Граф коммитов

1511 Коммитов

Автор SHA1 Сообщение Дата
Andrew McCreight 928e31b2f3 Bug 1038855, part 2 - Use an nsRefPtr for CompositorChild::SharedFrameMetricsData::mBuffer. r=bjacob 2014-07-16 12:39:03 -07:00
Wes Kocher a5c126599f Backed out changeset f4592864c982 (bug 1037360) for b2g mochitest-9 failures 2014-07-15 15:45:38 -07:00
Sotaro Ikeda 11459c436e Bug 1037360 - Fix SharedBufferManagerParent's destruction r=jmuizelaar 2014-07-15 13:33:26 -07:00
Benoit Girard 1465e09a99 Bug 1037691 - Fix instances where refcounted classes expose a public destructor. r=ehsan,bent,mayhemer,jrmuizel 2014-07-15 11:37:45 -04:00
Sotaro Ikeda 225b222537 Bug 1035755 - Fix RemoveTextureFromCompositableTracker::ReleaseTextureClient() r=nical 2014-07-15 06:22:35 -07:00
Sotaro Ikeda f84bfd597e Bug 1036419 - Add lock to GrallocReporter::CollectReports() r=nical 2014-07-14 06:35:23 -07:00
Jeff Gilbert f032a2a672 Bug 1037151 - Merge SharedSurface_GL into SharedSurface. - r=mattwoodrow 2014-07-11 15:10:49 -07:00
Sotaro Ikeda c0c6b14e58 Bug 1036561 - Reduce log spam from SharedBufferManagerChild r=nical 2014-07-10 07:24:07 -07:00
Nicolas Silva 88c17fcf92 Bug 1027601 - Create and Allocate TextureClients in a single step in SharedPlanarYCbCrImage. r=sotaro 2014-07-10 13:48:29 +02:00
Carsten "Tomcat" Book 388e2958d7 Backed out changeset ef24cd472cfb (bug 1027601) 2014-07-09 13:22:50 +02:00
Nicolas Silva e9e3887261 Bug 1027601 - Create and Allocate TextureClients in a single step in SharedPlanarYCbCrImage. r=sotaro 2014-07-09 11:59:53 +02:00
Sotaro Ikeda dbd3cbb58a Bug 1035281 - Fix potential file descriptor leak problem r=nical 2014-07-08 18:35:42 -07:00
Sotaro Ikeda e445b313ec Bug 1034294 - Fix SharedBufferManagerParent r=nical 2014-07-08 06:46:17 -07:00
Wes Kocher c76c770c79 Backed out changeset 476485e3e30a (bug 1034294) for b2g m8 orange on a CLOSED TREE 2014-07-07 11:24:14 -07:00
Sotaro Ikeda 9cf8d7b48d Bug 1034294 - Fix SharedBufferManagerParent 2014-07-07 07:11:40 -07:00
Benoit Jacob 31db832664 Bug 774388 - Patch 14: block on ImageBridgeParent shutdown before we destroy the compositor thread - r=mattwoodrow 2014-07-06 21:30:15 -04:00
Benoit Jacob 1a2d3576a2 Bug 774388 - Patch 13: Temporarily addref to prevent destruction during handling of sync Stop message, to avoid triggering the 'mismatched CxxStackFrames' assertion of MessageChannel - r=nical 2014-07-04 14:04:12 -04:00
Benoit Jacob 26180d9fe5 Bug 774388 - Patch 12: make ImageBridgeParent 's refcounting implementation automatically defer destruction to the main thread - r=nical 2014-07-04 14:04:12 -04:00
Benoit Jacob 0c6a7dfedf Bug 774388 - Patch 11: Use a plain RunnableFunction to release the ImageBridgeParent singleton - r=mattwoodrow 2014-07-04 14:04:12 -04:00
Benoit Jacob c083b322d1 Bug 774388 - Patch 10: destroy the ImageBridgeChild singleton on the main thread (and sprinkle some assertions) - r=mattwoodrow 2014-07-04 14:04:12 -04:00
Benoit Jacob 61671a4f88 Bug 774388 - Patch 9: Introduce NS_INLINE_DECL_THREADSAFE_REFCOUNTING_WITH_MAIN_THREAD_DESTRUCTION - r=mattwoodrow 2014-07-04 14:04:12 -04:00
