Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 1 | #include "cache.h" |
Stefan Beller | 57a6a50 | 2018-04-25 11:21:04 -0700 | [diff] [blame] | 2 | #include "repository.h" |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 3 | #include "pack.h" |
Nicolas Pitre | 70f5d5d | 2008-02-28 00:25:19 -0500 | [diff] [blame] | 4 | #include "pack-revindex.h" |
Nguyễn Thái Ngọc Duy | 1e49f22 | 2011-11-07 09:59:26 +0700 | [diff] [blame] | 5 | #include "progress.h" |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 6 | #include "packfile.h" |
Stefan Beller | a80d72d | 2018-03-23 18:20:59 +0100 | [diff] [blame] | 7 | #include "object-store.h" |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 8 | |
Jonathan Nieder | 9cba13c | 2011-03-16 02:08:34 -0500 | [diff] [blame] | 9 | struct idx_entry { |
Alexandre Julliard | 3af5192 | 2007-06-03 20:21:41 +0200 | [diff] [blame] | 10 | off_t offset; |
brian m. carlson | 9fd7504 | 2017-05-06 22:10:20 +0000 | [diff] [blame] | 11 | union idx_entry_object { |
| 12 | const unsigned char *hash; |
| 13 | struct object_id *oid; |
| 14 | } oid; |
Nicolas Pitre | c41a4a9 | 2008-06-24 23:19:02 -0400 | [diff] [blame] | 15 | unsigned int nr; |
Alexandre Julliard | 3af5192 | 2007-06-03 20:21:41 +0200 | [diff] [blame] | 16 | }; |
| 17 | |
| 18 | static int compare_entries(const void *e1, const void *e2) |
| 19 | { |
| 20 | const struct idx_entry *entry1 = e1; |
| 21 | const struct idx_entry *entry2 = e2; |
| 22 | if (entry1->offset < entry2->offset) |
| 23 | return -1; |
| 24 | if (entry1->offset > entry2->offset) |
| 25 | return 1; |
| 26 | return 0; |
| 27 | } |
| 28 | |
Nicolas Pitre | c41a4a9 | 2008-06-24 23:19:02 -0400 | [diff] [blame] | 29 | int check_pack_crc(struct packed_git *p, struct pack_window **w_curs, |
| 30 | off_t offset, off_t len, unsigned int nr) |
| 31 | { |
| 32 | const uint32_t *index_crc; |
Stephen Boyd | 1e4cd68 | 2011-04-03 00:06:54 -0700 | [diff] [blame] | 33 | uint32_t data_crc = crc32(0, NULL, 0); |
Nicolas Pitre | c41a4a9 | 2008-06-24 23:19:02 -0400 | [diff] [blame] | 34 | |
| 35 | do { |
Junio C Hamano | ef49a7a | 2011-06-10 11:52:15 -0700 | [diff] [blame] | 36 | unsigned long avail; |
Nicolas Pitre | c41a4a9 | 2008-06-24 23:19:02 -0400 | [diff] [blame] | 37 | void *data = use_pack(p, w_curs, offset, &avail); |
| 38 | if (avail > len) |
| 39 | avail = len; |
| 40 | data_crc = crc32(data_crc, data, avail); |
| 41 | offset += avail; |
| 42 | len -= avail; |
| 43 | } while (len); |
| 44 | |
| 45 | index_crc = p->index_data; |
brian m. carlson | ccc12e0 | 2018-02-01 02:18:43 +0000 | [diff] [blame] | 46 | index_crc += 2 + 256 + p->num_objects * (the_hash_algo->rawsz/4) + nr; |
Nicolas Pitre | c41a4a9 | 2008-06-24 23:19:02 -0400 | [diff] [blame] | 47 | |
| 48 | return data_crc != ntohl(*index_crc); |
| 49 | } |
| 50 | |
Nguyễn Thái Ngọc Duy | 94e1082 | 2018-11-10 06:49:07 +0100 | [diff] [blame] | 51 | static int verify_packfile(struct repository *r, |
| 52 | struct packed_git *p, |
Nguyễn Thái Ngọc Duy | c9486eb | 2011-11-07 09:59:25 +0700 | [diff] [blame] | 53 | struct pack_window **w_curs, |
