Commit | Line | Data |
---|---|---|
c906108c | 1 | /* Implement a cached obstack. |
c2d11a7d JM |
2 | Written by Fred Fish <fnf@cygnus.com> |
3 | Rewritten by Jim Blandy <jimb@cygnus.com> | |
2c7ef074 AC |
4 | |
5 | Copyright 1999, 2000, 2002 Free Software Foundation, Inc. | |
c906108c | 6 | |
c5aa993b | 7 | This file is part of GDB. |
c906108c | 8 | |
c5aa993b JM |
9 | This program is free software; you can redistribute it and/or modify |
10 | it under the terms of the GNU General Public License as published by | |
11 | the Free Software Foundation; either version 2 of the License, or | |
12 | (at your option) any later version. | |
c906108c | 13 | |
c5aa993b JM |
14 | This program is distributed in the hope that it will be useful, |
15 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
16 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
17 | GNU General Public License for more details. | |
c906108c | 18 | |
c5aa993b JM |
19 | You should have received a copy of the GNU General Public License |
20 | along with this program; if not, write to the Free Software | |
21 | Foundation, Inc., 59 Temple Place - Suite 330, | |
22 | Boston, MA 02111-1307, USA. */ | |
c906108c SS |
23 | |
24 | #include "defs.h" | |
25 | #include "obstack.h" | |
26 | #include "bcache.h" | |
27 | #include "gdb_string.h" /* For memcpy declaration */ | |
28 | ||
2c7ef074 AC |
29 | #include <stddef.h> |
30 | #include <stdlib.h> | |
31 | ||
357e46e7 DB |
32 | /* The old hash function was stolen from SDBM. This is what DB 3.0 uses now, |
33 | * and is better than the old one. | |
34 | */ | |
c2d11a7d | 35 | \f |
c2d11a7d | 36 | unsigned long |
d85a5daf | 37 | hash(const void *addr, int length) |
c2d11a7d | 38 | { |
357e46e7 DB |
39 | const unsigned char *k, *e; |
40 | unsigned long h; | |
41 | ||
42 | k = (const unsigned char *)addr; | |
43 | e = k+length; | |
44 | for (h=0; k< e;++k) | |
45 | { | |
46 | h *=16777619; | |
47 | h ^= *k; | |
48 | } | |
49 | return (h); | |
c906108c | 50 | } |
c2d11a7d JM |
51 | \f |
52 | /* Growing the bcache's hash table. */ | |
53 | ||
54 | /* If the average chain length grows beyond this, then we want to | |
55 | resize our hash table. */ | |
56 | #define CHAIN_LENGTH_THRESHOLD (5) | |
57 | ||
58 | static void | |
59 | expand_hash_table (struct bcache *bcache) | |
c906108c | 60 | { |
c2d11a7d JM |
61 | /* A table of good hash table sizes. Whenever we grow, we pick the |
62 | next larger size from this table. sizes[i] is close to 1 << (i+10), | |
63 | so we roughly double the table size each time. After we fall off | |
64 | the end of this table, we just double. Don't laugh --- there have | |
65 | been executables sighted with a gigabyte of debug info. */ | |
66 | static unsigned long sizes[] = { | |
67 | 1021, 2053, 4099, 8191, 16381, 32771, | |
68 | 65537, 131071, 262144, 524287, 1048573, 2097143, | |
69 | 4194301, 8388617, 16777213, 33554467, 67108859, 134217757, | |
70 | 268435459, 536870923, 1073741827, 2147483659UL | |
71 | }; | |
745b8ca0 | 72 | unsigned int new_num_buckets; |
c2d11a7d | 73 | struct bstring **new_buckets; |
745b8ca0 | 74 | unsigned int i; |
c2d11a7d JM |
75 | |
76 | /* Find the next size. */ | |
745b8ca0 | 77 | new_num_buckets = bcache->num_buckets * 2; |
c2d11a7d JM |
78 | for (i = 0; i < (sizeof (sizes) / sizeof (sizes[0])); i++) |
79 | if (sizes[i] > bcache->num_buckets) | |
80 | { | |
81 | new_num_buckets = sizes[i]; | |
82 | break; | |
83 | } | |
c2d11a7d JM |
84 | |
85 | /* Allocate the new table. */ | |
86 | { | |
87 | size_t new_size = new_num_buckets * sizeof (new_buckets[0]); | |
88 | new_buckets = (struct bstring **) xmalloc (new_size); | |
89 | memset (new_buckets, 0, new_size); | |
90 | ||
91 | bcache->structure_size -= (bcache->num_buckets | |
