git.ozlabs.org Git - ccan/blob - ccan/tdb2/summary.c

   1  /*
   2    Trivial Database 2: 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 "private.h"
  19 #include <assert.h>
  20 #include <ccan/tally/tally.h>
  21
  22 static void sizes_for_bucket(unsigned bucket, size_t *min, size_t *max)
  23 {
  24         if (bucket <= 8) {
  25                 *min = *max = TDB_MIN_DATA_LEN + bucket * 8;
  26         } else if (bucket == 9) {
  27                 /* FIXME: This is twisted; fix size_to_bucket. */
  28                 *min = TDB_MIN_DATA_LEN + (1ULL << (bucket - 3)) + 8;
  29                 *max = TDB_MIN_DATA_LEN + (1ULL << (bucket - 2)) - 8;
  30         } else {
  31                 *min = TDB_MIN_DATA_LEN + (1ULL << (bucket - 3));
  32                 *max = TDB_MIN_DATA_LEN + (1ULL << (bucket - 2)) - 8;
  33         }
  34         assert(size_to_bucket(63, *min) == bucket);
  35         assert(size_to_bucket(63, *max) == bucket);
  36         if (bucket > 8)
  37                 assert(size_to_bucket(63, *min - 8) == bucket - 1);
  38         assert(size_to_bucket(63, *max + 8) == bucket + 1);
  39 }
  40
  41 static int count_hash(struct tdb_context *tdb,
  42                       tdb_off_t hash_off, unsigned bits)
  43 {
  44         const tdb_off_t *h;
  45         unsigned int i, count = 0;
  46
  47         h = tdb_access_read(tdb, hash_off, sizeof(*h) << bits, true);
  48         if (!h)
  49                 return -1;
  50         for (i = 0; i < (1 << bits); i++)
  51                 count += (h[i] != 0);
  52
  53         tdb_access_release(tdb, h);
  54         return count;
  55 }
  56
  57 static tdb_len_t summarize_zone(struct tdb_context *tdb, tdb_off_t zone_off,
  58                                 struct tally *zones,
  59                                 struct tally *hashes,
  60                                 struct tally *free,
  61                                 struct tally *keys,
  62                                 struct tally *data,
  63                                 struct tally *extra,
  64                                 struct tally *uncoal,
  65                                 uint64_t bucketlen[],
  66                                 unsigned int *num_buckets)
  67 {
  68         struct free_zone_header zhdr;
  69         tdb_off_t off, end;
  70         tdb_len_t len;
  71         unsigned int hdrlen;
  72         tdb_len_t unc = 0;
  73
  74         if (tdb_read_convert(tdb, zone_off, &zhdr, sizeof(zhdr)) == -1)
  75                 return TDB_OFF_ERR;
  76
  77         tally_add(zones, 1ULL << zhdr.zone_bits);
  78         *num_buckets = BUCKETS_FOR_ZONE(zhdr.zone_bits);
  79
  80         hdrlen = sizeof(zhdr)
  81                 + (BUCKETS_FOR_ZONE(zhdr.zone_bits) + 1) * sizeof(tdb_off_t);
  82
  83         end = zone_off + (1ULL << zhdr.zone_bits);
  84         if (end > tdb->map_size)
  85                 end = tdb->map_size;
  86
  87         for (off = zone_off + hdrlen; off < end; off += len) {
  88                 union {
  89                         struct tdb_used_record u;
  90                         struct tdb_free_record f;
  91                 } pad, *p;
  92                 p = tdb_get(tdb, off, &pad, sizeof(pad));
  93                 if (!p)
  94                         return TDB_OFF_ERR;
  95                 if (rec_magic(&p->u) != TDB_MAGIC) {
  96                         len = p->f.data_len;
  97                         tally_add(free, len);
  98                         bucketlen[size_to_bucket(frec_zone_bits(&p->f), len)]++;
  99                         len += sizeof(p->u);
 100                         unc++;
 101                 } else {
 102                         if (unc) {
 103                                 tally_add(uncoal, unc);
