OSDN Git Service

android-x86/frameworks-native.git
11 years agoam 1cd763f1: am cd91024c: Merge "[MIPS] Avoid unaligned load/store for 64-bit doubles."
Keun-young Park [Tue, 26 Mar 2013 18:45:01 +0000 (11:45 -0700)]
am 1cd763f1: am cd91024c: Merge "[MIPS] Avoid unaligned load/store for 64-bit doubles."

* commit '1cd763f123b9f1a2772ddff15ace81ec07b7660c':
  [MIPS] Avoid unaligned load/store for 64-bit doubles.

11 years agoam cd91024c: Merge "[MIPS] Avoid unaligned load/store for 64-bit doubles."
Keun-young Park [Tue, 26 Mar 2013 18:43:05 +0000 (11:43 -0700)]
am cd91024c: Merge "[MIPS] Avoid unaligned load/store for 64-bit doubles."

* commit 'cd91024ca16db96daa697e22f7a088b416a424e3':
  [MIPS] Avoid unaligned load/store for 64-bit doubles.

11 years agoMerge "[MIPS] Avoid unaligned load/store for 64-bit doubles."
Keun-young Park [Tue, 26 Mar 2013 17:47:19 +0000 (17:47 +0000)]
Merge "[MIPS] Avoid unaligned load/store for 64-bit doubles."

11 years agoMerge "Add seinfo parsing to PackageManagerService." into jb-mr2-dev
Geremy Condra [Tue, 26 Mar 2013 01:39:21 +0000 (01:39 +0000)]
Merge "Add seinfo parsing to PackageManagerService." into jb-mr2-dev

11 years agoAdd seinfo parsing to PackageManagerService.
Robert Craig [Mon, 25 Mar 2013 10:33:03 +0000 (06:33 -0400)]
Add seinfo parsing to PackageManagerService.

This patch set allows the PMS to parse the
mac_permissions.xml file which contains the
seinfo values. Each package that is installed
on the device will be assigned an seinfo value
based on policy. This seinfo value will help label
the app process and data directory.  Modifications
include adjustments to ApplicationInfo.java
to store the seinfo tag per package as well as
adjustments to installd to communicate the seinfo
tag to libselinux.

Change-Id: I61ad1ea12fb6a9a6d0b108ec163bc4bf4c954b58
Signed-off-by: rpcraig <rpcraig@tycho.ncsc.mil>
11 years agoMerge "Bugreport broadcasts are only useful to owners." into jb-mr2-dev
Jeff Sharkey [Tue, 26 Mar 2013 00:18:36 +0000 (00:18 +0000)]
Merge "Bugreport broadcasts are only useful to owners." into jb-mr2-dev

11 years agoBugreport broadcasts are only useful to owners.
Jeff Sharkey [Tue, 26 Mar 2013 00:10:45 +0000 (17:10 -0700)]
Bugreport broadcasts are only useful to owners.

Bug: 8415156
Change-Id: Ia73969262abe9f715f038dad934651d7e5d65c87

11 years agoMerge "Fix virtual displays for HWC<=1.1" into jb-mr2-dev
Jesse Hall [Mon, 25 Mar 2013 22:37:03 +0000 (22:37 +0000)]
Merge "Fix virtual displays for HWC<=1.1" into jb-mr2-dev

11 years agoMerge changes I98927540,I95c71e8d into jb-mr2-dev
Jesse Hall [Mon, 25 Mar 2013 22:36:38 +0000 (22:36 +0000)]
Merge changes I98927540,I95c71e8d into jb-mr2-dev

* changes:
  Fix dump when virtual display exists
  Isolate knowledge that fb target == output buffer

11 years agoImplement #8323587, #8323342, #8323590: new features.
Dianne Hackborn [Sat, 23 Mar 2013 00:24:57 +0000 (17:24 -0700)]
Implement #8323587, #8323342, #8323590: new features.

8323587: Add feature for supporting app widgets
8323342: Add feature for replacing the home screen
8323590: Add feature for supporting input methods

The app widget service looks for the app widget feature
and refuses to work if it doesn't exist.  I didn't do
this for the input method service because some devices
will probably want to still make use of that mechanism
without supporting third party input methods.

