I believe I found what is a memory leak in the audio of FireFox 31.0. I don't think it existed in the previous version. I originally emailed dev-developer-tools about this, but at this point I think this mailing list may be more appropriate?
I haven't figured out how to reproduce this in a sample yet. I'm hoping someone here will know of this as a known issue. I'll inline the original email here. The details as I've discovered them are in the rest of the history: * I'm working on an HTML5 2d canvas game. If I play it from my website, it runs fine, but for some reason if I play it off of my hard drive (eg: file://...) it makes the CPU memory usage spike to 60% and the Memory usage seems to yoyo between 1-3 gigs of usage over and over again. Weirder still, there are multiple sounds I'm playing with Howler.js. Locally, only one of these sounds plays. When played from my website, they all play as expected. When I use the profiler on the local version, it tells me that the game is mostly idle. It's telling me this even though the game is stuttering like crazy. Here's a screenshot: http://imgur.com/I0pbedE <http://imgur.com/I0pbedE> It seems like this is telling me that everything's ok. Meanwhile, my CPU is at 60%, my memory usage has jumped 2 gigs, and Firefox is crashing every 20 minutes. I think this is somehow related to version 31.0. I didn't notice any of these issues until I upgraded to 31.0. I'm using windows 8. How can I debug this and provide better feedback? I suspect this is a heap issue more than a CPU issue, but it doesn't seem that there's a heap profiler in FF? I'm still trying to reproduce this locally. Is there a more appropriate mailing list to post this in?* ---------- Forwarded message ---------- From: Dan Kaplan <[email protected]> Date: Wed, Jul 30, 2014 at 9:55 PM Subject: Re: Performance issues that the dev tools are not helping me with To: Russ Thomas <[email protected]> Cc: dev-developer-tools <[email protected]> Here's a crash report: AdapterDeviceID: 0x1187 AdapterVendorID: 0x10de Add-ons: %7Bab91efd4-6975-4081-8552-1b3922ed79e2%7D:1.0.16.0,%7Be4a8a97b-f2ed-450b-b12d-ee082ba24781%7D:2.1,%7B73a6fe31-595d-460b-a920-fcc0f8843232%7D:2.6.8.36,artur.dubovoy%40gmail.com:6 .1.2,%7B972ce4c6-7e08-4474-a285-3208198ce6fd%7D:31.0,firebug%40software.joehewitt.com:2 .0.2,%7Bd10d0bf8-f5b5-c8b4-a8b2-2b9879e08c5d%7D:2.6.4 AvailablePageFile: 6332538880 AvailablePhysicalMemory: 6691201024 AvailableVirtualMemory: 267755520 BIOS_Manufacturer: AMI BlockedDllList: BreakpadReserveAddress: 44957696 BreakpadReserveSize: 41943040 BuildID: 20140716183446 CrashTime: 1406782450 EMCheckCompatibility: true EventLoopNestingLevel: 1 FramePoisonBase: 00000000f0de0000 FramePoisonSize: 65536 InstallTime: 1406698646 Notes: AdapterVendorID: 0x10de, AdapterDeviceID: 0x1187, AdapterSubsysID: 0984196e, AdapterDriverVersion: 9.18.13.4052 D2D? D2D+ DWrite? DWrite+ D3D10 Layers? D3D10 Layers+ ProductID: {ec8030f7-c20a-464f-9b0e-13a3a9e97384} ProductName: Firefox ReleaseChannel: release SecondsSinceLastCrash: 7156 StartupTime: 1406779574 SystemMemoryUsePercentage: 47 Theme: classic/1.0 Throttleable: 1 TotalPageFile: 13255356416 TotalPhysicalMemory: 12835926016 TotalVirtualMemory: 4294836224 Vendor: Mozilla Version: 31.0 Winsock_LSP: MSAFD Tcpip [TCP/IP] : 2 : 1 : MSAFD Tcpip [UDP/IP] : 2 : 2 : %SystemRoot%\system32\mswsock.dll MSAFD Tcpip [RAW/IP] : 2 : 3 : MSAFD Tcpip [TCP/IPv6] : 2 : 1 : %SystemRoot%\system32\mswsock.dll MSAFD Tcpip [UDP/IPv6] : 2 : 2 : MSAFD Tcpip [RAW/IPv6] : 2 : 3 : %SystemRoot%\system32\mswsock.dll RSVP TCPv6 Service Provider : 2 : 1 : RSVP TCP Service Provider : 2 : 1 : %SystemRoot%\system32\mswsock.dll RSVP UDPv6 Service Provider : 2 : 2 : RSVP UDP Service Provider : 2 : 2 : %SystemRoot%\system32\mswsock.dll MSAFD RfComm [Bluetooth] : 2 : 1 : useragent_locale: en-US This report also contains technical information about the state of the application when it crashed. On Wed, Jul 30, 2014 at 9:44 PM, Dan Kaplan <[email protected]> wrote: > In the previous email I meant to say: With the dev tools visible, this > only jumps another 5% in *CPU*. > > OK I actually figured out there's a point of time in my source code where > this issue doesn't occur. This problem was introduced the commit that I > added sound to my game. That was 7/8/2014. That was before Firefox 31.0 > was available, right? I think I would have noticed on the day of that this > was slowing down the browser to a crash. > > I'll continue to look into it. I'm still confused why I can't reproduce > this with a simple test. > > > On Wed, Jul 30, 2014 at 9:10 PM, Dan Kaplan <[email protected]> wrote: > >> I'm following along with the issue you created and using it as a guide. >> I started in safe mode and my local game ran at about 30-40% CPU and 500mb >> memory for a very long time. It was still going way slower than I remember >> in the previous version of FF and there was still the sound issue, but it >> seemed consistent. >> >> I then opened dev tools and memory immediately jumped to 1500 mb and 50% >> CPU and both kept climbing. It seems that opening devtools is somehow >> related to the issues I'm seeing? >> >> FWIW, when I play my game on my website, the CPU stays in the 5-10% range >> and the memory is in 400-600mb range the whole time. With the dev tools >> visible, this only jumps another 5% in memory. >> >> I hope these little updates aren't annoying, I'm kind of thinking out >> loud hoping someone can give me more advice on how to reproduce this for a >> bug report. >> >> >> On Wed, Jul 30, 2014 at 8:56 PM, Dan Kaplan <[email protected]> wrote: >> >>> My first thought is it's not as simple as that bug report because I >>> can't reproduce this on my site, only locally. Here's a project that works >>> without issue even when run locally. >>> https://github.com/tieTYT/firefox_31.0_sound_bug I haven't pinned it >>> down yet. >>> >>> >>> On Wed, Jul 30, 2014 at 8:19 PM, Russ Thomas <[email protected]> >>> wrote: >>> >>>> https://bugzilla.mozilla.org/show_bug.cgi?id=936784 >>>> >>>> Sincerely hope this hasn't returned but since I was using howler, and >>>> you're seeing memory issues... >>>> >>>> >>>> >>>> On 30 July 2014 22:00, Dan Kaplan <[email protected]> wrote: >>>> >>>>> I've not been able to reproduce this with a simple example, yet. I'm >>>>> hoping someone else has already discovered that this is a known issue. >>>>> >>>>> I'm working on an HTML5 2d canvas game. If I play it from my website, >>>>> it >>>>> runs fine, but for some reason if I play it off of my hard drive (eg: >>>>> file://...) it makes the CPU memory usage spike to 60% and the Memory >>>>> usage >>>>> seems to yoyo between 1-3 gigs of usage over and over again. Weirder >>>>> still, there are multiple sounds I'm playing with Howler.js. Locally, >>>>> only >>>>> one of these sounds plays. When played from my website, they all play >>>>> as >>>>> expected. >>>>> >>>>> When I use the profiler on the local version, it tells me that the >>>>> game is >>>>> mostly idle. It's telling me this even though the game is stuttering >>>>> like >>>>> crazy. Here's a screenshot: http://imgur.com/I0pbedE It seems like >>>>> this >>>>> is telling me that everything's ok. Meanwhile, my CPU is at 60%, my >>>>> memory >>>>> usage has jumped 2 gigs, and Firefox is crashing every 20 minutes. >>>>> >>>>> I think this is somehow related to version 31.0. I didn't notice any >>>>> of >>>>> these issues until I upgraded to 31.0. I'm using windows 8. >>>>> >>>>> How can I debug this and provide better feedback? I suspect this is a >>>>> heap >>>>> issue more than a CPU issue, but it doesn't seem that there's a heap >>>>> profiler in FF? >>>>> >>>>> I'm still trying to reproduce this locally. Is there a more >>>>> appropriate >>>>> mailing list to post this in? >>>>> _______________________________________________ >>>>> dev-developer-tools mailing list >>>>> [email protected] >>>>> https://lists.mozilla.org/listinfo/dev-developer-tools >>>>> >>>> >>>> >>> >> > _______________________________________________ dev-media mailing list [email protected] https://lists.mozilla.org/listinfo/dev-media