Benoit Jacob 4b5305dd44 Bug 774388 - Patch 8: Avoid a deadlock between shutdown of CompositorParent and of CompositorThreadHolder - r=mattwoodrow 2014-07-04 14:04:11 -04:00
Benoit Jacob 4ecdd340bf Bug 774388 - Patch 7: Block the main thread on compositor shutdown - r=mattwoodrow 2014-07-04 14:04:11 -04:00
Benoit Jacob ef93653aa4 Bug 774388 - Patch 5: Wait for [CrossProcess]CompositorParent's to be gone before we tear down the compositor thread - r=mattwoodrow 2014-07-04 14:04:11 -04:00
Benoit Jacob 15eb88c866 back out bug 774388 and bug 1028383 for intermittent e10s mochitest-2 crashes 2014-07-04 21:24:32 -04:00
Benoit Jacob 408bebb7c2 Bug 774388 - Patch 13: Temporarily addref to prevent destruction during handling of sync Stop message, to avoid triggering the 'mismatched CxxStackFrames' assertion of MessageChannel - r=nical 2014-07-04 14:04:12 -04:00
Benoit Jacob f15b491b47 Bug 774388 - Patch 12: make ImageBridgeParent 's refcounting implementation automatically defer destruction to the main thread - r=nical 2014-07-04 14:04:12 -04:00
Benoit Jacob da7d384505 Bug 774388 - Patch 11: Use a plain RunnableFunction to release the ImageBridgeParent singleton - r=mattwoodrow 2014-07-04 14:04:12 -04:00
Benoit Jacob d8d7a86cab Bug 774388 - Patch 10: destroy the ImageBridgeChild singleton on the main thread (and sprinkle some assertions) - r=mattwoodrow 2014-07-04 14:04:12 -04:00
Benoit Jacob 688107fdd1 Bug 774388 - Patch 9: Introduce NS_INLINE_DECL_THREADSAFE_REFCOUNTING_WITH_MAIN_THREAD_DESTRUCTION - r=mattwoodrow 2014-07-04 14:04:12 -04:00
Benoit Jacob 34a843e71a Bug 774388 - Patch 8: Avoid a deadlock between shutdown of CompositorParent and of CompositorThreadHolder - r=mattwoodrow 2014-07-04 14:04:11 -04:00
Benoit Jacob e981c4f333 Bug 774388 - Patch 7: Block the main thread on compositor shutdown - r=mattwoodrow 2014-07-04 14:04:11 -04:00
Benoit Jacob e69d16b7ba Bug 774388 - Patch 5: Wait for [CrossProcess]CompositorParent's to be gone before we tear down the compositor thread - r=mattwoodrow 2014-07-04 14:04:11 -04:00
Ryan VanderMeulen 6d9424856d Backed out 10 changesets (bug 1033358, bug 774388, bug 1028383) for causing frequent shutdown crashes on a CLOSED TREE.
Backed out changeset a54b05c9e4a1 (bug 1028383)
Backed out changeset 37985f79e0c2 (bug 774388)
Backed out changeset d07521729077 (bug 774388)
Backed out changeset 9f14b17f358c (bug 774388)
Backed out changeset 2d347d6aa9bc (bug 774388)
Backed out changeset 99581dfb5ec4 (bug 774388)
Backed out changeset 2532e22d6135 (bug 774388)
Backed out changeset 719844108f1a (bug 774388)
Backed out changeset 7829c78348a4 (bug 1033358)
Backed out changeset c571df9a85de (bug 1033358)
2014-07-03 20:37:05 -04:00
Benoit Jacob 616ab38143 Bug 774388 - Patch 11: Use a plain RunnableFunction to release the ImageBridgeParent singleton - r=mattwoodrow 2014-07-03 14:53:34 -04:00
Benoit Jacob 6b1ce47d0e Bug 774388 - Patch 10: destroy the ImageBridgeChild singleton on the main thread (and sprinkle some assertions) - r=mattwoodrow 2014-07-03 14:53:32 -04:00
Benoit Jacob 0ce7e6bc43 Bug 774388 - Patch 9: Introduce NS_INLINE_DECL_THREADSAFE_REFCOUNTING_WITH_MAIN_THREAD_DESTRUCTION - r=mattwoodrow 2014-07-03 14:53:31 -04:00
