Closed Bug 1252273 Opened 5 years ago Closed 2 years ago

[e10s] Frequent dom/events failures on OSX (test_bug607464.html & test_bug946632.html timeouts)


(Core :: DOM: Events, defect, P3)




Tracking Status
e10s + ---
firefox47 --- wontfix


(Reporter: RyanVM, Unassigned)


(Blocks 1 open bug, )


Component: DOM: Events → Graphics: Layers
The leaks are actually happening after dom/media, so let's keep this bug focused on the two failing tests.
Component: Graphics: Layers → DOM: Events
No longer depends on: 1246529
Summary: [e10s] Frequent dom/events failures on OSX (test_bug607464.html & test_bug946632.html timeouts, IPC::Message leaks) → [e10s] Frequent dom/events failures on OSX (test_bug607464.html & test_bug946632.html timeouts)
Markus, can you take a look here about test_bug607464.html?
Mats, can you look at the test_bug946632.html failures?
Flags: needinfo?(mstange)
Flags: needinfo?(mats)
test_bug946632.html is started with SimpleTest.waitForFocus(...).  I don't see any
messages at all from this test in the log file so I suspect that we never got
a focus event for some reason.  The screenshot shows nothing unusual afaict.
It certainly looks like the window has focus.

Perhaps the root cause is to be found in one of the earlier tests that ran
before this one?  There are a few warnings in the log from those.
Or it could be an underlying bug in focus management under e10s?
(just wildly speculating though)

It seems unlikely that test_bug946632.html itself is to blame.
It does a "pushPrefEnv" before it runs its first test, and IIRC those are
Flags: needinfo?(mats)
mconley, you have any thoughts on focus management under e10s? (/me is extending Mats' wild speculation)
Flags: needinfo?(mconley)
Whiteboard: btpp-followup-2016-03-08
(In reply to Andrew Overholt [:overholt] from comment #7)
> mconley, you have any thoughts on focus management under e10s? (/me is
> extending Mats' wild speculation)

Focus is more Enn's domain, I think. At the very least, someone should add some logging in this test to confirm that this is where the "hang" is occurring.
Flags: needinfo?(mconley)
Only two of the logs in comment 0 show failures in test_bug607464.html:

And the failure is different. In the first log, SimpleTest.waitForFocus completes, and we send wheel events to the window, but those wheel events don't end up scrolling the window far enough (or they stop early for some reason).
In the second log, I'm not sure if SimpleTest.waitForFocus even completes.

Does this failure occur on Linux, too?
Flags: needinfo?(mstange)
See comment 10.
Flags: needinfo?(ryanvm)
The tests are run on Linux e10s and there are no active intermittent-failure bugs for either, so I'm going to say that Linux is unaffected :)
Flags: needinfo?(ryanvm)
Markus, see comment 12.
Flags: needinfo?(mstange)
The leave-open keyword is there and there is no activity for 6 months.
:hsinyi, maybe it's time to close this bug?
Flags: needinfo?(htsai)

The tests were disabled on e10s. This bug is still valid to keep track of re-enabling them.

Flags: needinfo?(htsai)
Priority: -- → P3

At this point, test_bug607464.html is still intermittently timing out across all platforms (tracked in bug 1400586) and test_bug946632.html runs fine now on OSX. Bug 1518478 will be re-enabling test_bug946632.html, so let's close this out and leave the remaining work to bug 1400586.

Closed: 2 years ago
Flags: needinfo?(mstange)
Keywords: leave-open
Resolution: --- → WORKSFORME
Whiteboard: btpp-followup-2016-03-08
You need to log in before you can comment on or make changes to this bug.