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 tdb_off_t count_hash(struct tdb_context *tdb,
  23                             tdb_off_t hash_off, unsigned bits)
  24 {
  25         const tdb_off_t *h;
  26         tdb_off_t count = 0;
  27         unsigned int i;
  28
  29         h = tdb_access_read(tdb, hash_off, sizeof(*h) << bits, true);
  30         if (TDB_PTR_IS_ERR(h)) {
  31                 return TDB_PTR_ERR(h);
  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 enum TDB_ERROR 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                         return TDB_PTR_ERR(p);
  65                 }
  66                 if (p->r.magic == TDB_RECOVERY_INVALID_MAGIC
  67                     || p->r.magic == TDB_RECOVERY_MAGIC) {
  68                         if (unc) {
  69                                 tally_add(uncoal, unc);
  70                                 unc = 0;
  71                         }
  72                         len = sizeof(p->r) + p->r.max_len;
  73                 } else if (frec_magic(&p->f) == TDB_FREE_MAGIC) {
  74                         len = frec_len(&p->f);
  75                         tally_add(fr, len);
  76                         tally_add(buckets, size_to_bucket(len));
  77                         len += sizeof(p->u);
  78                         unc++;
  79                 } else if (rec_magic(&p->u) == TDB_USED_MAGIC) {
  80                         if (unc) {
  81                                 tally_add(uncoal, unc);
  82                                 unc = 0;
  83                         }
  84                         len = sizeof(p->u)
  85                                 + rec_key_length(&p->u)
  86                                 + rec_data_length(&p->u)
  87                                 + rec_extra_padding(&p->u);
  88
  89                         tally_add(keys, rec_key_length(&p->u));
  90                         tally_add(data, rec_data_length(&p->u));
  91                         tally_add(extra, rec_extra_padding(&p->u));
  92                 } else if (rec_magic(&p->u) == TDB_HTABLE_MAGIC) {
  93                         tdb_off_t count = count_hash(tdb,
  94                                                      off + sizeof(p->u),
  95                                                      TDB_SUBLEVEL_HASH_BITS);
  96                         if (TDB_OFF_IS_ERR(count)) {
  97                                 return count;
  98                         }
  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                         if (TDB_OFF_IS_ERR(len)) {
 119                                 return len;
 120                         }
 121                 }
 122                 tdb_access_release(tdb, p);
 123         }
 124         if (unc)
 125                 tally_add(uncoal, unc);
 126         return TDB_SUCCESS;
 127 }
 128
 129 #define SUMMARY_FORMAT \
 130         "Size of file/data: %zu/%zu\n" \
 131         "Number of records: %zu\n" \
 132         "Smallest/average/largest keys: %zu/%zu/%zu\n%s" \
 133         "Smallest/average/largest data: %zu/%zu/%zu\n%s" \
 134         "Smallest/average/largest padding: %zu/%zu/%zu\n%s" \
 135         "Number of free records: %zu\n" \
 136         "Smallest/average/largest free records: %zu/%zu/%zu\n%s" \
 137         "Number of uncoalesced records: %zu\n" \
 138         "Smallest/average/largest uncoalesced runs: %zu/%zu/%zu\n%s" \
 139         "Number of free lists: %zu\n%s" \
 140         "Toplevel hash used: %u of %u\n" \
 141         "Number of chains: %zu\n" \
 142         "Number of subhashes: %zu\n" \
 143         "Smallest/average/largest subhash entries: %zu/%zu/%zu\n%s" \
 144         "Percentage keys/data/padding/free/rechdrs/freehdrs/hashes: %.0f/%.0f/%.0f/%.0f/%.0f/%.0f/%.0f\n"
 145
 146 #define BUCKET_SUMMARY_FORMAT_A                                 \
 147         "Free bucket %zu: total entries %zu.\n"                 \
 148         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
 149 #define BUCKET_SUMMARY_FORMAT_B                                 \
 150         "Free bucket %zu-%zu: total entries %zu.\n"             \
 151         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
 152
 153 #define HISTO_WIDTH 70
 154 #define HISTO_HEIGHT 20
 155
 156 enum TDB_ERROR tdb_summary(struct tdb_context *tdb,
 157                            enum tdb_summary_flags flags,
 158                            char **summary)
 159 {
 160         tdb_len_t len;
 161         struct tally *ftables, *hashes, *freet, *keys, *data, *extra, *uncoal,
 162                 *buckets, *chains;
 163         char *hashesg, *freeg, *keysg, *datag, *extrag, *uncoalg, *bucketsg;
 164         enum TDB_ERROR ecode;
 165
 166         hashesg = freeg = keysg = datag = extrag = uncoalg = bucketsg = NULL;
 167
 168         ecode = tdb_allrecord_lock(tdb, F_RDLCK, TDB_LOCK_WAIT, false);
 169         if (ecode != TDB_SUCCESS) {
 170                 return ecode;
 171         }
 172
 173         ecode = tdb_lock_expand(tdb, F_RDLCK);
 174         if (ecode != TDB_SUCCESS) {
 175                 tdb_allrecord_unlock(tdb, F_RDLCK);
