lib/tdb/common/summary.c

   1  /*
   2    Trivial Database: human-readable summary code
   3    Copyright (C) Rusty Russell 2010
   4
   5    This library is free software; you can redistribute it and/or
   6    modify it under the terms of the GNU Lesser General Public
   7    License as published by the Free Software Foundation; either
   8    version 3 of the License, or (at your option) any later version.
   9
  10    This library is distributed in the hope that it will be useful,
  11    but WITHOUT ANY WARRANTY; without even the implied warranty of
  12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13    Lesser General Public License for more details.
  14
  15    You should have received a copy of the GNU Lesser General Public
  16    License along with this library; if not, see <http://www.gnu.org/licenses/>.
  17 */
  18 #include "tdb_private.h"
  19
  20 #define SUMMARY_FORMAT \
  21         "Size of file/data: %llu/%zu\n" \
  22         "Header offset/logical size: %zu/%zu\n" \
  23         "Number of records: %zu\n" \
  24         "Incompatible hash: %s\n" \
  25         "Active/supported feature flags: 0x%08x/0x%08x\n" \
  26         "Robust mutexes locking: %s\n" \
  27         "Smallest/average/largest keys: %zu/%zu/%zu\n" \
  28         "Smallest/average/largest data: %zu/%zu/%zu\n" \
  29         "Smallest/average/largest padding: %zu/%zu/%zu\n" \
  30         "Number of dead records: %zu\n" \
  31         "Smallest/average/largest dead records: %zu/%zu/%zu\n" \
  32         "Number of free records: %zu\n" \
  33         "Smallest/average/largest free records: %zu/%zu/%zu\n" \
  34         "Number of hash chains: %zu\n" \
  35         "Smallest/average/largest hash chains: %zu/%zu/%zu\n" \
  36         "Number of uncoalesced records: %zu\n" \
  37         "Smallest/average/largest uncoalesced runs: %zu/%zu/%zu\n" \
  38         "Percentage keys/data/padding/free/dead/rechdrs&tailers/hashes: %.0f/%.0f/%.0f/%.0f/%.0f/%.0f/%.0f\n"
  39
  40 /* We don't use tally module, to keep upstream happy. */
  41 struct tally {
  42         size_t min, max, total;
  43         size_t num;
  44 };
  45
  46 static void tally_init(struct tally *tally)
  47 {
  48         tally->total = 0;
  49         tally->num = 0;
  50         tally->min = tally->max = 0;
  51 }
  52
  53 static void tally_add(struct tally *tally, size_t len)
  54 {
  55         if (tally->num == 0)
  56                 tally->max = tally->min = len;
  57         else if (len > tally->max)
  58                 tally->max = len;
  59         else if (len < tally->min)
  60                 tally->min = len;
  61         tally->num++;
  62         tally->total += len;
  63 }
  64
  65 static size_t tally_mean(const struct tally *tally)
  66 {
  67         if (!tally->num)
  68                 return 0;
  69         return tally->total / tally->num;
  70 }
  71
  72 static size_t get_hash_length(struct tdb_context *tdb, unsigned int i)
  73 {
  74         tdb_off_t rec_ptr;
  75         struct tdb_chainwalk_ctx chainwalk;
  76         size_t count = 0;
  77
  78         if (tdb_ofs_read(tdb, TDB_HASH_TOP(i), &rec_ptr) == -1)
  79                 return 0;
  80
  81         tdb_chainwalk_init(&chainwalk, rec_ptr);
  82
  83         /* keep looking until we find the right record */
  84         while (rec_ptr) {
  85                 struct tdb_record r;
  86                 bool ok;
  87                 ++count;
  88                 if (tdb_rec_read(tdb, rec_ptr, &r) == -1)
  89                         return 0;
  90                 rec_ptr = r.next;
  91                 ok = tdb_chainwalk_check(tdb, &chainwalk, rec_ptr);
  92                 if (!ok) {
  93                         return SIZE_MAX;
  94                 }
  95         }
  96         return count;
  97 }
  98
  99 _PUBLIC_ char *tdb_summary(struct tdb_context *tdb)
 100 {
 101         off_t file_size;
 102         tdb_off_t off, rec_off;
 103         struct tally freet, keys, data, dead, extra, hashval, uncoal;
 104         struct tdb_record rec;
 105         char *ret = NULL;
 106         bool locked;
 107         size_t unc = 0;
 108         int len;
 109         struct tdb_record recovery;
 110
 111         /* Read-only databases use no locking at all: it's best-effort.