Nguyễn Thái Ngọc Duy | 1e49f22 | 2011-11-07 09:59:26 +0700 | [diff] [blame] | 54 | verify_fn fn, |
| 55 | struct progress *progress, uint32_t base_count) |
| 56 | |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 57 | { |
Shawn O. Pearce | c4001d9 | 2007-03-06 20:44:30 -0500 | [diff] [blame] | 58 | off_t index_size = p->index_size; |
Nicolas Pitre | 4287307 | 2007-03-16 16:42:50 -0400 | [diff] [blame] | 59 | const unsigned char *index_base = p->index_data; |
brian m. carlson | ccc12e0 | 2018-02-01 02:18:43 +0000 | [diff] [blame] | 60 | git_hash_ctx ctx; |
brian m. carlson | 9fd7504 | 2017-05-06 22:10:20 +0000 | [diff] [blame] | 61 | unsigned char hash[GIT_MAX_RAWSZ], *pack_sig; |
Mike Hommey | 4277c67 | 2009-01-18 09:04:26 +0100 | [diff] [blame] | 62 | off_t offset = 0, pack_sig_ofs = 0; |
Shawn O. Pearce | 326bf39 | 2007-03-06 20:44:19 -0500 | [diff] [blame] | 63 | uint32_t nr_objects, i; |
Nicolas Pitre | 6241360 | 2008-05-29 17:34:50 -0400 | [diff] [blame] | 64 | int err = 0; |
Alexandre Julliard | 3af5192 | 2007-06-03 20:21:41 +0200 | [diff] [blame] | 65 | struct idx_entry *entries; |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 66 | |
Jeff King | a9445d8 | 2016-09-21 23:49:05 -0400 | [diff] [blame] | 67 | if (!is_pack_valid(p)) |
| 68 | return error("packfile %s cannot be accessed", p->pack_name); |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 69 | |
brian m. carlson | ccc12e0 | 2018-02-01 02:18:43 +0000 | [diff] [blame] | 70 | the_hash_algo->init_fn(&ctx); |
Mike Hommey | 4277c67 | 2009-01-18 09:04:26 +0100 | [diff] [blame] | 71 | do { |
Junio C Hamano | ef49a7a | 2011-06-10 11:52:15 -0700 | [diff] [blame] | 72 | unsigned long remaining; |
Shawn O. Pearce | 079afb1 | 2006-12-23 02:34:13 -0500 | [diff] [blame] | 73 | unsigned char *in = use_pack(p, w_curs, offset, &remaining); |
| 74 | offset += remaining; |
Mike Hommey | 4277c67 | 2009-01-18 09:04:26 +0100 | [diff] [blame] | 75 | if (!pack_sig_ofs) |
brian m. carlson | ccc12e0 | 2018-02-01 02:18:43 +0000 | [diff] [blame] | 76 | pack_sig_ofs = p->pack_size - the_hash_algo->rawsz; |
Nicolas Pitre | 6241360 | 2008-05-29 17:34:50 -0400 | [diff] [blame] | 77 | if (offset > pack_sig_ofs) |
| 78 | remaining -= (unsigned int)(offset - pack_sig_ofs); |
brian m. carlson | ccc12e0 | 2018-02-01 02:18:43 +0000 | [diff] [blame] | 79 | the_hash_algo->update_fn(&ctx, in, remaining); |
Mike Hommey | 4277c67 | 2009-01-18 09:04:26 +0100 | [diff] [blame] | 80 | } while (offset < pack_sig_ofs); |
brian m. carlson | ccc12e0 | 2018-02-01 02:18:43 +0000 | [diff] [blame] | 81 | the_hash_algo->final_fn(hash, &ctx); |
Nicolas Pitre | 6241360 | 2008-05-29 17:34:50 -0400 | [diff] [blame] | 82 | pack_sig = use_pack(p, w_curs, pack_sig_ofs, NULL); |
Jeff King | 67947c3 | 2018-08-28 17:22:52 -0400 | [diff] [blame] | 83 | if (!hasheq(hash, pack_sig)) |
brian m. carlson | ccc12e0 | 2018-02-01 02:18:43 +0000 | [diff] [blame] | 84 | err = error("%s pack checksum mismatch", |
