until JS_DestroyRuntime is called (68450, r=rginda, sr=jband).
- NUL-terminate tagbuf in tagify, for the HTML helpers such as string.big()
(66648, r=timeless, sr=jband).
git-svn-id: svn://10.0.0.236/trunk@86863 18797224-902f-48f8-a5cc-f745e15eee43
- Optimize compile (parse+emit) operation to generate code for each top-level
statement or function in turn, recycling JSParseNodes as we go for greatly
reduced "long linear script" footprint.
- Fix O(n**2) growth problems in bytecode and srcnote generation.
- Add js_ParseTokenStream entry point to compiler, for tree-generation without
code-generation. Move JSOP_EVAL instruction selection from code-generator to
parser, to match other such specializations and enable js_ParseTokenStream.
- Fix js_CompileTokenStream (and get it right in new js_ParseTokenStream) to
respect JSOPTION_VAROBJFIX.
- Clean up bracing, multi-line conditions, and overlong lines.
git-svn-id: svn://10.0.0.236/trunk@85619 18797224-902f-48f8-a5cc-f745e15eee43
especially to jband for his great stress-test setup and particularly helpful
(in terms of reproducing bugs in draft patches) MP and laptop machines.
- Radical(*) object (scope) locking optimization: don't lock if a scope is
accessed on the context that exclusively owns it (initially, the context
on which the scope was created). Once a scope becomes shared among more
than one owner-context, give it the usual thin or fat lock, per existing
jslock.c code.
I did this at the memory cost of another word per JSScope, ownercx, which
raised scope size from 12 to 13 words if !DEBUG. I also added a linked
list head pointer, rt->scopeSharingTodo, and a scopeSharingDone condition
variable to JSRuntime, and a scopeToShare pointer to JSContext that's
necessary for deadlock avoidance.
The rt->scopeSharingTodo list links JSScopes through the scope->u.link
union arm, which overlays the pre-existing scope->count (now u.count)
member. This list holds scopes still exclusively owned by a context, but
wanted by js_LockScope calls active on other threads. Those calls wait
on the rt->scopeSharingDone condition, which is notified every time an
owner-context ends the request running on it, in which code active on
that context may be using scope freely until end of request.
The code that waits on rt->scopeSharingDone must first suspend any and
all requests active on the calling context, and resume those contexts
after the wait is notified. This means a GC could happen while the
thread locking a scope owned by another thread's context blocks; all
calls to JS_LOCK_OBJ must therefore first home fp->sp above any live
operands, e.g. The interpreter takes care to do that already.
To avoid AB-BA deadlocks, if a js_LockScope attempt on one context finds
that the owner-context of the scope is already waiting on a scope owned
by the current context (or indirectly depending on such a scope lock),
the attempt converts the scope from lock-free exclusive ownership to
shared ownership (thin or fat lock).
- Fix js_SetupLocks and the js_LockGlobal/js_UnlockGlobal code to avoid
divmod instruction costs, strength-reducing to bit-mask instructions.
- The radical lock-free scope change required care in handling the 0=>1
and 1=>0 transitions of cx->requestDepth, which was till now thread-local
because part of the JSContext not manipulated by other threads. It's
still updated only by cx's thread, but it is read by other threads in
the course of attempting to claim exclusive ownership of a scope for more
lock-free JS object operations.
- The JS_SuspendRequest and JS_ResumeRequest APIs have changed incompatibly
to require their caller to save and restore the requestCount found when
JS_SuspendRequest is called. This is necessary to avoid deadlock; sorry
for the incompatible change.
- Fixed various nits in jslock.[ch], including using Init/Finish rather
than New/Destroy for the methods that take a JSThinLock and initialize
and finish/free its members. Another example: JS_ATOMIC_ADDREF is now
JS_ATOMIC_INCREMENT and JS_ATOMIC_DECREMENT, so the two cases can be
mapped to PR_AtomicIncrement and PR_AtomicDecrement. This entailed
changing jsrefcount from jsword to int32 (PRInt32).
- No need to use JS_ATOMIC_INCREMENT on JSScopeProperty.nrefs, as it is
always and everywhere protected by the property's JSScope.lock.
