Bug 82964 - REGRESSION (r112217): H&R Block tax site won't load
Summary: REGRESSION (r112217): H&R Block tax site won't load
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: Page Loading (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Nobody
URL: http://taxes.hrblock.com
Keywords: InRadar, NeedsReduction, Regression
Depends on: sam
Blocks:
  Show dependency treegraph
 
Reported: 2012-04-02 16:12 PDT by Andy Estes
Modified: 2012-04-03 15:42 PDT (History)
6 users (show)

See Also:


Attachments
Proposed Patch (2.34 KB, patch)
2012-04-02 16:51 PDT, Bill Budge
no flags Details | Formatted Diff | Diff
Proposed Patch (7.69 KB, patch)
2012-04-02 17:33 PDT, Bill Budge
webkit.review.bot: commit-queue-
Details | Formatted Diff | Diff
Archive of layout-test-results from ec2-cr-linux-03 (6.45 MB, application/zip)
2012-04-02 19:21 PDT, WebKit Review Bot
no flags Details
Proposed Patch (17.55 KB, patch)
2012-04-02 22:46 PDT, Bill Budge
no flags Details | Formatted Diff | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Andy Estes 2012-04-02 16:12:05 PDT
* SUMMARY
H&R Block's tax website will not load in WebKit nightly builds that contain r112217.

* STEPS TO REPRODUCE
1. Go to http://taxes.hrblock.com
2. Under 'New User', click 'Start Without an Account'.
3. Accept the license agreement and privacy policy and click 'Next'.

* RESULTS
Expected: A secure connection should be established and the site's main menu should be displayed.
Actual: Nothing happens after step #3.

* REGRESSION
Caused by <http://trac.webkit.org/changeset/112217>.
Comment 1 Andy Estes 2012-04-02 16:13:47 PDT
<rdar://problem/11167741>
Comment 2 Bill Budge 2012-04-02 16:28:46 PDT
I am looking at it right now.
Comment 3 Bill Budge 2012-04-02 16:51:03 PDT
Created attachment 135229 [details]
Proposed Patch
Comment 4 Adam Barth 2012-04-02 16:53:23 PDT
Comment on attachment 135229 [details]
Proposed Patch

I see.  Maybe try setting a custom header and having a same-origin redirect.  That will probably fail the CORS check but be allowed because it's same-origin.
Comment 5 Bill Budge 2012-04-02 17:33:49 PDT
Created attachment 135241 [details]
Proposed Patch

Added a test case for this. I manually verified that the code change fixes the website. I still need to verify the test.
Comment 6 WebKit Review Bot 2012-04-02 19:21:48 PDT
Comment on attachment 135241 [details]
Proposed Patch

Attachment 135241 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/12310822

New failing tests:
http/tests/xmlhttprequest/access-control-and-redirects-async.html
Comment 7 WebKit Review Bot 2012-04-02 19:21:54 PDT
Created attachment 135259 [details]
Archive of layout-test-results from ec2-cr-linux-03

The attached test failures were seen while running run-webkit-tests on the chromium-ews.
Bot: ec2-cr-linux-03  Port: <class 'webkitpy.common.config.ports.ChromiumXVFBPort'>  Platform: Linux-2.6.35-28-virtual-x86_64-with-Ubuntu-10.10-maverick
Comment 8 Bill Budge 2012-04-02 22:46:37 PDT
Created attachment 135276 [details]
Proposed Patch

I had to do fairly extensive reworking of this patch, both code and tests.

If you prefer to revert the change, don't forget the later change I made to AssociatedURLLoader and tests.
http://trac.webkit.org/changeset/112485
Comment 9 Bill Budge 2012-04-02 23:23:26 PDT
The problem with 57600 is that it breaks same-origin redirects when using access control. This patch rearranges things so that same origin requests (and loaders with universal access) can proceed as before. Only if that fails do we check using access control.

I tested this with the replication steps above and the website now loads.
Comment 10 Adam Barth 2012-04-03 00:55:03 PDT
Comment on attachment 135276 [details]
Proposed Patch

Thanks Bill!
Comment 11 WebKit Review Bot 2012-04-03 01:45:44 PDT
Comment on attachment 135276 [details]
Proposed Patch

Clearing flags on attachment: 135276

Committed r112997: <http://trac.webkit.org/changeset/112997>
Comment 12 Alexey Proskuryakov 2012-04-03 13:19:51 PDT
This bug is still open, can it be closed now?
Comment 13 Adam Barth 2012-04-03 15:42:35 PDT
Yes.  The bot didn't close it because the 2nd patch was still marked for review.