When using OMTC, the lock screen sticks during dragging and disappears a moment after it's unlocked.

This patch turns on Tiled Thebes Layers for MOZ_HAVA_COMPOSITOR or gonk builds. Using the tiled thebes layer makes the problem go away in B2G. Still need to determine root cause.

Try server results with the patch applied: https://tbpl.mozilla.org/?tree=Try&rev=3b1b2d179a93

Comment on attachment 630342 [details] [diff] [review] Turns on Tiled Thebes Perfect. We can defer investigating the bug in the other path until after gralloc support lands. It's not clear a priori which path will be faster, and not worth investigating until gralloc. Please file a followup bug for that and have the gralloc bug block the followup.

Followup bug is bug 726101

Err: followup bug is bug 762101

dhylands, whups! You want to set the Keyword, not the Whiteboard! :) (Yes, that's pretty confusing, sorry.)

I tried this patch on the device and saw a new assertion: I/Gecko ( 2449): [Child 2449] WARNING: NS_ENSURE_TRUE(IsChromeProcess()) failed: file /Volumes/mac/moz/iB2G/content/base/src/nsFrameMessageManager.cpp, line 555 I/Gecko ( 2449): [Child 2449] WARNING: nsWindow::GetNativeData not implemented for this type: file /Volumes/mac/moz/iB2G/widget/xpwidgets/PuppetWidget.cpp, line 578 I/Gecko ( 2449): [Child 2449] WARNING: nsWindow::GetNativeData not implemented for this type: file /Volumes/mac/moz/iB2G/widget/xpwidgets/PuppetWidget.cpp, line 578 I/Gecko ( 2369): [Parent 2369] WARNING: NS_ENSURE_SUCCESS(status, status) failed with result 0x804B0046: file /Volumes/mac/moz/iB2G/content/base/src/nsCrossSiteListenerProxy.cpp, line 508 I/Gecko ( 2449): [Child 2449] ###!!! ABORT: PaintedTiledLayerBuffer must be made IPC safe (not share pointers): file /Volumes/mac/moz/iB2G/gfx/layers/ipc/ShadowLayers.cpp, line 228 F/libc ( 2449): Fatal signal 11 (SIGSEGV) at 0x00000000 (code=1) I/wpa_supplicant( 2427): wlan0: Trying to associate with d8:c7:c8:eb:0c:a1 (SSID='Mozilla Guest' freq=2462 MHz) I/wpa_supplicant( 2427): wlan0: Association request to the driver failed I/DEBUG ( 1823): *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** I/DEBUG ( 1823): Build fingerprint: 'unknown' I/DEBUG ( 1823): pid: 2449, tid: 2449 >>> /system/b2g/plugin-container <<< I/DEBUG ( 1823): signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 00000000

Out-of-process content won the race with this patch due to the unfortunate checkin-needed issue, so we have to wait on bug 747811 now.

Closing this bug since it isn't relevant any more.