blob: 2cc3603189b8cd27781fd16cba2eb10a16817038 [file] [log] [blame]
Junio C Hamanof3bf9222005-06-30 17:15:39 -07001#include "cache.h"
Stefan Beller57a6a502018-04-25 11:21:04 -07002#include "repository.h"
Junio C Hamanof3bf9222005-06-30 17:15:39 -07003#include "pack.h"
Nicolas Pitre70f5d5d2008-02-28 00:25:19 -05004#include "pack-revindex.h"
Nguyễn Thái Ngọc Duy1e49f222011-11-07 09:59:26 +07005#include "progress.h"
Jonathan Tan0317f452017-08-18 15:20:19 -07006#include "packfile.h"
Stefan Bellera80d72d2018-03-23 18:20:59 +01007#include "object-store.h"
Junio C Hamanof3bf9222005-06-30 17:15:39 -07008
Jonathan Nieder9cba13c2011-03-16 02:08:34 -05009struct idx_entry {
Alexandre Julliard3af51922007-06-03 20:21:41 +020010 off_t offset;
brian m. carlson9fd75042017-05-06 22:10:20 +000011 union idx_entry_object {
12 const unsigned char *hash;
13 struct object_id *oid;
14 } oid;
Nicolas Pitrec41a4a92008-06-24 23:19:02 -040015 unsigned int nr;
Alexandre Julliard3af51922007-06-03 20:21:41 +020016};
17
18static 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 Pitrec41a4a92008-06-24 23:19:02 -040029int 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 Boyd1e4cd682011-04-03 00:06:54 -070033 uint32_t data_crc = crc32(0, NULL, 0);
Nicolas Pitrec41a4a92008-06-24 23:19:02 -040034
35 do {
Junio C Hamanoef49a7a2011-06-10 11:52:15 -070036 unsigned long avail;
Nicolas Pitrec41a4a92008-06-24 23:19:02 -040037 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. carlsonccc12e02018-02-01 02:18:43 +000046 index_crc += 2 + 256 + p->num_objects * (the_hash_algo->rawsz/4) + nr;
Nicolas Pitrec41a4a92008-06-24 23:19:02 -040047
48 return data_crc != ntohl(*index_crc);
49}
50
Nguyễn Thái Ngọc Duy94e10822018-11-10 06:49:07 +010051static int verify_packfile(struct repository *r,
52 struct packed_git *p,
Nguyễn Thái Ngọc Duyc9486eb2011-11-07 09:59:25 +070053 struct pack_window **w_curs,
Nguyễn Thái Ngọc Duy1e49f222011-11-07 09:59:26 +070054 verify_fn fn,
55 struct progress *progress, uint32_t base_count)
56
Junio C Hamanof3bf9222005-06-30 17:15:39 -070057{
Shawn O. Pearcec4001d92007-03-06 20:44:30 -050058 off_t index_size = p->index_size;
Nicolas Pitre42873072007-03-16 16:42:50 -040059 const unsigned char *index_base = p->index_data;
brian m. carlsonccc12e02018-02-01 02:18:43 +000060 git_hash_ctx ctx;
brian m. carlson9fd75042017-05-06 22:10:20 +000061 unsigned char hash[GIT_MAX_RAWSZ], *pack_sig;
Mike Hommey4277c672009-01-18 09:04:26 +010062 off_t offset = 0, pack_sig_ofs = 0;
Shawn O. Pearce326bf392007-03-06 20:44:19 -050063 uint32_t nr_objects, i;
Nicolas Pitre62413602008-05-29 17:34:50 -040064 int err = 0;
Alexandre Julliard3af51922007-06-03 20:21:41 +020065 struct idx_entry *entries;
Junio C Hamanof3bf9222005-06-30 17:15:39 -070066
Jeff Kinga9445d82016-09-21 23:49:05 -040067 if (!is_pack_valid(p))
68 return error("packfile %s cannot be accessed", p->pack_name);
Junio C Hamanof3bf9222005-06-30 17:15:39 -070069
brian m. carlsonccc12e02018-02-01 02:18:43 +000070 the_hash_algo->init_fn(&ctx);
Mike Hommey4277c672009-01-18 09:04:26 +010071 do {
Junio C Hamanoef49a7a2011-06-10 11:52:15 -070072 unsigned long remaining;
Shawn O. Pearce079afb12006-12-23 02:34:13 -050073 unsigned char *in = use_pack(p, w_curs, offset, &remaining);
74 offset += remaining;
Mike Hommey4277c672009-01-18 09:04:26 +010075 if (!pack_sig_ofs)
