1268440 - crash in mozilla::layers::DXGIYCbCrTextureHostD3D11::SetCompositor

Status: RESOLVED FIXED

(Core :: Graphics: Layers, defect)

Description

[[:philipp]]

[Tracking Requested - why for this release]:
maybe a regression, signature is jumping up in 47.0b1 data

This bug was filed from the Socorro interface and is 
report bp-c43ead15-5002-40c7-974f-ff8fa2160428.
=============================================================
Crashing Thread (34)
Frame 	Module 	Signature 	Source
0 	xul.dll 	mozilla::layers::DXGIYCbCrTextureHostD3D11::SetCompositor(mozilla::layers::Compositor*) 	gfx/layers/d3d11/TextureD3D11.cpp
1 	xul.dll 	mozilla::layers::CompositableHost::Attach(mozilla::layers::Layer*, mozilla::layers::Compositor*, unsigned int) 	gfx/layers/composite/CompositableHost.h
2 	xul.dll 	mozilla::layers::ImageHost::Attach(mozilla::layers::Layer*, mozilla::layers::Compositor*, unsigned int) 	gfx/layers/composite/ImageHost.cpp
3 	xul.dll 	mozilla::layers::LayerTransactionParent::Attach(mozilla::layers::ShadowLayerParent*, mozilla::layers::CompositableHost*, bool) 	gfx/layers/ipc/LayerTransactionParent.cpp
4 	xul.dll 	mozilla::layers::LayerTransactionParent::RecvUpdate(nsTArray<mozilla::layers::Edit>&&, nsTArray<mozilla::layers::OpDestroy>&&, unsigned __int64 const&, mozilla::layers::TargetConfig const&, nsTArray<mozilla::layers::PluginWindowData>&&, bool const&, bool const&, unsigned int const&, bool const&, mozilla::TimeStamp const&, int const&, nsTArray<mozilla::layers::EditReply>*) 	gfx/layers/ipc/LayerTransactionParent.cpp
5 	xul.dll 	mozilla::layers::PLayerTransactionParent::OnMessageReceived(IPC::Message const&, IPC::Message*&) 	obj-firefox/ipc/ipdl/PLayerTransactionParent.cpp
6 	xul.dll 	mozilla::layers::PCompositorParent::OnMessageReceived(IPC::Message const&, IPC::Message*&) 	obj-firefox/ipc/ipdl/PCompositorParent.cpp
7 	xul.dll 	mozilla::ipc::MessageChannel::DispatchSyncMessage(IPC::Message const&, IPC::Message*&) 	ipc/glue/MessageChannel.cpp
8 	xul.dll 	mozilla::ipc::MessageChannel::DispatchMessageW(IPC::Message const&) 	ipc/glue/MessageChannel.cpp
9 	xul.dll 	mozilla::ipc::MessageChannel::OnMaybeDequeueOne() 	ipc/glue/MessageChannel.cpp

there have been (literally) a handful of those crashes since 43.0 but in early 47.0b1 data the signature seems to be on the rise.

Comment 1

[Edwin Flores [inactive from 2016-12-01] [:eflores] [:edwin]]

This appears to have been fixed in bug 1245813. Nical, would it be possible to uplift 2e636d5fcda8 to 47?

Comment 2

[Edwin Flores [inactive from 2016-12-01] [:eflores] [:edwin]]

oops. see above comment.

Comment 3

[Nicolas Silva [:nical]]

(In reply to Edwin Flores [:eflores] [:edwin] from comment #1)
> This appears to have been fixed in bug 1245813. Nical, would it be possible
> to uplift 2e636d5fcda8 to 47?

Yeah, I'll rebase the patch.

Comment 4

[Nicolas Silva [:nical]]

Attachment: Fix

Turns out the relevant patches were already uplifted in beta 47, but I made a mistake while rebasing one of the patches on top of 47 in bug 1258768, and introduced this crash. This patch fixes the rebasing mistake so that DXGIYCbCrTextureHostD3D11::SetCompositor behaves the same way it should, which is equivalent to what it does in nightly and aurora.

Technically this change has already been reviewed (I just mistakenly uplifted something different from the reviewed code due to the rebasing mistake), but I suppose it will be less confusing if this patch flagged r+ here since it is on a different bug. Likewise, it has already been approved for beta.

Approval Request Comment
[Feature/regressing bug #]:
[User impact if declined]: Crashes.
[Describe test coverage new/current, TreeHerder]:
[Risks and why]: none, this code is already on nightly and aurora, and should have been on beta if I hadn't messed up when uplifting bug 1258768.
[String/UUID change made/needed]: None.

Comment 5

[Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)]

Based on comment 4, this code is already on Fx48.

Comment 6

[Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)]

Comment on attachment 8747698 [details] [diff] [review]
Fix

Fix crash (rebased beta patch issue), Beta47+

Comment 7

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/releases/mozilla-beta/rev/0c49cd1c3701

Comment 8

[u279076]

Although this is fixed in code I don't believe it's been shipped yet in a Beta build. Currently this is #15 in Beta and rising so I'm tagging it as a topcrash. We'll want to keep an eye on it to make sure volume starts dropping once the next Beta is out.