This crash is noted in a few different bugs currently. I want a new bug to clarify which tests fail for this reason. Here's a sample: A crash on shutdown following testInputConnection (but this is not always on shutdown, and affects several tests): https://tbpl.mozilla.org/php/getParsedLog.php?id=41795166&tree=Try&full=1#error0 08:25:37 INFO - 71 INFO TEST-END | testInputConnection | finished in 27643ms 08:25:37 INFO - 72 INFO TEST-START | Shutdown 08:25:37 INFO - 73 INFO Passed: 69 08:25:37 INFO - 74 INFO Failed: 0 08:25:37 INFO - 75 INFO Todo: 0 08:25:37 INFO - 76 INFO SimpleTest FINISHED 08:25:37 INFO - INFO | automation.py | Application ran for: 0:00:32.391246 08:25:37 INFO - INFO | zombiecheck | Reading PID log: /tmp/tmp4tShDYpidlog 08:25:37 INFO - /data/anr/traces.txt not found 08:25:37 INFO - mozcrash INFO | Downloading symbols from: https://ftp-ssl.mozilla.org/pub/mozilla.org/firefox/try-builds/gbrown@mozilla.com-06e1c01790b5/try-android/fennec-33.0a1.en-US.android-arm.crashreporter-symbols.zip 08:25:37 WARNING - PROCESS-CRASH | Shutdown | application crashed [@ libui.so + 0x1befe] 08:25:37 INFO - Crash dump filename: /tmp/tmpYTBpV_/7e84623c-09b5-88d4-65e720b2-5e30e215.dmp 08:25:37 INFO - Operating system: Android 08:25:37 INFO - 0.0.0 Linux 2.6.29-ge3d684d #1 Mon Dec 16 22:26:51 UTC 2013 armv7l generic/sdk/generic:2.3.7/GINGERBREAD/eng.ubuntu.20140123.014351:eng/test-keys 08:25:37 INFO - CPU: arm 08:25:37 INFO - 0 CPUs 08:25:37 INFO - 08:25:37 INFO - Crash reason: SIGSEGV 08:25:37 INFO - Crash address: 0x200 08:25:37 INFO - 08:25:37 INFO - Thread 39 (crashed) 08:25:37 INFO - 0 libui.so + 0x1befe 08:25:37 INFO - r4 = 0x0021ede8 r5 = 0x00000200 r6 = 0x00000001 r7 = 0xffffffed 08:25:37 INFO - r8 = 0x00000000 r9 = 0x00000000 r10 = 0x00000200 fp = 0x00000002 08:25:37 INFO - sp = 0x536ffa50 lr = 0xac712083 pc = 0xab91befe 08:25:37 INFO - Found by: given as instruction pointer in context 08:25:37 INFO - 1 libsurfaceflinger_client.so + 0x12081 08:25:37 INFO - sp = 0x536ffa68 pc = 0xac712083 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 2 libsurfaceflinger_client.so + 0x12847 08:25:37 INFO - sp = 0x536ffab0 pc = 0xac712849 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 3 libsurfaceflinger_client.so + 0xffff 08:25:37 INFO - sp = 0x536ffac8 pc = 0xac710001 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 4 libsurfaceflinger_client.so + 0x12b89 08:25:37 INFO - sp = 0x536ffb00 pc = 0xac712b8b 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 5 libEGL_emulation.so + 0x6d89 08:25:37 INFO - sp = 0x536ffb08 pc = 0x80406d8b 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 6 libEGL_emulation.so + 0x6d03 08:25:37 INFO - sp = 0x536ffb10 pc = 0x80406d05 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 7 libEGL_emulation.so + 0xc746 08:25:37 INFO - sp = 0x536ffb14 pc = 0x8040c748 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 