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 (TDB_PTR_IS_ERR(h)) {
  30                 tdb->ecode = TDB_PTR_ERR(h);
  31                 return -1;
  32         }
  33         for (i = 0; i < (1 << bits); i++)
  34                 count += (h[i] != 0);
  35
  36         tdb_access_release(tdb, h);
  37         return count;
  38 }
  39
  40 static bool summarize(struct tdb_context *tdb,
  41                       struct tally *hashes,
  42                       struct tally *ftables,
  43                       struct tally *fr,
  44                       struct tally *keys,
  45                       struct tally *data,
  46                       struct tally *extra,
  47                       struct tally *uncoal,
  48                       struct tally *buckets,
  49                       struct tally *chains)
  50 {
  51         tdb_off_t off;
  52         tdb_len_t len;
  53         tdb_len_t unc = 0;
  54
  55         for (off = sizeof(struct tdb_header); off < tdb->map_size; off += len) {
  56                 const union {
  57                         struct tdb_used_record u;
  58                         struct tdb_free_record f;
  59                         struct tdb_recovery_record r;
  60                 } *p;
  61                 /* We might not be able to get the whole thing. */
  62                 p = tdb_access_read(tdb, off, sizeof(p->f), true);
  63                 if (TDB_PTR_IS_ERR(p)) {
  64                         tdb->ecode = TDB_PTR_ERR(p);
  65                         return false;
  66                 }
  67                 if (p->r.magic == TDB_RECOVERY_INVALID_MAGIC
  68                     || p->r.magic == TDB_RECOVERY_MAGIC) {
  69                         if (unc) {
  70                                 tally_add(uncoal, unc);
  71                                 unc = 0;
  72                         }
  73                         len = sizeof(p->r) + p->r.max_len;
  74                 } else if (frec_magic(&p->f) == TDB_FREE_MAGIC) {
  75                         len = frec_len(&p->f);
  76                         tally_add(fr, len);
  77                         tally_add(buckets, size_to_bucket(len));
  78                         len += sizeof(p->u);
  79                         unc++;
  80                 } else if (rec_magic(&p->u) == TDB_USED_MAGIC) {
  81                         if (unc) {
  82                                 tally_add(uncoal, unc);
  83                                 unc = 0;
  84                         }
  85                         len = sizeof(p->u)
  86                                 + rec_key_length(&p->u)
  87                                 + rec_data_length(&p->u)
  88                                 + rec_extra_padding(&p->u);
  89
  90                         tally_add(keys, rec_key_length(&p->u));
  91                         tally_add(data, rec_data_length(&p->u));
  92                         tally_add(extra, rec_extra_padding(&p->u));
  93                 } else if (rec_magic(&p->u) == TDB_HTABLE_MAGIC) {
  94                         int count = count_hash(tdb,
  95                                                off + sizeof(p->u),
  96                                                TDB_SUBLEVEL_HASH_BITS);
  97                         if (count == -1)
  98                                 return false;
  99                         tally_add(hashes, count);
 100                         tally_add(extra, rec_extra_padding(&p->u));
 101                         len = sizeof(p->u)
 102                                 + rec_data_length(&p->u)
 103                                 + rec_extra_padding(&p->u);
 104                 } else if (rec_magic(&p->u) == TDB_FTABLE_MAGIC) {
 105                         len = sizeof(p->u)
 106                                 + rec_data_length(&p->u)
 107                                 + rec_extra_padding(&p->u);
 108                         tally_add(ftables, rec_data_length(&p->u));
 109                         tally_add(extra, rec_extra_padding(&p->u));
