Add periodic check to see if context has been reset

Add periodic check to see if context has been reset.
Context may be set to NULL when shutdown is requested
while trying to send ping. With this CL, rlz will cancel
the request when it detects the NULL context.

non chromeos chrome used to wait even after context is set
to NULL, so maybe I should do this only for chromeos. Please
let me know what you think.

This also fixes the edge race case where the g_context
may be set to NULL in the middle of FinancialPing::PingServer.

BUG=261377
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=229114

[Roger Tawa OOO till Jul 10th, 2013-10-17 00:18:18]

Looks good Mitsuru.

Since g_context is being read and written from different threads, maybe it
should be protected by a lock?

[oshima, 2013-10-17 00:55:56]

On 2013/10/17 00:18:18, Roger Tawa wrote:
> Looks good Mitsuru.
> 
> Since g_context is being read and written from different threads, maybe it
> should be protected by a lock?

Changed to use AtomicWord because it doesn't really need a lock.
(URLRequestContextGetter itself is thread safe)

[Roger Tawa OOO till Jul 10th, 2013-10-17 01:03:21]

lgtm

Thanks Mitsuru!  A couple of comments below.

https://codereview.chromium.org/26929004/diff/21001/rlz/lib/financial_ping.cc
File rlz/lib/financial_ping.cc (right):

https://codereview.chromium.org/26929004/diff/21001/rlz/lib/financial_ping.cc...
rlz/lib/financial_ping.cc:193: reinterpret_cast<AtomicWord>(context));
Do you need the first reinterpret_cast<>?

https://codereview.chromium.org/26929004/diff/21001/rlz/lib/financial_ping.cc...
rlz/lib/financial_ping.cc:228: if (!g_context) {
Should use nobarrier_load here too.

[oshima, 2013-10-17 01:09:06]

https://codereview.chromium.org/26929004/diff/21001/rlz/lib/financial_ping.cc
File rlz/lib/financial_ping.cc (right):

https://codereview.chromium.org/26929004/diff/21001/rlz/lib/financial_ping.cc...
rlz/lib/financial_ping.cc:193: reinterpret_cast<AtomicWord>(context));
On 2013/10/17 01:03:22, Roger Tawa wrote:
> Do you need the first reinterpret_cast<>?

Done.

https://codereview.chromium.org/26929004/diff/21001/rlz/lib/financial_ping.cc...
rlz/lib/financial_ping.cc:228: if (!g_context) {
On 2013/10/17 01:03:22, Roger Tawa wrote:
> Should use nobarrier_load here too.

Done.

[commit-bot: I haz the power, 2013-10-17 01:46:15]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/oshima@chromium.org/26929004/29001

[commit-bot: I haz the power, 2013-10-17 03:33:22]

Step "update" is always a major failure.
Look at the try server FAQ for more details.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[commit-bot: I haz the power, 2013-10-17 03:49:18]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/oshima@chromium.org/26929004/29001

[commit-bot: I haz the power, 2013-10-17 05:29:12]

Step "update" is always a major failure.
Look at the try server FAQ for more details.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[commit-bot: I haz the power, 2013-10-17 05:32:29]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/oshima@chromium.org/26929004/29001

[commit-bot: I haz the power, 2013-10-17 13:12:25]

Change committed as 229114