Change-Id: Ia358ccc6059d43f163d74810c7cbe9568a67ede9

11 years agoFix virtual displays for HWC<=1.1
Jesse Hall [Fri, 22 Mar 2013 22:13:48 +0000 (15:13 -0700)]
Fix virtual displays for HWC<=1.1

If we're using a HWC that doesn't support virtual displays, or we have
more virtual displays than HWC supports concurrently, the
VirtualDisplaySurface should simply be a passthrough from source
(GLES) to sink.

This change also tries to distinguish between display types and HWC
display IDs a little better, though there's more to do here. Probably
needs a higher-level rethink; it's too error-prone now.

Bug: 8446838
Change-Id: I708d2cf262ec30177042304f174ca5b8da701df1

11 years agoFix dump when virtual display exists
Jesse Hall [Thu, 21 Mar 2013 21:06:52 +0000 (14:06 -0700)]
Fix dump when virtual display exists

SurfaceFlinger::getLayerSortedByZForHwcDisplay only worked for builtin
displays.

Bug: 8384764
Change-Id: I989275407fb2f06d166a6e70321c3211e27e562e

11 years agoIsolate knowledge that fb target == output buffer
Jesse Hall [Wed, 20 Mar 2013 20:44:00 +0000 (13:44 -0700)]
Isolate knowledge that fb target == output buffer

HWComposer didn't allow the virtual display output buffer to be set
directly, instead it always used the framebuffer target buffer.
DisplayDevice was only providing the framebuffer release fence to
DisplaySurfaces after a commit.

This change fixes both of these, so both HWComposer and DisplayDevice
should continue to work if VirtualDisplaySurface changes to use
separate framebuffer and output buffers. It's also more correct since
VirtualDisplaySurface uses the correct release fence when queueing the
buffer to the sink.

Bug: 8384764
Change-Id: I95c71e8d4f67705e23f122259ec8dd5dbce70dcf

11 years agoUse descriptive aliases instead of HWC_NUM_DISPLAY_TYPES
Jesse Hall [Wed, 20 Mar 2013 20:42:29 +0000 (13:42 -0700)]
Use descriptive aliases instead of HWC_NUM_DISPLAY_TYPES

Bug: 8384764
Change-Id: Ia2563fab19dbc6e4a95bed03445e609334841cca

11 years agoPass sp<Fence>s around instead of file descriptors
Jesse Hall [Wed, 20 Mar 2013 18:37:21 +0000 (11:37 -0700)]
Pass sp<Fence>s around instead of file descriptors

Change-Id: Iac70584a2207940730e8f803a543e4e9a4000c47

11 years agoMerge "Release virtual display buffer immediately after HWC set" into jb-mr2-dev
Jesse Hall [Fri, 22 Mar 2013 16:51:31 +0000 (16:51 +0000)]
Merge "Release virtual display buffer immediately after HWC set" into jb-mr2-dev

11 years agomake sure screenshot are in a format supported by Bitmap.java
Mathias Agopian [Fri, 22 Mar 2013 07:24:39 +0000 (00:24 -0700)]
make sure screenshot are in a format supported by Bitmap.java

Change-Id: I0fb9cc4088f9c1fd27e6c017b0a7c5617adb4660

11 years agoMerge "improved CallStack a bit" into jb-mr2-dev
Mathias Agopian [Fri, 22 Mar 2013 04:15:33 +0000 (04:15 +0000)]
Merge "improved CallStack a bit" into jb-mr2-dev

11 years agofix a typo that broke all screenshots
Mathias Agopian [Fri, 22 Mar 2013 04:13:21 +0000 (21:13 -0700)]
fix a typo that broke all screenshots

Bug: 8450197

Change-Id: I5b986cc0ff9c5e689d06a51ba68ab537d03d1f3d

11 years agoMerge "only use glReadPixels() when needed when taking screenshots" into jb-mr2-dev
Mathias Agopian [Fri, 22 Mar 2013 00:16:00 +0000 (00:16 +0000)]
Merge "only use glReadPixels() when needed when taking screenshots" into jb-mr2-dev

11 years agoimproved CallStack a bit
Mathias Agopian [Fri, 22 Mar 2013 00:12:40 +0000 (17:12 -0700)]
improved CallStack a bit

- added a ctor that updates and dumps the stack immediately
- added a "logtag" parameter to dump()

Change-Id: Ie51c256071d282591752243bdb4f68cf9ff8829d

11 years agoMerge "initialize planeAlpha (to 0xFF) for the FB_TARGET" into jb-mr2-dev
Mathias Agopian [Thu, 21 Mar 2013 23:58:38 +0000 (23:58 +0000)]
Merge "initialize planeAlpha (to 0xFF) for the FB_TARGET" into jb-mr2-dev