 104                                 unc = 0;
 105                         }
 106                         len = sizeof(p->u)
 107                                 + rec_key_length(&p->u)
 108                                 + rec_data_length(&p->u)
 109                                 + rec_extra_padding(&p->u);
 110
 111                         /* FIXME: Use different magic for hashes? */
 112                         if (!rec_key_length(&p->u) && !rec_hash(&p->u)) {
 113                                 int count = count_hash(tdb, off + sizeof(p->u),
 114                                                        TDB_SUBLEVEL_HASH_BITS);
 115                                 if (count == -1)
 116                                         return TDB_OFF_ERR;
 117                                 tally_add(hashes, count);
 118                         } else {
 119                                 tally_add(keys, rec_key_length(&p->u));
 120                                 tally_add(data, rec_data_length(&p->u));
 121                         }
 122                         tally_add(extra, rec_extra_padding(&p->u));
 123                 }
 124         }
 125         if (unc)
 126                 tally_add(uncoal, unc);
 127         return 1ULL << zhdr.zone_bits;
 128 }
 129
 130 #define SUMMARY_FORMAT \
 131         "Size of file/data: %zu/%zu\n" \
 132         "Number of zones: %zu\n" \
 133         "Smallest/average/largest zone size: %zu/%zu/%zu\n%s" \
 134         "Number of records: %zu\n" \
 135         "Smallest/average/largest keys: %zu/%zu/%zu\n%s" \
 136         "Smallest/average/largest data: %zu/%zu/%zu\n%s" \
 137         "Smallest/average/largest padding: %zu/%zu/%zu\n%s" \
 138         "Number of free records: %zu\n" \
 139         "Smallest/average/largest free records: %zu/%zu/%zu\n%s" \
 140         "Number of uncoalesced records: %zu\n" \
 141         "Smallest/average/largest uncoalesced runs: %zu/%zu/%zu\n%s" \
 142         "Toplevel hash used: %u of %u\n" \
 143         "Number of subhashes: %zu\n" \
 144         "Smallest/average/largest subhash entries: %zu/%zu/%zu\n%s" \
 145         "Percentage keys/data/padding/free/rechdrs/zonehdrs/hashes: %.0f/%.0f/%.0f/%.0f/%.0f/%.0f/%.0f\n"
 146
 147 #define BUCKET_SUMMARY_FORMAT_A                                 \
 148         "Free bucket %zu: total entries %zu.\n"                 \
 149         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
 150 #define BUCKET_SUMMARY_FORMAT_B                                 \
 151         "Free bucket %zu-%zu: total entries %zu.\n"             \
 152         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
 153
 154 #define HISTO_WIDTH 70
 155 #define HISTO_HEIGHT 20
 156
 157 char *tdb_summary(struct tdb_context *tdb, enum tdb_summary_flags flags)
 158 {
 159         tdb_off_t off;
 160         tdb_len_t len;
 161         unsigned int i, num_buckets, max_bucket = 0;
 162         uint64_t total_buckets = 0;
 163         struct tally *zones, *hashes, *freet, *keys, *data, *extra, *uncoal,
 164                 *buckets[BUCKETS_FOR_ZONE(63)+1] = { NULL };
 165         char *zonesg, *hashesg, *freeg, *keysg, *datag, *extrag, *uncoalg,
 166                 *bucketsg[BUCKETS_FOR_ZONE(63)+1] = { NULL };
 167         char *ret = NULL;
 168
 169         zonesg = hashesg = freeg = keysg = datag = extrag = uncoalg = NULL;
 170
 171         if (tdb_allrecord_lock(tdb, F_RDLCK, TDB_LOCK_WAIT, false) != 0)
 172                 return NULL;
 173
 174         if (tdb_lock_expand(tdb, F_RDLCK) != 0) {
 175                 tdb_allrecord_unlock(tdb, F_RDLCK);
 176                 return NULL;
 177         }
 178
 179         /* Start stats off empty. */
 180         zones = tally_new(HISTO_HEIGHT);
