Mercurial > hg > xemacs-beta

comparison src/lrecord.h @ 0:376386a54a3c r19-14

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Import from CVS: tag r19-14
author cvs
date Mon, 13 Aug 2007 08:45:50 +0200
parents
children ac2d302a0011
comparison
equal deleted inserted replaced
-1:000000000000 0:376386a54a3c
1 /* The "lrecord" structure (header of a compound lisp object).
2 Copyright (C) 1993, 1994, 1995 Free Software Foundation, Inc.
3 Copyright (C) 1996 Ben Wing.
4
5 This file is part of XEmacs.
6
7 XEmacs is free software; you can redistribute it and/or modify it
8 under the terms of the GNU General Public License as published by the
9 Free Software Foundation; either version 2, or (at your option) any
10 later version.
11
12 XEmacs is distributed in the hope that it will be useful, but WITHOUT
13 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
15 for more details.
16
17 You should have received a copy of the GNU General Public License
18 along with XEmacs; see the file COPYING. If not, write to
19 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
20 Boston, MA 02111-1307, USA. */
21
22 /* Synched up with: Not in FSF. */
23
24 #ifndef _XEMACS_LRECORD_H_
25 #define _XEMACS_LRECORD_H_
26
27 /* The "lrecord" type of Lisp object is used for all object types
28 other than a few simple ones. This allows many types to be
29 implemented but only a few bits required in a Lisp object for
30 type information. (The tradeoff is that each object has its
31 type marked in it, thereby increasing its size.) The first
32 four bytes of all lrecords is a pointer to a struct
33 lrecord_implementation, which contains methods describing
34 how to process this object.
35
36 lrecords are of two types: straight lrecords, and lcrecords.
37 Straight lrecords are used for those types of objects that
38 have their own allocation routines (typically allocated out
39 of 2K chunks of memory). These objects have a
40 `struct lrecord_header' at the top, containing only the
41 implementation pointer. There are special routines in alloc.c
42 to deal with each such object type.
43
44 Lcrecords are used for less common sorts of objects that don't
45 do their own allocation. Each such object is malloc()ed
46 individually, and the objects are chained together through
47 a `next' pointer. Lcrecords have a `struct lcrecord_header'
48 at the top, which contains an implementation pointer and
49 a `next' pointer, and are allocated using alloc_lcrecord().
50
51 Creating a new lcrecord type is fairly easy; just follow the
52 lead of some existing type (e.g. hashtables). Note that you
53 do not need to supply all the methods (see below); reasonable
54 defaults are provided for many of them. Alternatively, if you're
55 just looking for a way of encapsulating data (which possibly
56 could contain Lisp_Objects in it), you may well be able to use
57 the opaque type. */
58
59 struct lrecord_header
60 {
61 /* It would be better to put the mark-bit together with the
62 * following datatype identification field in an 8- or 16-bit integer
63 * rather than playing funny games with changing header->implementation
64 * and "wasting" 32 bits on the below pointer.
65 * The type-id would then be a 7 or 15
66 * bit index into a table of lrecord-implementations rather than a
67 * direct pointer. There would be 24 (or 16) bits left over for
68 * datatype-specific per-instance flags.
69 * The below is the simplest thing to do for the present,
70 * and doesn't incur that much overhead as most Emacs records
71 * are of such a size that the overhead isn't too bad.
72 * (The marker datatype is the worst case.)
73 * It also has the very very very slight advantage that type-checking
74 * involves one memory read (of the "implementation" slot) and a
75 * comparison against a link-time constant address rather than a
76 * read and a comparison against a variable value. (Variable since
77 * it is a very good idea to assign the indices into the hypothetical
78 * type-code table dynamically rather that pre-defining them.)
79 * I think I remember that Elk Lisp does something like this.
80 * Gee, I wonder if some cretin has patented it?
81 */
82 CONST struct lrecord_implementation *implementation;
83 };
84 #define set_lheader_implementation(header,imp) (header)->implementation=(imp)
85
86 struct lcrecord_header
87 {
88 struct lrecord_header lheader;
89 /* The "next" field is normally used to chain all lrecords together
90 * so that the GC can find (and free) all of them.
