[turbofan] Optimized lowering of DYNAMIC_GLOBAL lookup slot loads.

[turbofan] Optimized lowering of DYNAMIC_GLOBAL lookup slot loads.

This adds handling of JSLoadDynamicGlobal nodes to JSTypedLowering to
perform extension checks and an inline fast path. The fast path is a
global variable load from the global object.

R=bmeurer@chromium.org
BUG=v8:4131
LOG=N

Committed: https://crrev.com/440d099c5c2511836ba1b0be4823cd822b2e147f
Cr-Commit-Position: refs/heads/master@{#28750}

[Michael Starzinger, 2015-06-01 17:07:08]

Note that this is still missing a unit test for the typed lowering part. I will
add one, just wanted to get a first round of comments before/while I do so.
Thanks!

https://codereview.chromium.org/1150723005/diff/1/src/ast-numbering.cc
File src/ast-numbering.cc (right):

https://codereview.chromium.org/1150723005/diff/1/src/ast-numbering.cc#newcod...
src/ast-numbering.cc:158:
DisableCrankshaft(kReferenceToAVariableWhichRequiresDynamicLookup);
Please advise on whether I should land this together with this CL or separate.

[Benedikt Meurer, 2015-06-02 04:39:49]

LGTM modulo the vector on the operator, but that's probably separate.

https://codereview.chromium.org/1150723005/diff/1/src/ast-numbering.cc
File src/ast-numbering.cc (right):

https://codereview.chromium.org/1150723005/diff/1/src/ast-numbering.cc#newcod...
src/ast-numbering.cc:158:
DisableCrankshaft(kReferenceToAVariableWhichRequiresDynamicLookup);
I think that should go in separately.

https://codereview.chromium.org/1150723005/diff/1/src/compiler/js-operator.h
File src/compiler/js-operator.h (right):

https://codereview.chromium.org/1150723005/diff/1/src/compiler/js-operator.h#...
src/compiler/js-operator.h:117: VectorSlotPair(Handle<TypeFeedbackVector>
vector, FeedbackVectorICSlot slot)
The vector should be an input to the node. The operator should only have the
constant slot. But I guess that should be cleanup work later since it involves
changes to the global load, which Michael Stanton is working on as well.

[Michael Starzinger, 2015-06-02 07:38:29]

https://codereview.chromium.org/1150723005/diff/1/src/ast-numbering.cc
File src/ast-numbering.cc (right):

https://codereview.chromium.org/1150723005/diff/1/src/ast-numbering.cc#newcod...
src/ast-numbering.cc:158:
DisableCrankshaft(kReferenceToAVariableWhichRequiresDynamicLookup);
On 2015/06/02 04:39:49, Benedikt Meurer wrote:
> I think that should go in separately.

Done.

https://codereview.chromium.org/1150723005/diff/1/src/compiler/js-operator.h
File src/compiler/js-operator.h (right):

https://codereview.chromium.org/1150723005/diff/1/src/compiler/js-operator.h#...
src/compiler/js-operator.h:117: VectorSlotPair(Handle<TypeFeedbackVector>
vector, FeedbackVectorICSlot slot)
On 2015/06/02 04:39:49, Benedikt Meurer wrote:
> The vector should be an input to the node. The operator should only have the
> constant slot. But I guess that should be cleanup work later since it involves
> changes to the global load, which Michael Stanton is working on as well.

Acknowledged. Yes, I already talked to Michael Stanton yesterday and told him
that I'd add another operator that takes the vector. I agree that we should do
that all in one go and for now keep JSLoadNamed and JSDynamicGlobalAccess in
sync.

[Michael Starzinger, 2015-06-02 09:29:25]

The CQ bit was checked by mstarzinger@chromium.org

[Michael Starzinger, 2015-06-02 09:29:26]

The patchset sent to the CQ was uploaded after l-g-t-m from bmeurer@chromium.org
Link to the patchset: https://codereview.chromium.org/1150723005/#ps80001
(title: "Rebased.")

[commit-bot: I haz the power, 2015-06-02 09:29:42]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1150723005/80001

[commit-bot: I haz the power, 2015-06-02 09:37:55]

Committed patchset #5 (id:80001)

[commit-bot: I haz the power, 2015-06-02 09:38:11]

Patchset 5 (id:??) landed as
https://crrev.com/440d099c5c2511836ba1b0be4823cd822b2e147f
Cr-Commit-Position: refs/heads/master@{#28750}