- Cleaned up gratuitous casts in jscntxt.c by using &cx->links, etc.
- The lock used for mutual exclusion around both request begin and end vs.
GC synchronization is rt->gcLock, and this lock now also protects all
scope->ownercx pointer changes from non-null (exclusive) to null (shared),
the rt->scopeSharingTodo/scope->u.link list operations, and of course the
rt->scopeSharingDone condition.
But this means that js_GC cannot hold rt->gcLock across the bulk of its
body, in particular the mark phase, during which JS_GetPrivate calls,
e.g., may need to "promote" scope locks from lock-free to thin or fat,
because doing so would double-trip. There never was any good reason to
hold rt->gcLock so long, of course -- locks are for mutual exclusion, not
for waiting or notifying a thread -- those operations require a condition,
rt->gcDone, which we already use along with rt->gcLevel to keep racing GC
attempts at bay.
So now that rt->gcLock does not protect the mark phase, the enumeration
of rt->gcRootsHash can race badly with JS_RemoveRootRT, an API that may
legitimately be called outside of a request, without even a context. It
turns out that people may be cheating on the request model even with
JS_AddRoot, JS_AddNamedRoot, and JS_RemoveRoot calls, so we must make
all of those interlock with the GC using gcLevel and gcDone, unless they
are called on the gcThread.
Also, since bug 49816 was fixed, there has been no need for a separate
finalize phase, or for rt->gcFinalVec. Finalizers can no longer allocate
newborn GC-things that might be swept (because unmarked), or double-trip
on rt->gcLock (which is no longer held). So js_GC finalizes as it sweeps,
just as it did in days of old.
- I added comments to jslock.h making it plain that callers of JS_LOCK_OBJ
and JS_UNLOCK_OBJ must either be implementations of js_ObjectOps hooks,
or code reachable only from those hooks; or else must be predicated on
OBJ_IS_NATIVE tests. It turns out jsinterp.c's CACHED_GET and CACHED_SET
macros neglected to do such tests, limiting the ability of JS embeddings
to implement JSObjectOps with their own non-JSScope JSObjectMap subclass.
Fixed, small performance hit that the lock-free optimization should more
than make up for.
- jslock.c now gives a #error if you try to compile it on a platform that
lacks a compare-and-swap instruction. The #error says to use NSPR locks.
Before this change, some platforms would emulate compare-and-swap using
a global PRLock, which is always worse in runtime than using per-scope
PRLocks.
git-svn-id: svn://10.0.0.236/trunk@83229 18797224-902f-48f8-a5cc-f745e15eee43
- Tabs expanded as much as I could stand, without taking too much cvsblame.
git-svn-id: svn://10.0.0.236/trunk@59716 18797224-902f-48f8-a5cc-f745e15eee43
Courtesy Bill Gibbons <bill@gibbons.org>
His comments:
Here are the changes to JSRef to make it compile either as C or C++. Mostly the changes are to add missing casts (since C++ doesn't have implict conversion from void* to other pointer types nor implicit casts from ints to enumerations) plus a few random things like the use of "private" as a variable name.
There are a few other minor bug fixes; in particular:
* A long statement with and'ed conditions is reformatted to make it easier to remove other builtin objects (e.g. Date).
* A #if was added to jsscript.c for the JS_HAS_SCRIPT_OBJECT off case.
* In jsmath a #ifdef was changed to #if.
My notes also mention...
* jsobj.c should include jsopcode.h
* jsfun.c - doesn't link if JS_HAS_ARGS_OBJECT is off
* jsarray.c - a reference to js_ValueToSource should be conditional on JS_HAS_TOSOURCE
r=mccabe
git-svn-id: svn://10.0.0.236/trunk@59455 18797224-902f-48f8-a5cc-f745e15eee43
- Add JS1.5 getter/setter support in all its glory:
* getter function SN() {return ++x} at top-level or as a closure binds an SN
property getter than returns the incremented value of x. Likewise for
setter function SN(y) {return y = x}.
