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                 const union {
  54                         struct tdb_used_record u;
  55                         struct tdb_free_record f;
  56                         struct tdb_recovery_record r;
  57                 } *p;
  58                 /* We might not be able to get the whole thing. */
  59                 p = tdb_access_read(tdb, off, sizeof(p->f), true);
  60                 if (!p)
  61                         return false;
  62                 if (p->r.magic == TDB_RECOVERY_INVALID_MAGIC
  63                     || p->r.magic == TDB_RECOVERY_MAGIC) {
  64                         if (unc) {
  65                                 tally_add(uncoal, unc);
  66                                 unc = 0;
  67                         }
  68                         len = sizeof(p->r) + p->r.max_len;
  69                 } else if (frec_magic(&p->f) == TDB_FREE_MAGIC) {
  70                         len = frec_len(&p->f);
  71                         tally_add(free, len);
  72                         tally_add(buckets, size_to_bucket(len));
  73                         len += sizeof(p->u);
  74                         unc++;
  75                 } else if (rec_magic(&p->u) == TDB_MAGIC) {
  76                         if (unc) {
  77                                 tally_add(uncoal, unc);
  78                                 unc = 0;
  79                         }
  80                         len = sizeof(p->u)
  81                                 + rec_key_length(&p->u)
  82                                 + rec_data_length(&p->u)
  83                                 + rec_extra_padding(&p->u);
  84
  85                         /* FIXME: Use different magic for hashes, flists. */
  86                         if (!rec_key_length(&p->u) && rec_hash(&p->u) < 2) {
  87                                 if (rec_hash(&p->u) == 0) {
  88                                         int count = count_hash(tdb,
  89                                                         off + sizeof(p->u),
  90                                                         TDB_SUBLEVEL_HASH_BITS);
  91                                         if (count == -1)
  92                                                 return false;
  93                                         tally_add(hashes, count);
  94                                 } else {
  95                                         tally_add(flists,
  96                                                   rec_data_length(&p->u));
  97                                 }
  98                         } else {
  99                                 tally_add(keys, rec_key_length(&p->u));
 100                                 tally_add(data, rec_data_length(&p->u));
 101                         }
 102                         tally_add(extra, rec_extra_padding(&p->u));
 103                 } else
 104                         len = dead_space(tdb, off);
 105                 tdb_access_release(tdb, p);
 106         }
 107         if (unc)
 108                 tally_add(uncoal, unc);
 109         return true;
 110 }
 111
 112 #define SUMMARY_FORMAT \
 113         "Size of file/data: %zu/%zu\n" \
 114         "Number of records: %zu\n" \
 115         "Smallest/average/largest keys: %zu/%zu/%zu\n%s" \
 116         "Smallest/average/largest data: %zu/%zu/%zu\n%s" \
 117         "Smallest/average/largest padding: %zu/%zu/%zu\n%s" \
 118         "Number of free records: %zu\n" \
 119         "Smallest/average/largest free records: %zu/%zu/%zu\n%s" \
 120         "Number of uncoalesced records: %zu\n" \
 121         "Smallest/average/largest uncoalesced runs: %zu/%zu/%zu\n%s" \
 122         "Number of free lists: %zu\n%s" \
 123         "Toplevel hash used: %u of %u\n" \
 124         "Number of subhashes: %zu\n" \
 125         "Smallest/average/largest subhash entries: %zu/%zu/%zu\n%s" \
 126         "Percentage keys/data/padding/free/rechdrs/freehdrs/hashes: %.0f/%.0f/%.0f/%.0f/%.0f/%.0f/%.0f\n"
 127
 128 #define BUCKET_SUMMARY_FORMAT_A                                 \
 129         "Free bucket %zu: total entries %zu.\n"                 \
 130         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
 131 #define BUCKET_SUMMARY_FORMAT_B                                 \
 132         "Free bucket %zu-%zu: total entries %zu.\n"             \
 133         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
 134
 135 #define HISTO_WIDTH 70
 136 #define HISTO_HEIGHT 20
 137
 138 char *tdb_summary(struct tdb_context *tdb, enum tdb_summary_flags flags)
 139 {
 140         tdb_len_t len;
 141         struct tally *flists, *hashes, *freet, *keys, *data, *extra, *uncoal,
 142                 *buckets;
 143         char *hashesg, *freeg, *keysg, *datag, *extrag, *uncoalg, *bucketsg;
 144         char *ret = NULL;
 145
 146         hashesg = freeg = keysg = datag = extrag = uncoalg = bucketsg = NULL;
 147
 148         if (tdb_allrecord_lock(tdb, F_RDLCK, TDB_LOCK_WAIT, false) != 0)
