![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to jsscope.h CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [Synchronet] / sbbs / include / mozilla / js|
Return to jsscope.h CVS log | Up to [Synchronet] / sbbs / include / mozilla / js |
1.1 root 1: /* -*- Mode: C; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 4 -*-
2: *
3: * ***** BEGIN LICENSE BLOCK *****
4: * Version: MPL 1.1/GPL 2.0/LGPL 2.1
5: *
6: * The contents of this file are subject to the Mozilla Public License Version
7: * 1.1 (the "License"); you may not use this file except in compliance with
8: * the License. You may obtain a copy of the License at
9: * http://www.mozilla.org/MPL/
10: *
11: * Software distributed under the License is distributed on an "AS IS" basis,
12: * WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
13: * for the specific language governing rights and limitations under the
14: * License.
15: *
16: * The Original Code is Mozilla Communicator client code, released
17: * March 31, 1998.
18: *
19: * The Initial Developer of the Original Code is
20: * Netscape Communications Corporation.
21: * Portions created by the Initial Developer are Copyright (C) 1998
22: * the Initial Developer. All Rights Reserved.
23: *
24: * Contributor(s):
25: *
26: * Alternatively, the contents of this file may be used under the terms of
27: * either of the GNU General Public License Version 2 or later (the "GPL"),
28: * or the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
29: * in which case the provisions of the GPL or the LGPL are applicable instead
30: * of those above. If you wish to allow use of your version of this file only
31: * under the terms of either the GPL or the LGPL, and not to allow others to
32: * use your version of this file under the terms of the MPL, indicate your
33: * decision by deleting the provisions above and replace them with the notice
34: * and other provisions required by the GPL or the LGPL. If you do not delete
35: * the provisions above, a recipient may use your version of this file under
36: * the terms of any one of the MPL, the GPL or the LGPL.
37: *
38: * ***** END LICENSE BLOCK ***** */
39:
40: #ifndef jsscope_h___
41: #define jsscope_h___
42: /*
43: * JS symbol tables.
44: */
45: #include "jstypes.h"
46: #include "jsobj.h"
47: #include "jsprvtd.h"
48: #include "jspubtd.h"
49:
50: #ifdef JS_THREADSAFE
51: # include "jslock.h"
52: #endif
53:
54: /*
55: * Given P independent, non-unique properties each of size S words mapped by
56: * all scopes in a runtime, construct a property tree of N nodes each of size
57: * S+L words (L for tree linkage). A nominal L value is 2 for leftmost-child
58: * and right-sibling links. We hope that the N < P by enough that the space
59: * overhead of L, and the overhead of scope entries pointing at property tree
60: * nodes, is worth it.
61: *
62: * The tree construction goes as follows. If any empty scope in the runtime
63: * has a property X added to it, find or create a node under the tree root
64: * labeled X, and set scope->lastProp to point at that node. If any non-empty
65: * scope whose most recently added property is labeled Y has another property
66: * labeled Z added, find or create a node for Z under the node that was added
67: * for Y, and set scope->lastProp to point at that node.
68: *
69: * A property is labeled by its members' values: id, getter, setter, slot,
70: * attributes, tiny or short id, and a field telling for..in order. Note that
71: * labels are not unique in the tree, but they are unique among a node's kids
72: * (barring rare and benign multi-threaded race condition outcomes, see below)
73: * and along any ancestor line from the tree root to a given leaf node (except
74: * for the hard case of duplicate formal parameters to a function).
75: *
76: * Thus the root of the tree represents all empty scopes, and the first ply
77: * of the tree represents all scopes containing one property, etc. Each node
78: * in the tree can stand for any number of scopes having the same ordered set
79: * of properties, where that node was the last added to the scope. (We need
80: * not store the root of the tree as a node, and do not -- all we need are
81: * links to its kids.)
82: *
83: * Sidebar on for..in loop order: ECMA requires no particular order, but this
84: * implementation has promised and delivered property definition order, and
85: * compatibility is king. We could use an order number per property, which
86: * would require a sort in js_Enumerate, and an entry order generation number
87: * per scope. An order number beats a list, which should be doubly-linked for
88: * O(1) delete. An even better scheme is to use a parent link in the property
89: * tree, so that the ancestor line can be iterated from scope->lastProp when
90: * filling in a JSIdArray from back to front. This parent link also helps the
91: * GC to sweep properties iteratively.
