object-store: move packed_git and packed_git_mru to object store
[git/git.git] / pack-check.c
CommitLineData
f9253394
JH
1#include "cache.h"
2#include "pack.h"
70f5d5d3 3#include "pack-revindex.h"
1e49f22f 4#include "progress.h"
0317f455 5#include "packfile.h"
a80d72db 6#include "object-store.h"
f9253394 7
9cba13ca 8struct idx_entry {
3af51928 9 off_t offset;
9fd75046 10 union idx_entry_object {
11 const unsigned char *hash;
12 struct object_id *oid;
13 } oid;
c41a4a94 14 unsigned int nr;
3af51928
AJ
15};
16
17static int compare_entries(const void *e1, const void *e2)
18{
19 const struct idx_entry *entry1 = e1;
20 const struct idx_entry *entry2 = e2;
21 if (entry1->offset < entry2->offset)
22 return -1;
23 if (entry1->offset > entry2->offset)
24 return 1;
25 return 0;
26}
27
c41a4a94
NP
28int check_pack_crc(struct packed_git *p, struct pack_window **w_curs,
29 off_t offset, off_t len, unsigned int nr)
30{
31 const uint32_t *index_crc;
1e4cd68c 32 uint32_t data_crc = crc32(0, NULL, 0);
c41a4a94
NP
33
34 do {
ef49a7a0 35 unsigned long avail;
c41a4a94
NP
36 void *data = use_pack(p, w_curs, offset, &avail);
37 if (avail > len)
38 avail = len;
39 data_crc = crc32(data_crc, data, avail);
40 offset += avail;
41 len -= avail;
42 } while (len);
43
44 index_crc = p->index_data;
45 index_crc += 2 + 256 + p->num_objects * (20/4) + nr;
46
47 return data_crc != ntohl(*index_crc);
48}
49
03e79c88 50static int verify_packfile(struct packed_git *p,
c9486eb0 51 struct pack_window **w_curs,
1e49f22f
NTND
52 verify_fn fn,
53 struct progress *progress, uint32_t base_count)
54
f9253394 55{
c4001d92 56 off_t index_size = p->index_size;
42873078 57 const unsigned char *index_base = p->index_data;
9126f009 58 git_SHA_CTX ctx;
9fd75046 59 unsigned char hash[GIT_MAX_RAWSZ], *pack_sig;
4277c670 60 off_t offset = 0, pack_sig_ofs = 0;
326bf396 61 uint32_t nr_objects, i;
62413604 62 int err = 0;
3af51928 63 struct idx_entry *entries;
f9253394 64
a9445d85
JK
65 if (!is_pack_valid(p))
66 return error("packfile %s cannot be accessed", p->pack_name);
f9253394 67
9126f009 68 git_SHA1_Init(&ctx);
4277c670 69 do {
ef49a7a0 70 unsigned long remaining;
079afb18
SP
71 unsigned char *in = use_pack(p, w_curs, offset, &remaining);
72 offset += remaining;
4277c670
MH
73 if (!pack_sig_ofs)
74 pack_sig_ofs = p->pack_size - 20;
62413604
NP
75 if (offset > pack_sig_ofs)
76 remaining -= (unsigned int)(offset - pack_sig_ofs);
9126f009 77 git_SHA1_Update(&ctx, in, remaining);
4277c670 78 } while (offset < pack_sig_ofs);
9fd75046 79 git_SHA1_Final(hash, &ctx);
62413604 80 pack_sig = use_pack(p, w_curs, pack_sig_ofs, NULL);
9fd75046 81 if (hashcmp(hash, pack_sig))
62413604
NP
82 err = error("%s SHA1 checksum mismatch",
83 p->pack_name);
84 if (hashcmp(index_base + index_size - 40, pack_sig))
22e5e58a 85 err = error("%s SHA1 does not match its index",
62413604 86 p->pack_name);
079afb18 87 unuse_pack(w_curs);
f3bf9224
JH
88
89 /* Make sure everything reachable from idx is valid. Since we
90 * have verified that nr_objects matches between idx and pack,
91 * we do not do scan-streaming check on the pack file.
