Convert to immutable heap number when materializing arguments object.

Convert to immutable heap number when materializing arguments object.

BUG=chromium:457935
LOG=n
R=ishell@chromium.org

Committed: https://crrev.com/3f3558f3657286fc45480eec44fc699286723280
Cr-Commit-Position: refs/heads/master@{#26704}

[Jarin, 2015-02-17 15:00:56]

Could you take a look, please?

[Igor Sheludko, 2015-02-17 15:07:54]

https://codereview.chromium.org/935623002/diff/1/src/deoptimizer.cc
File src/deoptimizer.cc (right):

https://codereview.chromium.org/935623002/diff/1/src/deoptimizer.cc#newcode3393
src/deoptimizer.cc:3393: if (value->IsMutableHeapNumber()) {
I think we could save some cycles if we put this code into TAGGED case of
SlotRef::GetValue().

https://codereview.chromium.org/935623002/diff/1/test/mjsunit/regress/regress...
File test/mjsunit/regress/regress-457935.js (right):

https://codereview.chromium.org/935623002/diff/1/test/mjsunit/regress/regress...
test/mjsunit/regress/regress-457935.js:9: function __f_25() {
I think it would be better to get rid of __X_ ugliness.

[Jarin, 2015-02-17 15:20:48]

https://codereview.chromium.org/935623002/diff/1/src/deoptimizer.cc
File src/deoptimizer.cc (right):

https://codereview.chromium.org/935623002/diff/1/src/deoptimizer.cc#newcode3393
src/deoptimizer.cc:3393: if (value->IsMutableHeapNumber()) {
On 2015/02/17 15:07:54, Igor Sheludko wrote:
> I think we could save some cycles if we put this code into TAGGED case of
> SlotRef::GetValue().

Done. (Even though there is so much stuff going on here that couple of cycles
cannot make any difference.)

https://codereview.chromium.org/935623002/diff/1/test/mjsunit/regress/regress...
File test/mjsunit/regress/regress-457935.js (right):

https://codereview.chromium.org/935623002/diff/1/test/mjsunit/regress/regress...
test/mjsunit/regress/regress-457935.js:9: function __f_25() {
On 2015/02/17 15:07:54, Igor Sheludko wrote:
> I think it would be better to get rid of __X_ ugliness.

Done (but we do keep the underscores in other tests and I use it as a clue that
I should not try too hard to understand the code.)

[Jarin, 2015-02-17 15:32:08]

On 2015/02/17 15:20:48, jarin wrote:
> On 2015/02/17 15:07:54, Igor Sheludko wrote:
> > I think we could save some cycles if we put this code into TAGGED case of
> > SlotRef::GetValue().

As discussed offline, I had misunderstood (and just moved the check around in
the same method). GetValue is indeed a better home for the check. Fixed now.

[Igor Sheludko, 2015-02-17 15:34:09]

lgtm

[Jarin, 2015-02-17 16:55:12]

The CQ bit was checked by jarin@chromium.org

[commit-bot: I haz the power, 2015-02-17 16:55:25]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/935623002/40001

[commit-bot: I haz the power, 2015-02-17 18:08:59]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-02-17 18:09:08]

Patchset 3 (id:??) landed as
https://crrev.com/3f3558f3657286fc45480eec44fc699286723280
Cr-Commit-Position: refs/heads/master@{#26704}