| 1 | /* Include file cached obstack implementation. |
| 2 | Written by Fred Fish <fnf@cygnus.com> |
| 3 | Rewritten by Jim Blandy <jimb@cygnus.com> |
| 4 | Copyright 1999, 2000 Free Software Foundation, Inc. |
| 5 | |
| 6 | This file is part of GDB. |
| 7 | |
| 8 | This program is free software; you can redistribute it and/or modify |
| 9 | it under the terms of the GNU General Public License as published by |
| 10 | the Free Software Foundation; either version 2 of the License, or |
| 11 | (at your option) any later version. |
| 12 | |
| 13 | This program is distributed in the hope that it will be useful, |
| 14 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 15 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 16 | GNU General Public License for more details. |
| 17 | |
| 18 | You should have received a copy of the GNU General Public License |
| 19 | along with this program; if not, write to the Free Software |
| 20 | Foundation, Inc., 59 Temple Place - Suite 330, |
| 21 | Boston, MA 02111-1307, USA. */ |
| 22 | |
| 23 | #ifndef BCACHE_H |
| 24 | #define BCACHE_H 1 |
| 25 | |
| 26 | /* A bcache is a data structure for factoring out duplication in |
| 27 | read-only structures. You give the bcache some string of bytes S. |
| 28 | If the bcache already contains a copy of S, it hands you back a |
| 29 | pointer to its copy. Otherwise, it makes a fresh copy of S, and |
| 30 | hands you back a pointer to that. In either case, you can throw |
| 31 | away your copy of S, and use the bcache's. |
| 32 | |
| 33 | The "strings" in question are arbitrary strings of bytes --- they |
| 34 | can contain zero bytes. You pass in the length explicitly when you |
| 35 | call the bcache function. |
| 36 | |
| 37 | This means that you can put ordinary C objects in a bcache. |
| 38 | However, if you do this, remember that structs can contain `holes' |
| 39 | between members, added for alignment. These bytes usually contain |
| 40 | garbage. If you try to bcache two objects which are identical from |
| 41 | your code's point of view, but have different garbage values in the |
| 42 | structure's holes, then the bcache will treat them as separate |
| 43 | strings, and you won't get the nice elimination of duplicates you |
| 44 | were hoping for. So, remember to memset your structures full of |
| 45 | zeros before bcaching them! |
| 46 | |
| 47 | You shouldn't modify the strings you get from a bcache, because: |
| 48 | |
| 49 | - You don't necessarily know who you're sharing space with. If I |
| 50 | stick eight bytes of text in a bcache, and then stick an |
| 51 | eight-byte structure in the same bcache, there's no guarantee |
| 52 | those two objects don't actually comprise the same sequence of |
| 53 | bytes. If they happen to, the bcache will use a single byte |
| 54 | string for both of them. Then, modifying the structure will |
| 55 | change the string. In bizarre ways. |
| 56 | |
| 57 | - Even if you know for some other reason that all that's okay, |
| 58 | there's another problem. A bcache stores all its strings in a |
| 59 | hash table. If you modify a string's contents, you will probably |
| 60 | change its hash value. This means that the modified string is |
| 61 | now in the wrong place in the hash table, and future bcache |
| 62 | probes will never find it. So by mutating a string, you give up |
| 63 | any chance of sharing its space with future duplicates. */ |
| 64 | |
| 65 | |
| 66 | /* The type used to hold a single bcache string. The user data is |
| 67 | stored in d.data. Since it can be any type, it needs to have the |
| 68 | same alignment as the most strict alignment of any type on the host |
| 69 | machine. I don't know of any really correct way to do this in |
| 70 | stock ANSI C, so just do it the same way obstack.h does. |
| 71 | |
| 72 | It would be nicer to have this stuff hidden away in bcache.c, but |
| 73 | struct objstack contains a struct bcache directly --- not a pointer |
| 74 | to one --- and then the memory-mapped stuff makes this a real pain. |
| 75 | We don't strictly need to expose struct bstring, but it's better to |
| 76 | have it all in one place. */ |
| 77 | |
| 78 | struct bstring { |
| 79 | struct bstring *next; |
| 80 | size_t length; |
| 81 | |
| 82 | union |
| 83 | { |
| 84 | char data[1]; |
| 85 | double dummy; |
| 86 | } |
| 87 | d; |
| 88 | }; |
| 89 | |
| 90 | |
| 91 | /* The structure for a bcache itself. |
| 92 | To initialize a bcache, just fill it with zeros. */ |
| 93 | struct bcache { |
| 94 | /* All the bstrings are allocated here. */ |
| 95 | struct obstack cache; |
| 96 | |
| 97 | /* How many hash buckets we're using. */ |
| 98 | unsigned int num_buckets; |
| 99 | |
| 100 | /* Hash buckets. This table is allocated using malloc, so when we |
| 101 | grow the table we can return the old table to the system. */ |
| 102 | struct bstring **bucket; |
| 103 | |
| 104 | /* Statistics. */ |
| 105 | unsigned long unique_count; /* number of unique strings */ |
| 106 | long total_count; /* total number of strings cached, including dups */ |
| 107 | long unique_size; /* size of unique strings, in bytes */ |
| 108 | long total_size; /* total number of bytes cached, including dups */ |
| 109 | long structure_size; /* total size of bcache, including infrastructure */ |
| 110 | }; |
| 111 | |
| 112 | |
| 113 | /* Find a copy of the LENGTH bytes at ADDR in BCACHE. If BCACHE has |
| 114 | never seen those bytes before, add a copy of them to BCACHE. In |
| 115 | either case, return a pointer to BCACHE's copy of that string. */ |
| 116 | extern void *bcache (const void *addr, int length, struct bcache *bcache); |
| 117 | |
| 118 | /* Free all the storage that BCACHE refers to. The result is a valid, |
| 119 | but empty, bcache. This does not free BCACHE itself, since that |
| 120 | might be part of some larger object. */ |
| 121 | extern void free_bcache (struct bcache *bcache); |
| 122 | |
| 123 | /* Print statistics on BCACHE's memory usage and efficacity at |
| 124 | eliminating duplication. TYPE should be a string describing the |
| 125 | kind of data BCACHE holds. Statistics are printed using |
| 126 | `printf_filtered' and its ilk. */ |
| 127 | extern void print_bcache_statistics (struct bcache *bcache, char *type); |
| 128 | /* The hash function */ |
| 129 | extern unsigned long hash(const void *addr, int length); |
| 130 | #endif /* BCACHE_H */ |