tdb2: add write flag to tdb_direct
[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 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                       struct tally *chains)
48 {
49         tdb_off_t off;
50         tdb_len_t len;
51         tdb_len_t unc = 0;
52
53         for (off = sizeof(struct tdb_header); off < tdb->map_size; off += len) {
54                 const union {
55                         struct tdb_used_record u;
56                         struct tdb_free_record f;
57                         struct tdb_recovery_record r;
58                 } *p;
59                 /* We might not be able to get the whole thing. */
60                 p = tdb_access_read(tdb, off, sizeof(p->f), true);
61                 if (!p)
62                         return false;
63                 if (p->r.magic == TDB_RECOVERY_INVALID_MAGIC
64                     || p->r.magic == TDB_RECOVERY_MAGIC) {
65                         if (unc) {
66                                 tally_add(uncoal, unc);
67                                 unc = 0;
68                         }
69                         len = sizeof(p->r) + p->r.max_len;
70                 } else if (frec_magic(&p->f) == TDB_FREE_MAGIC) {
71                         len = frec_len(&p->f);
72                         tally_add(free, len);
73                         tally_add(buckets, size_to_bucket(len));
74                         len += sizeof(p->u);
75                         unc++;
76                 } else if (rec_magic(&p->u) == TDB_MAGIC) {
77                         if (unc) {
78                                 tally_add(uncoal, unc);
79                                 unc = 0;
80                         }
81                         len = sizeof(p->u)
82                                 + rec_key_length(&p->u)
83                                 + rec_data_length(&p->u)
84                                 + rec_extra_padding(&p->u);
85
86                         /* FIXME: Use different magic for hashes, flists. */
87                         if (!rec_key_length(&p->u) && rec_hash(&p->u) < 3) {
88                                 if (rec_hash(&p->u) == 0) {
89                                         int count = count_hash(tdb,
90                                                         off + sizeof(p->u),
91                                                         TDB_SUBLEVEL_HASH_BITS);
92                                         if (count == -1)
93                                                 return false;
94                                         tally_add(hashes, count);
95                                 } else if (rec_hash(&p->u) == 1) {
96                                         tally_add(flists,
97                                                   rec_data_length(&p->u));
98                                 } else if (rec_hash(&p->u) == 2) {
99                                         tally_add(chains, 1);
100                                 }
101                         } else {
102                                 tally_add(keys, rec_key_length(&p->u));
103                                 tally_add(data, rec_data_length(&p->u));
104                         }
105                         tally_add(extra, rec_extra_padding(&p->u));
106                 } else
107                         len = dead_space(tdb, off);
108                 tdb_access_release(tdb, p);
109         }
110         if (unc)
111                 tally_add(uncoal, unc);
112         return true;
113 }
114
115 #define SUMMARY_FORMAT \
116         "Size of file/data: %zu/%zu\n" \
117         "Number of records: %zu\n" \
118         "Smallest/average/largest keys: %zu/%zu/%zu\n%s" \
119         "Smallest/average/largest data: %zu/%zu/%zu\n%s" \
120         "Smallest/average/largest padding: %zu/%zu/%zu\n%s" \
121         "Number of free records: %zu\n" \
122         "Smallest/average/largest free records: %zu/%zu/%zu\n%s" \
123         "Number of uncoalesced records: %zu\n" \
124         "Smallest/average/largest uncoalesced runs: %zu/%zu/%zu\n%s" \
125         "Number of free lists: %zu\n%s" \
126         "Toplevel hash used: %u of %u\n" \
127         "Number of chains: %zu\n" \
128         "Number of subhashes: %zu\n" \
129         "Smallest/average/largest subhash entries: %zu/%zu/%zu\n%s" \
130         "Percentage keys/data/padding/free/rechdrs/freehdrs/hashes: %.0f/%.0f/%.0f/%.0f/%.0f/%.0f/%.0f\n"
131
132 #define BUCKET_SUMMARY_FORMAT_A                                 \
133         "Free bucket %zu: total entries %zu.\n"                 \
134         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
135 #define BUCKET_SUMMARY_FORMAT_B                                 \
136         "Free bucket %zu-%zu: total entries %zu.\n"             \
137         "Smallest/average/largest length: %zu/%zu/%zu\n%s"
138
139 #define HISTO_WIDTH 70
140 #define HISTO_HEIGHT 20
141
142 char *tdb_summary(struct tdb_context *tdb, enum tdb_summary_flags flags)
143 {
144         tdb_len_t len;
145         struct tally *flists, *hashes, *freet, *keys, *data, *extra, *uncoal,
146                 *buckets, *chains;
147         char *hashesg, *freeg, *keysg, *datag, *extrag, *uncoalg, *bucketsg;
148         char *ret = NULL;
149
150         hashesg = freeg = keysg = datag = extrag = uncoalg = bucketsg = NULL;