91 * "alloc_lcrecord" threads records together.
92 * The "next" field may be used for other purposes as long as some
93 * other mechanism is provided for letting the GC do its work.
94 * (For example, the event and marker datatypes allocates members out
95 * of memory chunks, and it are able to find all unmarked
96 * events by sweeping through the elements of the list of chunks)
97 */
98 struct lcrecord_header *next;
99 /* This is just for debugging/printing convenience.
100 Having this slot doesn't hurt us much spacewise, since an lcrecord
101 already has the above slots together with malloc overhead. */
102 int uid :31;
103 /* A flag that indicates whether this lcrecord is on a "free list".
104 Free lists are used to minimize the number of calls to malloc()
105 when we're repeatedly allocating and freeing a number of the
106 same sort of lcrecord. Lcrecords on a free list always get
107 marked in a different fashion, so we can use this flag as a
108 sanity check to make sure that free lists only have freed lcrecords
109 and no freed lcrecords are elsewhere. */
110 int free :1;
111 };
112
113 /* Used for lcrecords in an lcrecord-list. */
114 struct free_lcrecord_header
115 {
116 struct lcrecord_header lcheader;
117 Lisp_Object chain;
118 };
119
120 /* This as the value of lheader->implementation->finalizer
121 * means that this record is already marked */
122 extern void this_marks_a_marked_record (void *, int);
123
124 /* see alloc.c for an explanation */
125 extern Lisp_Object this_one_is_unmarkable (Lisp_Object obj,
126 void (*markobj) (Lisp_Object));
127
128 struct lrecord_implementation
129 {
130 CONST char *name;
131 /* This function is called at GC time, to make sure that all Lisp_Objects
132 pointed to by this object get properly marked. It should call
133 the mark_object function on all Lisp_Objects in the object. If
134 the return value is non-nil, it should be a Lisp_Object to be
135 marked (don't call the mark_object function explicitly on it,
136 because the GC routines will do this). Doing it this way reduces
137 recursion, so the object returned should preferably be the one
138 with the deepest level of Lisp_Object pointers. This function
139 can be NULL, meaning no GC marking is necessary. */
140 Lisp_Object (*marker) (Lisp_Object, void (*mark_object) (Lisp_Object));
141 /* This can be NULL if the object is an lcrecord; the
142 default_object_printer() in print.c will be used. */
143 void (*printer) (Lisp_Object, Lisp_Object printcharfun, int escapeflag);
144 /* This function is called at GC time when the object is about to
145 be freed, and at dump time (FOR_DISKSAVE will be non-zero in this
146 case). It should perform any necessary cleanup (e.g. freeing
147 malloc()ed memory. This can be NULL, meaning no special
148 finalization is necessary.
149
150 WARNING: remember that the finalizer is called at dump time even
151 though the object is not being freed. */
152 void (*finalizer) (void *header, int for_disksave);
153 /* This can be NULL, meaning compare objects with EQ(). */
154 int (*equal) (Lisp_Object obj1, Lisp_Object obj2, int depth);
155 /* This can be NULL, meaning use the Lisp_Object itself as the hash;
156 but *only* if the `equal' function is EQ (if two objects are
157 `equal', they *must* hash to the same value or the hashing won't
158 work). */
159 unsigned long (*hash) (Lisp_Object, int);
160 Lisp_Object (*getprop) (Lisp_Object obj, Lisp_Object prop);
161 int (*putprop) (Lisp_Object obj, Lisp_Object prop, Lisp_Object val);
162 int (*remprop) (Lisp_Object obj, Lisp_Object prop);
163 Lisp_Object (*plist) (Lisp_Object obj);
164
165 /* Only one of these is non-0. If both are 0, it means that this type
166 is not instantiable by alloc_lcrecord(). */
167 unsigned int static_size;
168 unsigned int (*size_in_bytes_method) (CONST void *header);
169 /* A unique subtag-code (dynamically) assigned to this datatype. */
170 /* (This is a pointer so the rest of this structure can be read-only.) */
171 int *lrecord_type_index;
172 /* A "basic" lrecord is any lrecord that's not an lcrecord, i.e.
