Fix bug when a NavigationThrottle blocks a request during a redirect.

Fix bug when a NavigationThrottle blocks a request during a redirect.

When PlzNavigate is **disabled**, when a request is blocked on a
redirect and the browser asks the renderer to commit an error page,
blink doesn't know about the latest redirect and use the post-redirect
URL. This cause a mismatch in TakeNavigationHandleForCommit.

This CL doesn't make blink aware of the redirect, but allows the
NavigationHandle to be matched with the navigation.

BUG=695421
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

[arthursonzogni, 2017-02-24 09:57:26]

Description was changed from

==========
Fix bug when a NavigationThrottle block a request during a redirect.

When PlzNavigate is **disabled**, when a request is blocked on a
redirect and the browser asks the renderer to commit an error page,
blink doesn't know about the latest redirect and use the post-redirect
URL. This cause a mismatch in TakeNavigationHandleForCommit.

This CL doesn't make blink aware of the redirect, but allows the
NavigationHandle to be matched with the navigation.

BUG=695421
==========

to

==========
Fix bug when a NavigationThrottle block a request during a redirect.

When PlzNavigate is **disabled**, when a request is blocked on a
redirect and the browser asks the renderer to commit an error page,
blink doesn't know about the latest redirect and use the post-redirect
URL. This cause a mismatch in TakeNavigationHandleForCommit.

This CL doesn't make blink aware of the redirect, but allows the
NavigationHandle to be matched with the navigation.

BUG=695421
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation
==========

[arthursonzogni, 2017-02-24 09:57:27]

The CQ bit was checked by arthursonzogni@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-02-24 09:57:41]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-02-24 10:47:53]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-02-24 10:47:55]

Dry run: This issue passed the CQ dry run.

[arthursonzogni, 2017-02-24 13:47:58]

arthursonzogni@chromium.org changed reviewers:
+ creis@chromium.org

[arthursonzogni, 2017-02-24 13:47:58]

Hi Charlie,

Please take a look, Thanks!

+CC clamy@

[Charlie Reis, 2017-02-24 22:19:11]

Hmm, I'm concerned about having the renderer commit the wrong URL.  (Sounds like
this isn't a problem today but will be after some upcoming CLs?)  See below.

https://codereview.chromium.org/2712123002/diff/1/content/browser/frame_host/...
File content/browser/frame_host/navigation_handle_impl.cc (right):

https://codereview.chromium.org/2712123002/diff/1/content/browser/frame_host/...
content/browser/frame_host/navigation_handle_impl.cc:656: DCHECK(url_ ==
params.url ||
nit: We probably shouldn't downgrade these from CHECKs to DCHECKs.

https://codereview.chromium.org/2712123002/diff/1/content/browser/frame_host/...
content/browser/frame_host/navigation_handle_impl.cc:658: params.url) !=
redirect_chain_.end());
I don't think this is safe.  According to the bug, params.url is the
pre-redirect URL, but that's what NavigationController will use when creating a
NavigationEntry (which affects what is shown in the omnibox, used for
same-origin checks within NavigationController, etc).

Can we find a way to send a ResourceMsg_ReceivedRedirect instead so that the
renderer knows to commit the right URL?

https://codereview.chromium.org/2712123002/diff/1/content/browser/frame_host/...
File content/browser/frame_host/render_frame_host_impl.cc (right):

https://codereview.chromium.org/2712123002/diff/1/content/browser/frame_host/...
content/browser/frame_host/render_frame_host_impl.cc:3411: // with the
post-redirect url.
This doesn't match what's written on the bug:

"Hence the renderer never learns of the redirect, so the provisional load fails
on the renderer side with the pre-redirect URL.

This means that the error page is also committed under the pre-redirect URL,
which causes a mismatch in TakeNavigationHandleForCommit, because the
NavigationHandle already has the post-redirect URL."

Did you mean "pre-redirect url" on line 3411?

[arthursonzogni, 2017-02-26 20:49:58]

Description was changed from

==========
Fix bug when a NavigationThrottle block a request during a redirect.

When PlzNavigate is **disabled**, when a request is blocked on a
redirect and the browser asks the renderer to commit an error page,
blink doesn't know about the latest redirect and use the post-redirect
URL. This cause a mismatch in TakeNavigationHandleForCommit.

This CL doesn't make blink aware of the redirect, but allows the
NavigationHandle to be matched with the navigation.

BUG=695421
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation
==========

to

==========
Fix bug when a NavigationThrottle blocks a request during a redirect.

When PlzNavigate is **disabled**, when a request is blocked on a
redirect and the browser asks the renderer to commit an error page,
blink doesn't know about the latest redirect and use the post-redirect
URL. This cause a mismatch in TakeNavigationHandleForCommit.

This CL doesn't make blink aware of the redirect, but allows the
NavigationHandle to be matched with the navigation.

BUG=695421
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation
==========

[arthursonzogni, 2017-03-06 15:19:22]

Thanks Charlie,

This CL will be merged with:
https://codereview.chromium.org/2698623006
I will close this CL.

In the meantime, @clamy makes the CL move forward while I was OOO in:
https://codereview.chromium.org/2729433003/