1041527 - browser_tabview_*_perwindowpb.js

Status: RESOLVED FIXED

Description

[Joel Maher ( :jmaher ) (UTC-4)]

we have 3 tests which are seem to cause a crash in mochitest browser-chrome while running just the directory (not a full set of directories and subdirectories).  This is the "components/tabview/test/" directory.

Look at the bc1 column for all platforms for these 3 try runs where we have uncommented a single test at a time to show the error:

browser_tabview_bug624727_perwindowpb.js:
https://tbpl.mozilla.org/?tree=Try&rev=e77cc4c2768b

browser_tabview_bug650280_perwindowpb.js:
https://tbpl.mozilla.org/?tree=Try&rev=5266cd7a73cd

browser_tabview_privatebrowsing_perwindowpb.js:
https://tbpl.mozilla.org/?tree=Try&rev=b261c33d240a

These all seem to be the same error:
https://tbpl.mozilla.org/php/getParsedLog.php?id=44177751&tree=Try

We are planning on switching the browser-chrome tests to run per directory which would mean these would fail and either need to be fixed or disabled until we can fix them.

Comment 1

[Joel Maher ( :jmaher ) (UTC-4)]

Gavin, can you help find an owner and prioritize this?

Comment 2

[Tim Taubert [:ttaubert] (inactive)]

That's pretty much the same as bug 1027084. Not using about:home would help.

Comment 3

[Tim Taubert [:ttaubert] (inactive)]

I don't see us using about:home directly so it's probably loaded when opening new windows.

Comment 4

[Joel Maher ( :jmaher ) (UTC-4)]

is there a way to use about:blank or something else?

Comment 5

[Tim Taubert [:ttaubert] (inactive)]

Yup :)

Comment 6

[Tim Taubert [:ttaubert] (inactive)]

Attachment: 0001-Bug-1041527-Ensure-that-about-home-isn-t-the-initial.patch

Tests run locally, they did crash before. Joel, can you please give this a try run with your patches on top?

Comment 7

[:Ehsan Akhgari]

Comment on attachment 8459621 [details] [diff] [review]
0001-Bug-1041527-Ensure-that-about-home-isn-t-the-initial.patch

Stealing the review.

Comment 8

[Joel Maher ( :jmaher ) (UTC-4)]

oh, let me add this to my queue and see what happens :)

Comment 9

[Joel Maher ( :jmaher ) (UTC-4)]

Pushed to try:
https://tbpl.mozilla.org/?tree=Try&rev=9a2b271154eb

I will retrigger this afternoon and see the results tonight or tomorrow morning.

Comment 10

[Joel Maher ( :jmaher ) (UTC-4)]

given the fact that we had perma orange prior to this patch and now bc1 has full green for 3 of each build, I think this does the trick!

https://tbpl.mozilla.org/?tree=Try&rev=9a2b271154eb

Thanks for doing this and being fast about it.  Any chance the linux bc3 failures for privatebrowsing could benefit from this same fix?

Comment 11

[Tim Taubert [:ttaubert] (inactive)]

https://hg.mozilla.org/integration/fx-team/rev/1e9c0444b588

Comment 12

[Tim Taubert [:ttaubert] (inactive)]

(In reply to Joel Maher (:jmaher) from comment #10)
> Thanks for doing this and being fast about it.  Any chance the linux bc3
> failures for privatebrowsing could benefit from this same fix?

Pushed a possible fix for those to try, let's see how that works out. Couldn't reproduce those locally unfortunately...

https://tbpl.mozilla.org/?tree=Try&rev=e7d620089705

Comment 13

[Tim Taubert [:ttaubert] (inactive)]

Oh, wait. Wouldn't I have to include the patch that enables --run-by-dir?

Comment 14

[Joel Maher ( :jmaher ) (UTC-4)]

did that here:
https://tbpl.mozilla.org/?tree=Try&rev=ad4eeefd37f9

(fyi- I didn't update my repo to compare this change to the previous try push)

Comment 15

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/1e9c0444b588

Comment 16

[Joel Maher ( :jmaher ) (UTC-4)]

it seems as though this magic trick doesn't work for the ssl test, but we gave it a try!

Comment 17

[Tim Taubert [:ttaubert] (inactive)]

Guess we should take those leaks to a different bug.

Comment 18

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/24e2207ded77
https://hg.mozilla.org/releases/mozilla-beta/rev/c340fefc0fe8
https://hg.mozilla.org/releases/mozilla-esr31/rev/98acfcb6a938