 110                 } else if (rec_magic(&p->u) == TDB_CHAIN_MAGIC) {
 111                         len = sizeof(p->u)
 112                                 + rec_data_length(&p->u)
 113                                 + rec_extra_padding(&p->u);
 114                         tally_add(chains, 1);
 115                         tally_add(extra, rec_extra_padding(&p->u));
 116                 } else
 117                         len = dead_space(tdb, off);
 118                 tdb_access_release(tdb, p);
 119         }
 120         if (unc)
 121                 tally_add(uncoal, unc);
 122         return true;
 123 }
 124
 125 #define SUMMARY_FORMAT \
 126         "Size of file/data: %zu/%zu\n" \
 127         "Number of records: %zu\n" \
 128         "Smallest/average/largest keys: %zu/%zu/%zu\n%s" \
 129         "Smallest/average/largest data: %zu/%zu/%zu\n%s" \
 130         "Smallest/average/largest padding: %zu/%zu/%zu\n%s" \
 131         "Number of free records: %zu\n" \
 132         "Smallest/average/largest free records: %zu/%zu/%zu\n%s" \
 133         "Number of uncoalesced records: %zu\n" \
 134         "Smallest/average/largest uncoalesced runs: %zu/%zu/%zu\n%s" \
 135         "Number of free lists: %zu\n%s" \
 136         "Toplevel hash used: %u of %u\n" \
 137         "Number of chains: %zu\n" \
 138         "Number of subhashes: %zu\n" \
 139         "Smallest/average/largest subhash entries: %zu/%zu/%zu\n%s" \
 140         "Percentage keys/data/padding/free/rechdrs/freehdrs/hashes: %.0f/%.0f/%.0f/%.0f/%.0f/%.0f/%.0f\n"
 141
 142 #define BUCKET_SUMMARY_FORMAT_A                                 \
 143         "Free bucket %zu: total entries %zu.\n"                 \
 144         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
 145 #define BUCKET_SUMMARY_FORMAT_B                                 \
 146         "Free bucket %zu-%zu: total entries %zu.\n"             \
 147         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
 148
 149 #define HISTO_WIDTH 70
 150 #define HISTO_HEIGHT 20
 151
 152 char *tdb_summary(struct tdb_context *tdb, enum tdb_summary_flags flags)
 153 {
 154         tdb_len_t len;
 155         struct tally *ftables, *hashes, *freet, *keys, *data, *extra, *uncoal,
 156                 *buckets, *chains;
 157         char *hashesg, *freeg, *keysg, *datag, *extrag, *uncoalg, *bucketsg;
 158         char *ret = NULL;
 159         enum TDB_ERROR ecode;
 160
 161         hashesg = freeg = keysg = datag = extrag = uncoalg = bucketsg = NULL;
 162
 163         ecode = tdb_allrecord_lock(tdb, F_RDLCK, TDB_LOCK_WAIT, false);
 164         if (ecode != TDB_SUCCESS) {
 165                 tdb->ecode = ecode;
 166                 return NULL;
 167         }
 168
 169         ecode = tdb_lock_expand(tdb, F_RDLCK);
 170         if (ecode != TDB_SUCCESS) {
 171                 tdb->ecode = ecode;
 172                 tdb_allrecord_unlock(tdb, F_RDLCK);
 173                 return NULL;
 174         }
 175
 176         /* Start stats off empty. */
 177         ftables = tally_new(HISTO_HEIGHT);
 178         hashes = tally_new(HISTO_HEIGHT);
 179         freet = tally_new(HISTO_HEIGHT);
 180         keys = tally_new(HISTO_HEIGHT);
 181         data = tally_new(HISTO_HEIGHT);
 182         extra = tally_new(HISTO_HEIGHT);
 183         uncoal = tally_new(HISTO_HEIGHT);
 184         buckets = tally_new(HISTO_HEIGHT);
 185         chains = tally_new(HISTO_HEIGHT);
 186         if (!ftables || !hashes || !freet || !keys || !data || !extra
 187             || !uncoal || !buckets || !chains) {
 188                 tdb_logerr(tdb, TDB_ERR_OOM, TDB_LOG_ERROR,
 189                            "tdb_summary: failed to allocate tally structures");
 190                 goto unlock;
 191         }
 192
 193         if (!summarize(tdb, hashes, ftables, freet, keys, data, extra, uncoal,
 194                        buckets, chains))
 195                 goto unlock;
 196
 197         if (flags & TDB_SUMMARY_HISTOGRAMS) {
 198                 hashesg = tally_histogram(hashes, HISTO_WIDTH, HISTO_HEIGHT);
