Crash when a page in a suborigin sets window.location.hash |
|||
Issue descriptionVersion: 56.0.2902.0 dev (64-bit) OS: Ubuntu 14.04.5 LTS, Trusty Tahr What steps will reproduce the problem? (1) Page sets a Suborigin header. (2) Page assigns a string to window.location.hash. Example: http://lingro.com:81/cgi-bin/suborigin-testing/bug1.py #!/usr/bin/python import cgi print """\ Content-Type: text/html; charset=utf-8 Suborigin: foo <script>window.location.hash = '#bar'</script> """ What is the expected output? The URL fragment is changed, no crash. What do you see instead? The "Aw, Snap!" crashed tab. Some quick testing reveals that this happens for non-navigating hash assignments, including directly setting window.location = "#...", but not for navigations, e.g. window.location = "/foo#bar".
,
Oct 31 2016
Sure, sorry! The most recent one is bf90c04700000000
,
Nov 2 2016
After some thought I'm bumping it to P1 because the crash is blocking suborigin testing/adoption in several of the applications we've had on our radar.
,
Nov 3 2016
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/25ec43232e79ebfe0c4fdd2e301d957268c66e26 commit 25ec43232e79ebfe0c4fdd2e301d957268c66e26 Author: mkwst <mkwst@chromium.org> Date: Thu Nov 03 13:47:47 2016 Suborigins should not crash fragment navigations. We should be checking for physical origin matches when committing URLs, as the URL we commit doesn't include the suborigin. BUG= 660683 R=jochen@chromium.org Review-Url: https://codereview.chromium.org/2472963002 Cr-Commit-Position: refs/heads/master@{#429563} [modify] https://crrev.com/25ec43232e79ebfe0c4fdd2e301d957268c66e26/content/renderer/render_frame_impl.cc [add] https://crrev.com/25ec43232e79ebfe0c4fdd2e301d957268c66e26/third_party/WebKit/LayoutTests/http/tests/security/suborigins/suborigin-in-page-navigation.php
,
Nov 14 2016
If this isn't fixed, Artur, ping me.
,
Nov 14 2016
I just tried it in Canary and everything seems to work fine now. Thanks for the fix, Mike! |
|||
►
Sign in to add a comment |
|||
Comment 1 by jochen@chromium.org
, Oct 31 2016