brian m. carlsonccc12e02018-02-01 02:18:43 +000076 pack_sig_ofs = p->pack_size - the_hash_algo->rawsz;
Nicolas Pitre62413602008-05-29 17:34:50 -040077 if (offset > pack_sig_ofs)
78 remaining -= (unsigned int)(offset - pack_sig_ofs);
brian m. carlsonccc12e02018-02-01 02:18:43 +000079 the_hash_algo->update_fn(&ctx, in, remaining);
Mike Hommey4277c672009-01-18 09:04:26 +010080 } while (offset < pack_sig_ofs);
brian m. carlsonccc12e02018-02-01 02:18:43 +000081 the_hash_algo->final_fn(hash, &ctx);
Nicolas Pitre62413602008-05-29 17:34:50 -040082 pack_sig = use_pack(p, w_curs, pack_sig_ofs, NULL);
Jeff King67947c32018-08-28 17:22:52 -040083 if (!hasheq(hash, pack_sig))
brian m. carlsonccc12e02018-02-01 02:18:43 +000084 err = error("%s pack checksum mismatch",
Nicolas Pitre62413602008-05-29 17:34:50 -040085 p->pack_name);
Jeff King67947c32018-08-28 17:22:52 -040086 if (!hasheq(index_base + index_size - the_hash_algo->hexsz, pack_sig))
brian m. carlsonccc12e02018-02-01 02:18:43 +000087 err = error("%s pack checksum does not match its index",
Nicolas Pitre62413602008-05-29 17:34:50 -040088 p->pack_name);
Shawn O. Pearce079afb12006-12-23 02:34:13 -050089 unuse_pack(w_curs);
Junio C Hamanof3bf9222005-06-30 17:15:39 -070090
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 Pitre57059092007-04-09 01:06:28 -040095 nr_objects = p->num_objects;
Jeff Kingb32fa952016-02-22 17:44:25 -050096 ALLOC_ARRAY(entries, nr_objects + 1);
Nicolas Pitrec41a4a92008-06-24 23:19:02 -040097 entries[nr_objects].offset = pack_sig_ofs;
Alexandre Julliard3af51922007-06-03 20:21:41 +020098 /* first sort entries by pack offset, since unpacking them is more efficient that way */
99 for (i = 0; i < nr_objects; i++) {
brian m. carlson9fd75042017-05-06 22:10:20 +0000100 entries[i].oid.hash = nth_packed_object_sha1(p, i);
101 if (!entries[i].oid.hash)
Alexandre Julliard3af51922007-06-03 20:21:41 +0200102 die("internal error pack-check nth-packed-object");
Nicolas Pitre99093232008-06-24 23:17:12 -0400103 entries[i].offset = nth_packed_object_offset(p, i);
Nicolas Pitrec41a4a92008-06-24 23:19:02 -0400104 entries[i].nr = i;
Alexandre Julliard3af51922007-06-03 20:21:41 +0200105 }
René Scharfe9ed0d8d2016-09-29 17:27:31 +0200106 QSORT(entries, nr_objects, compare_entries);
Alexandre Julliard3af51922007-06-03 20:21:41 +0200107
Nicolas Pitre62413602008-05-29 17:34:50 -0400108 for (i = 0; i < nr_objects; i++) {
Junio C Hamanof3bf9222005-06-30 17:15:39 -0700109 void *data;
Nicolas Pitre21666f12007-02-26 14:55:59 -0500110 enum object_type type;
Shawn O. Pearcec4001d92007-03-06 20:44:30 -0500111 unsigned long size;
Nguyễn Thái Ngọc Duyec9d2242016-07-13 17:44:04 +0200112 off_t curpos;
113 int data_valid;
Junio C Hamanof3bf9222005-06-30 17:15:39 -0700114
Nicolas Pitrec41a4a92008-06-24 23:19:02 -0400115 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. carlson9fd75042017-05-06 22:10:20 +0000122 oid_to_hex(entries[i].oid.oid),
Nicolas Pitrec41a4a92008-06-24 23:19:02 -0400123 p->pack_name, (uintmax_t)offset);
124 }
Nguyễn Thái Ngọc Duyec9d2242016-07-13 17:44:04 +0200125
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. carlson17e65452018-03-12 02:27:39 +0000132 * Let check_object_signature() check it with
Nguyễn Thái Ngọc Duyec9d2242016-07-13 17:44:04 +0200133 * 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 Duy94e10822018-11-10 06:49:07 +0100139 data = unpack_entry(r, p, entries[i].offset, &type, &size);
