26cdd3e4fe2cbe7d171b0a190e11dc0148f0991a
[ccan] / 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 *chains)
49 {
50         tdb_off_t off;
51         tdb_len_t len;
52         tdb_len_t unc = 0;
53
54         for (off = sizeof(struct tdb_header);
55              off < tdb->file->map_size;
56              off += len) {
57                 const union {
58                         struct tdb_used_record u;
59                         struct tdb_free_record f;
60                         struct tdb_recovery_record r;
61                 } *p;
62                 /* We might not be able to get the whole thing. */
63                 p = tdb_access_read(tdb, off, sizeof(p->f), true);
64                 if (TDB_PTR_IS_ERR(p)) {
65                         return TDB_PTR_ERR(p);
66                 }
67                 if (frec_magic(&p->f) != TDB_FREE_MAGIC) {
68                         if (unc > 1) {
69                                 tally_add(uncoal, unc);
70                                 unc = 0;
71                         }
72                 }
73
74                 if (p->r.magic == TDB_RECOVERY_INVALID_MAGIC
75                     || p->r.magic == TDB_RECOVERY_MAGIC) {
76                         len = sizeof(p->r) + p->r.max_len;
77                 } else if (frec_magic(&p->f) == TDB_FREE_MAGIC) {
78                         len = frec_len(&p->f);
79                         tally_add(fr, len);
80                         len += sizeof(p->u);
81                         unc++;
82                 } else if (rec_magic(&p->u) == TDB_USED_MAGIC) {
83                         len = sizeof(p->u)
84                                 + rec_key_length(&p->u)
85                                 + rec_data_length(&p->u)
86                                 + rec_extra_padding(&p->u);
87
88                         tally_add(keys, rec_key_length(&p->u));
89                         tally_add(data, rec_data_length(&p->u));
90                         tally_add(extra, rec_extra_padding(&p->u));
91                 } else if (rec_magic(&p->u) == TDB_HTABLE_MAGIC) {
92                         tdb_off_t count = count_hash(tdb,
93                                                      off + sizeof(p->u),
94                                                      TDB_SUBLEVEL_HASH_BITS);
95                         if (TDB_OFF_IS_ERR(count)) {
96                                 return count;
97                         }
98                         tally_add(hashes, count);
99                         tally_add(extra, rec_extra_padding(&p->u));
100                         len = sizeof(p->u)
101                                 + rec_data_length(&p->u)
102                                 + rec_extra_padding(&p->u);
103                 } else if (rec_magic(&p->u) == TDB_FTABLE_MAGIC) {
104                         len = sizeof(p->u)
105                                 + rec_data_length(&p->u)
106                                 + rec_extra_padding(&p->u);
107                         tally_add(ftables, rec_data_length(&p->u));
108                         tally_add(extra, rec_extra_padding(&p->u));
109                 } else if (rec_magic(&p->u) == TDB_CHAIN_MAGIC) {
110                         len = sizeof(p->u)
111                                 + rec_data_length(&p->u)
112                                 + rec_extra_padding(&p->u);
113                         tally_add(chains, 1);
114                         tally_add(extra, rec_extra_padding(&p->u));
115                 } else {
116                         len = dead_space(tdb, off);
117                         if (TDB_OFF_IS_ERR(len)) {
118                                 return len;
119                         }
120                 }
121                 tdb_access_release(tdb, p);
122         }
123         if (unc)
124                 tally_add(uncoal, unc);
125         return TDB_SUCCESS;
126 }
127
128 #define SUMMARY_FORMAT \
129         "Size of file/data: %zu/%zu\n" \
130         "Number of records: %zu\n" \
131         "Smallest/average/largest keys: %zu/%zu/%zu\n%s" \
132         "Smallest/average/largest data: %zu/%zu/%zu\n%s" \
133         "Smallest/average/largest padding: %zu/%zu/%zu\n%s" \
134         "Number of free records: %zu\n" \
135         "Smallest/average/largest free records: %zu/%zu/%zu\n%s" \
136         "Number of uncoalesced records: %zu\n" \
137         "Smallest/average/largest uncoalesced runs: %zu/%zu/%zu\n%s" \
138         "Toplevel hash used: %u of %u\n" \
139         "Number of chains: %zu\n" \
140         "Number of subhashes: %zu\n" \
141         "Smallest/average/largest subhash entries: %zu/%zu/%zu\n%s" \
142         "Percentage keys/data/padding/free/rechdrs/freehdrs/hashes: %.0f/%.0f/%.0f/%.0f/%.0f/%.0f/%.0f\n"
143
144 #define BUCKET_SUMMARY_FORMAT_A                                 \
145         "Free bucket %zu: total entries %zu.\n"                 \
146         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
147 #define BUCKET_SUMMARY_FORMAT_B                                 \
148         "Free bucket %zu-%zu: total entries %zu.\n"             \
149         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
150
151 #define HISTO_WIDTH 70
152 #define HISTO_HEIGHT 20
153
154 enum TDB_ERROR tdb_summary(struct tdb_context *tdb,
155                            enum tdb_summary_flags flags,
156                            char **summary)
157 {
158         tdb_len_t len;
159         struct tally *ftables, *hashes, *freet, *keys, *data, *extra, *uncoal,
160                 *chains;