92: *
93: * What if a property Y is deleted from a scope? If Y is the last property in
94: * the scope, we simply adjust the scope's lastProp member after we remove the
95: * scope's hash-table entry pointing at that property node. The parent link
96: * mentioned in the for..in sidebar above makes this adjustment O(1). But if
97: * Y comes between X and Z in the scope, then we might have to "fork" the tree
98: * at X, leaving X->Y->Z in case other scopes have those properties added in
99: * that order; and to finish the fork, we'd add a node labeled Z with the path
100: * X->Z, if it doesn't exist. This could lead to lots of extra nodes, and to
101: * O(n^2) growth when deleting lots of properties.
102: *
103: * Rather, for O(1) growth all around, we should share the path X->Y->Z among
104: * scopes having those three properties added in that order, and among scopes
105: * having only X->Z where Y was deleted. All such scopes have a lastProp that
106: * points to the Z child of Y. But a scope in which Y was deleted does not
107: * have a table entry for Y, and when iterating that scope by traversing the
108: * ancestor line from Z, we will have to test for a table entry for each node,
109: * skipping nodes that lack entries.
110: *
111: * What if we add Y again? X->Y->Z->Y is wrong and we'll enumerate Y twice.
112: * Therefore we must fork in such a case, if not earlier. Because delete is
113: * "bursty", we should not fork eagerly. Delaying a fork till we are at risk
114: * of adding Y after it was deleted already requires a flag in the JSScope, to
115: * wit, SCOPE_MIDDLE_DELETE.
116: *
117: * What about thread safety? If the property tree operations done by requests
118: * are find-node and insert-node, then the only hazard is duplicate insertion.
119: * This is harmless except for minor bloat. When all requests have ended or
120: * been suspended, the GC is free to sweep the tree after marking all nodes
121: * reachable from scopes, performing remove-node operations as needed. Note
122: * also that the stable storage of the property nodes during active requests
123: * permits the property cache (see jsinterp.h) to dereference JSScopeProperty
124: * weak references safely.
125: *
126: * Is the property tree worth it compared to property storage in each table's
127: * entries? To decide, we must find the relation <> between the words used
128: * with a property tree and the words required without a tree.
129: *
130: * Model all scopes as one super-scope of capacity T entries (T a power of 2).
131: * Let alpha be the load factor of this double hash-table. With the property
132: * tree, each entry in the table is a word-sized pointer to a node that can be
133: * shared by many scopes. But all such pointers are overhead compared to the
134: * situation without the property tree, where the table stores property nodes
135: * directly, as entries each of size S words. With the property tree, we need
136: * L=2 extra words per node for siblings and kids pointers. Without the tree,
137: * (1-alpha)*S*T words are wasted on free or removed sentinel-entries required
138: * by double hashing.
139: *
140: * Therefore,
141: *
142: * (property tree) <> (no property tree)
143: * N*(S+L) + T <> S*T
144: * N*(S+L) + T <> P*S + (1-alpha)*S*T
145: * N*(S+L) + alpha*T + (1-alpha)*T <> P*S + (1-alpha)*S*T
146: *
147: * Note that P is alpha*T by definition, so
148: *
149: * N*(S+L) + P + (1-alpha)*T <> P*S + (1-alpha)*S*T
150: * N*(S+L) <> P*S - P + (1-alpha)*S*T - (1-alpha)*T
151: * N*(S+L) <> (P + (1-alpha)*T) * (S-1)
152: * N*(S+L) <> (P + (1-alpha)*P/alpha) * (S-1)
153: * N*(S+L) <> P * (1/alpha) * (S-1)
154: *
155: * Let N = P*beta for a compression ratio beta, beta <= 1:
156: *
157: * P*beta*(S+L) <> P * (1/alpha) * (S-1)
158: * beta*(S+L) <> (S-1)/alpha
159: * beta <> (S-1)/((S+L)*alpha)
160: *
161: * For S = 6 (32-bit architectures) and L = 2, the property tree wins iff
162: *
163: * beta < 5/(8*alpha)
164: *
165: * We ensure that alpha <= .75, so the property tree wins if beta < .83_. An
166: * average beta from recent Mozilla browser startups was around .6.
