388577 - FoxClocks 2.1.93 extension can leak nsGlobalWindows on trunk (but not on branch)

Status: RESOLVED INCOMPLETE

(Firefox :: General, defect, P4)

Description

[Steve England [:stevee]]

1. New Profile, start firefox
2. (If on trunk) about:config, set extensions.checkCompatibility to false
3. Install FoxClocks 2.1.93 from https://addons.mozilla.org/en-US/firefox/downloads/file/17041/foxclocks-2.1.93-fx+tb+sb.xpi
4. Restart firefox to complete installation.
5. Some dialog appears (Both options ticked). Just click OK. 
6. Close firefox and start back up with leak-logging enabled.
7. Note there are times in the status bar. Right click on them and choose "Foxclocks"
8. A dialog appears ("Zone Picker"/"Watch List"). Close it with [x] in the top right hand corner.
9. Close firefox any analyse nspr.log

Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9a7pre) Gecko/200707180303 Minefield/3.0a7pre reports:

Leaked outer window 1c57ab0 at address 1c57ab0.
Leaked inner window 1f63d98 (outer 1d74660) at address 1f63d98.
 ... with URI "about:blank".
Leaked outer window 1d74660 at address 1d74660.
Summary:
Leaked 3 out of 8 DOM Windows
Leaked 0 out of 41 documents
Leaked 0 out of 4 docshells

(Sometimes it seems 5 DOM Windows are reported to leak)


Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8.1.5pre) Gecko/2007071603 BonEcho/2.0.0.5pre reports:

Leaked 0 out of 9 DOM Windows
Leaked 0 out of 40 documents
Leaked 0 out of 4 docshells

Comment 1

[Mike Connor [:mconnor]]

moving out bugs that don't need to block b1

Comment 2

[Mike Connor [:mconnor]]

Does this still leak?

Comment 3

[Steve England [:stevee]]

Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9b2pre) Gecko/2007110913
Minefield/3.0b2pre ID:2007110913

Yes. STR in comment 0 and leak-gauge.pl still show similar leakage.

Comment 4

[Mike Schroepfer]

Moving Priority to P2 since this might affect other extensions as well

Comment 5

[Jesse Ruderman]

Several other extensions show similar trunk-only leaks (see bug 388573 comment 9) and we think it might be due to bug 398219 (see bug 388573 comment 6.

Comment 6

[Mike Schroepfer]

Peter can you take this?

Comment 7

[Carsten Book [:Tomcat]]

Attachment: leak log from a trace malloc build

nsStringStats
 => mAllocCount:          39535
 => mReallocCount:         3548
 => mFreeCount:           38758  --  LEAKED 777 !!!
 => mShareCount:          50748
 => mAdoptCount:           7000
 => mAdoptFreeCount:       6993  --  LEAKED 7 !!!

Comment 8

[Boris Zbarsky [:bzbarsky]]

The log shows nsGlobalWindows leaking.  The rest is corollaries...

Comment 9

[Steve England [:stevee]]

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9b3pre) Gecko/2008012104 Minefield/3.0b3pre ID:2008012104

I'm still seeing frequent (although not consistent) leakage in this case with the above build (post bug 412491)

Leaked outer window f65690 at address f65690.
Leaked outer window 17b9508 at address 17b9508.
Leaked inner window 24af880 (outer 17b9508) at address 24af880.
 ... with URI "about:blank".
Summary:
Leaked 3 out of 12 DOM Windows
Leaked 0 out of 46 documents
Leaked 0 out of 5 docshells

Leaked outer window 15ba418 at address 15ba418.
Leaked outer window 15fdfa8 at address 15fdfa8.
Leaked inner window 2a0f5e0 (outer 15ba418) at address 2a0f5e0.
 ... with URI "about:blank".
Leaked inner window f65658 (outer 15fdfa8) at address f65658.
 ... with URI "about:blank".
Summary:
Leaked 4 out of 9 DOM Windows
Leaked 0 out of 41 documents
Leaked 0 out of 4 docshells

Comment 10

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Doubt this extension is common enough that we want to hold the release.

Setting low priority for now, if you can get this to leak multiple windows (rather than just the one pair in the leak log) we should up the priority.

Comment 11

[Matthias Versen [:Matti]]

Steve: Is this still an issue ?

Comment 12

[Steve England [:stevee]]

Sorry Matthais I don't have the time to check this. Given that it's over 5 years old and both Firefox and FoxClocks have moved on, I'm happy if you mark it INVALID or WONTFIX.

Comment 13

[Matthias Versen [:Matti]]

marking incomplete, a lot has changed and it's very likely that this got fixed in the meantime