restart rendering process after it crashes

[Florin]

Hi,

I have wrote a small browser application on top of chromium's Content API. 

Currently, I have managed to detect if the rendering process has crashed, but I would like to restart it afterwards. 

How I can do this?

Thank you!

[Nasko Oskov]

Performing any navigation in the same WebContents should recreate a renderer process. If you want the same document, you can just reload it.

--
--
Chromium Developers mailing list: chromi...@chromium.org
View archives, change email options, or unsubscribe:
http://groups.google.com/a/chromium.org/group/chromium-dev

[Florin Maticu]

Hi,

I have managed to make it work by posting a task to content::BrowserThread::UI. The task is  loading an URL after the render process has crashed.

content::BrowserThread::PostTask(content::BrowserThread::UI, FROM_HERE,

    base::Bind(&loadURLAfterRenderCrash,this));

On the other hand, I get a stack trace if I try to load the URL from the RenderProcessGone(base::TerminationStatus status). The method is implemented by my WebContentsDelegate class.

In both cases I am using the same content::WebContents object (I have only one that is needed for the entire application).

I was expecting that the RenderProcessGone method is executed in the BrowserThread::UI context also.

Can someone explain why is not the case?

Thank you!

Florin

[Nasko Oskov]

On Tue, Apr 12, 2016 at 4:29 AM, Florin Maticu <flm...@airtame.com> wrote:

Hi,

I have managed to make it work by posting a task to content::BrowserThread::UI. The task is  loading an URL after the render process has crashed.

content::BrowserThread::PostTask(content::BrowserThread::UI, FROM_HERE,

    base::Bind(&loadURLAfterRenderCrash,this));

On the other hand, I get a stack trace if I try to load the URL from the RenderProcessGone(base::TerminationStatus status). The method is implemented by my WebContentsDelegate class.

Initiating navigation while handling notifications from WebContentsObserver must be avoided. It leads to problems. Your approach of posting a task is the right way to do it. 

Also, having a single object be both WebContentsDelegate and WebContentsObserver seems suboptimal and can lead to problems if you ever want to add more than one WebContents to your app.

 

In both cases I am using the same content::WebContents object (I have only one that is needed for the entire application).

I was expecting that the RenderProcessGone method is executed in the BrowserThread::UI context also.

It only runs on the UI thread. 

 

Can someone explain why is not the case?

Can you show me a stack trace where this is actually the case? WebContentsObserver methods are only called on the UI thread.

In general, WebContentsObserver::RenderProcessGone is a problematic API and it is on our list of APIs to remove. You are better off using RenderProcessHostObserver to monitor for the lifetime of renderer processes.

[Florin Maticu]

Hi,

I have attached my stack trace after the rendering process is killed (the case when I am trying to load an url from : RenderProcessGone(base::TerminationStatus status)

[0413/102558:ERROR:web_contents.cc(97)] Render process has terminated : (render killed)

[0413/102558:FATAL:raw_channel_posix.cc(328)] Check failed: message_loop_for_io()->WatchFileDescriptor( fd_.get().fd, true, base::MessageLoopForIO::WATCH_READ, read_watcher_.get(), this). 

#0 0x000000058d38 base::debug::StackTrace::StackTrace()

#1 0x00000005efac logging::LogMessage::~LogMessage()

#2 0x0000004682c6 mojo::system::(anonymous namespace)::RawChannelPosix::OnInit()

#3 0x0000004666f4 mojo::system::RawChannel::Init()

#4 0x00000046c16a mojo::system::Channel::Init()

#5 0x000000459e92 mojo::system::ChannelManager::CreateChannelOnIOThreadHelper()

#6 0x00000045a254 mojo::system::ChannelManager::CreateChannelHelper()

#7 0x00000045ac00 base::internal::RunnableAdapter<>::Run()

#8 0x00000045ab0c _ZN4base8internal12InvokeHelperILb0EvNS0_15RunnableAdapterIMN4mojo6system14ChannelManagerEFvyNS3_8embedder20ScopedPlatformHandleE13scoped_refptrINS4_15ChannelEndpointEERKNS_8CallbackIFvvEEES8_INS_10TaskRunnerEEEEENS0_8TypeListIJPS5_RKyS7_PS9_SF_PSG_EEEE8MakeItSoESK_SM_SO_S7_SP_SF_SQ_

#9 0x00000045aa6c _ZN4base8internal7InvokerINS_13IndexSequenceIJLj0ELj1ELj2ELj3ELj4ELj5EEEENS0_9BindStateINS0_15RunnableAdapterIMN4mojo6system14ChannelManagerEFvyNS6_8embedder20ScopedPlatformHandleE13scoped_refptrINS7_15ChannelEndpointEERKNS_8CallbackIFvvEEESB_INS_10TaskRunnerEEEEEFvPS8_ySA_SD_SI_SK_ENS0_8TypeListIJNS0_17UnretainedWrapperIS8_EEyNS0_13PassedWrapperISA_EESD_SG_SK_EEEEENSQ_IJNS0_12UnwrapTraitsISS_EENSX_IyEENSX_ISU_EENSX_ISD_EENSX_ISG_EENSX_ISK_EEEEENS0_12InvokeHelperILb0EvSN_NSQ_IJSO_RKySA_PSC_SI_PSJ_EEEEESF_E3RunEPNS0_13BindStateBaseE

#10 0x000000090560 base::debug::TaskAnnotator::RunTask()

#11 0x000000060a74 base::MessageLoop::RunTask()

#12 0x000000060d4c base::MessageLoop::DeferOrRunPendingTask()

#13 0x000000060ef2 base::MessageLoop::DoWork()

#14 0x00000008f5d4 base::MessagePumpLibevent::Run()

#15 0x00000006a7d8 base::RunLoop::Run()

#16 0x000000060538 base::MessageLoop::Run()

#17 0x00000007bb8a base::Thread::ThreadMain()

#18 0x000000078906 base::(anonymous namespace)::ThreadFunc()

#19 0x0000769f6100 start_thread

Best regards,

Florin

[Anand]

That code is ancient (was deleted in January, I think). You should sync. Mojo stuff is under very active development.