 149                 return NULL;
 150
 151         if (tdb_lock_expand(tdb, F_RDLCK) != 0) {
 152                 tdb_allrecord_unlock(tdb, F_RDLCK);
 153                 return NULL;
 154         }
 155
 156         /* Start stats off empty. */
 157         flists = tally_new(HISTO_HEIGHT);
 158         hashes = tally_new(HISTO_HEIGHT);
 159         freet = tally_new(HISTO_HEIGHT);
 160         keys = tally_new(HISTO_HEIGHT);
 161         data = tally_new(HISTO_HEIGHT);
 162         extra = tally_new(HISTO_HEIGHT);
 163         uncoal = tally_new(HISTO_HEIGHT);
 164         buckets = tally_new(HISTO_HEIGHT);
 165         if (!flists || !hashes || !freet || !keys || !data || !extra
 166             || !uncoal || !buckets) {
 167                 tdb->ecode = TDB_ERR_OOM;
 168                 goto unlock;
 169         }
 170
 171         if (!summarize(tdb, hashes, flists, freet, keys, data, extra, uncoal,
 172                        buckets))
 173                 goto unlock;
 174
 175         if (flags & TDB_SUMMARY_HISTOGRAMS) {
 176                 hashesg = tally_histogram(hashes, HISTO_WIDTH, HISTO_HEIGHT);
 177                 freeg = tally_histogram(freet, HISTO_WIDTH, HISTO_HEIGHT);
 178                 keysg = tally_histogram(keys, HISTO_WIDTH, HISTO_HEIGHT);
 179                 datag = tally_histogram(data, HISTO_WIDTH, HISTO_HEIGHT);
 180                 extrag = tally_histogram(extra, HISTO_WIDTH, HISTO_HEIGHT);
 181                 uncoalg = tally_histogram(uncoal, HISTO_WIDTH, HISTO_HEIGHT);
 182                 bucketsg = tally_histogram(buckets, HISTO_WIDTH, HISTO_HEIGHT);
 183         }
 184
 185         /* 20 is max length of a %llu. */
 186         len = strlen(SUMMARY_FORMAT) + 33*20 + 1
 187                 + (hashesg ? strlen(hashesg) : 0)
 188                 + (freeg ? strlen(freeg) : 0)
 189                 + (keysg ? strlen(keysg) : 0)
 190                 + (datag ? strlen(datag) : 0)
 191                 + (extrag ? strlen(extrag) : 0)
 192                 + (uncoalg ? strlen(uncoalg) : 0)
 193                 + (bucketsg ? strlen(bucketsg) : 0);
 194
 195         ret = malloc(len);
 196         if (!ret)
 197                 goto unlock;
 198
 199         len = sprintf(ret, SUMMARY_FORMAT,
 200                       (size_t)tdb->map_size,
 201                       tally_num(keys) + tally_num(data),
 202                       tally_num(keys),
 203                       tally_min(keys), tally_mean(keys), tally_max(keys),
 204                       keysg ? keysg : "",
 205                       tally_min(data), tally_mean(data), tally_max(data),
 206                       datag ? datag : "",
 207                       tally_min(extra), tally_mean(extra), tally_max(extra),
 208                       extrag ? extrag : "",
 209                       tally_num(freet),
 210                       tally_min(freet), tally_mean(freet), tally_max(freet),
 211                       freeg ? freeg : "",
 212                       tally_total(uncoal, NULL),
 213                       tally_min(uncoal), tally_mean(uncoal), tally_max(uncoal),
 214                       uncoalg ? uncoalg : "",
 215                       tally_num(buckets),
 216                       bucketsg ? bucketsg : "",
 217                       count_hash(tdb, offsetof(struct tdb_header, hashtable),
 218                                  TDB_TOPLEVEL_HASH_BITS),
 219                       1 << TDB_TOPLEVEL_HASH_BITS,
 220                       tally_num(hashes),
 221                       tally_min(hashes), tally_mean(hashes), tally_max(hashes),
 222                       hashesg ? hashesg : "",
 223                       tally_total(keys, NULL) * 100.0 / tdb->map_size,
 224                       tally_total(data, NULL) * 100.0 / tdb->map_size,
 225                       tally_total(extra, NULL) * 100.0 / tdb->map_size,
 226                       tally_total(freet, NULL) * 100.0 / tdb->map_size,
 227                       (tally_num(keys) + tally_num(freet) + tally_num(hashes))
 228                       * sizeof(struct tdb_used_record) * 100.0 / tdb->map_size,
 229                       tally_num(flists) * sizeof(struct tdb_freelist)
 230                       * 100.0 / tdb->map_size,
 231                       (tally_num(hashes)
 232                        * (sizeof(tdb_off_t) << TDB_SUBLEVEL_HASH_BITS)
 233                        + (sizeof(tdb_off_t) << TDB_TOPLEVEL_HASH_BITS))
 234                       * 100.0 / tdb->map_size);
 235
 236 unlock:
 237         free(hashesg);
 238         free(freeg);
 239         free(keysg);
 240         free(datag);
 241         free(extrag);
 242         free(uncoalg);
 243         free(bucketsg);
 244         free(hashes);
 245         free(buckets);
 246         free(freet);
 247         free(keys);
 248         free(data);
 249         free(extra);
 250         free(uncoal);
 251
 252         tdb_allrecord_unlock(tdb, F_RDLCK);
 253         tdb_unlock_expand(tdb, F_RDLCK);
 254         return ret;
 255 }