173 one that does not have an lcrecord_header at the front and which
174 is (usually) allocated in frob blocks. We only use this flag
175 for some consistency checking, and that only when error-checking
176 is enabled. */
177 int basic_p;
178 };
179
180 extern int gc_in_progress;
181
182 #define MARKED_RECORD_P(obj) (gc_in_progress && \
183 XRECORD_LHEADER (obj)->implementation->finalizer == \
184 this_marks_a_marked_record)
185
186 /* Declaring the following structures as const puts them in the
187 text (read-only) segment, which makes debugging inconvenient
188 because this segment is not mapped when processing a core-
189 dump file */
190
191 #ifdef DEBUG_XEMACS
192 #define CONST_IF_NOT_DEBUG
193 #else
194 #define CONST_IF_NOT_DEBUG CONST
195 #endif
196
197 /* DEFINE_LRECORD_IMPLEMENTATION is for objects with constant size.
198 DEFINE_LRECORD_SEQUENCE_IMPLEMENTATION is for objects whose size varies.
199 */
200
201 #if defined (ERROR_CHECK_TYPECHECK)
202 # define DECLARE_ERROR_CHECK_TYPECHECK(c_name, structtype)
203 #else
204 # define DECLARE_ERROR_CHECK_TYPECHECK(c_name, structtype)
205 #endif
206
207 #define DEFINE_BASIC_LRECORD_IMPLEMENTATION(name,c_name,marker,printer,nuker,equal,hash,structtype) \
208 DECLARE_ERROR_CHECK_TYPECHECK(c_name, structtype) \
209 static int lrecord_##c_name##_lrecord_type_index; \
210 CONST_IF_NOT_DEBUG struct lrecord_implementation lrecord_##c_name[2] = \
211 { { name, marker, printer, nuker, equal, hash, \
212 0, 0, 0, 0, sizeof (structtype), 0, \
213 &(lrecord_##c_name##_lrecord_type_index), 1 }, \
214 { 0, 0, 0, this_marks_a_marked_record, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1 } }
215
216 #define DEFINE_BASIC_LRECORD_IMPLEMENTATION_WITH_PROPS(name,c_name,marker,printer,nuker,equal,hash,getprop,putprop,remprop,props,structtype) \
217 DECLARE_ERROR_CHECK_TYPECHECK(c_name, structtype) \
218 static int lrecord_##c_name##_lrecord_type_index; \
219 CONST_IF_NOT_DEBUG struct lrecord_implementation lrecord_##c_name[2] = \
220 { { name, marker, printer, nuker, equal, hash, \
221 getprop, putprop, remprop, props, sizeof (structtype), 0, \
222 &(lrecord_##c_name##_lrecord_type_index), 1 }, \
223 { 0, 0, 0, this_marks_a_marked_record, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1 } }
224
225 #define DEFINE_LRECORD_IMPLEMENTATION(name,c_name,marker,printer,nuker,equal,hash,structtype) \
226 DECLARE_ERROR_CHECK_TYPECHECK(c_name, structtype) \
227 static int lrecord_##c_name##_lrecord_type_index; \
228 CONST_IF_NOT_DEBUG struct lrecord_implementation lrecord_##c_name[2] = \
229 { { name, marker, printer, nuker, equal, hash, \
230 0, 0, 0, 0, sizeof (structtype), 0, \
231 &(lrecord_##c_name##_lrecord_type_index), 0 }, \
232 { 0, 0, 0, this_marks_a_marked_record, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }
233
234 #define DEFINE_LRECORD_IMPLEMENTATION_WITH_PROPS(name,c_name,marker,printer,nuker,equal,hash,getprop,putprop,remprop,props,structtype) \
235 DECLARE_ERROR_CHECK_TYPECHECK(c_name, structtype) \
236 static int lrecord_##c_name##_lrecord_type_index; \
237 CONST_IF_NOT_DEBUG struct lrecord_implementation lrecord_##c_name[2] = \
238 { { name, marker, printer, nuker, equal, hash, \