Nicolas Pitre | 6241360 | 2008-05-29 17:34:50 -0400 | [diff] [blame] | 85 | p->pack_name); |
Jeff King | 67947c3 | 2018-08-28 17:22:52 -0400 | [diff] [blame] | 86 | if (!hasheq(index_base + index_size - the_hash_algo->hexsz, pack_sig)) |
brian m. carlson | ccc12e0 | 2018-02-01 02:18:43 +0000 | [diff] [blame] | 87 | err = error("%s pack checksum does not match its index", |
Nicolas Pitre | 6241360 | 2008-05-29 17:34:50 -0400 | [diff] [blame] | 88 | p->pack_name); |
Shawn O. Pearce | 079afb1 | 2006-12-23 02:34:13 -0500 | [diff] [blame] | 89 | unuse_pack(w_curs); |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 90 | |
| 91 | /* Make sure everything reachable from idx is valid. Since we |
| 92 | * have verified that nr_objects matches between idx and pack, |
| 93 | * we do not do scan-streaming check on the pack file. |
| 94 | */ |
Nicolas Pitre | 5705909 | 2007-04-09 01:06:28 -0400 | [diff] [blame] | 95 | nr_objects = p->num_objects; |
Jeff King | b32fa95 | 2016-02-22 17:44:25 -0500 | [diff] [blame] | 96 | ALLOC_ARRAY(entries, nr_objects + 1); |
Nicolas Pitre | c41a4a9 | 2008-06-24 23:19:02 -0400 | [diff] [blame] | 97 | entries[nr_objects].offset = pack_sig_ofs; |
Alexandre Julliard | 3af5192 | 2007-06-03 20:21:41 +0200 | [diff] [blame] | 98 | /* first sort entries by pack offset, since unpacking them is more efficient that way */ |
| 99 | for (i = 0; i < nr_objects; i++) { |
brian m. carlson | 9fd7504 | 2017-05-06 22:10:20 +0000 | [diff] [blame] | 100 | entries[i].oid.hash = nth_packed_object_sha1(p, i); |
| 101 | if (!entries[i].oid.hash) |
Alexandre Julliard | 3af5192 | 2007-06-03 20:21:41 +0200 | [diff] [blame] | 102 | die("internal error pack-check nth-packed-object"); |
Nicolas Pitre | 9909323 | 2008-06-24 23:17:12 -0400 | [diff] [blame] | 103 | entries[i].offset = nth_packed_object_offset(p, i); |
Nicolas Pitre | c41a4a9 | 2008-06-24 23:19:02 -0400 | [diff] [blame] | 104 | entries[i].nr = i; |
Alexandre Julliard | 3af5192 | 2007-06-03 20:21:41 +0200 | [diff] [blame] | 105 | } |
René Scharfe | 9ed0d8d | 2016-09-29 17:27:31 +0200 | [diff] [blame] | 106 | QSORT(entries, nr_objects, compare_entries); |
Alexandre Julliard | 3af5192 | 2007-06-03 20:21:41 +0200 | [diff] [blame] | 107 | |
Nicolas Pitre | 6241360 | 2008-05-29 17:34:50 -0400 | [diff] [blame] | 108 | for (i = 0; i < nr_objects; i++) { |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 109 | void *data; |
Nicolas Pitre | 21666f1 | 2007-02-26 14:55:59 -0500 | [diff] [blame] | 110 | enum object_type type; |
Shawn O. Pearce | c4001d9 | 2007-03-06 20:44:30 -0500 | [diff] [blame] | 111 | unsigned long size; |
Nguyễn Thái Ngọc Duy | ec9d224 | 2016-07-13 17:44:04 +0200 | [diff] [blame] | 112 | off_t curpos; |
| 113 | int data_valid; |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 114 | |
Nicolas Pitre | c41a4a9 | 2008-06-24 23:19:02 -0400 | [diff] [blame] | 115 | if (p->index_version > 1) { |
| 116 | off_t offset = entries[i].offset; |
| 117 | off_t len = entries[i+1].offset - offset; |
| 118 | unsigned int nr = entries[i].nr; |