8 libEGL_emulation.so + 0xc746 08:25:37 INFO - sp = 0x536ffb24 pc = 0x8040c748 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 9 libEGL_emulation.so + 0x71f9 08:25:37 INFO - sp = 0x536ffb28 pc = 0x804071fb 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 10 libEGL.so + 0x3767 08:25:37 INFO - sp = 0x536ffb34 pc = 0xabe03769 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 11 libEGL.so + 0x37b7 08:25:37 INFO - sp = 0x536ffb50 pc = 0xabe037b9 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 12 libEGL.so + 0x3767 08:25:37 INFO - sp = 0x536ffb5c pc = 0xabe03769 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 13 libxul.so!mozilla::gl::GLContextEGL::SwapBuffers() [GLLibraryEGL.h:06e1c01790b5 : 284 + 0xb] 08:25:37 INFO - sp = 0x536ffb68 pc = 0x4de004f5 08:25:37 INFO - Found by: stack scanning 08:25:37 INFO - 14 libxul.so!mozilla::layers::CompositorOGL::EndFrame() [CompositorOGL.cpp:06e1c01790b5 : 1317 + 0x5] 08:25:37 INFO - sp = 0x536ffb70 pc = 0x4de62bd7 08:25:37 INFO - Found by: call frame info 08:25:37 INFO - 15 libxul.so!mozilla::layers::LayerManagerComposite::Render() [LayerManagerComposite.cpp:06e1c01790b5 : 475 + 0x5] 08:25:37 INFO - r4 = 0x536ffc00 r5 = 0x536ffbcc sp = 0x536ffb90 pc = 0x4de5b56b 08:25:37 INFO - Found by: call frame info 08:25:37 INFO - 16 libxul.so!mozilla::layers::LayerManagerComposite::EndTransaction(void (*)(mozilla::layers::ThebesLayer*, gfxContext*, nsIntRegion const&, mozilla::layers::DrawRegionClip, nsIntRegion const&, void*), void*, mozilla::layers::LayerManager::EndTransactionFlags) [LayerManagerComposite.cpp:06e1c01790b5 : 248 + 0x5] 08:25:37 INFO - r4 = 0x55241a80 r5 = 0x57409400 r6 = 0x4de49b39 r7 = 0x4de368db 08:25:37 INFO - r8 = 0x00000000 r9 = 0x52ed1190 sp = 0x536ffc50 pc = 0x4de5b6df 08:25:37 INFO - Found by: call frame info 08:25:37 INFO - 17 libxul.so!mozilla::layers::LayerManagerComposite::EndEmptyTransaction(mozilla::layers::LayerManager::EndTransactionFlags) [LayerManagerComposite.cpp:06e1c01790b5 : 198 + 0xb] 08:25:37 INFO - r4 = 0x4de5b67d r5 = 0x0000011c r6 = 0x551ee800 r7 = 0x00000000 08:25:37 INFO - r8 = 0x00000000 r9 = 0x52ed1190 sp = 0x536ffcb0 pc = 0x4de52cb1 08:25:37 INFO - Found by: call frame info 08:25:37 INFO - 18 libxul.so!mozilla::layers::CompositorParent::CompositeToTarget(mozilla::gfx::DrawTarget*, nsIntRect const*) [CompositorParent.cpp:06e1c01790b5 : 639 + 0x5] 08:25:37 INFO - r4 = 0xbabdff80 r5 = 0x0000011c r6 = 0x551ee800 r7 = 0x00000000 08:25:37 INFO - r8 = 0x00000000 r9 = 0x52ed1190 sp = 0x536ffcb8 pc = 0x4de67639 08:25:37 INFO - Found by: call frame info 08:25:37 INFO - 19 libxul.so!mozilla::layers::CompositorParent::ResumeComposition() [CompositorParent.cpp:06e1c01790b5 : 415 + 0x5] 08:25:37 INFO - r4 = 0x536ffd24 r5 = 0x551ee800 r6 = 0x55143af0 r7 = 0x536ffe04 08:25:37 INFO - r8 = 0x4ec86f0f r9 = 0x52ed1190 sp = 0x536ffd20 pc = 0x4de6776d 