Closed Bug 1067510 Opened 11 years ago Closed 11 years ago

iframe javascript memory leak

Categories

(Core :: JavaScript Engine, defect)

32 Branch
x86_64
Windows 7
defect
Not set
normal

Tracking

()

RESOLVED INVALID

People

(Reporter: jordan.blumenthal, Unassigned)

Details

Attachments

(1 file)

Attached image debug_1.jpg
User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/37.0.2062.120 Safari/537.36 Steps to reproduce: Many embedded iframes eventually slow Firefox to a crawl. I'm thinking this is a javascript engine/garbage collection issue. Use case: 1) Visit: http://www-stage.wongdoody.com/mangles 2) Click on any of the words in the clouds. 3) Laugh at hilarious video(s) 4) Repeat about 12-20 times (dependent on RAM) 5) Eventually FF performance significantly degrades to the point where the program becomes unresponsive Actual results: This can be viewed here: http://www-stage.wongdoody.com/mangles Click through about 12-16 (depending on RAM) and see what happens, after awhile performance slows down and eventually you get the "script is unrespsonsive" error window. This is not happening on any other browsers, it's specific to Firefox and is confirmed to be happening on all versions and OS we tested on. My post on stackoverflow for more detailed explanation: http://stackoverflow.com/questions/25816657/firefox-javascript-and-iframe-performance-with-jquery Expected results: Ideally, embedding any number of iframes would trigger js garbage collection, but that does not seem to be happening.
I can't reproduce the problem in a current nightly (on Mac, though). Memory usage is stable as I watch the videos, and in particular when a video ends memory usage drops to the level it was at before the video started. Jordan, do you see the problem in safe mode?
Flags: needinfo?(jordan.blumenthal)
Unfortunately I do see the problem, even in safe mode. And I can confirm this is still happening in Safe Mode, on Windows 7 at least. I know the issue was happening across multiple OS including OSX. Perhaps the issue was cleared up on a nightly build? Thanks.
Flags: needinfo?(jordan.blumenthal)
Jordan, what exact Firefox version are you using? I just tried the Firefox 32 release (still on Mac) and I still don't see the sort of issue you describe.... When you get this problem, what does about:memory have to say?
Flags: needinfo?(jordan.blumenthal)
I'm also using Firefox 32 release. Here's what I'm getting: 377.05 MB ── private 393.37 MB ── resident 784.43 MB ── vsize 1,805.94 MB ── vsize-max-contiguous 190.05 MB (100.0%) -- js-main-runtime If you want I can export the full memory measurement. Physical memory isn't showing huge issues, but javascript performance is really, really poor after a certain # of clicks. I'm using jQuery javascript library.
Flags: needinfo?(jordan.blumenthal)
It might not be a memory leak then, and something that I'm doing that's causing problems for the JS engine.
Status: UNCONFIRMED → RESOLVED
Closed: 11 years ago
Resolution: --- → INVALID
Marking this as invalid, it's an issue with firefox and vimeo js embeds which are causing performance issues. Confirmed not a memory leak.
Jordan, I have yet to reproduce the sort of slow script dialog, or any particular performance issue on that site in Firefox 32 (still on Mac, in case that matters). If you have reliable steps to reproduce, they would be very much appreciated.
Unfortunately I can confirm this is still happening on multiple machines, all on 32.0 and 32.0.1, but now I'm interested as to why the error cannot be replicated. All of our test cases happen on Windows, so perhaps that may point us to the cause of the issue? Here are the steps I used to replicate this morning: visit: http://www-stage.wongdoody.com/mangles Click through 20 or so of the mangle items (wingdaddy, wandduty, etc) Eventually after going through a bunch of them (literally, 20+), you should see decreased performance.
Just to be clear: Are you clicking through some of those items more than once? Because I only see 14 items there...
Yep, that's the idea. We were testing what would happen if we added 20+ and we found this weird slowdown issue.
Also if you have someone with a windows box running 32.0 that may help, most of our test cases are on windows.
Alice0775, are you able to reproduce this?
Flags: needinfo?(alice0775)
(In reply to Boris Zbarsky [:bz] from comment #12) > Alice0775, are you able to reproduce this? I can reproduce the high CPU usage and unrespsonsive script dialog after 15-20 times(click word and watch video). https://hg.mozilla.org/releases/mozilla-release/rev/1727a19db430 Mozilla/5.0 (Windows NT 6.1; WOW64; rv:32.0) Gecko/20100101 Firefox/32.0 ID:20140911151253
Flags: needinfo?(alice0775)
In Google Chrome37, video player would not launch when I click a word after I repeat the step more than 20 times. So, I think this is the site problem.
Jordan, it might help to have the site use unminified scripts in terms of figuring out what's going on here.
This is confirmed as my error, jQuery wasn't wiping away the bindings, so on each call the event handlers were piling up. Sorry to waste your time, my apologies.
This is confirmed as my error, jQuery wasn't wiping away the bindings, so on each call the event handlers were piling up. Sorry to waste your time, my apologies.
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: