Use a WeakCell in the CallIC type vector.

Use a WeakCell in the CallIC type vector.

This allows us to clear the IC on a more sedate schedule, just
like Load and Store ICs.

R=ulan@chromium.org
BUG=

Committed: https://crrev.com/bcc79d33ca6d97d9ecfcfcf110a6ea84a0225389
Cr-Commit-Position: refs/heads/master@{#26332}

[mvstanton, 2015-01-28 13:11:01]

Patchset #3 (id:40001) has been deleted

[mvstanton, 2015-01-28 13:11:22]

Hi Ulan,
Here is the weak cell handling for CallICs. Thanks for the look!
--Michael

[ulan, 2015-01-28 13:35:41]

LGTM!

[mvstanton, 2015-01-29 16:16:25]

Thanks for the look at this Ulan.
As you see, I had to repair a regression.
I need to preserve the behavior that avoids a MISS to the
runtime on the UNINITIALIZED to MONOMORPHIC transition,
even though setting up the vector for MONOMORPHIC is a
bit more complicated now.

This seems to do the job...thanks for having another glance,
--Michael

[ulan, 2015-01-29 16:36:28]

On 2015/01/29 16:16:25, mvstanton wrote:
> Thanks for the look at this Ulan.
> As you see, I had to repair a regression.
> I need to preserve the behavior that avoids a MISS to the
> runtime on the UNINITIALIZED to MONOMORPHIC transition,
> even though setting up the vector for MONOMORPHIC is a
> bit more complicated now.
> 
> This seems to do the job...thanks for having another glance,
> --Michael

New patch set lgtm.

[mvstanton, 2015-01-29 16:52:31]

The CQ bit was checked by mvstanton@chromium.org

[commit-bot: I haz the power, 2015-01-29 16:52:52]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/881433002/80001

[commit-bot: I haz the power, 2015-01-29 17:36:37]

Committed patchset #4 (id:80001)

[commit-bot: I haz the power, 2015-01-29 17:37:00]

Patchset 4 (id:??) landed as
https://crrev.com/bcc79d33ca6d97d9ecfcfcf110a6ea84a0225389
Cr-Commit-Position: refs/heads/master@{#26332}

[Michael Achenbach, 2015-01-29 18:00:01]

A revert of this CL (patchset #4 id:80001) has been created in
https://codereview.chromium.org/886663004/ by machenbach@chromium.org.

The reason for reverting is: Breaks a regression test on linux isloates, and
linux with gcc 4.8..