92 | * sizeof (bcache->bucket[0])); | |
93 | bcache->structure_size += new_size; | |
94 | } | |
c906108c | 95 | |
c2d11a7d JM |
96 | /* Rehash all existing strings. */ |
97 | for (i = 0; i < bcache->num_buckets; i++) | |
c906108c | 98 | { |
c2d11a7d JM |
99 | struct bstring *s, *next; |
100 | ||
101 | for (s = bcache->bucket[i]; s; s = next) | |
c906108c | 102 | { |
c2d11a7d JM |
103 | struct bstring **new_bucket; |
104 | next = s->next; | |
105 | ||
106 | new_bucket = &new_buckets[(hash (&s->d.data, s->length) | |
107 | % new_num_buckets)]; | |
108 | s->next = *new_bucket; | |
109 | *new_bucket = s; | |
c906108c SS |
110 | } |
111 | } | |
c2d11a7d JM |
112 | |
113 | /* Plug in the new table. */ | |
114 | if (bcache->bucket) | |
b8c9b27d | 115 | xfree (bcache->bucket); |
c2d11a7d JM |
116 | bcache->bucket = new_buckets; |
117 | bcache->num_buckets = new_num_buckets; | |
c906108c SS |
118 | } |
119 | ||
c2d11a7d JM |
120 | \f |
121 | /* Looking up things in the bcache. */ | |
122 | ||
123 | /* The number of bytes needed to allocate a struct bstring whose data | |
124 | is N bytes long. */ | |
125 | #define BSTRING_SIZE(n) (offsetof (struct bstring, d.data) + (n)) | |
126 | ||
127 | /* Find a copy of the LENGTH bytes at ADDR in BCACHE. If BCACHE has | |
128 | never seen those bytes before, add a copy of them to BCACHE. In | |
129 | either case, return a pointer to BCACHE's copy of that string. */ | |
c906108c | 130 | void * |
d85a5daf | 131 | bcache (const void *addr, int length, struct bcache *bcache) |
c906108c | 132 | { |
c2d11a7d JM |
133 | int hash_index; |
134 | struct bstring *s; | |
c906108c | 135 | |
c2d11a7d JM |
136 | /* If our average chain length is too high, expand the hash table. */ |
137 | if (bcache->unique_count >= bcache->num_buckets * CHAIN_LENGTH_THRESHOLD) | |
138 | expand_hash_table (bcache); | |
139 | ||
140 | bcache->total_count++; | |
141 | bcache->total_size += length; | |
142 | ||
143 | hash_index = hash (addr, length) % bcache->num_buckets; | |
144 | ||
145 | /* Search the hash bucket for a string identical to the caller's. */ | |
146 | for (s = bcache->bucket[hash_index]; s; s = s->next) | |
147 | if (s->length == length | |
148 | && ! memcmp (&s->d.data, addr, length)) | |
149 | return &s->d.data; | |
150 | ||
151 | /* The user's string isn't in the list. Insert it after *ps. */ | |
152 | { | |
153 | struct bstring *new | |
154 | = obstack_alloc (&bcache->cache, BSTRING_SIZE (length)); | |
155 | memcpy (&new->d.data, addr, length); | |
156 | new->length = length; | |
157 | new->next = bcache->bucket[hash_index]; | |
158 | bcache->bucket[hash_index] = new; | |
159 | ||
160 | bcache->unique_count++; | |
161 | bcache->unique_size += length; | |
162 | bcache->structure_size += BSTRING_SIZE (length); | |
163 | ||
164 | return &new->d.data; | |
165 | } | |
c906108c SS |
166 | } |
167 | ||
c2d11a7d JM |
168 | \f |
169 | /* Freeing bcaches. */ | |
170 | ||
171 | /* Free all the storage associated with BCACHE. */ | |
c906108c | 172 | void |
c2d11a7d | 173 | free_bcache (struct bcache *bcache) |
c906108c | 174 | { |
c2d11a7d | 175 | obstack_free (&bcache->cache, 0); |
df02e9ed | 176 | if (bcache->bucket) |
b8c9b27d | 177 | xfree (bcache->bucket); |
c906108c | 178 | |
c2d11a7d JM |
179 | /* This isn't necessary, but at least the bcache is always in a |
180 | consistent state. */ | |
181 | memset (bcache, 0, sizeof (*bcache)); | |
182 | } | |
183 | ||
184 | ||
185 | \f | |
186 | /* Printing statistics. */ | |
187 | ||
188 | static int | |
189 | compare_ints (const void *ap, const void *bp) | |
190 | { | |
191 | /* Because we know we're comparing two ints which are positive, | |
192 | there's no danger of overflow here. */ | |
193 | return * (int *) ap - * (int *) bp; | |
194 | } | |
