More instrumentation to track down net/ jank.

More instrumentation to track down net/ jank.

BUG=455942
Committed: https://crrev.com/e186724963e66587af264ef6c37701c70f34658e
Cr-Commit-Position: refs/heads/master@{#326687}

[eroman, 2015-04-23 21:21:56]

eroman@chromium.org changed reviewers:
+ mmenke@chromium.org

[mmenke, 2015-04-23 21:32:51]

LGTM

https://codereview.chromium.org/1105743002/diff/1/net/proxy/proxy_script_fetc...
File net/proxy/proxy_script_fetcher_impl.cc (right):

https://codereview.chromium.org/1105743002/diff/1/net/proxy/proxy_script_fetc...
net/proxy/proxy_script_fetcher_impl.cc:182: cur_request_->Start();
Hrm...Some poor soul is also stuck looking into jank here.  Suppose proxies are
rare enough that if it were just the same cause of jank, it probably wouldn't be
enough to show up on the dashboard, though.

[eroman, 2015-04-23 22:46:03]

The CQ bit was checked by eroman@chromium.org

[commit-bot: I haz the power, 2015-04-23 22:47:38]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1105743002/1

[eroman, 2015-04-23 22:50:46]

https://codereview.chromium.org/1105743002/diff/1/net/proxy/proxy_script_fetc...
File net/proxy/proxy_script_fetcher_impl.cc (right):

https://codereview.chromium.org/1105743002/diff/1/net/proxy/proxy_script_fetc...
net/proxy/proxy_script_fetcher_impl.cc:182: cur_request_->Start();
On 2015/04/23 21:32:51, mmenke (Out 4-27 to 5-1) wrote:
> Hrm...Some poor soul is also stuck looking into jank here.  Suppose proxies
are
> rare enough that if it were just the same cause of jank, it probably wouldn't
be
> enough to show up on the dashboard, though.

Heh, I wonder who that poor soul might be...
I guess we will see what the data reveals. I am actually curious to see what
this one turns out to be, since I would have guessed URLRequest::Start() would
be the slow thing in this function (but that is already being measured).

Also got me thinking, what happens with ScopedTracker for re-entrency? Maybe it
is posible for URLRequest::Start() to call URLRequest::Start() again by way of
ProxyScriptFetcherImpl. Hmm.

[mmenke, 2015-04-23 23:04:39]

On 2015/04/23 22:50:46, eroman wrote:
>
https://codereview.chromium.org/1105743002/diff/1/net/proxy/proxy_script_fetc...
> File net/proxy/proxy_script_fetcher_impl.cc (right):
> 
>
https://codereview.chromium.org/1105743002/diff/1/net/proxy/proxy_script_fetc...
> net/proxy/proxy_script_fetcher_impl.cc:182: cur_request_->Start();
> On 2015/04/23 21:32:51, mmenke (Out 4-27 to 5-1) wrote:
> > Hrm...Some poor soul is also stuck looking into jank here.  Suppose proxies
> are
> > rare enough that if it were just the same cause of jank, it probably
wouldn't
> be
> > enough to show up on the dashboard, though.
> 
> Heh, I wonder who that poor soul might be...
> I guess we will see what the data reveals. I am actually curious to see what
> this one turns out to be, since I would have guessed URLRequest::Start() would
> be the slow thing in this function (but that is already being measured).
> 
> Also got me thinking, what happens with ScopedTracker for re-entrency? Maybe
it
> is posible for URLRequest::Start() to call URLRequest::Start() again by way of
> ProxyScriptFetcherImpl. Hmm.

My guess, assuming you don't get some sort of singularity that sucks the entire
planet in, is that you just get two different profiler data points, which cover
disjoint times.

[commit-bot: I haz the power, 2015-04-23 23:07:52]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-04-23 23:08:54]

Patchset 1 (id:??) landed as
https://crrev.com/e186724963e66587af264ef6c37701c70f34658e
Cr-Commit-Position: refs/heads/master@{#326687}