11 years agoinitialize planeAlpha (to 0xFF) for the FB_TARGET
Mathias Agopian [Thu, 21 Mar 2013 23:25:12 +0000 (16:25 -0700)]
initialize planeAlpha (to 0xFF) for the FB_TARGET

Bug: 8422625
Change-Id: I7b72456c2e01ed92e675cc1bfa0cfd0d2ce5064e

11 years agoatrace: use creat instead of truncate
Jamie Gennis [Thu, 21 Mar 2013 21:06:31 +0000 (14:06 -0700)]
atrace: use creat instead of truncate

Change-Id: Ie25c704f33c419c0c542249ae2841393862df0ab

11 years agoonly use glReadPixels() when needed when taking screenshots
Mathias Agopian [Thu, 21 Mar 2013 01:36:41 +0000 (18:36 -0700)]
only use glReadPixels() when needed when taking screenshots

some drivers don't support this yet, so we use a system
property to enable the glReadPixels "workaround" for them:

  ro.bq.gpu_to_cpu_unsupported=1

Change-Id: I74d6a3a8f0cee8d5a507b72c760cf247e39195e0

11 years agoMerge "CpuConsumer: Properly track acquired buffers" into jb-mr2-dev
Eino-Ville Talvala [Thu, 21 Mar 2013 16:39:13 +0000 (16:39 +0000)]
Merge "CpuConsumer: Properly track acquired buffers" into jb-mr2-dev

11 years agoMerge "Revert "Revert "Change SurfaceControl setPosition to take floats""" into jb...
Ramanan Rajeswaran [Thu, 21 Mar 2013 15:52:48 +0000 (15:52 +0000)]
Merge "Revert "Revert "Change SurfaceControl setPosition to take floats""" into jb-mr2-dev

11 years agoRevert "Revert "Change SurfaceControl setPosition to take floats""
Ramanan Rajeswaran [Thu, 21 Mar 2013 15:49:59 +0000 (15:49 +0000)]
Revert "Revert "Change SurfaceControl setPosition to take floats""

This reverts commit bbb57f3331c7182399ed82e9c4f93a965677dde3

Change-Id: Ib8f039f530d6b8175b5205c6c0008b135d4001a2

11 years agofix build. update test to new API
Mathias Agopian [Wed, 20 Mar 2013 22:56:04 +0000 (15:56 -0700)]
fix build. update test to new API

Change-Id: I31ca84c85edff3e2ff90c554769e058110f27d49

11 years agoMerge "rework how we take screenshots for a CPU consumer" into jb-mr2-dev
Mathias Agopian [Wed, 20 Mar 2013 22:14:55 +0000 (22:14 +0000)]
Merge "rework how we take screenshots for a CPU consumer" into jb-mr2-dev

11 years agoMerge "Move screenshots near beginning of dumpstate." into jb-mr2-dev
Jeff Sharkey [Wed, 20 Mar 2013 19:09:44 +0000 (19:09 +0000)]
Merge "Move screenshots near beginning of dumpstate." into jb-mr2-dev

11 years agoRelease virtual display buffer immediately after HWC set
Jesse Hall [Wed, 20 Mar 2013 00:18:09 +0000 (17:18 -0700)]
Release virtual display buffer immediately after HWC set

Previously we only queued a virtual display buffer to the sink when
the next frame was about to be displayed. This may delay the "last"
frame of an animation indefinitely. Now we queue the buffer as soon as
HWC set() returns and gives us the release fence.

Bug: 8384764
Change-Id: I3844a188e0f6ef6ff28f3e11477cfa063a924b1a

11 years agoMerge "Updated MediaDrm API based on review input." into jb-mr2-dev
Jeff Tinker [Wed, 20 Mar 2013 16:47:34 +0000 (16:47 +0000)]
Merge "Updated MediaDrm API based on review input." into jb-mr2-dev

11 years agorework how we take screenshots for a CPU consumer
Mathias Agopian [Wed, 20 Mar 2013 05:22:21 +0000 (22:22 -0700)]
rework how we take screenshots for a CPU consumer

We're not using IMemoryHeap as a transport anymore,
instead we're providing a CpuConsumer and use the
IGraphicBufferProducer version of the screenshot API.