| 119 | if (check_pack_crc(p, w_curs, offset, len, nr)) |
| 120 | err = error("index CRC mismatch for object %s " |
| 121 | "from %s at offset %"PRIuMAX"", |
brian m. carlson | 9fd7504 | 2017-05-06 22:10:20 +0000 | [diff] [blame] | 122 | oid_to_hex(entries[i].oid.oid), |
Nicolas Pitre | c41a4a9 | 2008-06-24 23:19:02 -0400 | [diff] [blame] | 123 | p->pack_name, (uintmax_t)offset); |
| 124 | } |
Nguyễn Thái Ngọc Duy | ec9d224 | 2016-07-13 17:44:04 +0200 | [diff] [blame] | 125 | |
| 126 | curpos = entries[i].offset; |
| 127 | type = unpack_object_header(p, w_curs, &curpos, &size); |
| 128 | unuse_pack(w_curs); |
| 129 | |
| 130 | if (type == OBJ_BLOB && big_file_threshold <= size) { |
| 131 | /* |
brian m. carlson | 17e6545 | 2018-03-12 02:27:39 +0000 | [diff] [blame] | 132 | * Let check_object_signature() check it with |
Nguyễn Thái Ngọc Duy | ec9d224 | 2016-07-13 17:44:04 +0200 | [diff] [blame] | 133 | * the streaming interface; no point slurping |
| 134 | * the data in-core only to discard. |
| 135 | */ |
| 136 | data = NULL; |
| 137 | data_valid = 0; |
| 138 | } else { |
Nguyễn Thái Ngọc Duy | 94e1082 | 2018-11-10 06:49:07 +0100 | [diff] [blame] | 139 | data = unpack_entry(r, p, entries[i].offset, &type, &size); |
Nguyễn Thái Ngọc Duy | ec9d224 | 2016-07-13 17:44:04 +0200 | [diff] [blame] | 140 | data_valid = 1; |
| 141 | } |
| 142 | |
| 143 | if (data_valid && !data) |
Nicolas Pitre | 6241360 | 2008-05-29 17:34:50 -0400 | [diff] [blame] | 144 | err = error("cannot unpack %s from %s at offset %"PRIuMAX"", |
brian m. carlson | 9fd7504 | 2017-05-06 22:10:20 +0000 | [diff] [blame] | 145 | oid_to_hex(entries[i].oid.oid), p->pack_name, |
Nicolas Pitre | 6241360 | 2008-05-29 17:34:50 -0400 | [diff] [blame] | 146 | (uintmax_t)entries[i].offset); |
brian m. carlson | 17e6545 | 2018-03-12 02:27:39 +0000 | [diff] [blame] | 147 | else if (check_object_signature(entries[i].oid.oid, data, size, type_name(type))) |
Junio C Hamano | 1038f0c | 2005-07-07 15:12:20 -0700 | [diff] [blame] | 148 | err = error("packed %s from %s is corrupt", |
brian m. carlson | 9fd7504 | 2017-05-06 22:10:20 +0000 | [diff] [blame] | 149 | oid_to_hex(entries[i].oid.oid), p->pack_name); |
Nguyễn Thái Ngọc Duy | c9486eb | 2011-11-07 09:59:25 +0700 | [diff] [blame] | 150 | else if (fn) { |
| 151 | int eaten = 0; |
brian m. carlson | 9fd7504 | 2017-05-06 22:10:20 +0000 | [diff] [blame] | 152 | err |= fn(entries[i].oid.oid, type, size, data, &eaten); |
Nguyễn Thái Ngọc Duy | c9486eb | 2011-11-07 09:59:25 +0700 | [diff] [blame] | 153 | if (eaten) |
| 154 | data = NULL; |
| 155 | } |
Nguyễn Thái Ngọc Duy | 1e49f22 | 2011-11-07 09:59:26 +0700 | [diff] [blame] | 156 | if (((base_count + i) & 1023) == 0) |
| 157 | display_progress(progress, base_count + i); |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 158 | free(data); |
Nguyễn Thái Ngọc Duy | 1e49f22 | 2011-11-07 09:59:26 +0700 | [diff] [blame] | 159 | |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 160 | } |
Nguyễn Thái Ngọc Duy | 1e49f22 | 2011-11-07 09:59:26 +0700 | [diff] [blame] | 161 | display_progress(progress, base_count + i); |