 112          * We may have a write lock already, so skip that case too. */
 113         if (tdb->read_only || tdb->allrecord_lock.count != 0) {
 114                 locked = false;
 115         } else {
 116                 if (tdb_lockall_read(tdb) == -1)
 117                         return NULL;
 118                 locked = true;
 119         }
 120
 121         if (tdb_recovery_area(tdb, tdb->methods, &rec_off, &recovery) != 0) {
 122                 goto unlock;
 123         }
 124
 125         tally_init(&freet);
 126         tally_init(&keys);
 127         tally_init(&data);
 128         tally_init(&dead);
 129         tally_init(&extra);
 130         tally_init(&hashval);
 131         tally_init(&uncoal);
 132
 133         for (off = TDB_DATA_START(tdb->hash_size);
 134              off < tdb->map_size - 1;
 135              off += sizeof(rec) + rec.rec_len) {
 136                 if (tdb->methods->tdb_read(tdb, off, &rec, sizeof(rec),
 137                                            DOCONV()) == -1)
 138                         goto unlock;
 139                 switch (rec.magic) {
 140                 case TDB_MAGIC:
 141                         tally_add(&keys, rec.key_len);
 142                         tally_add(&data, rec.data_len);
 143                         tally_add(&extra, rec.rec_len - (rec.key_len
 144                                                          + rec.data_len));
 145                         if (unc > 1)
 146                                 tally_add(&uncoal, unc - 1);
 147                         unc = 0;
 148                         break;
 149                 case TDB_FREE_MAGIC:
 150                         tally_add(&freet, rec.rec_len);
 151                         unc++;
 152                         break;
 153                 /* If we crash after ftruncate, we can get zeroes or fill. */
 154                 case TDB_RECOVERY_INVALID_MAGIC:
 155                 case 0x42424242:
 156                         unc++;
 157                         /* If it's a valid recovery, we can trust rec_len. */
 158                         if (off != rec_off) {
 159                                 rec.rec_len = tdb_dead_space(tdb, off)
 160                                         - sizeof(rec);
 161                         }
 162
 163                         FALL_THROUGH;
 164                 case TDB_DEAD_MAGIC:
 165                         tally_add(&dead, rec.rec_len);
 166                         break;
 167                 default:
 168                         TDB_LOG((tdb, TDB_DEBUG_ERROR,
 169                                  "Unexpected record magic 0x%x at offset %u\n",
 170                                  rec.magic, off));
 171                         goto unlock;
 172                 }
 173         }
 174         if (unc > 1)
 175                 tally_add(&uncoal, unc - 1);
 176
 177         for (off = 0; off < tdb->hash_size; off++)
 178                 tally_add(&hashval, get_hash_length(tdb, off));
 179
 180         file_size = tdb->hdr_ofs + tdb->map_size;
 181
 182         len = asprintf(&ret, SUMMARY_FORMAT,
 183                  (unsigned long long)file_size, keys.total+data.total,
 184                  (size_t)tdb->hdr_ofs, (size_t)tdb->map_size,
 185                  keys.num,
 186                  (tdb->hash_fn == tdb_jenkins_hash)?"yes":"no",
 187                  (unsigned)tdb->feature_flags, TDB_SUPPORTED_FEATURE_FLAGS,
 188                  (tdb->feature_flags & TDB_FEATURE_FLAG_MUTEX)?"yes":"no",
 189                  keys.min, tally_mean(&keys), keys.max,
 190                  data.min, tally_mean(&data), data.max,
 191                  extra.min, tally_mean(&extra), extra.max,
 192                  dead.num,
 193                  dead.min, tally_mean(&dead), dead.max,
 194                  freet.num,
 195                  freet.min, tally_mean(&freet), freet.max,
 196                  hashval.num,
 197                  hashval.min, tally_mean(&hashval), hashval.max,
 198                  uncoal.total,
 199                  uncoal.min, tally_mean(&uncoal), uncoal.max,
 200                  keys.total * 100.0 / file_size,
 201                  data.total * 100.0 / file_size,
 202                  extra.total * 100.0 / file_size,
 203                  freet.total * 100.0 / file_size,
 204                  dead.total * 100.0 / file_size,
 205                  (keys.num + freet.num + dead.num)
 206                  * (sizeof(struct tdb_record) + sizeof(uint32_t))
 207                  * 100.0 / file_size,
 208                  tdb->hash_size * sizeof(tdb_off_t)
 209                  * 100.0 / file_size);
 210         if (len == -1) {
 211                 goto unlock;
 212         }
 213
 214 unlock:
 215         if (locked) {
 216                 tdb_unlockall_read(tdb);
 217         }
 218         return ret;
 219 }