However, some GPU drivers don't support properly
a GPU to CPU path, to work around this, we use a
temporary BufferQueue on the server side for the
GL rendering, and we use glReadPixels into the
CpuConsumer (we're now using a CPU to CPU path
which is always supported).

Currently this "wrapping" is always performed,
but it can be bypassed on devices that support
the GPU to CPU path.

This also addresses a DoS attack vector on
SurfaceFlinger, where an application could
consume all of SF's filedescriptors by creating
a lot of screenshots in a row.

Bug: 8390553

Change-Id: I9e81514c2a7711b9bb393f74305be7d2abe08f1c

11 years agodisable RefBase consistency checks (NDEBUG)
Mathias Agopian [Wed, 20 Mar 2013 00:36:57 +0000 (17:36 -0700)]
disable RefBase consistency checks (NDEBUG)

Bug: 8328715
Change-Id: Ib57646ff909fd8744610f37f3b50d90d884dff31

11 years agoFix a crasher with RefBase debugging and vectors of wp<>
Mathias Agopian [Tue, 19 Mar 2013 05:27:41 +0000 (22:27 -0700)]
Fix a crasher with RefBase debugging and vectors of wp<>

background:
we have some code to fix-up the IDs of references when
using RefBase's DEBUG_REFS when those refs are managed by
arrays wp<> or sp<> (this is because wp<> / sp<> don't have
a trivial ctor when DEBUG_REFS is enabled, and Vector
treats them as trivial for obvious performance reasons)

this is complicated by the fact that we don't want to have
to recompile everything when enabling DEBUG_REFs (i.e.: the
Vector code cannot know wheter it's enabled or not for its
template stuff).

problem:
there was a bug in the fix-up code for wp<> which was trying
to access the weakref_impl from the RefBase* however, this was
moronic since RefBase could have been destroyed if there wasn't
any more strong refs -- and this happned. Instead we need to get
the weakref_impl directly from the wp<>

Change-Id: Ie16e334204205fdbff142acb9faff8479a78450b

11 years agoMerge "A few tweaks to RefBase debugging" into jb-mr2-dev
Mathias Agopian [Tue, 19 Mar 2013 23:00:40 +0000 (23:00 +0000)]
Merge "A few tweaks to RefBase debugging" into jb-mr2-dev

11 years agoMove screenshots near beginning of dumpstate.
Jeff Sharkey [Tue, 19 Mar 2013 22:05:19 +0000 (15:05 -0700)]
Move screenshots near beginning of dumpstate.

Bug: 8420215
Change-Id: I6adffd1182ed8f99b8d11d43785a4eb4a388ceb5

11 years agoDon't log an error on HWC1.1+ devices with no FB HAL
Jesse Hall [Mon, 18 Mar 2013 18:28:50 +0000 (11:28 -0700)]
Don't log an error on HWC1.1+ devices with no FB HAL

Change-Id: I015e5a1a1f926181e51f82525f69beff71bd70cd

11 years agoMerge changes I61ae54f3,I57cb668e,I7a3f1e1a,Id28a2f9b into jb-mr2-dev
Jesse Hall [Tue, 19 Mar 2013 17:45:29 +0000 (17:45 +0000)]
Merge changes I61ae54f3,I57cb668e,I7a3f1e1a,Id28a2f9b into jb-mr2-dev

* changes:
  Add BufferQueueInterposer and use it for virtual displays
  Add DisplaySurface abstraction
  Fix argument types in IGraphicBufferProducer methods
  Minor cleanups/fixes before virtual display refactoring

11 years agoA few tweaks to RefBase debugging
Mathias Agopian [Tue, 19 Mar 2013 03:31:18 +0000 (20:31 -0700)]
A few tweaks to RefBase debugging

- stacks are now saved in /data/debug which must be
  created and writable by the user.

- removed "always fatal" DEBUG_REFS option, it wasn't
  really needed.

