870705 - epic citadel html5 demo does not work, browser freezes then the web console throws an error

Status: NEW

(Core :: JavaScript Engine, defect)

Description

[Mathieu Pellerin]

When pushing on the play button of epic's citadel html5 demo using Firefox Nightly build (Mozilla/5.0 (Windows NT 6.1; rv:23.0) Gecko/20130509 Firefox/23.0), the browser freezes and shortly after an error message in the web console pops up, saying: "out of memory".

My computer's processor is a i7 3.20GHz, with 3GB of usable RAM (due to the machine running a 32-bit version of windows 7).

about:support's Graphics section:
Adapter Description	NVIDIA GeForce GTX 260
Adapter Drivers	nvd3dum nvwgf2um,nvwgf2um
Adapter RAM	896
Device ID	0x05e2
Direct2D Enabled	true
DirectWrite Enabled	true (6.1.7600.16699)
Driver Date	2-9-2013
Driver Version	9.18.13.1407
GPU #2 Active	false
GPU Accelerated Windows	1/1 Direct3D 10
Vendor ID	0x10de
WebGL Renderer	Google Inc. -- ANGLE (NVIDIA GeForce GTX 260)
AzureCanvasBackend	direct2d
AzureContentBackend	direct2d
AzureFallbackCanvasBackend	cairo

Comment 1

[Luke Wagner [:luke]]

Is this reproducible even after closing the browser and loading Citadel in its own tab?  (We have some work underway to reduce memory usage during compilation.)

Comment 2

[Mathieu Pellerin]

You mean only loading a citadel tab in a newly started Firefox browser?

Comment 3

[Luke Wagner [:luke]]

That's right

Comment 4

[Mathieu Pellerin]

Sorry just realised I had forgotten to follow up on this. 

Opening page in a clean new browser session still led to out of memory error.

With yesterday's nightly build, browser simply crashes before outputting oom error.

Comment 5

[Mathieu Pellerin]

No crash anymore, but still running into the out of memory message.

Luke your mentioning work on reducing memory, is there a bug to follow to keep track of this effort?

Comment 6

[Mathieu Pellerin]

Still haven't been able to successfully load the Epic Citadel HTML5 / asm.js demo. The errors I now get:

--
[14:03:05.733] Error: successfully compiled asm.js code (total compilation time 4839ms; 1 functions compiled slowly: dtK:9937:1392949 (350ms))
[14:03:05.738] uncaught exception: out of memory
--

Comment 7

[Luke Wagner [:luke]]

That's strange.  We generally haven't seen this problem on 32-bit Firefoxen unless they have multiple copies open (note: page continue to be alive in the back/forward history for a while).  I wonder what is different here.  Do you have any addons installed?

Comment 8

[Cornel Ionce [:noni] [Hubs QA]]

User Agents:
Mozilla/5.0 (Windows NT 6.3; rv:25.0) Gecko/20100101 Firefox/25.0
Mozilla/5.0 (Windows NT 5.1; rv:25.0) Gecko/20100101 Firefox/25.0

Build ID: 20130926170421

I've encountered the same error like Mathieu in comment 6, using Firefox 25 beta 3 on Windows 8.1 x32 and and Windows XP.

On Windows 8.1 it first crashed (please see bug 868905#c7). After the crash I encountered this problem (until the OS was restarted). After the restart, it crashed again and got the same error.
On Windows Xp no crash was encountered, only this error.

Comment 9

[Simona Badau, Desktop QA]

The issue is reproducible also on Windows Vista - after pressing the play button, Firefox 25 beta 4 freezes for a few seconds and shorter after the "uncaught exception: out of memory" error is displayed in the Browser Console.

Mozilla/5.0 (Windows NT 6.0; rv:25.0) Gecko/20100101 Firefox/25.0
Build ID: 20131001024718

Comment 10

[u279076]

Nominating for tracking since this is related to a high-profile project. That said this is the state since at least Firefox 23 so I can understand if we decide to drop tracking.

Comment 11

[Lukas Blakk [:lsblakk] use ?needinfo]

Naveed - do you have concern about this shipping in FF25?  As mentioned in comment 10, this has been present in product for a while.  If there's concern about this for a FF25 feature, please let us know otherwise we will not track.

Comment 12

[Alex Keybl [:akeybl]]

Also pinged mbest to make sure he has a heads up.

Comment 13

[Naveed Ihsanullah [:naveed]]

I don't think this issue is 100%. I haven't been able to reproduce it either in VM or on a native 32bit win8 box. My guess is mbest will be ok letting it go but I will try to get his attention so he weighs in.

Comment 14

[Martin Best (:mbest)]

Hey guys.  Yes, memory usage is a focus in the short term.  Luke and I discussed these issues. New tech will run into these types of issues and we need to smooth out the rough edges quickly.  We need to crush the bugs but I wouldn't block shipping as it effects a subset of the population and there is very little content on the web as extream as Citadel.  These extream tests are ment to strain the browser and highlight what we need to fix as a priority.  With that said it would be good to fast track the fixes should they prove relatively safe to do so as this is a high profile demo.  Our goal is to expand the number of users that can see it without issue.  The crash is a worry, an oom should kill the game not the browser so it would be awesome if we can get a reliable repro case on that.

Right now we have no solid way to see if the user has the minimum requirements of a given game.  This is a problem without a good solution at the moment.  We need a way to fail gracefully and let the user know why.

Comment 15

[Alex Keybl [:akeybl]]

Thanks Naveed and Martin. No need to track then, we'll just consider a low risk fix for uplift if/when found.

Comment 16

[Martin Best (:mbest)]

Just wanted to follow up on this, ran tests this morning and the live demo is running well on Epic's site in Beta.