195 | ||
196 | ||
197 | static void | |
198 | print_percentage (int portion, int total) | |
199 | { | |
200 | if (total == 0) | |
201 | printf_filtered ("(not applicable)\n"); | |
c906108c | 202 | else |
c2d11a7d JM |
203 | printf_filtered ("%3d%%\n", portion * 100 / total); |
204 | } | |
205 | ||
206 | ||
207 | /* Print statistics on BCACHE's memory usage and efficacity at | |
208 | eliminating duplication. NAME should describe the kind of data | |
209 | BCACHE holds. Statistics are printed using `printf_filtered' and | |
210 | its ilk. */ | |
211 | void | |
212 | print_bcache_statistics (struct bcache *c, char *type) | |
213 | { | |
214 | int occupied_buckets; | |
215 | int max_chain_length; | |
216 | int median_chain_length; | |
217 | ||
218 | /* Count the number of occupied buckets, and measure chain lengths. */ | |
219 | { | |
745b8ca0 | 220 | unsigned int b; |
c2d11a7d JM |
221 | int *chain_length |
222 | = (int *) alloca (c->num_buckets * sizeof (*chain_length)); | |
223 | ||
224 | occupied_buckets = 0; | |
225 | ||
226 | for (b = 0; b < c->num_buckets; b++) | |
227 | { | |
228 | struct bstring *s = c->bucket[b]; | |
229 | ||
230 | chain_length[b] = 0; | |
231 | ||
232 | if (s) | |
233 | { | |
234 | occupied_buckets++; | |
235 | ||
236 | while (s) | |
237 | { | |
238 | chain_length[b]++; | |
239 | s = s->next; | |
240 | } | |
241 | } | |
242 | } | |
243 | ||
244 | /* To compute the median, we need the set of chain lengths sorted. */ | |
245 | qsort (chain_length, c->num_buckets, sizeof (chain_length[0]), | |
246 | compare_ints); | |
247 | ||
248 | if (c->num_buckets > 0) | |
249 | { | |
250 | max_chain_length = chain_length[c->num_buckets - 1]; | |
251 | median_chain_length = chain_length[c->num_buckets / 2]; | |
252 | } | |
253 | else | |
254 | { | |
255 | max_chain_length = 0; | |
256 | median_chain_length = 0; | |
257 | } | |
258 | } | |
259 | ||
260 | printf_filtered (" Cached '%s' statistics:\n", type); | |
261 | printf_filtered (" Total object count: %ld\n", c->total_count); | |
745b8ca0 | 262 | printf_filtered (" Unique object count: %lu\n", c->unique_count); |
c2d11a7d JM |
263 | printf_filtered (" Percentage of duplicates, by count: "); |
264 | print_percentage (c->total_count - c->unique_count, c->total_count); | |
265 | printf_filtered ("\n"); | |
266 | ||
267 | printf_filtered (" Total object size: %ld\n", c->total_size); | |
268 | printf_filtered (" Unique object size: %ld\n", c->unique_size); | |
269 | printf_filtered (" Percentage of duplicates, by size: "); | |
270 | print_percentage (c->total_size - c->unique_size, c->total_size); | |
271 | printf_filtered ("\n"); | |
272 | ||
273 | printf_filtered (" Total memory used by bcache, including overhead: %ld\n", | |
274 | c->structure_size); | |
275 | printf_filtered (" Percentage memory overhead: "); | |
276 | print_percentage (c->structure_size - c->unique_size, c->unique_size); | |
277 | printf_filtered (" Net memory savings: "); | |
278 | print_percentage (c->total_size - c->structure_size, c->total_size); | |
279 | printf_filtered ("\n"); | |
280 | ||
281 | printf_filtered (" Hash table size: %3d\n", c->num_buckets); | |
282 | printf_filtered (" Hash table population: "); | |
283 | print_percentage (occupied_buckets, c->num_buckets); | |
284 | printf_filtered (" Median hash chain length: %3d\n", | |
285 | median_chain_length); | |
286 | printf_filtered (" Average hash chain length: "); | |
287 | if (c->num_buckets > 0) | |
745b8ca0 | 288 | printf_filtered ("%3lu\n", c->unique_count / c->num_buckets); |
c906108c | 289 | else |
c2d11a7d JM |
290 | printf_filtered ("(not applicable)\n"); |
291 | printf_filtered (" Maximum hash chain length: %3d\n", max_chain_length); | |
292 | printf_filtered ("\n"); | |
c906108c | 293 | } |