Refactor compilation dependency handling.

Refactor compilation dependency handling.

Extract a new data structure CompilationDependencies and move (most) logic there.

R=mstarzinger@chromium.org,verwaest@chromium.org
BUG=

Committed: https://crrev.com/b882479f1c84a48961b8aec81fa1bb1225034784
Cr-Commit-Position: refs/heads/master@{#27892}

[titzer, 2015-04-16 09:52:00]

PTAL much goodness

[Toon Verwaest, 2015-04-16 13:38:19]

the readability is unbearable!
lgtm with suggestion

https://codereview.chromium.org/1095433002/diff/1/src/dependencies.cc
File src/dependencies.cc (right):

https://codereview.chromium.org/1095433002/diff/1/src/dependencies.cc#newcode37
src/dependencies.cc:37: if (object->IsMap()) {
What about adding something similar to DependentCode::ForObject?

[titzer, 2015-04-16 14:32:10]

On 2015/04/16 13:38:19, Toon Verwaest wrote:
> the readability is unbearable!
> lgtm with suggestion
> 
> https://codereview.chromium.org/1095433002/diff/1/src/dependencies.cc
> File src/dependencies.cc (right):
> 
>
https://codereview.chromium.org/1095433002/diff/1/src/dependencies.cc#newcode37
> src/dependencies.cc:37: if (object->IsMap()) {
> What about adding something similar to DependentCode::ForObject?

I've moved that method over and factored out the setting so it's symmetric.

[Michael Starzinger, 2015-04-16 14:36:43]

LGTM. I love it.

https://codereview.chromium.org/1095433002/diff/40001/src/dependencies.h
File src/dependencies.h (right):

https://codereview.chromium.org/1095433002/diff/40001/src/dependencies.h#newc...
src/dependencies.h:5: #ifndef V8_DEPENDENCIES_H_
suggestion: What do you think about calling this file
"compilation-dependencies.h" instead?

https://codereview.chromium.org/1095433002/diff/40001/src/dependencies.h#newc...
src/dependencies.h:34: void AssumePropertyCell(Handle<PropertyCell> cell) {
nit: Maybe AssumePropertyCellValue() or AssumePropertyCellContent() is clearer?

[titzer, 2015-04-16 14:47:27]

https://codereview.chromium.org/1095433002/diff/40001/src/dependencies.h
File src/dependencies.h (right):

https://codereview.chromium.org/1095433002/diff/40001/src/dependencies.h#newc...
src/dependencies.h:5: #ifndef V8_DEPENDENCIES_H_
On 2015/04/16 14:36:42, Michael Starzinger wrote:
> suggestion: What do you think about calling this file
> "compilation-dependencies.h" instead?

Done.

https://codereview.chromium.org/1095433002/diff/40001/src/dependencies.h#newc...
src/dependencies.h:34: void AssumePropertyCell(Handle<PropertyCell> cell) {
On 2015/04/16 14:36:42, Michael Starzinger wrote:
> nit: Maybe AssumePropertyCellValue() or AssumePropertyCellContent() is
clearer?

As discussed in person, we take dependencies on property cells if the code
depends either on it being a constant or just existing (i.e. not being deleted).
We could separate the cases in the future.

[titzer, 2015-04-16 14:48:28]

The CQ bit was checked by titzer@chromium.org

[titzer, 2015-04-16 14:48:28]

The patchset sent to the CQ was uploaded after l-g-t-m from
verwaest@chromium.org, mstarzinger@chromium.org
Link to the patchset: https://codereview.chromium.org/1095433002/#ps60001
(title: " ")

[commit-bot: I haz the power, 2015-04-16 14:48:36]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1095433002/60001

[commit-bot: I haz the power, 2015-04-16 16:04:48]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2015-04-16 16:05:00]

Patchset 4 (id:??) landed as
https://crrev.com/b882479f1c84a48961b8aec81fa1bb1225034784
Cr-Commit-Position: refs/heads/master@{#27892}

[dcarney, 2015-04-16 19:02:03]

> As discussed in person, we take dependencies on property cells if the code
> depends either on it being a constant or just existing (i.e. not being
deleted).
> We could separate the cases in the future.

I just ran across this name and found it confusing as hell. 
AssumePropertyCellValid might be better here, as most of the things that trigger
deopt go through PropertyCell::InvalidateEntry.

[Michael Achenbach, 2015-04-17 07:46:54]

A revert of this CL (patchset #4 id:60001) has been created in
https://codereview.chromium.org/1093783002/ by machenbach@chromium.org.

The reason for reverting is: [Sheriff] Causes crashes in laout tests:
http://build.chromium.org/p/client.v8/builders/V8-Blink%20Linux%2064%20%28dbg...

Extra bisect run:
http://build.chromium.org/p/client.v8/builders/V8-Blink%20Linux%2064%20%28dbg....