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 int count_hash(struct tdb_context *tdb,
  23                       tdb_off_t hash_off, unsigned bits)
  24 {
  25         const tdb_off_t *h;
  26         unsigned int i, count = 0;
  27
  28         h = tdb_access_read(tdb, hash_off, sizeof(*h) << bits, true);
  29         if (!h)
  30                 return -1;
  31         for (i = 0; i < (1 << bits); i++)
  32                 count += (h[i] != 0);
  33
  34         tdb_access_release(tdb, h);
  35         return count;
  36 }
  37
  38 static bool summarize(struct tdb_context *tdb,
  39                       struct tally *hashes,
  40                       struct tally *flists,
  41                       struct tally *free,
  42                       struct tally *keys,
  43                       struct tally *data,
  44                       struct tally *extra,
  45                       struct tally *uncoal,
  46                       struct tally *buckets)
  47 {
  48         tdb_off_t off;
  49         tdb_len_t len;
  50         tdb_len_t unc = 0;
  51
  52         for (off = sizeof(struct tdb_header); off < tdb->map_size; off += len) {
  53                 union {
  54                         struct tdb_used_record u;
  55                         struct tdb_free_record f;
  56                 } pad, *p;
  57                 p = tdb_get(tdb, off, &pad, sizeof(pad));
  58                 if (!p)
  59                         return false;
  60                 if (rec_magic(&p->u) != TDB_MAGIC) {
  61                         len = p->f.data_len;
  62                         tally_add(free, len);
  63                         tally_add(buckets, size_to_bucket(len));
  64                         len += sizeof(p->u);
  65                         unc++;
  66                 } else {
  67                         if (unc) {
  68                                 tally_add(uncoal, unc);
  69                                 unc = 0;
  70                         }
  71                         len = sizeof(p->u)
  72                                 + rec_key_length(&p->u)
  73                                 + rec_data_length(&p->u)
  74                                 + rec_extra_padding(&p->u);
  75
  76                         /* FIXME: Use different magic for hashes, flists. */
  77                         if (!rec_key_length(&p->u) && rec_hash(&p->u) < 2) {
  78                                 if (rec_hash(&p->u) == 0) {
  79                                         int count = count_hash(tdb,
  80                                                         off + sizeof(p->u),
  81                                                         TDB_SUBLEVEL_HASH_BITS);
  82                                         if (count == -1)
  83                                                 return false;
  84                                         tally_add(hashes, count);
  85                                 } else {
  86                                         tally_add(flists,
  87                                                   rec_data_length(&p->u));
  88                                 }
  89                         } else {
  90                                 tally_add(keys, rec_key_length(&p->u));
  91                                 tally_add(data, rec_data_length(&p->u));
  92                         }
  93                         tally_add(extra, rec_extra_padding(&p->u));
  94                 }
  95         }
  96         if (unc)
  97                 tally_add(uncoal, unc);
  98         return true;
  99 }
 100
 101 #define SUMMARY_FORMAT \
 102         "Size of file/data: %zu/%zu\n" \
 103         "Number of records: %zu\n" \
 104         "Smallest/average/largest keys: %zu/%zu/%zu\n%s" \
 105         "Smallest/average/largest data: %zu/%zu/%zu\n%s" \
 106         "Smallest/average/largest padding: %zu/%zu/%zu\n%s" \
 107         "Number of free records: %zu\n" \
 108         "Smallest/average/largest free records: %zu/%zu/%zu\n%s" \
 109         "Number of uncoalesced records: %zu\n" \
 110         "Smallest/average/largest uncoalesced runs: %zu/%zu/%zu\n%s" \
 111         "Number of free lists: %zu\n%s" \
 112         "Toplevel hash used: %u of %u\n" \
 113         "Number of subhashes: %zu\n" \
 114         "Smallest/average/largest subhash entries: %zu/%zu/%zu\n%s" \
 115         "Percentage keys/data/padding/free/rechdrs/freehdrs/hashes: %.0f/%.0f/%.0f/%.0f/%.0f/%.0f/%.0f\n"
 116
 117 #define BUCKET_SUMMARY_FORMAT_A                                 \
 118         "Free bucket %zu: total entries %zu.\n"                 \
 119         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
 120 #define BUCKET_SUMMARY_FORMAT_B                                 \
 121         "Free bucket %zu-%zu: total entries %zu.\n"             \
 122         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
 123
 124 #define HISTO_WIDTH 70
 125 #define HISTO_HEIGHT 20
 126
 127 char *tdb_summary(struct tdb_context *tdb, enum tdb_summary_flags flags)
 128 {
 129         tdb_len_t len;
 130         struct tally *flists, *hashes, *freet, *keys, *data, *extra, *uncoal,
 131                 *buckets;
 132         char *hashesg, *freeg, *keysg, *datag, *extrag, *uncoalg, *bucketsg;
