track global accesses to constant types

track global accesses to constant types

BUG=

Committed: https://crrev.com/7bcc3d10835105fadc8ecf42a877dec479f0f80d
Cr-Commit-Position: refs/heads/master@{#27966}

[dcarney, 2015-04-21 07:20:07]

https://codereview.chromium.org/1062163005/diff/260001/src/code-stubs-hydroge...
File src/code-stubs-hydrogen.cc (right):

https://codereview.chromium.org/1062163005/diff/260001/src/code-stubs-hydroge...
src/code-stubs-hydrogen.cc:1371: // store_mode = STORE_TO_INITIALIZED_ENTRY;
i had to comment these out because there was some issues with smi not doing what
i expected them to do. is there something else i need to do?

https://codereview.chromium.org/1062163005/diff/260001/src/code-stubs-hydroge...
src/code-stubs-hydrogen.cc:1375: access =
access.WithRepresentation(Representation::Smi());
here i have a smi check and a withrepresentation. do we want to convert doubles
into smis in this code path or just deopt?

https://codereview.chromium.org/1062163005/diff/260001/src/hydrogen.cc
File src/hydrogen.cc (right):

https://codereview.chromium.org/1062163005/diff/260001/src/hydrogen.cc#newcod...
src/hydrogen.cc:6633: Add<HCheckSmi>(value);
same questions as in the code stub

https://codereview.chromium.org/1062163005/diff/260001/src/hydrogen.cc#newcod...
src/hydrogen.cc:6645: // TODO(dcarney): bailout here or deopt?
what to do here and above?

[Toon Verwaest, 2015-04-21 08:06:22]

verwaest@chromium.org changed reviewers:
+ verwaest@chromium.org

[Toon Verwaest, 2015-04-21 08:06:23]

lgtm
the deopt on unstable map is annoying, but I presume doesn't really happen all
that often.

https://codereview.chromium.org/1062163005/diff/260001/src/code-stubs-hydroge...
File src/code-stubs-hydrogen.cc (right):

https://codereview.chromium.org/1062163005/diff/260001/src/code-stubs-hydroge...
src/code-stubs-hydrogen.cc:1374: Add<HCheckSmi>(value);
This shouldn't be necessary; there'll be (should be) a t->s change due to
access.WithRepresentation(Smi) which will check.

https://codereview.chromium.org/1062163005/diff/260001/src/code-stubs-hydroge...
src/code-stubs-hydrogen.cc:1383: Add<HCheckHeapObject>(value);
This OTOH is necessary since there's no such thing as t->heapobject.

https://codereview.chromium.org/1062163005/diff/260001/src/code-stubs.h
File src/code-stubs.h (right):

https://codereview.chromium.org/1062163005/diff/260001/src/code-stubs.h#newco...
src/code-stubs.h:1136:
pattern.Add(Handle<Map>(global_map_placeholder(isolate())->map()),
handle(...) should work fine. Below the same.

[dcarney, 2015-04-21 13:35:23]

The CQ bit was checked by dcarney@chromium.org

[dcarney, 2015-04-21 13:35:23]

The patchset sent to the CQ was uploaded after l-g-t-m from
verwaest@chromium.org
Link to the patchset: https://codereview.chromium.org/1062163005/#ps280001
(title: " ")

[commit-bot: I haz the power, 2015-04-21 13:35:30]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1062163005/280001

[commit-bot: I haz the power, 2015-04-21 13:37:21]

Committed patchset #15 (id:280001)

[Michael Achenbach, 2015-04-21 18:02:48]

A revert of this CL (patchset #15 id:280001) has been created in
https://codereview.chromium.org/1094313003/ by machenbach@chromium.org.

The reason for reverting is: [Sheriff] Speculative revert for maybe breaking
layout tests, e.g. 
http://build.chromium.org/p/client.v8/builders/V8-Blink%20Linux%2032/builds/2589

Will reland if it doesn't help..

[Michael Achenbach, 2015-04-21 19:31:56]

The revert helped. More examples of the failure:
http://build.chromium.org/p/client.v8/builders/V8-Blink%20Win/builds/3016
http://build.chromium.org/p/client.v8/builders/V8-Blink%20Mac/builds/2896

[commit-bot: I haz the power, 2015-04-22 04:10:03]

Patchset 15 (id:??) landed as
https://crrev.com/7bcc3d10835105fadc8ecf42a877dec479f0f80d
Cr-Commit-Position: refs/heads/master@{#27966}