* getters and setters may be defined in an object literal:
o = {p getter:function() {return ++this.x},
p setter:function(y){return this.x = y},
x:42};
* getter= and setter= operators (compound tokens) may be used to bind getter
and setter properties dynamically:
o = new Object;
o.p getter= function() {return ++this.x};
o.p setter= function(y){return this.x = y};
o.x = 42;
Waldemar is concerned that this form will collide semantically with JS2, so
I am not committing to keeping it in JS1.5. I'd like to check my code in
ASAP so shaver can use it, and I'd also like to see this form get used (or
not) during Mozilla betas. Caveat emptor, and if you find this "dynamic"
or "imperative" form necessary and hard to substitute, please let me know.
If this proves important to users, then I think JS1.5 should keep it.
- Cleaned up property flags (in a binary-incompatible fashion -- who cares?) by
eliminating JSPROP_ASSIGNHACK and JSPROP_TINYIDHACK.
- Added JS_DONT_PRETTY_PRINT flag to be ORed with the indent argument to the
several JS_Decompile*() API calls. This avoids any newlines or identation in
the decompiled string.
- Improved and extended (for getter/setter non-reservation) scanner lookahead
by using a circular (power-of-2 sized) token buffer.
- Fix ECMA Edition 3 deviation where function f(){function g(){}} bound f.g by
mistake (it should arrange to make a closure named g in activations of f, but
it should not bind a property of function f).
git-svn-id: svn://10.0.0.236/trunk@48436 18797224-902f-48f8-a5cc-f745e15eee43
- Fix js_GetAtom fallibility by returning &dummy on assert-botch "can't happen"
index out of range case.
- js_InitAtomMap needn't bother nulling ale->next with tmp save
- js_InitAtomState explicit tail fusion for FROB via goto bad, and early memset
(I know, JSRuntime is cleared already and it contains the atom state ... but
jsatom.c doesn't know that).
- Clear all ATOM_ flags save ATOM_PINNED when creating a new atom.
- Cleanup xtra, ALIGNNUM, etc. useless variables, use JSVAL_ALIGN and JS_MAX.
git-svn-id: svn://10.0.0.236/trunk@37314 18797224-902f-48f8-a5cc-f745e15eee43
+ merging of js/src and js/ref
+ elimination of most dependencies on NSPR
+ JS1.4 feature additions and accumulated bug fixes
More details are in last week's mozilla status report.
git-svn-id: svn://10.0.0.236/trunk@12797 18797224-902f-48f8-a5cc-f745e15eee43
and development branches, including but
not limited to:
- Preliminary exception handling per
ECMA proposal; try, multiple
catchblocks, and finally. Catchblocks
are of the form catch (v) or
catch(v:<guard>), where guard is an
optional boolean expression that is
evaluated to determine whether the
exception is to be caught by that block.
- ECMA-proposed 'in' operator; "'foo' in
o" or "4 in o" asks if o has property
foo or element 4.
- Added a new set of defines in
jsconfig.h for js 1.4
features-in-progress. (in, instanceof,
exception handling.) Default build
version is now 1.4. Fixed a few
conditional features that had become
broken.
- Progress towards porting to FreeBSD
and Alpha; casts of NaN and friends to
int are a little more localized. Not
there yet...
- New config files to compile on more
OSes; various fixes to improve
portability.
git-svn-id: svn://10.0.0.236/trunk@6907 18797224-902f-48f8-a5cc-f745e15eee43
and tested in that release):
#114564: Fix JS_Enumerate to return an empty id array instead of null
#115395: Fix JS garbage collection
#115200: Security dialog no longer blows assertion
#123211: Make sure output of sort array function is -1, 0, or 1.
#116195: Fix ImportProperty(), cope with an existing local variable
of the same name as the imported property
Unbusticate JS_invoke() so that closures work again
#115384,#115395: Handle bugs in toSource that strike when
getProperty is non-idempotent. Fix crash during JavaScript
garbage collection after enumerating object properties.
#??????: mjudge - Win16 files to make ptrdiff_t be 32 bits on Win16
git-svn-id: svn://10.0.0.236/trunk@3573 18797224-902f-48f8-a5cc-f745e15eee43