Benoit Jacob 4426b5bf8c Bug 774388 - Patch 8: Avoid a deadlock between shutdown of CompositorParent and of CompositorThreadHolder - r=mattwoodrow 2014-07-03 14:53:29 -04:00
Benoit Jacob 5d921050a5 Bug 774388 - Patch 7: Block the main thread on compositor shutdown - r=mattwoodrow 2014-07-03 14:53:28 -04:00
Benoit Jacob 5792743acd Bug 774388 - Patch 5: Wait for [CrossProcess]CompositorParent's to be gone before we tear down the compositor thread - r=mattwoodrow 2014-07-03 14:53:27 -04:00
Sotaro Ikeda f87d695780 Bug 1032364 - Change gralloc buffer's key to 64bit r=jmuizelaar 2014-07-02 14:45:59 -07:00
Sotaro Ikeda 40254e3c7f Bug 1031527 - remove dup fd from ParamTraits<MagicGrallocBufferHandle>::Read() r=jrmuizel 2014-06-30 16:36:52 -07:00
Chris Peterson 146b02353e Bug 1031165 - Fix non-unified OS X build's warnings-as-errors in gfx/layers. r=bjacob 2014-06-26 23:36:27 -07:00
Sotaro Ikeda e17cd37863 Bug 1029719 - Release TextureClient on correct thread r=jmuizelaar 2014-06-27 06:26:51 -07:00
Markus Stange 76bee9f7aa Bug 1018522 - Move the call to ShadowLayersUpdated before EndTransaction so that EndTransaction can pick up layer tree changes caused by ShadowLayersUpdated. r=mattwoodrow
--HG--
extra : rebase_source : 640af9f3ef2488e8bf10315787a7ed389771568e
2014-06-26 14:40:10 +02:00
Benoit Jacob 360bdb42d9 Bug 1028588 - Fix dangerous public destructors in gfx - r=jrmuizel 2014-06-23 14:49:07 -04:00
Sotaro Ikeda 1ad1ef7739 Bug 1026580 - Remove an infinite loop possibility from AsyncTransactionTracker::WaitComplete() r=nical 2014-06-21 08:30:30 -07:00
Birunthan Mohanathas bc0233fe47 Bug 1026535 - Fix mismatched class/struct tags. r=ehsan 2014-06-18 17:57:51 -07:00
Benoit Jacob c957e59732 Bug 1027251 - Fix or whitelist dangerous public destructors in gfx/ - r=jrmuizel 2014-06-18 22:28:59 -04:00
Nicholas Nethercote aaf6fb2124 Bug 1010064 - Allow memory reports to be anonymized. r=bsmedberg.
--HG--
extra : rebase_source : 5eb756b17a97e751deb0f47e909fdb35533cb23b
2014-05-20 23:06:54 -07:00
Jonathan Watt 6200c18bc8 Bug 1025490 - Stop using gfxImageSurface in the layers code. r=Bas 2014-06-17 18:35:51 +01:00
Wes Kocher c2735215c7 Backed out changeset 66f9a0038c67 (bug 774388) for various frequent intermittent failures including mochitest-2 on a CLOSED TREE 2014-06-16 19:00:15 -07:00
Benoit Jacob 6d4f99fdba Bug 774388 - Patch 6: fix a silly double negation - r=nical 2014-06-06 09:51:27 -04:00
Benoit Jacob cea2702a66 Bug 774388 - Patch 5: Properly manage the lifetime of the compositor thread, by waiting for CrossProcessCompositorParents to die before destroying it - r=nical,mattwoodrow 2014-06-06 09:51:26 -04:00
Solomon Chiu 0e4402768f Bug 973777 - Breaking down the information of GrallocReporter. r=njn 2014-06-15 23:27:00 +08:00
"Kan-Ru Chen (陳侃如)" 888082a108 Bug 1020169 - Part 2. Register nested LayerTree. r=mattwoodrow
Use PCompositor::NotifyChildCreated which was ill-defined for
cross-process case.

Assume that for all LayerTrees managed by a CrossProcessCompositorParent
all share the same CompositorParent.
