Created attachment 829892 [details] about support.txt User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:28.0) Gecko/20100101 Firefox/28.0 (Beta/Release) Build ID: 20131110030205 Steps to reproduce: 1. Create fresh profile. 2. Go to http://www.scirra.com/labs/sbperfgl/ Actual results: firefox crashes Expected results: no crash excellent performance https://crash-stats.mozilla.com/report/index/c070ea91-6ba2-46c4-835b-5a4462131110 https://crash-stats.mozilla.com/report/index/badf1bc8-6ee4-495d-b08c-46adf2131110 https://crash-stats.mozilla.com/report/index/b97d2dae-d46e-4747-b212-610bb2131110
Regression Window(m-c) Good: http://hg.mozilla.org/mozilla-central/rev/f0d363d72753 Mozilla/5.0 (Windows NT 6.1; WOW64; rv:28.0) Gecko/20100101 Firefox/28.0 ID:20131030193803 Bad: http://hg.mozilla.org/mozilla-central/rev/e6b861af3117 Mozilla/5.0 (Windows NT 6.1; WOW64; rv:28.0) Gecko/20100101 Firefox/28.0 ID:20131031130944 Pushlog: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=f0d363d72753&tochange=e6b861af3117 Regression Window(b2g-i) Good: http://hg.mozilla.org/integration/b2g-inbound/rev/4ded57b1a15c Mozilla/5.0 (Windows NT 6.1; WOW64; rv:28.0) Gecko/20100101 Firefox/28.0 ID:20131031064700 Bad: http://hg.mozilla.org/integration/b2g-inbound/rev/d548421e14c5 Mozilla/5.0 (Windows NT 6.1; WOW64; rv:28.0) Gecko/20100101 Firefox/28.0 ID:20131031065459 Pushlog: http://hg.mozilla.org/integration/b2g-inbound/pushloghtml?fromchange=4ded57b1a15c&tochange=d548421e14c5
In local build Last good: d23ba68d326e First bad: 175163f053f7 regressed by: 175163f053f7 Jeff Gilbert — Bug 924241 - Don't force-present after post-resize clear. r=bjacob
Is this a test suite we care about, or one that represents a normal user scenario? Trying to understand whether we should track for release.
Yeah, we want to track this, though I'd be surprised if this was limited to just this page.
sse2_blt is #220 on the crash reports list for Firefox 28 for the last 7 days, with only 8 crashes. Jeff, do you still want to track that for Firefox 28? I am in the process of learning about crash report triage.
doesnt crash anymore.
Current Rank: * Firefox 29: #102 @ 0.13% * Firefox 28: #85 @ 0.17% * Firefox 27: N/A (32 crashes) * Firefox 26: N/A (86 crashes) Crashes definitely still exist but it's very low volume. Given comment 6 I'm inclined to resolve this WORKSFORME. @mayankleoboy1, can you please verify this no longer reproduces for you with either of the Firefox branches?
tried this with a new profile and a variety of options. For each test i cleared the complete history and restarted Nightly. default (omtc= off. textures = old) : load bench in new unfocussed tab, and dont switch till game loads completely = no crash load bench in current tab = crash FPS = 60. CPU use= ~10% OMTC on + old textures : load bench in new unfocussed tab = no crash load in current tab = no crash FPS= 12-30. CPU=9% OMTC+new textures : load bench in new tab = no crash FPS=21-30 . CPU use= 10% Lots of jerky animation Also, there is something related to caching. With the default mode (OMTC = off. textures=old). If i first open the bench in a new unfocussed tab, let it load, close it, then open the bench in the current tab, there is no crash. I dont remember in which condition i made comment 6 .
Since the crash is no longer reproducible and is low volume on crashstats I think we should just call this WORKSFORME. We can reopen the bug if the crash resurfaces.
reopened as per comment 8
Sorry, I missed that. > default (omtc= off. textures = old) : load bench in new unfocussed tab, and dont switch till game loads completely = no crash > load bench in current tab = crash > FPS = 60. CPU use= ~10% Can you please explain this in more detail?
1.create a new profile . Since it is default, layers.offmainthreadcomposition.enabled = false layers.use-deprecated-textures= true. 2. Open https://bugzilla.mozilla.org/show_bug.cgi?id=936928 3. Click on the benchmark URL. Crash 4. Now restart the browser, and open the benchmark by "ctrl+click" . Let the bench load/initialise in the unfocussed tab. You wont get a crash. The FPS and the CPU usage is as reported by the benchmark itself.
(In reply to mayankleoboy1 from comment #12) > 1. create a new profile . > Since it is default, > layers.offmainthreadcomposition.enabled = false > layers.use-deprecated-textures= true. > 2. Open https://bugzilla.mozilla.org/show_bug.cgi?id=936928 > 3. Click on the benchmark URL. Crash > 4. Now restart the browser, and open the benchmark by "ctrl+click" . Let the > bench load/initialise in the unfocussed tab. You wont get a crash. What if you ctrl+click the benchmark URL in step 3? In other words: 1. Start Firefox with a new profile 2. Open this bug report 3. CTRL+Click the benchmark URL
So step 4 is pretty much moot. The real bug here is that Scirra SpaceBlaster crashes when opened in the foreground with layers.offmainthreadcomposition.enabled = false and layers.use-deprecated-textures= true. In other words: 1. Start Firefox with a new profile 2. Ensure layers.offmainthreadcomposition.enabled = false and layers.use-deprecated-textures= true in about:config 3. Open this bug and left click the link in the URL field > http://www.scirra.com/labs/sbperfgl/ loads in the current tab and eventually crashes Is this correct?
bp-9b1c1d63-8ff4-4279-8b08-d141e2140117 STR 1. Start Nightly29.0a1 with a new profile 2. Open this bug report and click http://www.scirra.com/labs/sbperfgl/ Actual Results: Crash immediately
(In reply to Alice0775 White from comment #16) > 2. Open this bug report and click http://www.scirra.com/labs/sbperfgl/ When very low on memory I managed to get this: bp-eb210d96-ca25-4a42-b1c6-798572140401 by doing that. Probably not related. This bp-bb12ec9b-a7f2-4451-92ea-c19be2140401 has this Bug on it's "Related Bugs" List. It was not caused by visiting http://www.scirra.com/labs/sbperfgl/ . When this crash happened it triggered a WinXp BreakPoint Popup (from the OS) @ 0x002c:12cb in addition to popping up the MCR. The WinXP BPs have only started occurring this past week, prior to that I had not seen them in a few years. I am running Nightly 31.0a1 (2014-04-01), I too disagree with Comment 9. Bugs involving sse2_blt have only started (for me) in the past few days (and I do not visit Scirra or cimatron.partcommunity.com).
cant reproduce anymore.