92 */
57059091 93 nr_objects = p->num_objects;
b32fa95f 94 ALLOC_ARRAY(entries, nr_objects + 1);
c41a4a94 95 entries[nr_objects].offset = pack_sig_ofs;
3af51928
AJ
96 /* first sort entries by pack offset, since unpacking them is more efficient that way */
97 for (i = 0; i < nr_objects; i++) {
9fd75046 98 entries[i].oid.hash = nth_packed_object_sha1(p, i);
99 if (!entries[i].oid.hash)
3af51928 100 die("internal error pack-check nth-packed-object");
99093238 101 entries[i].offset = nth_packed_object_offset(p, i);
c41a4a94 102 entries[i].nr = i;
3af51928 103 }
9ed0d8d6 104 QSORT(entries, nr_objects, compare_entries);
3af51928 105
62413604 106 for (i = 0; i < nr_objects; i++) {
f3bf9224 107 void *data;
21666f1a 108 enum object_type type;
c4001d92 109 unsigned long size;
ec9d2249
NTND
110 off_t curpos;
111 int data_valid;
f3bf9224 112
c41a4a94
NP
113 if (p->index_version > 1) {
114 off_t offset = entries[i].offset;
115 off_t len = entries[i+1].offset - offset;
116 unsigned int nr = entries[i].nr;
117 if (check_pack_crc(p, w_curs, offset, len, nr))
118 err = error("index CRC mismatch for object %s "
119 "from %s at offset %"PRIuMAX"",
9fd75046 120 oid_to_hex(entries[i].oid.oid),
c41a4a94
NP
121 p->pack_name, (uintmax_t)offset);
122 }
ec9d2249
NTND
123
124 curpos = entries[i].offset;
125 type = unpack_object_header(p, w_curs, &curpos, &size);
126 unuse_pack(w_curs);
127
128 if (type == OBJ_BLOB && big_file_threshold <= size) {
129 /*
130 * Let check_sha1_signature() check it with
131 * the streaming interface; no point slurping
132 * the data in-core only to discard.
133 */
134 data = NULL;
135 data_valid = 0;
136 } else {
137 data = unpack_entry(p, entries[i].offset, &type, &size);
138 data_valid = 1;
139 }
140
141 if (data_valid && !data)
62413604 142 err = error("cannot unpack %s from %s at offset %"PRIuMAX"",
9fd75046 143 oid_to_hex(entries[i].oid.oid), p->pack_name,
62413604 144 (uintmax_t)entries[i].offset);
9fd75046 145 else if (check_sha1_signature(entries[i].oid.hash, data, size, typename(type)))
1038f0c0 146 err = error("packed %s from %s is corrupt",
9fd75046 147 oid_to_hex(entries[i].oid.oid), p->pack_name);
c9486eb0
NTND
148 else if (fn) {
149 int eaten = 0;
9fd75046 150 err |= fn(entries[i].oid.oid, type, size, data, &eaten);
c9486eb0
NTND
151 if (eaten)
152 data = NULL;
153 }
1e49f22f
NTND
154 if (((base_count + i) & 1023) == 0)
155 display_progress(progress, base_count + i);
f3bf9224 156 free(data);
1e49f22f 157
f3bf9224 158 }
1e49f22f 159 display_progress(progress, base_count + i);
3af51928 160 free(entries);
f3bf9224
JH
161
162 return err;
f9253394
JH
163}
164
9b0aa728 165int verify_pack_index(struct packed_git *p)
f9253394 166{
d079837e
SP
167 off_t index_size;
168 const unsigned char *index_base;
9126f009 169 git_SHA_CTX ctx;
f9253394 170 unsigned char sha1[20];
62413604 171 int err = 0;
f9253394 172
d079837e
SP
173 if (open_pack_index(p))
174 return error("packfile %s index not opened", p->pack_name);
175 index_size = p->index_size;
176 index_base = p->index_data;
177
f9253394 178 /* Verify SHA1 sum of the index file */
9126f009
NP
179 git_SHA1_Init(&ctx);
180 git_SHA1_Update(&ctx, index_base, (unsigned int)(index_size - 20));
181 git_SHA1_Final(sha1, &ctx);
42873078 182 if (hashcmp(sha1, index_base + index_size - 20))
62413604 183 err = error("Packfile index for %s SHA1 mismatch",
f3bf9224 184 p->pack_name);
9b0aa728
SP
185 return err;
186}
187
1e49f22f
NTND
188int verify_pack(struct packed_git *p, verify_fn fn,
189 struct progress *progress, uint32_t base_count)
9b0aa728
SP
190{
191 int err = 0;
192 struct pack_window *w_curs = NULL;
193
194 err |= verify_pack_index(p);
195 if (!p->index_data)
196 return -1;
f3bf9224 197
1e49f22f 198 err |= verify_packfile(p, &w_curs, fn, progress, base_count);
62413604 199 unuse_pack(&w_curs);
f3bf9224 200
62413604 201 return err;
f9253394 202}