08:25:37 INFO - Found by: call frame info 08:25:37 INFO - 20 libxul.so!RunnableMethod<mozilla::ipc::MessageChannel, void (mozilla::ipc::MessageChannel::*)(mozilla::ipc::MessageChannel*, mozilla::ipc::Side), Tuple2<mozilla::ipc::MessageChannel*, mozilla::ipc::Side> >::Run() [tuple.h:06e1c01790b5 : 400 + 0x17] 08:25:37 INFO - r4 = 0x4de67785 r5 = 0x536ffdf8 r6 = 0x55143af0 r7 = 0x536ffe04 08:25:37 INFO - r8 = 0x4ec86f0f r9 = 0x52ed1190 sp = 0x536ffd38 pc = 0x4dc3d1d7 08:25:37 INFO - Found by: call frame info 08:25:37 INFO - 21 libxul.so!MessageLoop::RunTask(Task*) [message_loop.cc:06e1c01790b5 : 357 + 0x5] 08:25:37 INFO - r4 = 0x573fefe0 r5 = 0x536ffdf8 r6 = 0x55143af0 r7 = 0x536ffe04 08:25:37 INFO - r8 = 0x4ec86f0f r9 = 0x52ed1190 sp = 0x536ffd48 pc = 0x4dc34b3d 08:25:37 INFO - Found by: call frame info 08:25:37 INFO - 22 libxul.so!MessageLoop::DeferOrRunPendingTask(MessageLoop::PendingTask const&) [message_loop.cc:06e1c01790b5 : 365 + 0x5] 08:25:37 INFO - r4 = 0x00000001 r5 = 0x536ffd78 r6 = 0x55143af0 r7 = 0x536ffe04 08:25:37 INFO - r8 = 0x4ec86f0f r9 = 0x52ed1190 sp = 0x536ffd58 pc = 0x4dc36c31 08:25:37 INFO - Found by: call frame info 08:25:37 INFO - 23 libxul.so!MessageLoop::DoWork() [message_loop.cc:06e1c01790b5 : 443 + 0x3] 08:25:37 INFO - r4 = 0x536ffdf8 r5 = 0x536ffd78 r6 = 0x55143af0 r7 = 0x536ffe04 08:25:37 INFO - r8 = 0x4ec86f0f r9 = 0x52ed1190 sp = 0x536ffd68 pc = 0x4dc387cf 08:25:37 INFO - Found by: call frame info 08:25:37 INFO - 24 libxul.so!base::MessagePumpDefault::Run(base::MessagePump::Delegate*) [message_pump_default.cc:06e1c01790b5 : 34 + 0x7] 08:25:37 INFO - r4 = 0x531dbaa0 r5 = 0x536ffdf8 r6 = 0x00000000 r7 = 0x4ec86f0f 08:25:37 INFO - r8 = 0x4ec86f0f r9 = 0x52ed1190 sp = 0x536ffd98 pc = 0x4dc34fe7 08:25:37 INFO - Found by: call frame info

This bug seems to be the root cause to many of our Android 2.3 (emulator) failures on Test Infra. I think snorp has looked at this a bit. What is our current assessment of the problem? Do we have any ideas to try? Is this a life cycle issue? It seems to happen on shutdown, but what kind of shutdown? Do we attempt a hard kill?

Folks run into this kind of thing: http://www.ogre3d.org/forums/viewtopic.php?f=21&t=79898 https://www.libreoffice.org/bugzilla/show_bug.cgi?id=75797 My entirely uneducated guess is that we're swapping buffers when we don't have a surface yet. But I really don't know what I'm talking about in this area.

Seems unlikely that we'd try to swap on a null (or garbage) surface, but I dunno. Geoff how does the harness try to do shutdown?

http://hg.mozilla.org/mozilla-central/annotate/cadcd47db610/mobile/android/base/tests/BaseTest.java#l178 http://hg.mozilla.org/mozilla-central/annotate/cadcd47db610/testing/mochitest/roboextender/bootstrap.js#l47

Ah. I think that's the same way we quit mochitest and reftests, so wth?

Android 2.3 is not supported in Firefox 48+.