 181         hashes = tally_new(HISTO_HEIGHT);
 182         freet = tally_new(HISTO_HEIGHT);
 183         keys = tally_new(HISTO_HEIGHT);
 184         data = tally_new(HISTO_HEIGHT);
 185         extra = tally_new(HISTO_HEIGHT);
 186         uncoal = tally_new(HISTO_HEIGHT);
 187         if (!zones || !hashes || !freet || !keys || !data || !extra
 188             || !uncoal) {
 189                 tdb->ecode = TDB_ERR_OOM;
 190                 goto unlock;
 191         }
 192
 193         for (i = 0; i < sizeof(buckets)/sizeof(buckets[0]); i++) {
 194                 buckets[i] = tally_new(HISTO_HEIGHT);
 195                 if (!buckets[i]) {
 196                         tdb->ecode = TDB_ERR_OOM;
 197                         goto unlock;
 198                 }
 199         }
 200
 201         for (off = sizeof(struct tdb_header);
 202              off < tdb->map_size - 1;
 203              off += len) {
 204                 uint64_t bucketlen[BUCKETS_FOR_ZONE(63)+1] = { 0 };
 205                 len = summarize_zone(tdb, off, zones, hashes, freet, keys,
 206                                      data, extra, uncoal, bucketlen,
 207                                      &num_buckets);
 208                 if (len == TDB_OFF_ERR)
 209                         goto unlock;
 210                 for (i = 0; i < num_buckets; i++)
 211                         tally_add(buckets[i], bucketlen[i]);
 212                 if (num_buckets > max_bucket)
 213                         max_bucket = num_buckets;
 214                 total_buckets += num_buckets;
 215         }
 216
 217         if (flags & TDB_SUMMARY_HISTOGRAMS) {
 218                 zonesg = tally_histogram(zones, HISTO_WIDTH, HISTO_HEIGHT);
 219                 hashesg = tally_histogram(hashes, HISTO_WIDTH, HISTO_HEIGHT);
 220                 freeg = tally_histogram(freet, HISTO_WIDTH, HISTO_HEIGHT);
 221                 keysg = tally_histogram(keys, HISTO_WIDTH, HISTO_HEIGHT);
 222                 datag = tally_histogram(data, HISTO_WIDTH, HISTO_HEIGHT);
 223                 extrag = tally_histogram(extra, HISTO_WIDTH, HISTO_HEIGHT);
 224                 uncoalg = tally_histogram(uncoal, HISTO_WIDTH, HISTO_HEIGHT);
 225                 for (i = 0; i < sizeof(buckets)/sizeof(buckets[0]); i++) {
 226                         bucketsg[i] = tally_histogram(buckets[i],
 227                                                       HISTO_WIDTH,
 228                                                       HISTO_HEIGHT);
 229                 }
 230         }
 231
 232         /* 20 is max length of a %llu. */
 233         len = strlen(SUMMARY_FORMAT) + 33*20 + 1
 234                 + (zonesg ? strlen(zonesg) : 0)
 235                 + (hashesg ? strlen(hashesg) : 0)
 236                 + (freeg ? strlen(freeg) : 0)
 237                 + (keysg ? strlen(keysg) : 0)
 238                 + (datag ? strlen(datag) : 0)
 239                 + (extrag ? strlen(extrag) : 0)
 240                 + (uncoalg ? strlen(uncoalg) : 0);
 241         for (i = 0; i < max_bucket; i++) {
 242                 len += strlen(BUCKET_SUMMARY_FORMAT_B) + 6 * 20
 243                         + (bucketsg[i] ? strlen(bucketsg[i]) : 0);
 244         }
 245
 246         ret = malloc(len);
 247         if (!ret)
 248                 goto unlock;
 249
 250         len = sprintf(ret, SUMMARY_FORMAT,
 251                       (size_t)tdb->map_size,
 252                       tally_num(keys) + tally_num(data),
 253                       tally_num(zones),
 254                       tally_min(zones), tally_mean(zones), tally_max(zones),
 255                       zonesg ? zonesg : "",
 256                       tally_num(keys),
 257                       tally_min(keys), tally_mean(keys), tally_max(keys),
