1354900 - Assertion failure in PluginInstanceChild::NPN_FinalizeAsyncSurface

Status: RESOLVED FIXED

(Core Graveyard :: Plug-ins, defect)

Description

[Honza Bambas (:mayhemer)]

MOZ_ASSERT(!surface || mCurrentDirectSurface != surface);

where mCurrentDirectSurface == surface and are non-null.


 	xul.dll!MOZ_NoReturn(2914) Line 213	C++
>	xul.dll!mozilla::plugins::PluginInstanceChild::NPN_FinalizeAsyncSurface(0x0422ec4c) Line 2914	C++
 	xul.dll!mozilla::plugins::child::_finalizeasyncsurface(0x0422dc40, 0x0422ec4c) Line 1808	C++
 	NPSWF32_25_0_0_127.dll!030587cc()	Unknown
 	[Frames below may be incorrect and/or missing, no symbols loaded for NPSWF32_25_0_0_127.dll]	
 	NPSWF32_25_0_0_127.dll!0307035f()	Unknown
 	NPSWF32_25_0_0_127.dll!0306e6f3()	Unknown
 	NPSWF32_25_0_0_127.dll!03075307()	Unknown
 	NPSWF32_25_0_0_127.dll!03071b1c()	Unknown
 	NPSWF32_25_0_0_127.dll!030727d7()	Unknown
 	NPSWF32_25_0_0_127.dll!0307377c()	Unknown
 	NPSWF32_25_0_0_127.dll!03070e6e()	Unknown
 	NPSWF32_25_0_0_127.dll!0306e562()	Unknown
 	NPSWF32_25_0_0_127.dll!035fd06d()	Unknown
 	NPSWF32_25_0_0_127.dll!0305caa8()	Unknown
 	NPSWF32_25_0_0_127.dll!0305cd0f()	Unknown
 	xul.dll!mozilla::plugins::PluginModuleChild::NPP_Destroy(0x00f53c00) Line 326	C++
 	xul.dll!mozilla::plugins::PluginInstanceChild::Destroy() Line 4309	C++
 	xul.dll!mozilla::plugins::PluginInstanceChild::AnswerNPP_Destroy(0x006ff7d8) Line 4364	C++
 	xul.dll!mozilla::plugins::PPluginInstanceChild::OnCallReceived({...}, 0x00000000) Line 3301	C++
 	xul.dll!mozilla::plugins::PPluginModuleChild::OnCallReceived({...}, 0x00000000) Line 1238	C++
 	xul.dll!mozilla::ipc::MessageChannel::DispatchInterruptMessage({...}, 0) Line 1942	C++
 	xul.dll!mozilla::ipc::MessageChannel::DispatchMessageW({...}) Line 1806	C++
 	xul.dll!mozilla::ipc::MessageChannel::RunMessage({...}) Line 1681	C++
 	xul.dll!mozilla::ipc::MessageChannel::MessageTask::Run() Line 1714	C++
 	xul.dll!MessageLoop::RunTask({...}) Line 362	C++
 	xul.dll!MessageLoop::DeferOrRunPendingTask({...}) Line 372	C++
 	xul.dll!MessageLoop::DoWork() Line 444	C++
 	xul.dll!base::MessagePumpForUI::DoRunLoop() Line 212	C++
 	xul.dll!base::MessagePumpWin::RunWithDispatcher(0x006ffdb0, 0x00000000) Line 58	C++
 	xul.dll!base::MessagePumpWin::Run(0x006ffdb0) Line 80	C++
 	xul.dll!MessageLoop::RunInternal() Line 239	C++
 	xul.dll!MessageLoop::RunHandler() Line 225	C++
 	xul.dll!MessageLoop::Run() Line 212	C++
 	xul.dll!XRE_InitChildProcess(5, 0x00f010a0, 0x006ffe88) Line 703	C++
 	xul.dll!mozilla::BootstrapImpl::XRE_InitChildProcess(7, 0x00f010a0, 0x006ffe88) Line 65	C++
 	plugin-container.exe!content_process_main(0x00f0e070, 7, 0x00f010a0) Line 64	C++
 	plugin-container.exe!NS_internal_main(8, 0x00f010a0) Line 25	C++
 	plugin-container.exe!wmain(8, 0x0093ceb0) Line 111	C++
 	plugin-container.exe!__scrt_common_main_seh() Line 253	C++


STR, 100% repro:
- debug build (32bit) of m-c
- clean profile
- load cnn.com

Comment 1

[Benjamin Smedberg]

David, is this a sign of a Flash bug? If so do we need this assertion or can we handle this more gracefully?

Does this need to block rollout of async painting to beta/release?

Comment 2

[Jim Mathies [:jimm]]

(In reply to Benjamin Smedberg [:bsmedberg] from comment #1)
> David, is this a sign of a Flash bug? If so do we need this assertion or can
> we handle this more gracefully?
> 
> Does this need to block rollout of async painting to beta/release?

We know about this, it's a side effect of some changes adobe made to the way they handle these surfaces for bug 1306698. It's not considered blocking. I need to to take this to adobe, but I wanted to wait until the other stuff we had out to them was closed out. There's more information in bug 1306698, comment 34.

Comment 3

[Benjamin Smedberg]

Can we remove the assertion, if this is semi-expected? Debug builds shouldn't be crashing on normal browsing like this.

Comment 4

[David Anderson [:dvander] - inactive, e-mail if emergency]

I wouldn't object to removing it. We always copy the texture and mCurrentDirectSurface might be pointless. It's only there to "enforce" the API.

Comment 5

[David Parks [:handyman]]

Attachment: Remove ASSERT when plugin incorrectly finalizes async surface

Removes the assert.  It would be a good idea to track this somehow so we can reverse it.

Comment 6

[David Parks [:handyman]]

Tests: https://treeherder.mozilla.org/#/jobs?repo=try&revision=900479a831b05a2c8097cb2cbfaa91721ebcac36

Comment 7

[David Parks [:handyman]]

That was fast ;)

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/e3b2db0bf090
Remove ASSERT when plugin incorrectly finalizes async surface. r=dvander

Comment 9

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/e3b2db0bf090

Comment 10

[Honza Bambas (:mayhemer)]

Thank you guys.