694689 - iframe compartment doesn't go away after removing iframe until tab is closed

Status: RESOLVED INCOMPLETE

(Core :: DOM: Core & HTML, defect, P5)

Description

[Eric Sh.]

Attachment: example.html

User Agent: Mozilla/5.0 (Windows NT 5.1; rv:10.0a1) Gecko/20111014 Firefox/10.0a1
Build ID: 20111014030948

Steps to reproduce:

1. Open up the example.html attachment
2 [review]. Click create once - this will create ten div's with and iframe in each(with google.com)
3. Click destroy - this will remove the iframes from the div's and the divs from the page


Actual results:

The compartment for google.com still exists after all iframes were deleted, and after pressing "minimize memory usage" multiple times.

This also happens with no add-ons enabled.


Expected results:

The compartment of the iframe should be removed when the iframe is removed.

here is the compartment:
│  ├───3,199,632 B (04.52%) -- compartment(http://www.google.com/)
│  │   ├──1,925,120 B (02.72%) -- gc-heap
│  │   │  ├────974,768 B (01.38%) -- arena-unused
│  │   │  ├────427,800 B (00.60%) -- objects
│  │   │  ├────257,856 B (00.36%) -- scripts
│  │   │  ├────245,640 B (00.35%) -- shapes
│  │   │  ├──────7,520 B (00.01%) -- arena-headers
│  │   │  ├──────5,440 B (00.01%) -- arena-padding
│  │   │  ├──────4,560 B (00.01%) -- strings
│  │   │  └──────1,536 B (00.00%) -- type-objects
│  │   ├────430,712 B (00.61%) -- script-data
│  │   ├────412,704 B (00.58%) -- property-tables
│  │   ├────327,680 B (00.46%) -- mjit-code
│  │   │    ├──324,300 B (00.46%) -- regexp
│  │   │    ├────3,300 B (00.00%) -- unused
│  │   │    └───────80 B (00.00%) -- method
│  │   ├─────55,544 B (00.08%) -- object-slots
│  │   ├─────31,072 B (00.04%) -- shape-kids
│  │   ├─────10,816 B (00.02%) -- type-inference
│  │   │     ├───9,280 B (00.01%) -- script-main
│  │   │     └───1,536 B (00.00%) -- object-main
│  │   ├──────4,016 B (00.01%) -- string-chars
│  │   └──────1,968 B (00.00%) -- object-empty-shapes

Comment 1

[Justin Lebar (not reading bugmail)]

Wow, that's pretty bad.  I can reproduce on trunk.

Comment 2

[Ed Morley [:emorley]]

Likewise on: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:10.0a1) Gecko/20111014 Firefox/10.0a1

Comment 3

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Justin, do you want this one?  If not, I'll take it.

Comment 4

[Justin Lebar (not reading bugmail)]

Ah, the compartment does go away when the tab is closed.  So this isn't as bad as I thought.

Be my guest, Kyle.  :)

Comment 5

[Olli Pettay [:smaug][[e-mail]]]

Seems like not all DOMWindows are deleted after clicking Destroy.

Comment 6

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

(In reply to Justin Lebar [:jlebar] from comment #4)
> Ah, the compartment does go away when the tab is closed.  So this isn't as
> bad as I thought.
> 
> Be my guest, Kyle.  :)

This is still pretty bad though.

I'll look into this this week if nobody beats me to it.

Comment 7

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

This is InstallTrigger related, so it's nowhere near as bad as we first thought.

Comment 8

[johnny_dipin]

Hi, 

I just wanted to thank Eric for creating the bug and also just to let you know that this memory consumption can be seen in IE as well.

Comment 9

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

I'm going to go ahead and call this a P2 since it's probably a dupe of a P2 bug.

Also, thanks for the bug report Eric.  Having a testcase made this pretty easy to debug.

Comment 10

[Peter Van der Beken [:peterv]]

The testcase doesn't work for me on trunk, I get a bunch of "Security Error: Content at http://www.google.com/ may not load data from https://bug694689.bugzilla.mozilla.org/attachment.cgi?id=567208" and the iframes are blank. Do I need to do something special to make it work?

Comment 11

[Eric Sh.]

Attachment: Example - Version 2

Comment 12

[Eric Sh.]

It appears google is either blocking us or using it in iframes :)

Comment 13

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

The only actionable thing here is fixing InstallTrigger, which is happening in another bug.  Unassigning myself, and we can verify that the fix for InstallTrigger fixes this testcase when that lands.

Thanks again for the nice bug report.

Comment 14

[Eric Sh.]

You should thank johnny_dipin he told me about large memory usage when doing this in his app.

Comment 15

[johnny_dipin]

Thank you Eric and thank you everyone

Comment 16

[Firefox Bug Husbandry Bot]

https://bugzilla.mozilla.org/show_bug.cgi?id=1472046

Move all DOM bugs that haven’t been updated in more than 3 years and has no one currently assigned to P5.

If you have questions, please contact :mdaly.

Comment 17

[Andrew McCreight [:mccr8]]

The browser has changed so much in the last dozen years that it is hard to say if this is still relevant. It sounds like some fix happened in another bug, but nobody came back and confirmed it.