.mailmap: update email address of Andrey Mazo
[git/git.git] / pack-check.c
CommitLineData
f9253394 1#include "cache.h"
57a6a500 2#include "repository.h"
f9253394 3#include "pack.h"
70f5d5d3 4#include "pack-revindex.h"
1e49f22f 5#include "progress.h"
0317f455 6#include "packfile.h"
a80d72db 7#include "object-store.h"
f9253394 8
9cba13ca 9struct idx_entry {
3af51928 10 off_t offset;
9fd75046 11 union idx_entry_object {
12 const unsigned char *hash;
13 struct object_id *oid;
14 } oid;
c41a4a94 15 unsigned int nr;
3af51928
AJ
16};
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
c41a4a94
NP
29int 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;
1e4cd68c 33 uint32_t data_crc = crc32(0, NULL, 0);
c41a4a94
NP
34
35 do {
ef49a7a0 36 unsigned long avail;
c41a4a94
NP
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;
ccc12e06 46 index_crc += 2 + 256 + p->num_objects * (the_hash_algo->rawsz/4) + nr;
c41a4a94
NP
47
48 return data_crc != ntohl(*index_crc);
49}
50
94e10825
NTND
51static int verify_packfile(struct repository *r,
52 struct packed_git *p,
c9486eb0 53 struct pack_window **w_curs,
1e49f22f
NTND
54 verify_fn fn,
55 struct progress *progress, uint32_t base_count)
56
f9253394 57{
c4001d92 58 off_t index_size = p->index_size;
42873078 59 const unsigned char *index_base = p->index_data;
ccc12e06 60 git_hash_ctx ctx;
9fd75046 61 unsigned char hash[GIT_MAX_RAWSZ], *pack_sig;
4277c670 62 off_t offset = 0, pack_sig_ofs = 0;
326bf396 63 uint32_t nr_objects, i;
62413604 64 int err = 0;
3af51928 65 struct idx_entry *entries;
f9253394 66
a9445d85
JK
67 if (!is_pack_valid(p))
68 return error("packfile %s cannot be accessed", p->pack_name);
f9253394 69
ccc12e06 70 the_hash_algo->init_fn(&ctx);
4277c670 71 do {
ef49a7a0 72 unsigned long remaining;
079afb18
SP
73 unsigned char *in = use_pack(p, w_curs, offset, &remaining);
74 offset += remaining;
4277c670 75 if (!pack_sig_ofs)
ccc12e06 76 pack_sig_ofs = p->pack_size - the_hash_algo->rawsz;
62413604
NP
77 if (offset > pack_sig_ofs)
78 remaining -= (unsigned int)(offset - pack_sig_ofs);
ccc12e06 79 the_hash_algo->update_fn(&ctx, in, remaining);
4277c670 80 } while (offset < pack_sig_ofs);
ccc12e06 81 the_hash_algo->final_fn(hash, &ctx);
62413604 82 pack_sig = use_pack(p, w_curs, pack_sig_ofs, NULL);
67947c34 83 if (!hasheq(hash, pack_sig))
ccc12e06 84 err = error("%s pack checksum mismatch",
62413604 85 p->pack_name);
67947c34 86 if (!hasheq(index_base + index_size - the_hash_algo->hexsz, pack_sig))
ccc12e06 87 err = error("%s pack checksum does not match its index",
62413604 88 p->pack_name);
079afb18 89 unuse_pack(w_curs);
f3bf9224
JH
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 */
57059091 95 nr_objects = p->num_objects;
b32fa95f 96 ALLOC_ARRAY(entries, nr_objects + 1);
c41a4a94 97 entries[nr_objects].offset = pack_sig_ofs;
3af51928
AJ
98 /* first sort entries by pack offset, since unpacking them is more efficient that way */
99 for (i = 0; i < nr_objects; i++) {
9fd75046 100 entries[i].oid.hash = nth_packed_object_sha1(p, i);
101 if (!entries[i].oid.hash)
3af51928 102 die("internal error pack-check nth-packed-object");
99093238 103 entries[i].offset = nth_packed_object_offset(p, i);
c41a4a94 104 entries[i].nr = i;
3af51928 105 }
9ed0d8d6 106 QSORT(entries, nr_objects, compare_entries);
3af51928 107
62413604 108 for (i = 0; i < nr_objects; i++) {
f3bf9224 109 void *data;
21666f1a 110 enum object_type type;
c4001d92 111 unsigned long size;
ec9d2249
NTND
112 off_t curpos;
113 int data_valid;
f3bf9224 114
c41a4a94
NP
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"",
9fd75046 122 oid_to_hex(entries[i].oid.oid),
c41a4a94
NP
123 p->pack_name, (uintmax_t)offset);
124 }
ec9d2249
NTND
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 /*
17e65451 132 * Let check_object_signature() check it with
ec9d2249
NTND
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 {
94e10825 139 data = unpack_entry(r, p, entries[i].offset, &type, &size);
ec9d2249
NTND
140 data_valid = 1;
141 }
142
143 if (data_valid && !data)
62413604 144 err = error("cannot unpack %s from %s at offset %"PRIuMAX"",
9fd75046 145 oid_to_hex(entries[i].oid.oid), p->pack_name,
62413604 146 (uintmax_t)entries[i].offset);
17e65451 147 else if (check_object_signature(entries[i].oid.oid, data, size, type_name(type)))
1038f0c0 148 err = error("packed %s from %s is corrupt",
9fd75046 149 oid_to_hex(entries[i].oid.oid), p->pack_name);
c9486eb0
NTND
150 else if (fn) {
151 int eaten = 0;
9fd75046 152 err |= fn(entries[i].oid.oid, type, size, data, &eaten);
c9486eb0
NTND
153 if (eaten)
154 data = NULL;
155 }
1e49f22f
NTND
156 if (((base_count + i) & 1023) == 0)
157 display_progress(progress, base_count + i);
f3bf9224 158 free(data);
1e49f22f 159
f3bf9224 160 }
1e49f22f 161 display_progress(progress, base_count + i);
3af51928 162 free(entries);
f3bf9224
JH
163
164 return err;
f9253394
JH
165}
166
9b0aa728 167int verify_pack_index(struct packed_git *p)
f9253394 168{
d079837e
SP
169 off_t index_size;
170 const unsigned char *index_base;
ccc12e06 171 git_hash_ctx ctx;
172 unsigned char hash[GIT_MAX_RAWSZ];
62413604 173 int err = 0;
f9253394 174
d079837e
SP
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
f9253394 180 /* Verify SHA1 sum of the index file */
ccc12e06 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);
67947c34 184 if (!hasheq(hash, index_base + index_size - the_hash_algo->rawsz))
ccc12e06 185 err = error("Packfile index for %s hash mismatch",
f3bf9224 186 p->pack_name);
9b0aa728
SP
187 return err;
188}
189
94e10825 190int verify_pack(struct repository *r, struct packed_git *p, verify_fn fn,
1e49f22f 191 struct progress *progress, uint32_t base_count)
9b0aa728
SP
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;
f3bf9224 199
94e10825 200 err |= verify_packfile(r, p, &w_curs, fn, progress, base_count);
62413604 201 unuse_pack(&w_curs);
f3bf9224 202
62413604 203 return err;
f9253394 204}