pack-check.c

   1 #include "cache.h"
   2 #include "pack.h"
   3 #include "pack-revindex.h"
   4 #include "progress.h"
   5
   6 struct idx_entry {
   7         off_t                offset;
   8         const unsigned char *sha1;
   9         unsigned int nr;
  10 };
  11
  12 static int compare_entries(const void *e1, const void *e2)
  13 {
  14         const struct idx_entry *entry1 = e1;
  15         const struct idx_entry *entry2 = e2;
  16         if (entry1->offset < entry2->offset)
  17                 return -1;
  18         if (entry1->offset > entry2->offset)
  19                 return 1;
  20         return 0;
  21 }
  22
  23 int check_pack_crc(struct packed_git *p, struct pack_window **w_curs,
  24                    off_t offset, off_t len, unsigned int nr)
  25 {
  26         const uint32_t *index_crc;
  27         uint32_t data_crc = crc32(0, NULL, 0);
  28
  29         do {
  30                 unsigned long avail;
  31                 void *data = use_pack(p, w_curs, offset, &avail);
  32                 if (avail > len)
  33                         avail = len;
  34                 data_crc = crc32(data_crc, data, avail);
  35                 offset += avail;
  36                 len -= avail;
  37         } while (len);
  38
  39         index_crc = p->index_data;
  40         index_crc += 2 + 256 + p->num_objects * (20/4) + nr;
  41
  42         return data_crc != ntohl(*index_crc);
  43 }
  44
  45 static int verify_packfile(struct packed_git *p,
  46                            struct pack_window **w_curs,
  47                            verify_fn fn,
  48                            struct progress *progress, uint32_t base_count)
  49
  50 {
  51         off_t index_size = p->index_size;
  52         const unsigned char *index_base = p->index_data;
  53         git_SHA_CTX ctx;
  54         unsigned char sha1[20], *pack_sig;
  55         off_t offset = 0, pack_sig_ofs = 0;
  56         uint32_t nr_objects, i;
  57         int err = 0;
  58         struct idx_entry *entries;
  59
  60         /* Note that the pack header checks are actually performed by
  61          * use_pack when it first opens the pack file.  If anything
  62          * goes wrong during those checks then the call will die out
  63          * immediately.
  64          */
  65
  66         git_SHA1_Init(&ctx);
  67         do {
  68                 unsigned long remaining;
  69                 unsigned char *in = use_pack(p, w_curs, offset, &remaining);
  70                 offset += remaining;
  71                 if (!pack_sig_ofs)
  72                         pack_sig_ofs = p->pack_size - 20;
  73                 if (offset > pack_sig_ofs)
  74                         remaining -= (unsigned int)(offset - pack_sig_ofs);
  75                 git_SHA1_Update(&ctx, in, remaining);
  76         } while (offset < pack_sig_ofs);
  77         git_SHA1_Final(sha1, &ctx);
  78         pack_sig = use_pack(p, w_curs, pack_sig_ofs, NULL);
  79         if (hashcmp(sha1, pack_sig))
  80                 err = error("%s SHA1 checksum mismatch",
  81                             p->pack_name);
  82         if (hashcmp(index_base + index_size - 40, pack_sig))
  83                 err = error("%s SHA1 does not match its index",
  84                             p->pack_name);
  85         unuse_pack(w_curs);
  86
  87         /* Make sure everything reachable from idx is valid.  Since we
  88          * have verified that nr_objects matches between idx and pack,
  89          * we do not do scan-streaming check on the pack file.
  90          */
  91         nr_objects = p->num_objects;
  92         entries = xmalloc((nr_objects + 1) * sizeof(*entries));
  93         entries[nr_objects].offset = pack_sig_ofs;
  94         /* first sort entries by pack offset, since unpacking them is more efficient that way */
  95         for (i = 0; i < nr_objects; i++) {
  96                 entries[i].sha1 = nth_packed_object_sha1(p, i);
  97                 if (!entries[i].sha1)
  98                         die("internal error pack-check nth-packed-object");
  99                 entries[i].offset = nth_packed_object_offset(p, i);
 100                 entries[i].nr = i;
 101         }
 102         qsort(entries, nr_objects, sizeof(*entries), compare_entries);
 103
 104         for (i = 0; i < nr_objects; i++) {
 105                 void *data;
 106                 enum object_type type;
 107                 unsigned long size;
 108
 109                 if (p->index_version > 1) {
 110                         off_t offset = entries[i].offset;
 111                         off_t len = entries[i+1].offset - offset;
 112                         unsigned int nr = entries[i].nr;
 113                         if (check_pack_crc(p, w_curs, offset, len, nr))
 114                                 err = error("index CRC mismatch for object %s "
 115                                             "from %s at offset %"PRIuMAX"",
 116                                             sha1_to_hex(entries[i].sha1),
 117                                             p->pack_name, (uintmax_t)offset);
 118                 }
 119                 data = unpack_entry(p, entries[i].offset, &type, &size);
 120                 if (!data)
 121                         err = error("cannot unpack %s from %s at offset %"PRIuMAX"",
 122                                     sha1_to_hex(entries[i].sha1), p->pack_name,
 123                                     (uintmax_t)entries[i].offset);
 124                 else if (check_sha1_signature(entries[i].sha1, data, size, typename(type)))
 125                         err = error("packed %s from %s is corrupt",
 126                                     sha1_to_hex(entries[i].sha1), p->pack_name);
 127                 else if (fn) {
 128                         int eaten = 0;
 129                         fn(entries[i].sha1, type, size, data, &eaten);
 130                         if (eaten)
 131                                 data = NULL;
 132                 }
 133                 if (((base_count + i) & 1023) == 0)
 134                         display_progress(progress, base_count + i);
 135                 free(data);
 136
 137         }
 138         display_progress(progress, base_count + i);
 139         free(entries);
 140
 141         return err;
 142 }
 143
 144 int verify_pack_index(struct packed_git *p)
 145 {
 146         off_t index_size;
 147         const unsigned char *index_base;
 148         git_SHA_CTX ctx;
 149         unsigned char sha1[20];
 150         int err = 0;
 151
 152         if (open_pack_index(p))
 153                 return error("packfile %s index not opened", p->pack_name);
 154         index_size = p->index_size;
 155         index_base = p->index_data;
 156
 157         /* Verify SHA1 sum of the index file */
 158         git_SHA1_Init(&ctx);
 159         git_SHA1_Update(&ctx, index_base, (unsigned int)(index_size - 20));
 160         git_SHA1_Final(sha1, &ctx);
 161         if (hashcmp(sha1, index_base + index_size - 20))
 162                 err = error("Packfile index for %s SHA1 mismatch",
 163                             p->pack_name);
 164         return err;
 165 }
 166
 167 int verify_pack(struct packed_git *p, verify_fn fn,
 168                 struct progress *progress, uint32_t base_count)
 169 {
 170         int err = 0;
 171         struct pack_window *w_curs = NULL;
 172
 173         err |= verify_pack_index(p);
 174         if (!p->index_data)
 175                 return -1;
 176
 177         err |= verify_packfile(p, &w_curs, fn, progress, base_count);
 178         unuse_pack(&w_curs);
 179
 180         return err;
 181 }