corysanin/tubestation
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Bug 882111 - Don't push an interpreter frame when calling into the JITs. r=djvj

Browse Source
This commit is contained in:
Jan de Mooij
2013-06-19 19:10:04 +02:00
parent d1859894fb
commit 9a7600d987
14 changed files with 640 additions and 368 deletions
Download Patch File Download Diff File Expand all files Collapse all files

82
js/src/jsiter.cpp
View File

@@ -1373,6 +1373,55 @@ SetGeneratorClosed(JSContext *cx, JSGenerator *gen)
gen->state = JSGEN_CLOSED;
}
GeneratorState::GeneratorState(JSContext *cx, JSGenerator *gen, JSGeneratorState futureState)
: RunState(cx, Generator, gen->fp->script()),
cx_(cx),
gen_(gen),
futureState_(futureState),
entered_(false)
{ }
GeneratorState::~GeneratorState()
{
if (entered_)
cx_->leaveGenerator(gen_);
}
StackFrame *
GeneratorState::pushInterpreterFrame(JSContext *cx)
{
gfg_.construct();
/*
* Write barrier is needed since the generator stack can be updated,
* and it's not barriered in any other way. We need to do it before
* gen->state changes, which can cause us to trace the generator
* differently.
*
* We could optimize this by setting a bit on the generator to signify
* that it has been marked. If this bit has already been set, there is no
* need to mark again. The bit would have to be reset before the next GC,
* or else some kind of epoch scheme would have to be used.
*/
GeneratorWriteBarrierPre(cx, gen_);
if (!cx->stack.pushGeneratorFrame(cx, gen_, gfg_.addr())) {
SetGeneratorClosed(cx, gen_);
return NULL;
}
/*
* Don't change the state until after the frame is successfully pushed
* or else we might fail to scan some generator values.
*/
gen_->state = futureState_;
gen_->regs = cx->stack.regs();
cx->enterGenerator(gen_); /* OOM check above. */
entered_ = true;
return gfg_.ref().fp();
}
static void
generator_trace(JSTracer *trc, JSObject *obj)
{
@@ -1493,19 +1542,6 @@ SendToGenerator(JSContext *cx, JSGeneratorOp op, HandleObject obj,
return false;
}
/*
* Write barrier is needed since the generator stack can be updated,
* and it's not barriered in any other way. We need to do it before
* gen->state changes, which can cause us to trace the generator
* differently.
*
* We could optimize this by setting a bit on the generator to signify
* that it has been marked. If this bit has already been set, there is no
* need to mark again. The bit would have to be reset before the next GC,
* or else some kind of epoch scheme would have to be used.
*/
GeneratorWriteBarrierPre(cx, gen);
JSGeneratorState futureState;
JS_ASSERT(gen->state == JSGEN_NEWBORN || gen->state == JSGEN_OPEN);
switch (op) {
@@ -1535,24 +1571,10 @@ SendToGenerator(JSContext *cx, JSGeneratorOp op, HandleObject obj,
JSBool ok;
{
GeneratorFrameGuard gfg;
if (!cx->stack.pushGeneratorFrame(cx, gen, &gfg)) {
SetGeneratorClosed(cx, gen);
GeneratorState state(cx, gen, futureState);
ok = RunScript(cx, state);
if (!ok && gen->state == JSGEN_CLOSED)
return false;
}
/*
* Don't change the state until after the frame is successfully pushed
* or else we might fail to scan some generator values.
*/
gen->state = futureState;
StackFrame *fp = gfg.fp();
gen->regs = cx->stack.regs();
cx->enterGenerator(gen); /* OOM check above. */
ok = RunScript(cx, fp);
cx->leaveGenerator(gen);
}
if (gen->fp->isYielding()) {