167: *
168: * Can we reduce L? Observe that the property tree degenerates into a list of
169: * lists if at most one property Y follows X in all scopes. In or near such a
170: * case, we waste a word on the right-sibling link outside of the root ply of
171: * the tree. Note also that the root ply tends to be large, so O(n^2) growth
172: * searching it is likely, indicating the need for hashing (but with increased
173: * thread safety costs).
174: *
175: * If only K out of N nodes in the property tree have more than one child, we
176: * could eliminate the sibling link and overlay a children list or hash-table
177: * pointer on the leftmost-child link (which would then be either null or an
178: * only-child link; the overlay could be tagged in the low bit of the pointer,
179: * or flagged elsewhere in the property tree node, although such a flag must
180: * not be considered when comparing node labels during tree search).
181: *
182: * For such a system, L = 1 + (K * averageChildrenTableSize) / N instead of 2.
183: * If K << N, L approaches 1 and the property tree wins if beta < .95.
184: *
185: * We observe that fan-out below the root ply of the property tree appears to
186: * have extremely low degree (see the MeterPropertyTree code that histograms
187: * child-counts in jsscope.c), so instead of a hash-table we use a linked list
188: * of child node pointer arrays ("kid chunks"). The details are isolated in
189: * jsscope.c; others must treat JSScopeProperty.kids as opaque. We leave it
190: * strongly typed for debug-ability of the common (null or one-kid) cases.
191: *
192: * One final twist (can you stand it?): the mean number of entries per scope
193: * in Mozilla is < 5, with a large standard deviation (~8). Instead of always
194: * allocating scope->table, we leave it null while initializing all the other
195: * scope members as if it were non-null and minimal-length. Until a property
196: * is added that crosses the threshold of 6 or more entries for hashing, or
197: * until a "middle delete" occurs, we use linear search from scope->lastProp
198: * to find a given id, and save on the space overhead of a hash table.
199: */
200:
201: struct JSScope {
202: JSObjectMap map; /* base class state */
203: JSObject *object; /* object that owns this scope */
204: uint16 flags; /* flags, see below */
205: int16 hashShift; /* multiplicative hash shift */
206: uint32 entryCount; /* number of entries in table */
207: uint32 removedCount; /* removed entry sentinels in table */
208: JSScopeProperty **table; /* table of ptrs to shared tree nodes */
209: JSScopeProperty *lastProp; /* pointer to last property added */
210: #ifdef JS_THREADSAFE
211: JSContext *ownercx; /* creating context, NULL if shared */
212: JSThinLock lock; /* binary semaphore protecting scope */
213: union { /* union lockful and lock-free state: */
214: jsrefcount count; /* lock entry count for reentrancy */
215: JSScope *link; /* next link in rt->scopeSharingTodo */
216: } u;
217: #ifdef DEBUG
218: const char *file[4]; /* file where lock was (re-)taken */
219: unsigned int line[4]; /* line where lock was (re-)taken */
220: #endif
221: #endif
222: };
223:
224: #define OBJ_SCOPE(obj) ((JSScope *)(obj)->map)
225:
226: /* By definition, hashShift = JS_DHASH_BITS - log2(capacity). */
227: #define SCOPE_CAPACITY(scope) JS_BIT(JS_DHASH_BITS-(scope)->hashShift)
228:
229: /* Scope flags and some macros to hide them from other files than jsscope.c. */
230: #define SCOPE_MIDDLE_DELETE 0x0001
231: #define SCOPE_SEALED 0x0002
232:
233: #define SCOPE_HAD_MIDDLE_DELETE(scope) ((scope)->flags & SCOPE_MIDDLE_DELETE)
234: #define SCOPE_SET_MIDDLE_DELETE(scope) ((scope)->flags |= SCOPE_MIDDLE_DELETE)
235: #define SCOPE_CLR_MIDDLE_DELETE(scope) ((scope)->flags &= ~SCOPE_MIDDLE_DELETE)
236:
237: #define SCOPE_IS_SEALED(scope) ((scope)->flags & SCOPE_SEALED)
238: #define SCOPE_SET_SEALED(scope) ((scope)->flags |= SCOPE_SEALED)
239: #if 0
240: /*
241: * Don't define this, it can't be done safely because JS_LOCK_OBJ will avoid
242: * taking the lock if the object owns its scope and the scope is sealed.
