472677 - mochitest-browser-chrome favicon load occasionally leaks

Status: RESOLVED WORKSFORME

(Firefox :: Tabbed Browser, defect)

Description

[Serge Gautherie (:sgautherie)]

[Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9.2a1pre) Gecko/20090108 Minefield/3.2a1pre] (home, optim default) (W2Ksp4)
(http://hg.mozilla.org/mozilla-central/rev/37ac7732142f)

While investigating bug 465952, I noticed that running browser_alltabslistener.js alone often leaks 5 to 8 KB.

I use
python runtests.py --browser-chrome --test-path=browser/base/content/test --leak-threshold=0 --autorun --close-when-done
with only alltabslistener.html and browser_alltabslistener.js.

It looks like running tests 1 to 3 never leaks;
but trying to add test 4 or 5 triggers the leak.


NB: I wonder if this leak could be causing the increased bug 465952 leak...

Comment 1

[Serge Gautherie (:sgautherie)]

Attachment: Example Windows leak log

Comment 2

[Serge Gautherie (:sgautherie)]

[Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9.1b2pre) Gecko/20090108 Minefield/3.1b2pre] (home, optim default) (W2Ksp4)
(http://hg.mozilla.org/mozilla-central/rev/20245c2d97d0)

Confirming that this bug was already there at bug 465952 comment 1 time.
(I must have missed it the way I tested back then :-()

Comment 3

[Serge Gautherie (:sgautherie)]

(In reply to comment #2)
> (I must have missed it the way I tested back then :-()

If I run this test with (for example) the 7 other browser_bug*.js then it seems I can't trigger the leak anymore.

This would explain why I missed it previously and why it can't be seen when running the full test suite :-/

Comment 4

[Serge Gautherie (:sgautherie)]

(In reply to comment #3)

Well, it's a little more complex than this:
if I run this test in addition to browser_sanitize-timespans.js, the (bug 465952) leak increase from 55k to 66k ... so they kind of add to each other...

Now, it's up to you.

Comment 5

[Serge Gautherie (:sgautherie)]

(In reply to comment #0)
> NB: I wonder if this leak could be causing the increased bug 465952 leak...

It used not to be needed, but now running this test is one way of triggering bug 465952.

(In reply to comment #3)
> This would explain why I missed it previously and why it can't be seen when
> running the full test suite :-/

This is not right:
I simply, and incorrectly, used to think this leak was part of bug 465952;
and the 70k leak reported by the tinderboxes is these two bugs (afaict fttb).

***

[Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9.2a1pre) Gecko/20090109 Minefield/3.2a1pre] (home, optim default) (W2Ksp4)
(http://hg.mozilla.org/mozilla-central/rev/0ff733711384)

(Simply confirming bug still there, up to this current rev..)

Comment 6

[Serge Gautherie (:sgautherie)]

[Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9.2a1pre) Gecko/20090117 Minefield/3.2a1pre] (home, optim default) (W2Ksp4)
(http://hg.mozilla.org/mozilla-central/rev/184eadf4e185 + bug 446300 patch)

Thanks to bug 429690, this is now easier to test:
python runtests.py --browser-chrome --test-path=browser/base/content/test/browser_alltabslistener.js --leak-threshold=0 --autorun --close-when-done

Comment 7

[Serge Gautherie (:sgautherie)]

[Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9.2a1pre) Gecko/20090122 Minefield/3.2a1pre] (home, optim default) (W2Ksp4)
(http://hg.mozilla.org/mozilla-central/rev/aa69e508e312)

Fwiw, bug 465952 was fixed, but this bug is still there.

Comment 8

[Serge Gautherie (:sgautherie)]

[Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9.2a1pre) Gecko/20090228 Minefield/3.2a1pre] (home, optim default) (W2Ksp4)
(http://hg.mozilla.org/mozilla-central/rev/f7f62131998d)

(Bug is still there.)

Comment 9

[Serge Gautherie (:sgautherie)]

|make NO_FAIL_ON_TEST_ERRORS=1 TEST_PATH=browser/base/content/test/browser_alltabslistener.js -C objdir mochitest-browser-chrome|

Can someone else reproduce?

Comment 10

[Serge Gautherie (:sgautherie)]

[Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9.2a1pre) Gecko/20090621 Minefield/3.6a1pre] (mozilla-central-win32-unittest/1245592825) (W2Ksp4)

The URLs are different from bug 486489 ones:
{
Leaked URLs:
  https://example.com/browser/browser/base/content/test/alltabslistener.html
  https://example.com/favicon.ico
  http://example.org/browser/browser/base/content/test/alltabslistener.html
  http://example.org/favicon.ico
  http://example.org/browser/browser/base/content/test/alltabslistener.html
  http://example.org/favicon.ico

TEST-UNEXPECTED-FAIL | runtests-leaks | leaked 8939 bytes during test execution
}

Comment 11

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1247643832.1247651664.24027.gz

this failure looks like same of this bug but the platform is Mac.

Comment 12

[Serge Gautherie (:sgautherie)]

(In reply to comment #11)
> http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1247643832.1247651664.24027.gz
> 
> this failure looks like same of this bug but the platform is Mac.

Not at all, afaict:
I eventually filed bug 504340.

Comment 13

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Dave, can you please check it? Both files are part of your patch on bug 463387.

Carsten can you run a leak test with the above test?

Comment 14

[Dave Townsend [:mossop]]

This appears to be just a temporary leak caused by something not releasing when loading a favicon. It goes away on its own so I don't think this is a big deal. Not caused by bug 463387

http://tinderbox.mozilla.org/Firefox-Unittest/1253005675.1253007827.29860.gz

Comment 15

[Ed Morley [:emorley]]

Mass marking whiteboard:[orange] bugs WFM (to clean up TBPL bug suggestions) that:
* Haven't changed in > 6months
* Whose whiteboard contains none of the strings: {disabled,marked,random,fuzzy,todo,fails,failing,annotated,leave open,time-bomb}
* Passed a (quick) manual inspection of bug summary/whiteboard to ensure they weren't a false positive.

I've also gone through and searched for cases where the whiteboard wasn't labelled correctly after test disabling, by using attachment description & basic comment searches. However if the test for which this bug was about has in fact been disabled/annotated/..., please accept my apologies & reopen/mark the whiteboard appropriately so this doesn't get re-closed in the future (and please ping me via IRC or email so I can try to tweak the saved searches to avoid more edge cases).

Sorry for the spam! Filter on: #FFA500