Add fallback mechanism to release Extension ports if the JS context has been destroyed.

Add fallback mechanism to release Extension ports if the JS context has been destroyed.

The bug is that chrome.runtime.sendMessage looks for when its callback is
garbage collected, and when it is, close the associated port. It's important to
close the port to notify the other end of the closure, cleanup renderer state
both locally and on that other end, and potentially browser state.

Unfortunately the port management is implemented in JS itself, and port
releasing needs to go through JS. The problem is that it's not possible to call
into JS while in the process of garbage collection, so we delay it, by which
point the JS context may have been destroyed and again it's not possible to
call into JS. We fixed the former case a while ago, and this patch fixes the
latter.

BUG=471599
R=rockot@chromium.org

Committed: https://crrev.com/70c00e241c68895fb245f53a6404866883ea0631
Cr-Commit-Position: refs/heads/master@{#330234}

[not at google - send to devlin, 2015-05-15 18:14:24]

LMK. I annotated the places where this differs.

https://codereview.chromium.org/1136953017/diff/20001/extensions/renderer/mes...
File extensions/renderer/messaging_bindings.cc (right):

https://codereview.chromium.org/1136953017/diff/20001/extensions/renderer/mes...
extensions/renderer/messaging_bindings.cc:63: class GCCallback : public
base::SupportsWeakPtr<GCCallback> {
Moved this up here because that's where it belongs, and I think it's easier to
review in isolation anyway?

Btw I believe SupportsWeakPtr here is fine because it's not part of a public
interface; and you'd need to something else funky to make link 89 work (function
is static).

https://codereview.chromium.org/1136953017/diff/20001/extensions/renderer/mes...
extensions/renderer/messaging_bindings.cc:74: context->AddInvalidationObserver(
Now watches for context invalidation.

https://codereview.chromium.org/1136953017/diff/20001/extensions/renderer/mes...
extensions/renderer/messaging_bindings.cc:89:
base::Bind(&GCCallback::RunCallback, data.GetParameter()->AsWeakPtr()));
Changed this from being base::Owned to a WeakPtr with explicit lifetime handling
(see "delete this").

https://codereview.chromium.org/1136953017/diff/20001/extensions/renderer/mes...
extensions/renderer/messaging_bindings.cc:96:
context_->CallFunction(v8::Local<v8::Function>::New(isolate, callback_));
Now uses the proper CallFunction method on context, rather than implementing it
itself.

https://codereview.chromium.org/1136953017/diff/20001/extensions/renderer/mes...
extensions/renderer/messaging_bindings.cc:268: if (port_id >= 0) {
The hack in question.

[Ken Rockot(use gerrit already), 2015-05-15 21:47:54]

lgtm

https://codereview.chromium.org/1136953017/diff/20001/extensions/renderer/mes...
File extensions/renderer/messaging_bindings.cc (right):

https://codereview.chromium.org/1136953017/diff/20001/extensions/renderer/mes...
extensions/renderer/messaging_bindings.cc:272: new GCCallback(context(),
args[0].As<v8::Object>(),
nit: maybe document that this kills itself

[not at google - send to devlin, 2015-05-15 21:53:12]

https://codereview.chromium.org/1136953017/diff/20001/extensions/renderer/mes...
File extensions/renderer/messaging_bindings.cc (right):

https://codereview.chromium.org/1136953017/diff/20001/extensions/renderer/mes...
extensions/renderer/messaging_bindings.cc:272: new GCCallback(context(),
args[0].As<v8::Object>(),
On 2015/05/15 21:47:54, Ken Rockot wrote:
> nit: maybe document that this kills itself

Done.

[not at google - send to devlin, 2015-05-15 21:53:17]

The CQ bit was checked by kalman@chromium.org

[not at google - send to devlin, 2015-05-15 21:53:17]

The patchset sent to the CQ was uploaded after l-g-t-m from rockot@chromium.org
Link to the patchset: https://codereview.chromium.org/1136953017/#ps40001
(title: "comments")

[commit-bot: I haz the power, 2015-05-15 21:55:24]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1136953017/40001

[commit-bot: I haz the power, 2015-05-15 23:42:33]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-05-18 11:29:14]

Patchset 3 (id:??) landed as
https://crrev.com/70c00e241c68895fb245f53a6404866883ea0631
Cr-Commit-Position: refs/heads/master@{#330234}