243: */
244: #define SCOPE_CLR_SEALED(scope) ((scope)->flags &= ~SCOPE_SEALED)
245: #endif
246:
247: /*
248: * A little information hiding for scope->lastProp, in case it ever becomes
249: * a tagged pointer again.
250: */
251: #define SCOPE_LAST_PROP(scope) ((scope)->lastProp)
252: #define SCOPE_REMOVE_LAST_PROP(scope) ((scope)->lastProp = \
253: (scope)->lastProp->parent)
254:
255: struct JSScopeProperty {
256: jsid id; /* int-tagged jsval/untagged JSAtom* */
257: JSPropertyOp getter; /* getter and setter hooks or objects */
258: JSPropertyOp setter;
259: uint32 slot; /* index in obj->slots vector */
260: uint8 attrs; /* attributes, see jsapi.h JSPROP_* */
261: uint8 flags; /* flags, see below for defines */
262: int16 shortid; /* tinyid, or local arg/var index */
263: JSScopeProperty *parent; /* parent node, reverse for..in order */
264: JSScopeProperty *kids; /* null, single child, or a tagged ptr
265: to many-kids data structure */
266: };
267:
268: /* JSScopeProperty pointer tag bit indicating a collision. */
269: #define SPROP_COLLISION ((jsuword)1)
270: #define SPROP_REMOVED ((JSScopeProperty *) SPROP_COLLISION)
271:
272: /* Macros to get and set sprop pointer values and collision flags. */
273: #define SPROP_IS_FREE(sprop) ((sprop) == NULL)
274: #define SPROP_IS_REMOVED(sprop) ((sprop) == SPROP_REMOVED)
275: #define SPROP_IS_LIVE(sprop) ((sprop) > SPROP_REMOVED)
276: #define SPROP_FLAG_COLLISION(spp,sprop) (*(spp) = (JSScopeProperty *) \
277: ((jsuword)(sprop) | SPROP_COLLISION))
278: #define SPROP_HAD_COLLISION(sprop) ((jsuword)(sprop) & SPROP_COLLISION)
279: #define SPROP_FETCH(spp) SPROP_CLEAR_COLLISION(*(spp))
280:
281: #define SPROP_CLEAR_COLLISION(sprop) \
282: ((JSScopeProperty *) ((jsuword)(sprop) & ~SPROP_COLLISION))
283:
284: #define SPROP_STORE_PRESERVING_COLLISION(spp, sprop) \
285: (*(spp) = (JSScopeProperty *) ((jsuword)(sprop) \
286: | SPROP_HAD_COLLISION(*(spp))))
287:
288: /* Bits stored in sprop->flags. */
289: #define SPROP_MARK 0x01
290: #define SPROP_IS_DUPLICATE 0x02
291: #define SPROP_IS_ALIAS 0x04
292: #define SPROP_HAS_SHORTID 0x08
293:
294: /*
295: * If SPROP_HAS_SHORTID is set in sprop->flags, we use sprop->shortid rather
296: * than id when calling sprop's getter or setter.