 176                 return ecode;
 177         }
 178
 179         /* Start stats off empty. */
 180         ftables = 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         buckets = tally_new(HISTO_HEIGHT);
 188         chains = tally_new(HISTO_HEIGHT);
 189         if (!ftables || !hashes || !freet || !keys || !data || !extra
 190             || !uncoal || !buckets || !chains) {
 191                 ecode = tdb_logerr(tdb, TDB_ERR_OOM, TDB_LOG_ERROR,
 192                                    "tdb_summary: failed to allocate"
 193                                    " tally structures");
 194                 goto unlock;
 195         }
 196
 197         ecode = summarize(tdb, hashes, ftables, freet, keys, data, extra,
 198                           uncoal, buckets, chains);
 199         if (ecode != TDB_SUCCESS) {
 200                 goto unlock;
 201         }
 202
 203         if (flags & TDB_SUMMARY_HISTOGRAMS) {
 204                 hashesg = tally_histogram(hashes, HISTO_WIDTH, HISTO_HEIGHT);
 205                 freeg = tally_histogram(freet, HISTO_WIDTH, HISTO_HEIGHT);
 206                 keysg = tally_histogram(keys, HISTO_WIDTH, HISTO_HEIGHT);
 207                 datag = tally_histogram(data, HISTO_WIDTH, HISTO_HEIGHT);
 208                 extrag = tally_histogram(extra, HISTO_WIDTH, HISTO_HEIGHT);
 209                 uncoalg = tally_histogram(uncoal, HISTO_WIDTH, HISTO_HEIGHT);
 210                 bucketsg = tally_histogram(buckets, HISTO_WIDTH, HISTO_HEIGHT);
 211         }
 212
 213         /* 20 is max length of a %llu. */
 214         len = strlen(SUMMARY_FORMAT) + 33*20 + 1
 215                 + (hashesg ? strlen(hashesg) : 0)
 216                 + (freeg ? strlen(freeg) : 0)
 217                 + (keysg ? strlen(keysg) : 0)
 218                 + (datag ? strlen(datag) : 0)
 219                 + (extrag ? strlen(extrag) : 0)
 220                 + (uncoalg ? strlen(uncoalg) : 0)
 221                 + (bucketsg ? strlen(bucketsg) : 0);
 222
 223         *summary = malloc(len);
 224         if (!*summary) {
 225                 ecode = tdb_logerr(tdb, TDB_ERR_OOM, TDB_LOG_ERROR,
 226                                    "tdb_summary: failed to allocate string");
 227                 goto unlock;
 228         }
 229
 230         sprintf(*summary, SUMMARY_FORMAT,
 231                 (size_t)tdb->map_size,
 232                 tally_num(keys) + tally_num(data),
 233                 tally_num(keys),
 234                 tally_min(keys), tally_mean(keys), tally_max(keys),
 235                 keysg ? keysg : "",
 236                 tally_min(data), tally_mean(data), tally_max(data),
 237                 datag ? datag : "",
 238                 tally_min(extra), tally_mean(extra), tally_max(extra),
 239                 extrag ? extrag : "",
 240                 tally_num(freet),
 241                 tally_min(freet), tally_mean(freet), tally_max(freet),
 242                 freeg ? freeg : "",
 243                 tally_total(uncoal, NULL),
 244                 tally_min(uncoal), tally_mean(uncoal), tally_max(uncoal),
 245                 uncoalg ? uncoalg : "",
 246                 tally_num(buckets),
 247                 bucketsg ? bucketsg : "",
 248                 (unsigned)count_hash(tdb, offsetof(struct tdb_header,
 249                                                    hashtable),
 250                                      TDB_TOPLEVEL_HASH_BITS),
 251                 1 << TDB_TOPLEVEL_HASH_BITS,
 252                 tally_num(chains),
 253                 tally_num(hashes),
 254                 tally_min(hashes), tally_mean(hashes), tally_max(hashes),
 255                 hashesg ? hashesg : "",
 256                 tally_total(keys, NULL) * 100.0 / tdb->map_size,
 257                 tally_total(data, NULL) * 100.0 / tdb->map_size,
 258                 tally_total(extra, NULL) * 100.0 / tdb->map_size,
 259                 tally_total(freet, NULL) * 100.0 / tdb->map_size,
 260                 (tally_num(keys) + tally_num(freet) + tally_num(hashes))
 261                 * sizeof(struct tdb_used_record) * 100.0 / tdb->map_size,
 262                 tally_num(ftables) * sizeof(struct tdb_freetable)
 263                 * 100.0 / tdb->map_size,
 264                 (tally_num(hashes)
 265                  * (sizeof(tdb_off_t) << TDB_SUBLEVEL_HASH_BITS)
 266                  + (sizeof(tdb_off_t) << TDB_TOPLEVEL_HASH_BITS)
 267                  + sizeof(struct tdb_chain) * tally_num(chains))
 268                 * 100.0 / tdb->map_size);
 269
 270 unlock:
 271         free(hashesg);
 272         free(freeg);
 273         free(keysg);
 274         free(datag);
 275         free(extrag);
 276         free(uncoalg);
 277         free(bucketsg);
 278         free(hashes);
 279         free(buckets);
 280         free(freet);
 281         free(keys);
 282         free(data);
 283         free(extra);
 284         free(uncoal);
 285         free(ftables);
 286         free(chains);
 287
 288         tdb_allrecord_unlock(tdb, F_RDLCK);
 289         tdb_unlock_expand(tdb, F_RDLCK);
 290         return ecode;
 291 }