151
152         if (tdb_allrecord_lock(tdb, F_RDLCK, TDB_LOCK_WAIT, false) != 0)
153                 return NULL;
154
155         if (tdb_lock_expand(tdb, F_RDLCK) != 0) {
156                 tdb_allrecord_unlock(tdb, F_RDLCK);
157                 return NULL;
158         }
159
160         /* Start stats off empty. */
161         flists = tally_new(HISTO_HEIGHT);
162         hashes = tally_new(HISTO_HEIGHT);
163         freet = tally_new(HISTO_HEIGHT);
164         keys = tally_new(HISTO_HEIGHT);
165         data = tally_new(HISTO_HEIGHT);
166         extra = tally_new(HISTO_HEIGHT);
167         uncoal = tally_new(HISTO_HEIGHT);
168         buckets = tally_new(HISTO_HEIGHT);
169         chains = tally_new(HISTO_HEIGHT);
170         if (!flists || !hashes || !freet || !keys || !data || !extra
171             || !uncoal || !buckets || !chains) {
172                 tdb_logerr(tdb, TDB_ERR_OOM, TDB_DEBUG_ERROR,
173                            "tdb_summary: failed to allocate tally structures");
174                 goto unlock;
175         }
176
177         if (!summarize(tdb, hashes, flists, freet, keys, data, extra, uncoal,
178                        buckets, chains))
179                 goto unlock;
180
181         if (flags & TDB_SUMMARY_HISTOGRAMS) {
182                 hashesg = tally_histogram(hashes, HISTO_WIDTH, HISTO_HEIGHT);
183                 freeg = tally_histogram(freet, HISTO_WIDTH, HISTO_HEIGHT);
184                 keysg = tally_histogram(keys, HISTO_WIDTH, HISTO_HEIGHT);
185                 datag = tally_histogram(data, HISTO_WIDTH, HISTO_HEIGHT);
186                 extrag = tally_histogram(extra, HISTO_WIDTH, HISTO_HEIGHT);
187                 uncoalg = tally_histogram(uncoal, HISTO_WIDTH, HISTO_HEIGHT);
188                 bucketsg = tally_histogram(buckets, HISTO_WIDTH, HISTO_HEIGHT);
189         }
190
191         /* 20 is max length of a %llu. */
192         len = strlen(SUMMARY_FORMAT) + 33*20 + 1
193                 + (hashesg ? strlen(hashesg) : 0)
194                 + (freeg ? strlen(freeg) : 0)
195                 + (keysg ? strlen(keysg) : 0)
196                 + (datag ? strlen(datag) : 0)
197                 + (extrag ? strlen(extrag) : 0)
198                 + (uncoalg ? strlen(uncoalg) : 0)
199                 + (bucketsg ? strlen(bucketsg) : 0);
200
201         ret = malloc(len);
202         if (!ret)
203                 goto unlock;
204
205         len = sprintf(ret, SUMMARY_FORMAT,
206                       (size_t)tdb->map_size,
207                       tally_num(keys) + tally_num(data),
208                       tally_num(keys),
209                       tally_min(keys), tally_mean(keys), tally_max(keys),
210                       keysg ? keysg : "",
211                       tally_min(data), tally_mean(data), tally_max(data),
212                       datag ? datag : "",
213                       tally_min(extra), tally_mean(extra), tally_max(extra),
214                       extrag ? extrag : "",
215                       tally_num(freet),
216                       tally_min(freet), tally_mean(freet), tally_max(freet),
217                       freeg ? freeg : "",
218                       tally_total(uncoal, NULL),
219                       tally_min(uncoal), tally_mean(uncoal), tally_max(uncoal),
220                       uncoalg ? uncoalg : "",
221                       tally_num(buckets),
222                       bucketsg ? bucketsg : "",
223                       count_hash(tdb, offsetof(struct tdb_header, hashtable),
224                                  TDB_TOPLEVEL_HASH_BITS),
225                       1 << TDB_TOPLEVEL_HASH_BITS,
226                       tally_num(chains),
227                       tally_num(hashes),
228                       tally_min(hashes), tally_mean(hashes), tally_max(hashes),
229                       hashesg ? hashesg : "",
230                       tally_total(keys, NULL) * 100.0 / tdb->map_size,
231                       tally_total(data, NULL) * 100.0 / tdb->map_size,
232                       tally_total(extra, NULL) * 100.0 / tdb->map_size,
233                       tally_total(freet, NULL) * 100.0 / tdb->map_size,
234                       (tally_num(keys) + tally_num(freet) + tally_num(hashes))
235                       * sizeof(struct tdb_used_record) * 100.0 / tdb->map_size,
236                       tally_num(flists) * sizeof(struct tdb_freelist)
237                       * 100.0 / tdb->map_size,
238                       (tally_num(hashes)
239                        * (sizeof(tdb_off_t) << TDB_SUBLEVEL_HASH_BITS)
240                        + (sizeof(tdb_off_t) << TDB_TOPLEVEL_HASH_BITS)
241                        + sizeof(struct tdb_chain) * tally_num(chains))
242                       * 100.0 / tdb->map_size);
243
244 unlock:
245         free(hashesg);
246         free(freeg);
247         free(keysg);
248         free(datag);
249         free(extrag);
250         free(uncoalg);
251         free(bucketsg);
252         free(hashes);
253         free(buckets);
254         free(freet);
255         free(keys);
256         free(data);
257         free(extra);
258         free(uncoal);
259         free(chains);
260
261         tdb_allrecord_unlock(tdb, F_RDLCK);
262         tdb_unlock_expand(tdb, F_RDLCK);
263         return ret;
264 }