889515 - Get rid of sCurrentCompositor

Status: RESOLVED FIXED

(Core :: Graphics: Layers, defect)

Description

[Matt Woodrow (:mattwoodrow)]

Attachment: Get rid of sCurrentCompositor

This breaks down when we have multiple windows.

Comment 1

[Nicolas Silva [:nical]]

Comment on attachment 770337 [details] [diff] [review]
Get rid of sCurrentCompositor

Review of attachment 770337 [details] [diff] [review]:
-----------------------------------------------------------------

::: gfx/layers/ipc/CompositorParent.cpp
@@ +680,5 @@
> +void
> +CompositorParent::NotifyChildCreated(uint64_t aChild)
> +{
> +  sIndirectLayerTrees[aChild].mParent = this;
> +}

Nit: Why not just put this line in RecvNotifyChildCreated ?

Comment 2

[Matt Woodrow (:mattwoodrow)]

(In reply to Nicolas Silva [:nical] from comment #1)
> Comment on attachment 770337 [details] [diff] [review]
> Get rid of sCurrentCompositor
> 
> Review of attachment 770337 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> ::: gfx/layers/ipc/CompositorParent.cpp
> @@ +680,5 @@
> > +void
> > +CompositorParent::NotifyChildCreated(uint64_t aChild)
> > +{
> > +  sIndirectLayerTrees[aChild].mParent = this;
> > +}
> 
> Nit: Why not just put this line in RecvNotifyChildCreated ?

Because the CrossProcessCompositorParent needs a function to call, I thought calling the RecvNotifyChildCreated version would be weird. The Recv*** functions are normally only called from auto-generated IPDL code.

Comment 3

[Nicolas Silva [:nical]]

ok, works for me

Comment 4

[Matt Woodrow (:mattwoodrow)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/cb34d80e6fc8

Comment 5

[Phil Ringnalda (:philor)]

Backed out in https://hg.mozilla.org/integration/mozilla-inbound/rev/a66fa3d6f6cb - 10.6 says (https://tbpl.mozilla.org/php/getParsedLog.php?id=25204404&tree=Mozilla-Inbound) "Assertion failure: lm->GetBackendType() == LAYERS_CLIENT, at ../../../layout/ipc/RenderFrameParent.cpp:608"

Comment 6

[Phil Ringnalda (:philor)]

Other platforms did eventually chip in, including opt crashes in browser-chrome, it was just that 10.6 was firstest and most thorough about failing.

Comment 7

[Matt Woodrow (:mattwoodrow)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/fa6ef0b63025

Comment 8

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

Backed out for bc bustage on OSX: http://hg.mozilla.org/integration/mozilla-inbound/rev/94e902f5f517
https://tbpl.mozilla.org/php/getParsedLog.php?id=25274228&tree=Mozilla-Inbound

Didn't happen on 10.6, but did on 10.7 and 10.8.

Comment 9

[Matt Woodrow (:mattwoodrow)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/dabdd7ceb43e

Comment 10

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/3e969b1777a7