Nguyễn Thái Ngọc Duyec9d2242016-07-13 17:44:04 +0200140 data_valid = 1;
141 }
142
143 if (data_valid && !data)
Nicolas Pitre62413602008-05-29 17:34:50 -0400144 err = error("cannot unpack %s from %s at offset %"PRIuMAX"",
brian m. carlson9fd75042017-05-06 22:10:20 +0000145 oid_to_hex(entries[i].oid.oid), p->pack_name,
Nicolas Pitre62413602008-05-29 17:34:50 -0400146 (uintmax_t)entries[i].offset);
brian m. carlson17e65452018-03-12 02:27:39 +0000147 else if (check_object_signature(entries[i].oid.oid, data, size, type_name(type)))
Junio C Hamano1038f0c2005-07-07 15:12:20 -0700148 err = error("packed %s from %s is corrupt",
brian m. carlson9fd75042017-05-06 22:10:20 +0000149 oid_to_hex(entries[i].oid.oid), p->pack_name);
Nguyễn Thái Ngọc Duyc9486eb2011-11-07 09:59:25 +0700150 else if (fn) {
151 int eaten = 0;
brian m. carlson9fd75042017-05-06 22:10:20 +0000152 err |= fn(entries[i].oid.oid, type, size, data, &eaten);
Nguyễn Thái Ngọc Duyc9486eb2011-11-07 09:59:25 +0700153 if (eaten)
154 data = NULL;
155 }
Nguyễn Thái Ngọc Duy1e49f222011-11-07 09:59:26 +0700156 if (((base_count + i) & 1023) == 0)
157 display_progress(progress, base_count + i);
Junio C Hamanof3bf9222005-06-30 17:15:39 -0700158 free(data);
Nguyễn Thái Ngọc Duy1e49f222011-11-07 09:59:26 +0700159
Junio C Hamanof3bf9222005-06-30 17:15:39 -0700160 }
Nguyễn Thái Ngọc Duy1e49f222011-11-07 09:59:26 +0700161 display_progress(progress, base_count + i);
Alexandre Julliard3af51922007-06-03 20:21:41 +0200162 free(entries);
Junio C Hamanof3bf9222005-06-30 17:15:39 -0700163
164 return err;
165}
166
Shawn O. Pearce9b0aa722010-04-19 07:23:07 -0700167int verify_pack_index(struct packed_git *p)
Junio C Hamanof3bf9222005-06-30 17:15:39 -0700168{
Shawn O. Pearced0798372007-05-26 01:24:19 -0400169 off_t index_size;
170 const unsigned char *index_base;
brian m. carlsonccc12e02018-02-01 02:18:43 +0000171 git_hash_ctx ctx;
172 unsigned char hash[GIT_MAX_RAWSZ];
Nicolas Pitre62413602008-05-29 17:34:50 -0400173 int err = 0;
Junio C Hamanof3bf9222005-06-30 17:15:39 -0700174
Shawn O. Pearced0798372007-05-26 01:24:19 -0400175 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 Hamanof3bf9222005-06-30 17:15:39 -0700180 /* Verify SHA1 sum of the index file */
brian m. carlsonccc12e02018-02-01 02:18:43 +0000181 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 King67947c32018-08-28 17:22:52 -0400184 if (!hasheq(hash, index_base + index_size - the_hash_algo->rawsz))
brian m. carlsonccc12e02018-02-01 02:18:43 +0000185 err = error("Packfile index for %s hash mismatch",
Junio C Hamanof3bf9222005-06-30 17:15:39 -0700186 p->pack_name);
Shawn O. Pearce9b0aa722010-04-19 07:23:07 -0700187 return err;
188}
Junio C Hamanof3bf9222005-06-30 17:15:39 -0700189
Nguyễn Thái Ngọc Duy94e10822018-11-10 06:49:07 +0100190int verify_pack(struct repository *r, struct packed_git *p, verify_fn fn,
Nguyễn Thái Ngọc Duy1e49f222011-11-07 09:59:26 +0700191 struct progress *progress, uint32_t base_count)
Shawn O. Pearce9b0aa722010-04-19 07:23:07 -0700192{
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 Duy94e10822018-11-10 06:49:07 +0100200 err |= verify_packfile(r, p, &w_curs, fn, progress, base_count);
Nicolas Pitre62413602008-05-29 17:34:50 -0400201 unuse_pack(&w_curs);
Junio C Hamanof3bf9222005-06-30 17:15:39 -0700202
Nicolas Pitre62413602008-05-29 17:34:50 -0400203 return err;
Junio C Hamanof3bf9222005-06-30 17:15:39 -0700204}