- DEBUG_REFS_ENABLED_BY_DEFAULT is not the default anymore
  (usually people want to target which refs they're tracking)

Change-Id: I37fae72e9dacde6ce1fa8f7dbe2bc01b1a1b95e5

11 years agoMerge "atrace: add support for tracing kernel functions" into jb-mr2-dev
Jamie Gennis [Mon, 18 Mar 2013 21:54:44 +0000 (21:54 +0000)]
Merge "atrace: add support for tracing kernel functions" into jb-mr2-dev

11 years agoAdd BufferQueueInterposer and use it for virtual displays
Jesse Hall [Fri, 15 Mar 2013 19:32:10 +0000 (12:32 -0700)]
Add BufferQueueInterposer and use it for virtual displays

BufferQueueInterposer allows a client to tap into a
IGraphicBufferProducer-based buffer queue, and modify buffers as they
pass from producer to consumer. VirtualDisplaySurface uses this to
layer HWC composition on top of GLES composition before passing the
buffer to the virtual display consumer.

Bug: 8384764
Change-Id: I61ae54f3d90de6a35f4f02bb5e64e7cc88e1cb83

11 years agoAdd DisplaySurface abstraction
Jesse Hall [Thu, 14 Mar 2013 21:29:29 +0000 (14:29 -0700)]
Add DisplaySurface abstraction

DisplayDevice now has a DisplaySurface instead of using
FramebufferSurface directly. FramebufferSurface implements
DisplaySurface, and so does the new VirtualDisplaySurface class.
DisplayDevice now always has a surface, not just for virtual displays.

In this change VirtualDisplaySurface is just a stub; buffers still go
directly from GLES to the final consumer.

Bug: 8384764
Change-Id: I57cb668edbc6c37bfebda90b9222d435bf589f37

11 years agoFix argument types in IGraphicBufferProducer methods
Jesse Hall [Sat, 16 Mar 2013 04:34:30 +0000 (21:34 -0700)]
Fix argument types in IGraphicBufferProducer methods

Bug: 8384764
Change-Id: I7a3f1e1a0584a70af04f9eafef900505389d2202

11 years agoMinor cleanups/fixes before virtual display refactoring
Jesse Hall [Thu, 7 Mar 2013 00:13:49 +0000 (16:13 -0800)]
Minor cleanups/fixes before virtual display refactoring

None of these should change behavior, except for removing some
incorrect log messages when using a virtual display.

- HWComposer::getAndResetReleaseFenceFd() checks the HWC version, so
  no need to do that in the DisplayDevice::onSwapBuffersCompleted().
  However, it should check that mFramebufferSurface is not NULL like
  it is for virtual displays.
- Comment that FramebufferSurface::dump() overrides the non-virtual
  ConsumerBase::dump(), and fix it so the right thing happens
  regardless of the static type of the pointer/reference the callee
  has. FramebufferSurface::dump() could be removed right now, but I'd
  need to bring it back in a later change.
- Use the right enum for validating display type ids.
- Don't try to send hotplug events for virtual displays.
- Mark virtual displays as connected so HWComposer::prepare() doesn't
  think something is wrong when it gets a non-NULL layer list.
- Remove unused FramebufferSurface methods.

Bug: 8384764
Change-Id: Id28a2f9be86b45f4bb7915fdf7752157035f4294

11 years agoatrace: add support for tracing kernel functions
Jamie Gennis [Tue, 12 Mar 2013 23:00:10 +0000 (16:00 -0700)]
atrace: add support for tracing kernel functions

This change adds support for tracing specific kernel function calls using the
function_graph tracer.  It adds a '-k' option to atrace that accepts a comma
separated list of kernel function names for which tracing will be enabled.

Change-Id: I872b2f1d474b8ebb904053853fc8cf8c0a98089c

11 years agoMerge "Add Vector::resize()" into jb-mr2-dev
Jesse Hall [Sat, 16 Mar 2013 03:23:12 +0000 (03:23 +0000)]
Merge "Add Vector::resize()" into jb-mr2-dev

11 years agodon't allow screenshots without the READ_FRAMEBUFFER permission
Mathias Agopian [Sat, 16 Mar 2013 01:31:56 +0000 (18:31 -0700)]
don't allow screenshots without the READ_FRAMEBUFFER permission

the recent screenshot rework allowed the older screenshot
interface to work without that permission

Change-Id: I6c4743f4591c81106e3b823d55a055f7b4907de1

11 years agoMerge "get rid of purgatory and fix QueuesToWindowComposer query" into jb-mr2-dev
Mathias Agopian [Sat, 16 Mar 2013 01:14:34 +0000 (01:14 +0000)]
Merge "get rid of purgatory and fix QueuesToWindowComposer query" into jb-mr2-dev

11 years agodon't capture hidden layers in screenshots
Mathias Agopian [Fri, 15 Mar 2013 23:08:10 +0000 (16:08 -0700)]
don't capture hidden layers in screenshots

