Fix unobservable constructor replacement on prototype maps

Fix unobservable constructor replacement on prototype maps

BUG=chromium:478522
LOG=y
R=verwaest@chromium.org

Committed: https://crrev.com/22f2b13fa8df33e4a0dc58860577a242a4aa5f86
Cr-Commit-Position: refs/heads/master@{#28126}

[Jakob Kummerow, 2015-04-20 17:20:53]

PTAL.

See the referenced bug for background. Apparently this was broken by
https://codereview.chromium.org/942493002. I'm not entirely sure this is the
right fix; feel free to suggest something else.
AFAICT there's indeed no way to test this behavior from JS; however the API
exposes it, and DevTools uses it.

I suspect this CL undoes most of the benefit of the original change. Would you
prefer to WontFix the bug?

[Toon Verwaest, 2015-04-20 23:00:30]

Yeah this indeed breaks the entire optimization. If we really want to support
the alternative, I guess we just need to revert the original change.
I'd prefer to what the spec requires though. If clients want to customize what
they see, they should rely on harmony_tostring (toStringTag) at most; not
internal implementation details (whether or not exposed by devtools).

[Jakob Kummerow, 2015-04-21 12:46:11]

OK, let's drop this approach then.

We could use the PrototypeInfo to store the original constructor's name on a
prototype map (and keep replacing the constructor pointer). WDYT?

By "clients [relying] on harmony_tostring", you mean individual scripts, right?
Sure, they could come up with any metadata that they want; however DevTools'
builtin console printing provided a nice convenience feature here, which V8 is
breaking. I can see the value in continuing to support the existing API.

[Toon Verwaest, 2015-04-21 13:12:53]

Good idea, that makes sense.

[Jakob Kummerow, 2015-04-27 13:31:55]

New approach implemented; unfortunately the patch set includes a rebase.

The testcase is unchanged, everything else is rewritten from scratch.

PTAL.

[Toon Verwaest, 2015-04-27 15:05:42]

lgtm

[Jakob Kummerow, 2015-04-28 14:12:24]

The CQ bit was checked by jkummerow@chromium.org

[commit-bot: I haz the power, 2015-04-28 14:12:33]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1097113003/40001

[commit-bot: I haz the power, 2015-04-28 14:25:45]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-04-28 14:25:45]

Try jobs failed on following builders:
  v8_linux_nodcheck_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_nodcheck_rel/build...)

[Jakob Kummerow, 2015-04-29 09:13:51]

The CQ bit was checked by jkummerow@chromium.org

[commit-bot: I haz the power, 2015-04-29 09:14:03]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1097113003/40001

[commit-bot: I haz the power, 2015-04-29 09:31:43]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-04-29 09:32:01]

Patchset 3 (id:??) landed as
https://crrev.com/22f2b13fa8df33e4a0dc58860577a242a4aa5f86
Cr-Commit-Position: refs/heads/master@{#28126}