New issue
Advanced search Search tips

Issue 889002 link

Starred by 3 users

Issue metadata

Status: Assigned
Owner:
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 1
Type: Bug

Blocking:
issue 899073



Sign in to add a comment

Strange failure on SitePerProcessHitTestBrowserTest.ScrolledNestedPopupMenuTest

Project Member Reported by wjmaclean@chromium.org, Sep 25

Issue description

kenrb@ -

There's a weird (sudden?) appearance of "D = http://baz.com/" and proxies for D where there didn't use to be any.

https://logs.chromium.org/v/?s=chrome%2Fbb%2Fofficial.desktop.continuous%2Fwin64_stable%2F1182%2F%2B%2Frecipes%2Fsteps%2Fcontent_browsertests%2F0%2Flogs%2FSitePerProcessHitTestBrowserTest.ScrolledNestedPopupMenuTest__x2f_2%2F0

[ RUN      ] SitePerProcessHitTestBrowserTest.ScrolledNestedPopupMenuTest/2
DevTools listening on ws://127.0.0.1:65129/devtools/browser/4af77d39-5b75-4878-9bef-9366960bff47
../../content/browser/site_per_process_hit_test_browsertest.cc(4076): error: Expected equality of these values:
  " Site A ------------ proxies for B C\n" "   +--Site B ------- proxies for A C\n" "        +--Site C -- proxies for A B\n" "Where A = http://a.com/\n" "      B = http://b.com/\n" "      C = http://c.com/"
    Which is: " Site A ------------ proxies for B C\n   +--Site B ------- proxies for A C\n        +--Site C -- proxies for A B\nWhere A = http://a.com/\n      B = http://b.com/\n      C = http://c.com/"
  DepictFrameTree(root)
    Which is: " Site A ------------ proxies for B C D\n   +--Site B ------- proxies for A C D\n        +--Site C -- proxies for A B D\nWhere A = http://a.com/\n      B = http://b.com/\n      C = http://c.com/\n      D = http://baz.com/"
With diff:
@@ -1,6 +1,7 @@
- Site A ------------ proxies for B C
-   +--Site B ------- proxies for A C
-        +--Site C -- proxies for A B
+ Site A ------------ proxies for B C D
+   +--Site B ------- proxies for A C D
+        +--Site C -- proxies for A B D
 Where A = http://a.com/
       B = http://b.com/
       C = http://c.com/
+      D = http://baz.com/

 
Cc: alex...@chromium.org
Interesting, it looks like sometimes the proxies aren't always cleaned up in time, following the navigation from baz.com to c.com on https://cs.chromium.org/chromium/src/content/browser/site_per_process_hit_test_browsertest.cc?l=4351&rcl=bb7e956646c46cda7d14c991e30066fdd6d99529.

alexmos@: I hadn't been aware that could happen. Have you seen this behavior before? It seems like it could potentially affect a lot of browser tests.
I was wondering if it's a regression somewhere else in the frame tree code ... if it is, we could be seeing more failed tests like this one fairly soon.
I think this is expected if we're navigating a frame cross-process and not waiting for RFH to be fully deleted - see issue 662550.  If the old RFH is still around in pending deletion state, the proxies will show up in DepictFrameTree. One solution is to use RenderFrameDeletedObserver on the old baz.com RFH to wait for it to fully shut down before calling DepictFrameTree.
Cc: creis@chromium.org
Components: Internals>Sandbox>SiteIsolation
That's right-- the pattern for fixing this is in r430474.
Maybe we can fix this in a more general way though?  See  https://crbug.com/888156#c11 
This test is flaky. See https://bugs.chromium.org/p/chromium/issues/detail?id=898628#c14 for how to make it flake reliably. I'll be disabling for now.
Blocking: 899073
Project Member

Comment 8 by bugdroid1@chromium.org, Nov 5

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/734653d3c6bb8afb9c2ed2cee5df11504c66a244

commit 734653d3c6bb8afb9c2ed2cee5df11504c66a244
Author: danakj <danakj@chromium.org>
Date: Mon Nov 05 21:02:26 2018

Disable the flaky ScrolledNestedPopupMenuTest tests.

R=kenrb@chromium.org

Change-Id: I24fcca54fe990bbacc19d892d868b3501739a8f5
Bug: 889002
Reviewed-on: https://chromium-review.googlesource.com/c/1318430
Reviewed-by: Ken Buchanan <kenrb@chromium.org>
Commit-Queue: Ken Buchanan <kenrb@chromium.org>
Commit-Queue: danakj <danakj@chromium.org>
Cr-Commit-Position: refs/heads/master@{#605463}
[modify] https://crrev.com/734653d3c6bb8afb9c2ed2cee5df11504c66a244/content/browser/site_per_process_hit_test_browsertest.cc

Status: Assigned (was: Untriaged)
This issue has an owner, a component and a priority, but is still listed as untriaged or unconfirmed. By definition, this bug is triaged. Changing status to "assigned". Please reach out to me if you disagree with how I've done this.

Sign in to add a comment