297: */
298: #define SPROP_USERID(sprop) \
299: (((sprop)->flags & SPROP_HAS_SHORTID) ? INT_TO_JSVAL((sprop)->shortid) \
300: : ID_TO_VALUE((sprop)->id))
301:
302: #define SPROP_INVALID_SLOT 0xffffffff
303:
304: #define SPROP_HAS_VALID_SLOT(sprop, scope) \
305: ((sprop)->slot < (scope)->map.freeslot)
306:
307: #define SPROP_HAS_STUB_GETTER(sprop) (!(sprop)->getter)
308: #define SPROP_HAS_STUB_SETTER(sprop) (!(sprop)->setter)
309:
310: #define SPROP_CALL_GETTER(cx,sprop,getter,obj,obj2,vp) \
311: (!(getter) || \
312: (getter)(cx, OBJ_THIS_OBJECT(cx,obj), SPROP_USERID(sprop), vp))
313: #define SPROP_CALL_SETTER(cx,sprop,setter,obj,obj2,vp) \
314: (!(setter) || \
315: (setter)(cx, OBJ_THIS_OBJECT(cx,obj), SPROP_USERID(sprop), vp))
316:
317: #define SPROP_GET(cx,sprop,obj,obj2,vp) \
318: (((sprop)->attrs & JSPROP_GETTER) \
319: ? js_InternalGetOrSet(cx, obj, (sprop)->id, \
320: OBJECT_TO_JSVAL((sprop)->getter), JSACC_READ, \
321: 0, 0, vp) \
322: : SPROP_CALL_GETTER(cx, sprop, (sprop)->getter, obj, obj2, vp))
323:
324: #define SPROP_SET(cx,sprop,obj,obj2,vp) \
325: (((sprop)->attrs & JSPROP_SETTER) \
326: ? js_InternalGetOrSet(cx, obj, (sprop)->id, \
327: OBJECT_TO_JSVAL((sprop)->setter), JSACC_WRITE, \
328: 1, vp, vp) \
329: : ((sprop)->attrs & JSPROP_GETTER) \
330: ? (JS_ReportErrorNumber(cx, js_GetErrorMessage, NULL, \
331: JSMSG_GETTER_ONLY, NULL), JS_FALSE) \
332: : SPROP_CALL_SETTER(cx, sprop, (sprop)->setter, obj, obj2, vp))
333:
334: /* Macro for common expression to test for shared permanent attributes. */
335: #define SPROP_IS_SHARED_PERMANENT(sprop) \
336: ((~(sprop)->attrs & (JSPROP_SHARED | JSPROP_PERMANENT)) == 0)
337:
338: extern JSScope *
339: js_GetMutableScope(JSContext *cx, JSObject *obj);
340:
341: extern JSScope *
342: js_NewScope(JSContext *cx, jsrefcount nrefs, JSObjectOps *ops, JSClass *clasp,
343: JSObject *obj);
344:
345: extern void
346: js_DestroyScope(JSContext *cx, JSScope *scope);
347:
348: #define ID_TO_VALUE(id) (((id) & JSVAL_INT) ? id : ATOM_KEY((JSAtom *)(id)))
349: #define HASH_ID(id) (((id) & JSVAL_INT) \
350: ? (jsatomid) JSVAL_TO_INT(id) \
351: : ((JSAtom *)id)->number)
352:
353: extern JS_FRIEND_API(JSScopeProperty **)
354: js_SearchScope(JSScope *scope, jsid id, JSBool adding);
355:
356: #define SCOPE_GET_PROPERTY(scope, id) \
357: SPROP_FETCH(js_SearchScope(scope, id, JS_FALSE))
358:
359: #define SCOPE_HAS_PROPERTY(scope, sprop) \
360: (SCOPE_GET_PROPERTY(scope, (sprop)->id) == (sprop))
361:
362: extern JSScopeProperty *
363: js_AddScopeProperty(JSContext *cx, JSScope *scope, jsid id,
364: JSPropertyOp getter, JSPropertyOp setter, uint32 slot,
365: uintN attrs, uintN flags, intN shortid);
366:
367: extern JSScopeProperty *
368: js_ChangeScopePropertyAttrs(JSContext *cx, JSScope *scope,
369: JSScopeProperty *sprop, uintN attrs, uintN mask,
370: JSPropertyOp getter, JSPropertyOp setter);
371:
372: extern JSBool
373: js_RemoveScopeProperty(JSContext *cx, JSScope *scope, jsid id);
374:
375: extern void
376: js_ClearScope(JSContext *cx, JSScope *scope);
377:
378: #define MARK_SCOPE_PROPERTY(sprop) ((sprop)->flags |= SPROP_MARK)
379:
380: extern void
381: js_SweepScopeProperties(JSRuntime *rt);
382:
383: extern JSBool
384: js_InitPropertyTree(JSRuntime *rt);
385:
386: extern void
387: js_FinishPropertyTree(JSRuntime *rt);
388:
389: #endif /* jsscope_h___ */
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.