got_object **,
211 2090a03d 2018-09-09 stsp struct got_pack *, struct got_packidx *, int, struct got_object_id *);
212 d582f26c 2020-03-18 stsp const struct got_error *got_pack_get_delta_chain_max_size(uint64_t *,
213 d582f26c 2020-03-18 stsp struct got_delta_chain *, struct got_pack *);
214 85a703fa 2019-01-13 stsp const struct got_error *got_pack_get_max_delta_object_size(uint64_t *,
215 42c69117 2019-11-10 stsp struct got_object *, struct got_pack *);
216 4788f1ce 2020-03-18 stsp const struct got_error *got_pack_dump_delta_chain_to_file(size_t *,
217 4788f1ce 2020-03-18 stsp struct got_delta_chain *, struct got_pack *, FILE *, FILE *, FILE *);
218 668a20f6 2020-03-18 stsp const struct got_error *got_pack_dump_delta_chain_to_mem(uint8_t **, size_t *,
219 668a20f6 2020-03-18 stsp struct got_delta_chain *, struct got_pack *);
220 24140570 2018-09-09 stsp const struct got_error *got_packfile_extract_object(struct got_pack *,
221 3840f4c9 2018-09-12 stsp struct got_object *, FILE *, FILE *, FILE *);
222 718b3ab0 2018-03-17 stsp const struct got_error *got_packfile_extract_object_to_mem(uint8_t **, size_t *,
223 7e212e3d 2018-09-09 stsp struct got_object *, struct got_pack *);
224 67fd6849 2022-02-13 stsp const struct got_error *got_packfile_extract_raw_delta(uint8_t **, size_t *,
225 24b7de1c 2022-12-03 stsp size_t *, off_t *, off_t *, off_t *, struct got_object_id *, uint64_t *,
226 24b7de1c 2022-12-03 stsp uint64_t *, struct got_pack *, struct got_packidx *, int);
