Add support for yield expressions

Add support for yield expressions

This CL extends the generator suspend and resume implementation to
capture values on the operand stack.

It factors out some helpers to measure and access the operand stack into
the JavaScriptFrame class.  It also refactors the suspend and resume
helpers to avoid handle allocation.

BUG=v8:2355
TEST=mjsunit/harmony/generators-iteration

Committed: http://code.google.com/p/v8/source/detail?r=14458

[wingo, 2013-04-18 10:00:07]

An initial patch, building on https://codereview.chromium.org/14066016/.  Stack
handlers aren't supported yet.

[wingo, 2013-04-25 10:32:09]

Updated patch applies on top of https://codereview.chromium.org/14416011/ (the
CL about yield inside with).

[Michael Starzinger, 2013-04-26 09:53:16]

Looking good already. One round of comments should suffice here.

https://codereview.chromium.org/14348003/diff/3001/src/frames-inl.h
File src/frames-inl.h (right):

https://codereview.chromium.org/14348003/diff/3001/src/frames-inl.h#newcode235
src/frames-inl.h:235: ASSERT(IsAligned(stack_size_in_bytes, kPointerSize));
Can we add an assert here that the type of the frame is "JAVA_SCRIPT" (as
opposed to "OPTIMIZED")? Because the notion of an operand stack doesn't apply to
optimized code.

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc
File src/runtime.cc (right):

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc#newcode2430
src/runtime.cc:2430: JSFunction *function = JSFunction::cast(frame->function());
style: The '*' sticks to the left. There are several other occurrences in the
code that I seemed to have missed in my earlier reviews.

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc#newcode2455
src/runtime.cc:2455: if (!alloc->To(&operand_stack))
nit: Conditional fits into one line.

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc#newcode2458
src/runtime.cc:2458: for (int i = 0; i < operands_count; i++)
style: Add curly braces around loop body.

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc#newcode2460
src/runtime.cc:2460: operand_stack->set(i, frame->GetOperand(i),
SKIP_WRITE_BARRIER);
The allocation function above doesn't guarantee the FixedArray to be in
new-space. We need a write-barrier here. Or as Boromir would put it: "One does
not simply skip a write-barrier".

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc#newcode2495
src/runtime.cc:2495: JavaScriptFrame *frame = stack_iterator.frame();
style: The '*' sticks to the left.

[wingo, 2013-04-26 10:11:42]

Updated patch addresses feedback.  Good catch with the write barrier issue,
thanks!

https://codereview.chromium.org/14348003/diff/3001/src/frames-inl.h
File src/frames-inl.h (right):

https://codereview.chromium.org/14348003/diff/3001/src/frames-inl.h#newcode235
src/frames-inl.h:235: ASSERT(IsAligned(stack_size_in_bytes, kPointerSize));
On 2013/04/26 09:53:16, Michael Starzinger wrote:
> Can we add an assert here that the type of the frame is "JAVA_SCRIPT" (as
> opposed to "OPTIMIZED")? Because the notion of an operand stack doesn't apply
to
> optimized code.

Done.

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc
File src/runtime.cc (right):

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc#newcode2430
src/runtime.cc:2430: JSFunction *function = JSFunction::cast(frame->function());
On 2013/04/26 09:53:16, Michael Starzinger wrote:
> style: The '*' sticks to the left. There are several other occurrences in the
> code that I seemed to have missed in my earlier reviews.

Fixed this and other locations.

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc#newcode2455
src/runtime.cc:2455: if (!alloc->To(&operand_stack))
On 2013/04/26 09:53:16, Michael Starzinger wrote:
> nit: Conditional fits into one line.

Done.

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc#newcode2458
src/runtime.cc:2458: for (int i = 0; i < operands_count; i++)
On 2013/04/26 09:53:16, Michael Starzinger wrote:
> style: Add curly braces around loop body.

Done.

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc#newcode2460
src/runtime.cc:2460: operand_stack->set(i, frame->GetOperand(i),
SKIP_WRITE_BARRIER);
On 2013/04/26 09:53:16, Michael Starzinger wrote:
> The allocation function above doesn't guarantee the FixedArray to be in
> new-space. We need a write-barrier here. Or as Boromir would put it: "One does
> not simply skip a write-barrier".

Good catch!

https://codereview.chromium.org/14348003/diff/3001/src/runtime.cc#newcode2495
src/runtime.cc:2495: JavaScriptFrame *frame = stack_iterator.frame();
On 2013/04/26 09:53:16, Michael Starzinger wrote:
> style: The '*' sticks to the left.

Done.

[Michael Starzinger, 2013-04-26 10:36:42]

LGTM. I'll land this after the dependent CL made it in.

[Michael Starzinger, 2013-04-26 12:09:40]

Committed patchset #3 manually as r14458 (presubmit successful).