Remove CanRetainOtherContext since embedded objects are now weak.

Remove CanRetainOtherContext since embedded objects are now weak.
Instead of CanRetainOtherContext, we now manually blacklist all access-checked objects.

BUG=

Committed: https://crrev.com/accbe2216eb6b4690e1ae7617450c99058f64ff5
Cr-Commit-Position: refs/heads/master@{#27473}

[Toon Verwaest, 2015-03-26 09:44:36]

verwaest@chromium.org changed reviewers:
+ dcarney@chromium.org, ulan@chromium.org

[Toon Verwaest, 2015-03-26 09:44:36]

PTAL

[Erik Corry, 2015-03-26 10:09:43]

erik.corry@gmail.com changed reviewers:
+ erik.corry@gmail.com

[Erik Corry, 2015-03-26 10:09:43]

https://codereview.chromium.org/1020803004/diff/20001/src/ic/stub-cache.cc
File src/ic/stub-cache.cc (right):

https://codereview.chromium.org/1020803004/diff/20001/src/ic/stub-cache.cc#ne...
src/ic/stub-cache.cc:120: if (entry(primary_, offset) == &primary_[i]) {
Apart from avoiding leaks this also served to remove maps from other contexts
that the context-specific optimized code would not be expected to see. As such
it could make code monomorphic where it would otherwise be polymorphic.  Perhaps
it just needs renaming?

[Toon Verwaest, 2015-03-26 10:19:06]

https://codereview.chromium.org/1020803004/diff/20001/src/ic/stub-cache.cc
File src/ic/stub-cache.cc (right):

https://codereview.chromium.org/1020803004/diff/20001/src/ic/stub-cache.cc#ne...
src/ic/stub-cache.cc:120: if (entry(primary_, offset) == &primary_[i]) {
On 2015/03/26 10:09:43, Erik Corry wrote:
> Apart from avoiding leaks this also served to remove maps from other contexts
> that the context-specific optimized code would not be expected to see. As such
> it could make code monomorphic where it would otherwise be polymorphic. 
Perhaps
> it just needs renaming?

That's true, but I presume we don't need the entire recursive mechanism though.
That's what I was trying to avoid, in addition to fixing the memory leak due to
the fact that we still inlined cross-context despite this method.

[Toon Verwaest, 2015-03-26 10:40:15]

Addressed comment, ptal again

[ulan, 2015-03-26 10:47:24]

lgtm

[Toon Verwaest, 2015-03-26 10:48:26]

The CQ bit was checked by verwaest@chromium.org

[commit-bot: I haz the power, 2015-03-26 10:48:38]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1020803004/60001

[Erik Corry, 2015-03-26 11:20:57]

lgtm

https://codereview.chromium.org/1020803004/diff/60001/src/hydrogen.cc
File src/hydrogen.cc (right):

https://codereview.chromium.org/1020803004/diff/60001/src/hydrogen.cc#newcode...
src/hydrogen.cc:8652: top_info()->closure()->context()->native_context()) {
You have this several places.

How about SameNativeContext(function, top_info->closure())

https://codereview.chromium.org/1020803004/diff/60001/src/type-info.cc
File src/type-info.cc (right):

https://codereview.chromium.org/1020803004/diff/60001/src/type-info.cc#newcod...
src/type-info.cc:432: Handle<Map> map(maps.at(i));
Perhaps you don't need this variable if it's only used one place.

https://codereview.chromium.org/1020803004/diff/60001/test/mjsunit/constant-f...
File test/mjsunit/constant-folding-2.js (right):

https://codereview.chromium.org/1020803004/diff/60001/test/mjsunit/constant-f...
test/mjsunit/constant-folding-2.js:29: // Flags: --nodead-code-elimination
--fold-constants --allow-natives-syntax --nostress-opt
Deliberate change?

[commit-bot: I haz the power, 2015-03-26 11:21:58]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2015-03-26 11:22:16]

Patchset 4 (id:??) landed as
https://crrev.com/accbe2216eb6b4690e1ae7617450c99058f64ff5
Cr-Commit-Position: refs/heads/master@{#27473}

[Toon Verwaest, 2015-03-26 11:40:16]

https://codereview.chromium.org/1020803004/diff/60001/test/mjsunit/constant-f...
File test/mjsunit/constant-folding-2.js (right):

https://codereview.chromium.org/1020803004/diff/60001/test/mjsunit/constant-f...
test/mjsunit/constant-folding-2.js:29: // Flags: --nodead-code-elimination
--fold-constants --allow-natives-syntax --nostress-opt
On 2015/03/26 11:20:57, Erik Corry wrote:
> Deliberate change?

Yes. If we stress-opt, there's no time for cross-context feedback to be cleared
on context disposal notification.