Avoid leaking a shared worker global scope on a failed script load.

Avoid leaking a shared worker global scope on a failed script load.

The spec tells us,

 http://whatwg.org/specs/web-apps/current-work/#run-a-worker

that in case of failure to load the script of a (shared) worker, we
should report the failure and abort (step 4.) The global scope's
closing flag will be left as false, meaning that we can safely shut
down and destroy the shared worker global scope. Do so.

R=
BUG=364390
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=177178

[sof, 2014-06-27 14:56:03]

Please take a look.

[sof, 2014-06-28 06:16:45]

On 2014/06/27 14:56:03, sof wrote:
> Please take a look.

Seeing one or two flaky linux_blink_dbg tests that needs investigation; please
don't spend time on this until I've understood those failures.

[sof, 2014-06-28 15:05:33]

On 2014/06/28 06:16:45, sof wrote:
> On 2014/06/27 14:56:03, sof wrote:
> > Please take a look.
> 
> Seeing one or two flaky linux_blink_dbg tests that needs investigation; please
> don't spend time on this until I've understood those failures.

It was caused by a shared worker being terminated before its script had loaded.
That will cancel the load and trigger a (cancellation) failure. That code path
better not delete the WebSharedWorkerImpl, as it is on the stack further up (in
stopWorkerThread()), but just let the shutdown of the WorkerGlobalScope handle
destruction of WebSharedWorkerImpl as per normal. Adjusted the CL.

Notice that for this very codepath ToT will delete the loader while a call to
cancel() is further up the stack (i.e., in stopWorkerThread()). That seems like
a risky thing to do; avoided now.

(The tree is currently broken, so not able to run a full set of tests on the
bots.)

[sof, 2014-06-28 19:51:19]

On 2014/06/28 15:05:33, sof wrote:
...
> 
> (The tree is currently broken, so not able to run a full set of tests on the
> bots.)

Bot results now available; nothing untowards showing.

[horo, 2014-06-30 01:16:21]

LGTM

Thank you for fixing this.

[tkent, 2014-06-30 01:17:04]

rslgtm

[tkent, 2014-06-30 01:17:09]

The CQ bit was checked by tkent@chromium.org

[commit-bot: I haz the power, 2014-06-30 01:17:57]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/sigbjornf@opera.com/361443002/40001

[commit-bot: I haz the power, 2014-06-30 02:03:23]

Change committed as 177178