 133         char *ret = NULL;
 134
 135         hashesg = freeg = keysg = datag = extrag = uncoalg = bucketsg = NULL;
 136
 137         if (tdb_allrecord_lock(tdb, F_RDLCK, TDB_LOCK_WAIT, false) != 0)
 138                 return NULL;
 139
 140         if (tdb_lock_expand(tdb, F_RDLCK) != 0) {
 141                 tdb_allrecord_unlock(tdb, F_RDLCK);
 142                 return NULL;
 143         }
 144
 145         /* Start stats off empty. */
 146         flists = tally_new(HISTO_HEIGHT);
 147         hashes = tally_new(HISTO_HEIGHT);
 148         freet = tally_new(HISTO_HEIGHT);
 149         keys = tally_new(HISTO_HEIGHT);
 150         data = tally_new(HISTO_HEIGHT);
 151         extra = tally_new(HISTO_HEIGHT);
 152         uncoal = tally_new(HISTO_HEIGHT);
 153         buckets = tally_new(HISTO_HEIGHT);
 154         if (!flists || !hashes || !freet || !keys || !data || !extra
 155             || !uncoal || !buckets) {
 156                 tdb->ecode = TDB_ERR_OOM;
 157                 goto unlock;
 158         }
 159
 160         if (!summarize(tdb, hashes, flists, freet, keys, data, extra, uncoal,
 161                        buckets))
 162                 goto unlock;
 163
 164         if (flags & TDB_SUMMARY_HISTOGRAMS) {
 165                 hashesg = tally_histogram(hashes, HISTO_WIDTH, HISTO_HEIGHT);
 166                 freeg = tally_histogram(freet, HISTO_WIDTH, HISTO_HEIGHT);
 167                 keysg = tally_histogram(keys, HISTO_WIDTH, HISTO_HEIGHT);
 168                 datag = tally_histogram(data, HISTO_WIDTH, HISTO_HEIGHT);
 169                 extrag = tally_histogram(extra, HISTO_WIDTH, HISTO_HEIGHT);
 170                 uncoalg = tally_histogram(uncoal, HISTO_WIDTH, HISTO_HEIGHT);
 171                 bucketsg = tally_histogram(buckets, HISTO_WIDTH, HISTO_HEIGHT);
 172         }
 173
 174         /* 20 is max length of a %llu. */
 175         len = strlen(SUMMARY_FORMAT) + 33*20 + 1
 176                 + (hashesg ? strlen(hashesg) : 0)
 177                 + (freeg ? strlen(freeg) : 0)
 178                 + (keysg ? strlen(keysg) : 0)
 179                 + (datag ? strlen(datag) : 0)
 180                 + (extrag ? strlen(extrag) : 0)
 181                 + (uncoalg ? strlen(uncoalg) : 0)
 182                 + (bucketsg ? strlen(bucketsg) : 0);
 183
 184         ret = malloc(len);
 185         if (!ret)
 186                 goto unlock;
 187
 188         len = sprintf(ret, SUMMARY_FORMAT,
 189                       (size_t)tdb->map_size,
 190                       tally_num(keys) + tally_num(data),
 191                       tally_num(keys),
 192                       tally_min(keys), tally_mean(keys), tally_max(keys),
 193                       keysg ? keysg : "",
 194                       tally_min(data), tally_mean(data), tally_max(data),
 195                       datag ? datag : "",
 196                       tally_min(extra), tally_mean(extra), tally_max(extra),
 197                       extrag ? extrag : "",
 198                       tally_num(freet),
 199                       tally_min(freet), tally_mean(freet), tally_max(freet),
 200                       freeg ? freeg : "",
 201                       tally_total(uncoal, NULL),
 202                       tally_min(uncoal), tally_mean(uncoal), tally_max(uncoal),
 203                       uncoalg ? uncoalg : "",
 204                       tally_num(buckets),
 205                       bucketsg ? bucketsg : "",
 206                       count_hash(tdb, offsetof(struct tdb_header, hashtable),
 207                                  TDB_TOPLEVEL_HASH_BITS),
 208                       1 << TDB_TOPLEVEL_HASH_BITS,
 209                       tally_num(hashes),
 210                       tally_min(hashes), tally_mean(hashes), tally_max(hashes),
 211                       hashesg ? hashesg : "",
 212                       tally_total(keys, NULL) * 100.0 / tdb->map_size,
 213                       tally_total(data, NULL) * 100.0 / tdb->map_size,
 214                       tally_total(extra, NULL) * 100.0 / tdb->map_size,
 215                       tally_total(freet, NULL) * 100.0 / tdb->map_size,
 216                       (tally_num(keys) + tally_num(freet) + tally_num(hashes))
 217                       * sizeof(struct tdb_used_record) * 100.0 / tdb->map_size,
 218                       tally_num(flists) * sizeof(struct tdb_freelist)
 219                       * 100.0 / tdb->map_size,
 220                       (tally_num(hashes)
 221                        * (sizeof(tdb_off_t) << TDB_SUBLEVEL_HASH_BITS)
 222                        + (sizeof(tdb_off_t) << TDB_TOPLEVEL_HASH_BITS))
 223                       * 100.0 / tdb->map_size);
 224
 225 unlock:
 226         free(hashesg);
 227         free(freeg);
 228         free(keysg);
 229         free(datag);
 230         free(extrag);
 231         free(uncoalg);
 232         free(bucketsg);
 233         free(hashes);
 234         free(buckets);
 235         free(freet);
 236         free(keys);
 237         free(data);
 238         free(extra);
 239         free(uncoal);
 240
 241         tdb_allrecord_unlock(tdb, F_RDLCK);
 242         tdb_unlock_expand(tdb, F_RDLCK);
 243         return ret;
 244 }