Closed feross closed 8 years ago
Crash stack trace:
base::MessagePumpCFRunLoopBase::RunWorkSource(void*) (in Electron Framework) (message_pump_mac.mm:309)
base::mac::CallWithEHFrame(void () block_pointer) (in Electron Framework) (call_with_eh_frame_asm.S:38)
base::MessagePumpCFRunLoopBase::RunWork() (in Electron Framework) (message_pump_mac.mm:330)
base::MessageLoop::DoWork() (in Electron Framework) (message_loop.cc:610)
base::MessageLoop::RunTask(base::PendingTask const&) (in Electron Framework) (message_loop.cc:487)
base::debug::TaskAnnotator::RunTask(char const*, base::PendingTask const&) (in Electron Framework) (task_annotator.cc:51)
jingle_glue::JingleThreadWrapper::ProcessPendingSends() (in Electron Framework) (thread_wrapper.cc:215)
(anonymous namespace)::MediaController::Construct_w(cricket::MediaEngineInterface*) (in Electron Framework) (mediacontroller.cc:74)
webrtc::Call::Create(webrtc::Call::Config const&) (in Electron Framework) (call.cc:176)
webrtc::internal::Call::Call(webrtc::Call::Config const&) (in Electron Framework) (call.cc:200)
webrtc::CongestionController::CongestionController(webrtc::ProcessThread*, webrtc::CallStats*, webrtc::BitrateObserver*) (in Electron Framework) (congestion_controller.cc:175)
webrtc::ProcessThreadImpl::Start() (in Electron Framework) (process_thread_impl.cc:81)
It looks like an internal webrtc bug.
Issue filed on the Chromium/WebRTC bug tracker here: https://bugs.chromium.org/p/webrtc/issues/detail?id=5766
@zcbenz This issue may be related to another webrtc issue that causes the CPU to run at 100%, which I've seen happen in the wild with WebTorrent Desktop. This issue was fixed in Chrome 51.
Any chance we can skip upgrading to Chrome 50 and go straight to 51 to fix this critical issue?
We will upgrade to Chrome 51 when it enters stable channel, using a beta release may introduce other bugs while fixing only one problem.
@zcbenz That's fair - I understand. Would it be possible to create a next
branch or something similar where the next version of Chromium/node.js can be tested prior to an Electron release? That might help to make Electron more stable overall and prevent some the bugs that have affected the recent releases.
That sounds a good idea to me, the only problem is upgrading to new versions of Chrome takes lots of work, and we don't have time to maintain two branches at the same time.
Closing since Electron is now on Chrome 51.
I get a EXC_CRASH (SIGABRT) crash. I have included the relevant parts of the stack trace below.
It looks like this crash is related to WebRTC (libjingle) in Electron. It's interesting that there is 2.1GB allocated to the stack. Could that be the source of the crash? That seems awfully high.
Any ideas?
Full stack trace is here: https://gist.github.com/feross/9ced83fac050bd927b8afea3939a99fa