 258                       keysg ? keysg : "",
 259                       tally_min(data), tally_mean(data), tally_max(data),
 260                       datag ? datag : "",
 261                       tally_min(extra), tally_mean(extra), tally_max(extra),
 262                       extrag ? extrag : "",
 263                       tally_num(freet),
 264                       tally_min(freet), tally_mean(freet), tally_max(freet),
 265                       freeg ? freeg : "",
 266                       tally_total(uncoal, NULL),
 267                       tally_min(uncoal), tally_mean(uncoal), tally_max(uncoal),
 268                       uncoalg ? uncoalg : "",
 269                       count_hash(tdb, offsetof(struct tdb_header, hashtable),
 270                                  TDB_TOPLEVEL_HASH_BITS),
 271                       1 << TDB_TOPLEVEL_HASH_BITS,
 272                       tally_num(hashes),
 273                       tally_min(hashes), tally_mean(hashes), tally_max(hashes),
 274                       hashesg ? hashesg : "",
 275                       tally_total(keys, NULL) * 100.0 / tdb->map_size,
 276                       tally_total(data, NULL) * 100.0 / tdb->map_size,
 277                       tally_total(extra, NULL) * 100.0 / tdb->map_size,
 278                       tally_total(freet, NULL) * 100.0 / tdb->map_size,
 279                       (tally_num(keys) + tally_num(freet) + tally_num(hashes))
 280                       * sizeof(struct tdb_used_record) * 100.0 / tdb->map_size,
 281                       (tally_num(zones) * sizeof(struct free_zone_header)
 282                        + total_buckets * sizeof(tdb_off_t))
 283                       * 100.0 / tdb->map_size,
 284                       (tally_num(hashes)
 285                        * (sizeof(tdb_off_t) << TDB_SUBLEVEL_HASH_BITS)
 286                        + (sizeof(tdb_off_t) << TDB_TOPLEVEL_HASH_BITS))
 287                       * 100.0 / tdb->map_size);
 288
 289         for (i = 0; i < max_bucket; i++) {
 290                 size_t min, max;
 291                 sizes_for_bucket(i, &min, &max);
 292                 if (min == max) {
 293                         len += sprintf(ret + len, BUCKET_SUMMARY_FORMAT_A,
 294                                        min, tally_total(buckets[i], NULL),
 295                                        tally_min(buckets[i]),
 296                                        tally_mean(buckets[i]),
 297                                        tally_max(buckets[i]),
 298                                        bucketsg[i] ? bucketsg[i] : "");
 299                 } else {
 300                         len += sprintf(ret + len, BUCKET_SUMMARY_FORMAT_B,
 301                                        min, max, tally_total(buckets[i], NULL),
 302                                        tally_min(buckets[i]),
 303                                        tally_mean(buckets[i]),
 304                                        tally_max(buckets[i]),
 305                                        bucketsg[i] ? bucketsg[i] : "");
 306                 }
 307         }
 308
 309 unlock:
 310         free(zonesg);
 311         free(hashesg);
 312         free(freeg);
 313         free(keysg);
 314         free(datag);
 315         free(extrag);
 316         free(uncoalg);
 317         free(zones);
 318         free(hashes);
 319         free(freet);
 320         free(keys);
 321         free(data);
 322         free(extra);
 323         free(uncoal);
 324         for (i = 0; i < sizeof(buckets)/sizeof(buckets[0]); i++) {
 325                 free(buckets[i]);
 326                 free(bucketsg[i]);
 327         }
 328
 329         tdb_allrecord_unlock(tdb, F_RDLCK);
 330         tdb_unlock_expand(tdb, F_RDLCK);
 331         return ret;
 332 }