239 getprop, putprop, remprop, props, sizeof (structtype), 0, \
240 &(lrecord_##c_name##_lrecord_type_index), 0 }, \
241 { 0, 0, 0, this_marks_a_marked_record, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }
242
243 #define DEFINE_LRECORD_SEQUENCE_IMPLEMENTATION(name,c_name,marker,printer,nuker,equal,hash,sizer,structtype) \
244 DECLARE_ERROR_CHECK_TYPECHECK(c_name, structtype) \
245 static int lrecord_##c_name##_lrecord_type_index; \
246 CONST_IF_NOT_DEBUG struct lrecord_implementation lrecord_##c_name[2] = \
247 { { name, marker, printer, nuker, equal, hash, \
248 0, 0, 0, 0, 0, sizer, \
249 &(lrecord_##c_name##_lrecord_type_index), 0 }, \
250 { 0, 0, 0, this_marks_a_marked_record, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }
251
252 #define DEFINE_LRECORD_SEQUENCE_IMPLEMENTATION_WITH_PROPS(name,c_name,marker,printer,nuker,equal,hash,getprop,putprop,remprop,props,sizer,structtype) \
253 DECLARE_ERROR_CHECK_TYPECHECK(c_name, structtype) \
254 static int lrecord_##c_name##_lrecord_type_index; \
255 CONST_IF_NOT_DEBUG struct lrecord_implementation lrecord_##c_name[2] = \
256 { { name, marker, printer, nuker, equal, hash, \
257 getprop, putprop, remprop, props, 0, sizer, \
258 &(lrecord_##c_name##_lrecord_type_index), 0 }, \
259 { 0, 0, 0, this_marks_a_marked_record, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }
260
261 #define LRECORDP(a) (XTYPE ((a)) == Lisp_Record)
262 #define XRECORD_LHEADER(a) ((struct lrecord_header *) XPNTR (a))
263 #define RECORD_TYPEP(x, ty) \
264 (LRECORDP (x) && XRECORD_LHEADER (x)->implementation == (ty))
265
266 /* NOTE: the DECLARE_LRECORD() must come before the associated
267 DEFINE_LRECORD_*() or you will get compile errors.
268
269 Furthermore, you always need to put the DECLARE_LRECORD() in a header
270 file, and make sure the header file is included in inline.c, even
271 if the type is private to a particular file. Otherwise, you will
272 get undefined references for the error_check_foo() inline function
273 under GCC. */
274
275 #ifdef ERROR_CHECK_TYPECHECK
276
277 # define DECLARE_LRECORD(c_name, structtype) \
278 extern CONST_IF_NOT_DEBUG struct lrecord_implementation \
279 lrecord_##c_name[]; \
280 INLINE structtype *error_check_##c_name (Lisp_Object _obj); \
281 INLINE structtype * \
282 error_check_##c_name (Lisp_Object _obj) \
283 { \
284 XUNMARK (_obj); \
285 assert (RECORD_TYPEP (_obj, lrecord_##c_name) || \
286 MARKED_RECORD_P (_obj)); \
287 return (structtype *) XPNTR (_obj); \
288 } \
289 extern Lisp_Object Q##c_name##p
290
291 # define DECLARE_NONRECORD(c_name, type_enum, structtype) \
292 INLINE structtype *error_check_##c_name (Lisp_Object _obj); \
293 INLINE structtype * \
294 error_check_##c_name (Lisp_Object _obj) \
295 { \
296 XUNMARK (_obj); \
297 assert (XGCTYPE (_obj) == type_enum); \
298 return (structtype *) XPNTR (_obj); \
299 } \
300 extern Lisp_Object Q##c_name##p
301
302 # define XRECORD(x, c_name, structtype) error_check_##c_name (x)
303 # define XNONRECORD(x, c_name, type_enum, structtype) error_check_##c_name (x)
304
305 # define XSETRECORD(var, p, c_name) do \
306 { \
307 XSETOBJ (var, Lisp_Record, p); \
308 assert (RECORD_TYPEP (var, lrecord_##c_name) || \