161         char *hashesg, *freeg, *keysg, *datag, *extrag, *uncoalg;
162         enum TDB_ERROR ecode;
163
164         hashesg = freeg = keysg = datag = extrag = uncoalg = NULL;
165
166         ecode = tdb_allrecord_lock(tdb, F_RDLCK, TDB_LOCK_WAIT, false);
167         if (ecode != TDB_SUCCESS) {
168                 return tdb->last_error = ecode;
169         }
170
171         ecode = tdb_lock_expand(tdb, F_RDLCK);
172         if (ecode != TDB_SUCCESS) {
173                 tdb_allrecord_unlock(tdb, F_RDLCK);
174                 return tdb->last_error = ecode;
175         }
176
177         /* Start stats off empty. */
178         ftables = tally_new(HISTO_HEIGHT);
179         hashes = tally_new(HISTO_HEIGHT);
180         freet = tally_new(HISTO_HEIGHT);
181         keys = tally_new(HISTO_HEIGHT);
182         data = tally_new(HISTO_HEIGHT);
183         extra = tally_new(HISTO_HEIGHT);
184         uncoal = tally_new(HISTO_HEIGHT);
185         chains = tally_new(HISTO_HEIGHT);
186         if (!ftables || !hashes || !freet || !keys || !data || !extra
187             || !uncoal || !chains) {
188                 ecode = tdb_logerr(tdb, TDB_ERR_OOM, TDB_LOG_ERROR,
189                                    "tdb_summary: failed to allocate"
190                                    " tally structures");
191                 goto unlock;
192         }
193
194         ecode = summarize(tdb, hashes, ftables, freet, keys, data, extra,
195                           uncoal, chains);
196         if (ecode != TDB_SUCCESS) {
197                 goto unlock;
198         }
199
200         if (flags & TDB_SUMMARY_HISTOGRAMS) {
201                 hashesg = tally_histogram(hashes, HISTO_WIDTH, HISTO_HEIGHT);
202                 freeg = tally_histogram(freet, HISTO_WIDTH, HISTO_HEIGHT);
203                 keysg = tally_histogram(keys, HISTO_WIDTH, HISTO_HEIGHT);
204                 datag = tally_histogram(data, HISTO_WIDTH, HISTO_HEIGHT);
205                 extrag = tally_histogram(extra, HISTO_WIDTH, HISTO_HEIGHT);
206                 uncoalg = tally_histogram(uncoal, HISTO_WIDTH, HISTO_HEIGHT);
207         }
208
209         /* 20 is max length of a %llu. */
210         len = strlen(SUMMARY_FORMAT) + 33*20 + 1
211                 + (hashesg ? strlen(hashesg) : 0)
212                 + (freeg ? strlen(freeg) : 0)
213                 + (keysg ? strlen(keysg) : 0)
214                 + (datag ? strlen(datag) : 0)
215                 + (extrag ? strlen(extrag) : 0)
216                 + (uncoalg ? strlen(uncoalg) : 0);
217
218         *summary = malloc(len);
219         if (!*summary) {
220                 ecode = tdb_logerr(tdb, TDB_ERR_OOM, TDB_LOG_ERROR,
221                                    "tdb_summary: failed to allocate string");
222                 goto unlock;
223         }
224
225         sprintf(*summary, SUMMARY_FORMAT,
226                 (size_t)tdb->file->map_size,
227                 tally_total(keys, NULL) + tally_total(data, NULL),
228                 tally_num(keys),
229                 tally_min(keys), tally_mean(keys), tally_max(keys),
230                 keysg ? keysg : "",
231                 tally_min(data), tally_mean(data), tally_max(data),
232                 datag ? datag : "",
233                 tally_min(extra), tally_mean(extra), tally_max(extra),
234                 extrag ? extrag : "",
235                 tally_num(freet),
236                 tally_min(freet), tally_mean(freet), tally_max(freet),
237                 freeg ? freeg : "",
238                 tally_total(uncoal, NULL),
239                 tally_min(uncoal), tally_mean(uncoal), tally_max(uncoal),
240                 uncoalg ? uncoalg : "",
241                 (unsigned)count_hash(tdb, offsetof(struct tdb_header,
242                                                    hashtable),
243                                      TDB_TOPLEVEL_HASH_BITS),
244                 1 << TDB_TOPLEVEL_HASH_BITS,
245                 tally_num(chains),
246                 tally_num(hashes),
247                 tally_min(hashes), tally_mean(hashes), tally_max(hashes),
248                 hashesg ? hashesg : "",
249                 tally_total(keys, NULL) * 100.0 / tdb->file->map_size,
250                 tally_total(data, NULL) * 100.0 / tdb->file->map_size,
251                 tally_total(extra, NULL) * 100.0 / tdb->file->map_size,
252                 tally_total(freet, NULL) * 100.0 / tdb->file->map_size,
253                 (tally_num(keys) + tally_num(freet) + tally_num(hashes))
254                 * sizeof(struct tdb_used_record) * 100.0 / tdb->file->map_size,
255                 tally_num(ftables) * sizeof(struct tdb_freetable)
256                 * 100.0 / tdb->file->map_size,
257                 (tally_num(hashes)
258                  * (sizeof(tdb_off_t) << TDB_SUBLEVEL_HASH_BITS)
259                  + (sizeof(tdb_off_t) << TDB_TOPLEVEL_HASH_BITS)
260                  + sizeof(struct tdb_chain) * tally_num(chains))
261                 * 100.0 / tdb->file->map_size);
262
263 unlock:
264         free(hashesg);
265         free(freeg);
266         free(keysg);
267         free(datag);
268         free(extrag);
269         free(uncoalg);
270         free(hashes);
271         free(freet);
272         free(keys);
273         free(data);
274         free(extra);
275         free(uncoal);
276         free(ftables);
277         free(chains);
278
279         tdb_allrecord_unlock(tdb, F_RDLCK);
280         tdb_unlock_expand(tdb, F_RDLCK);
281         return tdb->last_error = ecode;
282 }