 199                 freeg = tally_histogram(freet, HISTO_WIDTH, HISTO_HEIGHT);
 200                 keysg = tally_histogram(keys, HISTO_WIDTH, HISTO_HEIGHT);
 201                 datag = tally_histogram(data, HISTO_WIDTH, HISTO_HEIGHT);
 202                 extrag = tally_histogram(extra, HISTO_WIDTH, HISTO_HEIGHT);
 203                 uncoalg = tally_histogram(uncoal, HISTO_WIDTH, HISTO_HEIGHT);
 204                 bucketsg = tally_histogram(buckets, HISTO_WIDTH, HISTO_HEIGHT);
 205         }
 206
 207         /* 20 is max length of a %llu. */
 208         len = strlen(SUMMARY_FORMAT) + 33*20 + 1
 209                 + (hashesg ? strlen(hashesg) : 0)
 210                 + (freeg ? strlen(freeg) : 0)
 211                 + (keysg ? strlen(keysg) : 0)
 212                 + (datag ? strlen(datag) : 0)
 213                 + (extrag ? strlen(extrag) : 0)
 214                 + (uncoalg ? strlen(uncoalg) : 0)
 215                 + (bucketsg ? strlen(bucketsg) : 0);
 216
 217         ret = malloc(len);
 218         if (!ret)
 219                 goto unlock;
 220
 221         len = sprintf(ret, SUMMARY_FORMAT,
 222                       (size_t)tdb->map_size,
 223                       tally_num(keys) + tally_num(data),
 224                       tally_num(keys),
 225                       tally_min(keys), tally_mean(keys), tally_max(keys),
 226                       keysg ? keysg : "",
 227                       tally_min(data), tally_mean(data), tally_max(data),
 228                       datag ? datag : "",
 229                       tally_min(extra), tally_mean(extra), tally_max(extra),
 230                       extrag ? extrag : "",
 231                       tally_num(freet),
 232                       tally_min(freet), tally_mean(freet), tally_max(freet),
 233                       freeg ? freeg : "",
 234                       tally_total(uncoal, NULL),
 235                       tally_min(uncoal), tally_mean(uncoal), tally_max(uncoal),
 236                       uncoalg ? uncoalg : "",
 237                       tally_num(buckets),
 238                       bucketsg ? bucketsg : "",
 239                       count_hash(tdb, offsetof(struct tdb_header, hashtable),
 240                                  TDB_TOPLEVEL_HASH_BITS),
 241                       1 << TDB_TOPLEVEL_HASH_BITS,
 242                       tally_num(chains),
 243                       tally_num(hashes),
 244                       tally_min(hashes), tally_mean(hashes), tally_max(hashes),
 245                       hashesg ? hashesg : "",
 246                       tally_total(keys, NULL) * 100.0 / tdb->map_size,
 247                       tally_total(data, NULL) * 100.0 / tdb->map_size,
 248                       tally_total(extra, NULL) * 100.0 / tdb->map_size,
 249                       tally_total(freet, NULL) * 100.0 / tdb->map_size,
 250                       (tally_num(keys) + tally_num(freet) + tally_num(hashes))
 251                       * sizeof(struct tdb_used_record) * 100.0 / tdb->map_size,
 252                       tally_num(ftables) * sizeof(struct tdb_freetable)
 253                       * 100.0 / tdb->map_size,
 254                       (tally_num(hashes)
 255                        * (sizeof(tdb_off_t) << TDB_SUBLEVEL_HASH_BITS)
 256                        + (sizeof(tdb_off_t) << TDB_TOPLEVEL_HASH_BITS)
 257                        + sizeof(struct tdb_chain) * tally_num(chains))
 258                       * 100.0 / tdb->map_size);
 259
 260 unlock:
 261         free(hashesg);
 262         free(freeg);
 263         free(keysg);
 264         free(datag);
 265         free(extrag);
 266         free(uncoalg);
 267         free(bucketsg);
 268         free(hashes);
 269         free(buckets);
 270         free(freet);
 271         free(keys);
 272         free(data);
 273         free(extra);
 274         free(uncoal);
 275         free(ftables);
 276         free(chains);
 277
 278         tdb_allrecord_unlock(tdb, F_RDLCK);
 279         tdb_unlock_expand(tdb, F_RDLCK);
 280         return ret;
 281 }