1 /* String pool for GCC.
2 Copyright (C) 2000, 2001 Free Software Foundation, Inc.
4 This file is part of GNU CC.
6 GNU CC is free software; you can redistribute it and/or modify it
7 under the terms of the GNU General Public License as published by the
8 Free Software Foundation; either version 2, or (at your option) any
11 GNU CC is distributed in the hope that it will be useful, but WITHOUT
12 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
13 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
16 You should have received a copy of the GNU General Public License
17 along with GNU CC; see the file COPYING. If not, write to the Free
18 Software Foundation, 59 Temple Place - Suite 330, Boston, MA
21 /* String text, identifer text and identifier node allocator. Strings
22 allocated by ggc_alloc_string are stored in an obstack which is
23 never shrunk. Identifiers are uniquely stored in a hash table.
25 We have our own private hash table implementation. libiberty's
26 hashtab.c is not used because it requires 100% average space
27 overhead per string, which is unacceptable. Also, this algorithm
34 #include "hashtable.h"
38 #define IS_FE_IDENT(NODE) (TREE_CODE (NODE) == IDENTIFIER_NODE)
40 /* The "" allocated string. */
41 const char empty_string[] = "";
43 /* Character strings, each containing a single decimal digit.
44 Written this way to save space. */
45 const char digit_vector[] = {
46 '0', 0, '1', 0, '2', 0, '3', 0, '4', 0,
47 '5', 0, '6', 0, '7', 0, '8', 0, '9', 0
50 struct ht *ident_hash;
51 static struct obstack string_stack;
52 static int do_identifier_warnings;
54 static hashnode alloc_node PARAMS ((hash_table *));
55 static int mark_ident PARAMS ((struct cpp_reader *, hashnode, const PTR));
56 static void mark_ident_hash PARAMS ((void *));
57 static int scan_for_clashes PARAMS ((struct cpp_reader *, hashnode,
60 /* Initialize the string pool. */
64 /* Create with 16K (2^14) entries. */
65 ident_hash = ht_create (14);
66 ident_hash->alloc_node = alloc_node;
67 gcc_obstack_init (&string_stack);
68 ggc_add_root (&ident_hash, 1, sizeof ident_hash, mark_ident_hash);
71 /* Allocate a hash node. */
74 hash_table *table ATTRIBUTE_UNUSED;
76 return GCC_IDENT_TO_HT_IDENT (make_node (IDENTIFIER_NODE));
79 /* Allocate and return a string constant of length LENGTH, containing
80 CONTENTS. If LENGTH is -1, CONTENTS is assumed to be a
81 nul-terminated string, and the length is calculated using strlen.
82 If the same string constant has been allocated before, that copy is
83 returned this time too. */
86 ggc_alloc_string (contents, length)
91 length = strlen (contents);
95 if (length == 1 && contents[0] >= '0' && contents[0] <= '9')
96 return digit_string (contents[0] - '0');
98 obstack_grow0 (&string_stack, contents, length);
99 return obstack_finish (&string_stack);
102 /* NODE is an identifier known to the preprocessor. Make it known to
103 the front ends as well. */
106 make_identifier (node)
109 /* If this identifier is longer than the clash-warning length,
110 do a brute force search of the entire table for clashes. */
111 if (warn_id_clash && do_identifier_warnings
112 && IDENTIFIER_LENGTH (node) >= id_clash_len)
113 ht_forall (ident_hash, (ht_cb) scan_for_clashes,
114 IDENTIFIER_POINTER (node));
116 TREE_SET_CODE (node, IDENTIFIER_NODE);
117 #ifdef GATHER_STATISTICS
118 id_string_size += IDENTIFIER_LENGTH (node);
122 /* Return an IDENTIFIER_NODE whose name is TEXT (a null-terminated string).
123 If an identifier with that name has previously been referred to,
124 the same node is returned this time. */
127 get_identifier (text)
130 hashnode ht_node = ht_lookup (ident_hash,
131 (const unsigned char *) text,
132 strlen (text), HT_ALLOC);
134 /* ht_node can't be NULL here. */
135 return HT_IDENT_TO_GCC_IDENT (ht_node);
138 /* If an identifier with the name TEXT (a null-terminated string) has
139 previously been referred to, return that node; otherwise return
143 maybe_get_identifier (text)
148 size_t length = strlen (text);
150 ht_node = ht_lookup (ident_hash, (const unsigned char *) text,
151 length, HT_NO_INSERT);
154 node = HT_IDENT_TO_GCC_IDENT (ht_node);
155 if (IS_FE_IDENT (node))
162 /* If this identifier is longer than the clash-warning length,
163 do a brute force search of the entire table for clashes. */
166 scan_for_clashes (pfile, h, text)
167 struct cpp_reader *pfile ATTRIBUTE_UNUSED;
171 tree node = HT_IDENT_TO_GCC_IDENT (h);
173 if (IS_FE_IDENT (node)
174 && IDENTIFIER_LENGTH (node) >= id_clash_len
175 && !memcmp (IDENTIFIER_POINTER (node), text, id_clash_len))
177 warning ("\"%s\" and \"%s\" identical in first %d characters",
178 text, IDENTIFIER_POINTER (node), id_clash_len);
185 /* Record the size of an identifier node for the language in use.
186 SIZE is the total size in bytes.
187 This is called by the language-specific files. This must be
188 called before allocating any identifiers. */
191 set_identifier_size (size)
194 tree_code_length[(int) IDENTIFIER_NODE]
195 = (size - sizeof (struct tree_common)) / sizeof (tree);
198 /* Enable warnings on similar identifiers (if requested).
199 Done after the built-in identifiers are created. */
202 start_identifier_warnings ()
204 do_identifier_warnings = 1;
207 /* Report some basic statistics about the string pool. */
210 stringpool_statistics ()
212 ht_dump_statistics (ident_hash);
215 /* Mark an identifier for GC. */
218 mark_ident (pfile, h, v)
219 struct cpp_reader *pfile ATTRIBUTE_UNUSED;
221 const PTR v ATTRIBUTE_UNUSED;
223 ggc_mark_nonnull_tree (HT_IDENT_TO_GCC_IDENT (h));
227 /* Mark all identifiers for GC. */
230 mark_ident_hash (arg)
231 PTR arg ATTRIBUTE_UNUSED;
233 ht_forall (ident_hash, mark_ident, NULL);