Alexandre Julliard | 3af5192 | 2007-06-03 20:21:41 +0200 | [diff] [blame] | 162 | free(entries); |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 163 | |
| 164 | return err; |
| 165 | } |
| 166 | |
Shawn O. Pearce | 9b0aa72 | 2010-04-19 07:23:07 -0700 | [diff] [blame] | 167 | int verify_pack_index(struct packed_git *p) |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 168 | { |
Shawn O. Pearce | d079837 | 2007-05-26 01:24:19 -0400 | [diff] [blame] | 169 | off_t index_size; |
| 170 | const unsigned char *index_base; |
brian m. carlson | ccc12e0 | 2018-02-01 02:18:43 +0000 | [diff] [blame] | 171 | git_hash_ctx ctx; |
| 172 | unsigned char hash[GIT_MAX_RAWSZ]; |
Nicolas Pitre | 6241360 | 2008-05-29 17:34:50 -0400 | [diff] [blame] | 173 | int err = 0; |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 174 | |
Shawn O. Pearce | d079837 | 2007-05-26 01:24:19 -0400 | [diff] [blame] | 175 | if (open_pack_index(p)) |
| 176 | return error("packfile %s index not opened", p->pack_name); |
| 177 | index_size = p->index_size; |
| 178 | index_base = p->index_data; |
| 179 | |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 180 | /* Verify SHA1 sum of the index file */ |
brian m. carlson | ccc12e0 | 2018-02-01 02:18:43 +0000 | [diff] [blame] | 181 | the_hash_algo->init_fn(&ctx); |
| 182 | the_hash_algo->update_fn(&ctx, index_base, (unsigned int)(index_size - the_hash_algo->rawsz)); |
| 183 | the_hash_algo->final_fn(hash, &ctx); |
Jeff King | 67947c3 | 2018-08-28 17:22:52 -0400 | [diff] [blame] | 184 | if (!hasheq(hash, index_base + index_size - the_hash_algo->rawsz)) |
brian m. carlson | ccc12e0 | 2018-02-01 02:18:43 +0000 | [diff] [blame] | 185 | err = error("Packfile index for %s hash mismatch", |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 186 | p->pack_name); |
Shawn O. Pearce | 9b0aa72 | 2010-04-19 07:23:07 -0700 | [diff] [blame] | 187 | return err; |
| 188 | } |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 189 | |
Nguyễn Thái Ngọc Duy | 94e1082 | 2018-11-10 06:49:07 +0100 | [diff] [blame] | 190 | int verify_pack(struct repository *r, struct packed_git *p, verify_fn fn, |
Nguyễn Thái Ngọc Duy | 1e49f22 | 2011-11-07 09:59:26 +0700 | [diff] [blame] | 191 | struct progress *progress, uint32_t base_count) |
Shawn O. Pearce | 9b0aa72 | 2010-04-19 07:23:07 -0700 | [diff] [blame] | 192 | { |
| 193 | int err = 0; |
| 194 | struct pack_window *w_curs = NULL; |
| 195 | |
| 196 | err |= verify_pack_index(p); |
| 197 | if (!p->index_data) |
| 198 | return -1; |
| 199 | |
Nguyễn Thái Ngọc Duy | 94e1082 | 2018-11-10 06:49:07 +0100 | [diff] [blame] | 200 | err |= verify_packfile(r, p, &w_curs, fn, progress, base_count); |
Nicolas Pitre | 6241360 | 2008-05-29 17:34:50 -0400 | [diff] [blame] | 201 | unuse_pack(&w_curs); |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 202 | |
Nicolas Pitre | 6241360 | 2008-05-29 17:34:50 -0400 | [diff] [blame] | 203 | return err; |
Junio C Hamano | f3bf922 | 2005-06-30 17:15:39 -0700 | [diff] [blame] | 204 | } |