309 MARKED_RECORD_P (var)); \
310 } while (0)
311
312 #else /* not ERROR_CHECK_TYPECHECK */
313
314 # define DECLARE_LRECORD(c_name, structtype) \
315 extern Lisp_Object Q##c_name##p; \
316 extern CONST_IF_NOT_DEBUG struct lrecord_implementation \
317 lrecord_##c_name[]
318 # define DECLARE_NONRECORD(c_name, type_enum, structtype) \
319 extern Lisp_Object Q##c_name##p
320 # define XRECORD(x, c_name, structtype) ((structtype *) XPNTR (x))
321 # define XNONRECORD(x, c_name, type_enum, structtype) \
322 ((structtype *) XPNTR (x))
323 # define XSETRECORD(var, p, c_name) XSETOBJ (var, Lisp_Record, p)
324
325 #endif /* not ERROR_CHECK_TYPECHECK */
326
327 #define RECORDP(x, c_name) RECORD_TYPEP (x, lrecord_##c_name)
328 #define GC_RECORDP(x, c_name) gc_record_type_p (x, lrecord_##c_name)
329
330 /* Note: we now have two different kinds of type-checking macros.
331 The "old" kind has now been renamed CONCHECK_foo. The reason for
332 this is that the CONCHECK_foo macros signal a continuable error,
333 allowing the user (through debug-on-error) to subsitute a different
334 value and return from the signal, which causes the lvalue argument
335 to get changed. Quite a lot of code would crash if that happened,
336 because it did things like
337
338 foo = XCAR (list);
339 CHECK_STRING (foo);
340
341 and later on did XSTRING (XCAR (list)), assuming that the type
342 is correct (when it might be wrong, if the user substituted a
343 correct value in the debugger).
344
345 To get around this, I made all the CHECK_foo macros signal a
346 non-continuable error. Places where a continuable error is OK
347 (generally only when called directly on the argument of a Lisp
348 primitive) should be changed to use CONCHECK().
349
350 FSF Emacs does not have this problem because RMS took the cheesy
351 way out and disabled returning from a signal entirely. */
352
353 #define CONCHECK_RECORD(x, c_name) do \
354 { if (!RECORD_TYPEP (x, lrecord_##c_name)) \
355 x = wrong_type_argument (Q##c_name##p, x); } \
356 while (0)
357 #define CONCHECK_NONRECORD(x, lisp_enum, predicate) do \
358 { if (XTYPE (x) != lisp_enum) \
359 x = wrong_type_argument (predicate, x); } \
360 while (0)
361 #define CHECK_RECORD(x, c_name) do \
362 { if (!RECORD_TYPEP (x, lrecord_##c_name)) \
363 dead_wrong_type_argument (Q##c_name##p, x); } \
364 while (0)
365 #define CHECK_NONRECORD(x, lisp_enum, predicate) do \
366 { if (XTYPE (x) != lisp_enum) \
367 dead_wrong_type_argument (predicate, x); } \
368 while (0)
369
370 void *alloc_lcrecord (int size, CONST struct lrecord_implementation *);
371
372 int gc_record_type_p (Lisp_Object frob,
373 CONST struct lrecord_implementation *type);
374
375 /* Copy the data from one lcrecord structure into another, but don't
376 overwrite the header information. */
377
378 #define copy_lcrecord(dst, src) \
379 memcpy ((char *) dst + sizeof (struct lcrecord_header), \
380 (char *) src + sizeof (struct lcrecord_header), \
381 sizeof (*dst) - sizeof (struct lcrecord_header))
382
383 #define zero_lcrecord(lcr) \
384 memset ((char *) lcr + sizeof (struct lcrecord_header), 0, \
385 sizeof (*lcr) - sizeof (struct lcrecord_header))
386
387 #endif /* _XEMACS_LRECORD_H_ */