Skip the Service Worker CORS fallback for same origin requests. [2/2 chromium]

Skip the Service Worker CORS fallback for same origin requests. [2/2 chromium]

We are sending CORS fallback response from the browser process to the page's renderer process, when the ServiceWorker doesn't handle the CORS fetch request.
But when the request URL's origin is same as the page's security origin, we can skip the fallback and send the request from the browser process.

This change will improve the delay of the CORS fallback for same origin requests, especially when the renderer main thread is busy.

1/2 blink: https://codereview.chromium.org/1243353002/
2/2 chromium: this cl.

BUG=512764
Committed: https://crrev.com/97b13409b0c601f21f048744fa8e31852b476337
Cr-Commit-Position: refs/heads/master@{#341482}

[horo, 2015-07-22 09:06:16]

Patchset #1 (id:1) has been deleted

[horo, 2015-07-22 09:15:45]

Patchset #1 (id:20001) has been deleted

[horo, 2015-07-22 09:26:20]

horo@chromium.org changed reviewers:
+ falken@chromium.org

[horo, 2015-07-22 09:26:21]

falken@
Could you please review this?

[falken, 2015-07-22 22:12:42]

Why is request_mode CORS when the request is not cross-origin?

[horo, 2015-07-23 03:20:49]

On 2015/07/22 22:12:42, falken wrote:
> Why is request_mode CORS when the request is not cross-origin?

When the server returns 30x redirect response which location header is another
origin,

1. if the request_mode is CORS, Chrome sends the request to the new origin and
go the CORS algorithm (cf:check Access-Control-Allow-* headers).

2. if the request_mode is same-origin, the fetch results in error.

[falken, 2015-07-23 07:02:48]

falken@chromium.org changed reviewers:
+ tyoshino@chromium.org

[falken, 2015-07-23 07:02:48]

OK I think I see. If request mode were same-origin, a response that redirects
cross-origin would result in an error.

lgtm but I think a CORS expert (tyoshino?) better check that it's ok to skip the
CORS fallback.

[tyoshino (SeeGerritForStatus), 2015-07-23 07:29:59]

On 2015/07/23 07:02:48, falken wrote:
> OK I think I see. If request mode were same-origin, a response that redirects
> cross-origin would result in an error.
> 
> lgtm but I think a CORS expert (tyoshino?) better check that it's ok to skip
the
> CORS fallback.

I think it would work as far as FALLBACK_TO_NETWORK behaves as if
skipServiceWorker() were specified.

All we need to ensure in Blink is that the condition to proceed without fallback
is covered by the following condition in dispatchInitialRequest():

if (m_sameOriginRequest || m_options.crossOriginRequestPolicy ==
AllowCrossOriginRequests) {

[horo, 2015-07-31 02:26:09]

The CQ bit was checked by horo@chromium.org

[commit-bot: I haz the power, 2015-07-31 02:26:19]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1242953007/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1242953007/40001

[horo, 2015-07-31 02:27:39]

The CQ bit was unchecked by horo@chromium.org

[horo, 2015-08-03 02:10:28]

The CQ bit was checked by horo@chromium.org

[commit-bot: I haz the power, 2015-08-03 02:10:39]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1242953007/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1242953007/40001

[commit-bot: I haz the power, 2015-08-03 07:00:51]

Committed patchset #1 (id:40001)

[commit-bot: I haz the power, 2015-08-03 07:01:28]

Patchset 1 (id:??) landed as
https://crrev.com/97b13409b0c601f21f048744fa8e31852b476337
Cr-Commit-Position: refs/heads/master@{#341482}

[johnme, 2015-08-03 08:47:21]

A revert of this CL (patchset #1 id:40001) has been created in
https://codereview.chromium.org/1258893004/ by johnme@chromium.org.

The reason for reverting is: This broke
http/tests/serviceworker/fetch-request-fallback.html, causing failures on 3 bots
and timeouts on 18 bots.

See
http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=htt....