Reland: track global accesses to constant types

Reland: track global accesses to constant types

R=verwaest@chromium.org
BUG=468620
LOG=N

Committed: https://crrev.com/1d2be2af40d5741741c1614a6ae680dd2c604a6c
Cr-Commit-Position: refs/heads/master@{#28081}

[dcarney, 2015-04-22 11:36:15]

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

[dcarney, 2015-04-22 11:36:52]

ptal. this fixes that crash by bailing out instead of messing with the cell. it
rarely happens so should be okay

[dcarney, 2015-04-22 11:55:27]

Patchset #2 (id:20001) has been deleted

[Toon Verwaest, 2015-04-22 14:42:04]

lgtm

[dcarney, 2015-04-24 09:00:13]

Patchset #2 (id:40001) has been deleted

[dcarney, 2015-04-27 13:26:17]

The CQ bit was checked by dcarney@chromium.org

[dcarney, 2015-04-27 13:26:18]

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/1102543002/#ps80001
(title: " ")

[commit-bot: I haz the power, 2015-04-27 13:26:23]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1102543002/80001

[commit-bot: I haz the power, 2015-04-27 14:01:20]

Committed patchset #3 (id:80001)

[commit-bot: I haz the power, 2015-04-27 14:01:33]

Patchset 3 (id:??) landed as
https://crrev.com/1d2be2af40d5741741c1614a6ae680dd2c604a6c
Cr-Commit-Position: refs/heads/master@{#28081}

[brucedawson, 2015-04-29 20:55:12]

This change makes kUninitialized and kUndefined have the same value, where they
used to be distinct. I noticed this because the /analyze builder gave a new
warning pointing out that there was some redundant code. The warning is:

src\v8\src\objects.cc(17173) : warning C6287: Redundant code:  the left and
right sub-expressions are identical.

The code is this:

  if ((old_type == PropertyCellType::kUninitialized ||
       old_type == PropertyCellType::kUndefined) &&
      value->IsString()) {

This isn't necessarily a bug but I wanted to point it out. If nothing else the
code is now confusing. FWIW.

[dcarney, 2015-04-30 09:00:46]

On 2015/04/29 20:55:12, brucedawson wrote:
> This change makes kUninitialized and kUndefined have the same value, where
they
> used to be distinct. I noticed this because the /analyze builder gave a new
> warning pointing out that there was some redundant code. The warning is:
> 
> src\v8\src\objects.cc(17173) : warning C6287: Redundant code:  the left and
> right sub-expressions are identical.
> 
> The code is this:
> 
>   if ((old_type == PropertyCellType::kUninitialized ||
>        old_type == PropertyCellType::kUndefined) &&
>       value->IsString()) {
> 
> This isn't necessarily a bug but I wanted to point it out. If nothing else the
> code is now confusing. FWIW.

thanks, will fix