Turbofan: Make type feedback vector a Node.

Turbofan: Make type feedback vector a Node.

This improves code generation. Before, it was a constant loaded in a register
every time.

R=mstarzinger@chromium.org
BUG=

Committed: https://crrev.com/662a5585ee845e35460459c7b47e9750f1cdc0c1
Cr-Commit-Position: refs/heads/master@{#28874}

[mvstanton, 2015-06-09 12:37:18]

Hi Michael, here is the CL we discussed. Thanks for the help!
--Michael

[Michael Starzinger, 2015-06-09 13:23:30]

Looking good, mostly nits.

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/ast-graph-...
File src/compiler/ast-graph-builder.cc (right):

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:3451: Node*
AstGraphBuilder::BuildKeyedLoad(Node* object, Node* key, Node* vector,
nit: Instead of passing the vector as an argument, let's just call
GetFeedbackVector() within this helper.

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:3459: Node*
AstGraphBuilder::BuildNamedLoad(Node* object, Node* vector,
nit: Likewise.

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/ast-graph-...
File src/compiler/ast-graph-builder.h (right):

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.h:240: ResolvedFeedbackSlot
CreateResolvedFeedbackSlot(
nit: We could just call the function "ResolveFeedbackSlot" instead. WDYT?

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/js-operator.h
File src/compiler/js-operator.h (right):

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/js-operato...
src/compiler/js-operator.h:115: // A FeedbackVectorICSlot needs to query the
type feedback vector to get it's
nit: s/FeedbackVectorICSlot/ResolvedFeedbackSlot/ in the comment.

https://codereview.chromium.org/1175503002/diff/20001/test/unittests/compiler...
File test/unittests/compiler/js-typed-lowering-unittest.cc (right):

https://codereview.chromium.org/1175503002/diff/20001/test/unittests/compiler...
test/unittests/compiler/js-typed-lowering-unittest.cc:691: 
nit: Drop the empty newline for consistency.

[mvstanton, 2015-06-09 15:21:27]

Here you go, thanks for the look,
--Michael

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/ast-graph-...
File src/compiler/ast-graph-builder.cc (right):

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:3451: Node*
AstGraphBuilder::BuildKeyedLoad(Node* object, Node* key, Node* vector,
On 2015/06/09 13:23:30, Michael Starzinger wrote:
> nit: Instead of passing the vector as an argument, let's just call
> GetFeedbackVector() within this helper.

Ah, good point. Done.

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:3459: Node*
AstGraphBuilder::BuildNamedLoad(Node* object, Node* vector,
On 2015/06/09 13:23:30, Michael Starzinger wrote:
> nit: Likewise.

Done.

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/ast-graph-...
File src/compiler/ast-graph-builder.h (right):

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.h:240: ResolvedFeedbackSlot
CreateResolvedFeedbackSlot(
On 2015/06/09 13:23:30, Michael Starzinger wrote:
> nit: We could just call the function "ResolveFeedbackSlot" instead. WDYT?

Yep, done.

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/js-operator.h
File src/compiler/js-operator.h (right):

https://codereview.chromium.org/1175503002/diff/20001/src/compiler/js-operato...
src/compiler/js-operator.h:115: // A FeedbackVectorICSlot needs to query the
type feedback vector to get it's
On 2015/06/09 13:23:30, Michael Starzinger wrote:
> nit: s/FeedbackVectorICSlot/ResolvedFeedbackSlot/ in the comment.

doh! Done.

https://codereview.chromium.org/1175503002/diff/20001/test/unittests/compiler...
File test/unittests/compiler/js-typed-lowering-unittest.cc (right):

https://codereview.chromium.org/1175503002/diff/20001/test/unittests/compiler...
test/unittests/compiler/js-typed-lowering-unittest.cc:691: 
On 2015/06/09 13:23:30, Michael Starzinger wrote:
> nit: Drop the empty newline for consistency.

Done.

[Michael Starzinger, 2015-06-09 15:28:35]

LGTM. Thanks!

[Michael Starzinger, 2015-06-09 15:29:09]

The CQ bit was checked by mstarzinger@chromium.org

[commit-bot: I haz the power, 2015-06-09 15:29:28]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1175503002/60001

[commit-bot: I haz the power, 2015-06-09 16:02:32]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2015-06-09 16:02:40]

Patchset 4 (id:??) landed as
https://crrev.com/662a5585ee845e35460459c7b47e9750f1cdc0c1
Cr-Commit-Position: refs/heads/master@{#28874}