2014-06-14 15:25:02 +08:00
chiajung hung 27f22929f3 Bug 1012961 - ABORT: Invalid SurfaceDescriptor passed in: file gfx/layers/opengl/GrallocTextureHost.cpp, line 292. r=nical 2014-06-12 02:06:00 -04:00
Chris Lord 3c0175baa4 Bug 1011572 - Only do platform-sync on layer transactions when !IsSameProcess. r=nical
There's no need to do platform syncing when the compositor is in the same
process as the content thread, as they'll share a connection to the display
server.
2014-06-11 14:52:20 +01:00
Chris Lord 1188fbbd28 Bug 1011569 - Remove PreferMemoryOverShmem in favour of IsSameProcess. r=nical 2014-06-11 14:52:19 +01:00
Jonathan Watt a25ab6979d Bug 1022667, part 6 - Remove some unused code from ShadowLayerUtilsGralloc.cpp. r=mattwoodrow 2014-06-10 16:42:24 +01:00
Botond Ballo 727b1998f8 Bug 1014280 - Add a background color attribute to scrollable ContainerLayers. r=tn,BenWa
--HG--
extra : source : dff750e410e5f4e1ed019527ea17f37e53dd8fe6
2014-06-06 16:50:20 -04:00
Sotaro Ikeda 539d95f87e Bug 1001417 - Forward fence objects in SharedSurfaceGralloc to Compositor r=jgilbert,nical 2014-06-08 06:18:53 -07:00
Benoit Girard 297f25d502 Bug 1016678 - Add textured compositor benchmark. r=jrmuizel
--HG--
extra : rebase_source : 11465ad7c287db421fc68022269823b7da992835
2014-06-04 18:37:18 -04:00
Ed Morley c31807a83a Backed out changeset 6f2e001c5f39 (bug 774388) 2014-06-06 15:58:58 +01:00
Ed Morley 823c01e984 Backed out changeset 0f81ceab808a (bug 774388) 2014-06-06 15:58:54 +01:00
Benoit Jacob 166db540b6 Bug 774388 - Patch 6: fix a silly double negation - r=nical 2014-06-06 09:51:27 -04:00
Benoit Jacob 969d70bfa6 Bug 774388 - Patch 5: Wrap the global raw compositor thread pointer, and global raw refcount integer, into a proper refcounted singleton class - r=nical,mattwoodrow 2014-06-06 09:51:26 -04:00
Benoit Jacob c8affc25e2 Bug 774388 - Patch 3: remove sCompositorLoop, it's redundant - r=nical,mattwoodrow 2014-06-05 16:42:13 -04:00
Benoit Jacob 6e6c966de7 Bug 774388 - Patch 2: remove sCompositorThreadID, it's redundant - r=nical,mattwoodrow 2014-06-05 16:42:13 -04:00
Benoit Jacob 7008eef9b0 Bug 774388 - Patch 1: remove StartUpWithExistingThread, it's unused - r=nical 2014-06-05 16:42:13 -04:00
Ed Morley 8d9a53531a Backed out changeset d37aa03d00fb (bug 1016805) for OS X 10.6 reftest failures; CLOSED TREE 2014-06-04 17:52:55 +01:00
Sotaro Ikeda 43b3882d48 Bug 1016805 - fix AsyncTransactionTracker handling r=nical 2014-06-04 07:49:14 -07:00
Victor Porof 9ce7e9c089 Bug 1007203 - Always add categories when pushing to the pseudostack, r=djvj 2014-05-23 17:12:29 -04:00
Matt Woodrow 795c2136c0 Bug 854421 - Part 8: Try find a 'root' refresh driver to see if it's blocked. r=vlad 2014-05-29 09:43:41 +12:00
Matt Woodrow d78f4156e4 Bug 854421 - Part 7: Allocate and return transaction id's from the refresh driver when using OMTC to prevent over-production. r=roc 2014-05-29 09:43:39 +12:00
Matt Woodrow e9ea25188e Bug 854421 - Part 3: Add a way to detect PLayerTransaction objects that exist only for testing and won't composite. r=nical 2014-05-29 09:42:16 +12:00
Matt Woodrow 9c37506034 Bug 854421 - Part 2: If we cancel a composite when going into test mode, fire the DidComposite event anway in case someone is waiting on it. r=nical 2014-05-29 09:42:16 +12:00
Matt Woodrow b61c593824 Bug 854421 - Part 1: Add a transaction id number to DidComposite. r=nical 2014-05-29 09:42:14 +12:00
Nicolas Silva f10ccb8404 Bug 1016321 - Remove unwanted fprintfs. r=edmorley 2014-05-27 05:58:00 -04:00
Tatiana Meshkova 4a3f012888 Bug 1016795 - Make CompositorParent usable by Embedlite. r=dholbert 2014-05-28 10:44:00 -04:00
Ed Morley c3c9d1a417 Backed out changeset 1d4226ea1c1b (bug 854421) for causing bug 1017478 2014-05-29 10:57:03 +01:00
Ed Morley f645109fad Backed out changeset 0d658e89549b (bug 854421) 2014-05-29 10:56:08 +01:00
Ed Morley b65fcdc520 Backed out changeset 0ce26a65c8dd (bug 854421) 2014-05-29 10:56:04 +01:00
Ed Morley be1f5d2de8 Backed out changeset 3f58a7521a51 (bug 854421) 2014-05-29 10:55:51 +01:00
Ed Morley ced23842df Backed out changeset 123998cc3f92 (bug 854421) 2014-05-29 10:55:47 +01:00
Matt Woodrow c9c9b9cb80 Bug 854421 - Part 8: Try find a 'root' refresh driver to see if it's blocked. r=vlad 2014-05-29 09:43:41 +12:00
Matt Woodrow de268276b8 Bug 854421 - Part 7: Allocate and return transaction id's from the refresh driver when using OMTC to prevent over-production. r=roc 2014-05-29 09:43:39 +12:00
Matt Woodrow fbcf96b65f Bug 854421 - Part 3: Add a way to detect PLayerTransaction objects that exist only for testing and won't composite. r=nical 2014-05-29 09:42:16 +12:00
Matt Woodrow 24efc9cfe8 Bug 854421 - Part 2: If we cancel a composite when going into test mode, fire the DidComposite event anway in case someone is waiting on it. r=nical 2014-05-29 09:42:16 +12:00
Matt Woodrow 345fc54583 Bug 854421 - Part 1: Add a transaction id number to DidComposite. r=nical 2014-05-29 09:42:14 +12:00
Brian Birtles f3593b3f8a Bug 1015803 - Align Layers' Animation.numIterations with AnimationTiming.mIterationCount; r=dholbert
In the IPC Animation struct used in layers code we have a member called
'numIterations' where 'iterate forever' is represented by -1.

In layout/style however we have an AnimationTiming struct with an
mIterationCount member where 'iterate forever' is represented by
NS_IEEEPositiveInfinity().

This patch renames 'numIterations' to 'iterationCount' and uses infinity to
represent 'iterate forever'.
2014-05-28 16:51:50 +09:00
Matt Woodrow 0726340c00 Bug 1013767 - Restrict BasicCompositor snapshots to the bounds of the invalid region. r=nical 2014-05-28 13:21:32 +12:00
Sotaro Ikeda afdf8e01dd Bug 1006957 - Handle buffer ownership between SurfaceStream and CanvasClient r=nical 2014-05-24 09:14:45 -07:00
Steve Fink 12ba3a7a2a Backed out changeset 171edffe0641 (bug 1006957) for crash at BaseAutoLock<mozilla::Mutex>::BaseAutoLock(mozilla::Mutex&) from layers::AsyncTransactionTrackersHolder::ClearAllAsyncTransactionTrackers on a CLOSED TREE 2014-05-23 21:16:02 -07:00
Sotaro Ikeda ac5d74bdb6 Bug 1006957 - Handle buffer ownership between SurfaceStream and CanvasClient r=nical 2014-05-23 17:52:43 -07:00
Nicolas Silva 7c865568d2 Bug 924622 - Make sure gfx's ipc shutdown happens before shutting down xpcom threads. r=bsmedberg, sotaro 2014-05-07 11:06:19 +02:00
Nicolas Silva 69c73e6b6b Bug 1013292 - Fix some using namespace + unified build issues in gfx code. r=kats 2014-05-22 12:11:45 +02:00
Benoit Girard df3b7c99d5 Bug 1014042 - Add compositor benchmark. r=jrmuizel 2014-05-21 17:29:49 +00:00
Sotaro Ikeda ceac91078e Bug 1011327 - Add checks around SharedBufferManagerParent 2014-05-20 13:50:25 -07:00
Wes Kocher 738f9f9b41 Merge inbound to m-c 2014-05-15 15:59:45 -07:00