Bug: 8389956
Change-Id: I9ed836395258732c743c6fd44092bd01020dde13

11 years agoAdd Vector::resize()
Jesse Hall [Mon, 11 Mar 2013 17:16:48 +0000 (10:16 -0700)]
Add Vector::resize()

Bug: 8384764
Change-Id: Icee83d389f3e555eba7d419b64c8d52a9aa21b8b

11 years agoget rid of purgatory and fix QueuesToWindowComposer query
Mathias Agopian [Fri, 15 Mar 2013 02:18:13 +0000 (19:18 -0700)]
get rid of purgatory and fix QueuesToWindowComposer query

the purgatory list wasn't needed anymore; in fact it had no effect as
buffer life-time management is now handled by the BufferQueue.

For QueuesToWindowComposer we keep a list of wp<> on the IBinder
for IGraphicBufferProducers we hand over to clients so we can
easily check if an IGraphicBufferProducer is ours. We clean-up the
list when our IGraphicBufferProducer are destroyed.

Bug: 8349142
Change-Id: I1aa06652ade8c72d0004a3f5e6c3d6e8a82fc2ae

11 years agofix a couple race-conditions in RefBase::promote()
Dianne Hackborn [Thu, 14 Mar 2013 22:26:30 +0000 (15:26 -0700)]
fix a couple race-conditions in RefBase::promote()

Bug: 8390295
Change-Id: I7a48e3bf5b213cc1da2b8e844c6bb37ee24cb047

11 years agoam 7212ff29: am 20e154f1: Merge "Second try at adding a compatibility symbol for...
Elliott Hughes [Thu, 14 Mar 2013 17:31:38 +0000 (10:31 -0700)]
am 7212ff29: am 20e154f1: Merge "Second try at adding a compatibility symbol for the MemoryBase constructor."

* commit '7212ff29c6f4e4cd192fee6f072e80b36d8a728b':
  Second try at adding a compatibility symbol for the MemoryBase constructor.

11 years agoMerge "Increase screenshot timeout." into jb-mr2-dev
Jeff Sharkey [Thu, 14 Mar 2013 17:21:08 +0000 (17:21 +0000)]
Merge "Increase screenshot timeout." into jb-mr2-dev

11 years agoIncrease screenshot timeout.
Jeff Sharkey [Thu, 14 Mar 2013 17:20:31 +0000 (10:20 -0700)]
Increase screenshot timeout.

Devices with large displays take longer than 5 seconds to write.

Change-Id: If74bc4a3cec366ae28eca549bce28ef212e47d8f

11 years agoam 6250f6cb: am d8151306: Merge "Revert "Add a compatibility symbol for the MemoryBas...
Elliott Hughes [Thu, 14 Mar 2013 15:55:49 +0000 (08:55 -0700)]
am 6250f6cb: am d8151306: Merge "Revert "Add a compatibility symbol for the MemoryBase constructor.""

* commit '6250f6cb2be039be150de6c836f8595384838342':
  Revert "Add a compatibility symbol for the MemoryBase constructor."

11 years agoMerge "Support broadcast when dumpstate finishes." into jb-mr2-dev
Jeff Sharkey [Thu, 14 Mar 2013 15:44:15 +0000 (15:44 +0000)]
Merge "Support broadcast when dumpstate finishes." into jb-mr2-dev

11 years agoRevert "Add a compatibility symbol for the MemoryBase constructor."
Elliott Hughes [Thu, 14 Mar 2013 02:56:34 +0000 (02:56 +0000)]
Revert "Add a compatibility symbol for the MemoryBase constructor."

Cherry pick from AOSP to fix the build.
See https://android-review.googlesource.com/#/c/53690/

This reverts commit 7f974ff2f317ffbf9256579109f000ec876c3037

Change-Id: I1b123f446d0ff41e11ff81d4ef5f1472b59cac93

11 years agoam 9b57ad73: am 8facd1b2: Merge "Add a compatibility symbol for the MemoryBase constr...
Elliott Hughes [Thu, 14 Mar 2013 06:11:00 +0000 (23:11 -0700)]
am 9b57ad73: am 8facd1b2: Merge "Add a compatibility symbol for the MemoryBase constructor."

* commit '9b57ad737bceba9f8d4f5be303c70fa72be00eaf':
  Add a compatibility symbol for the MemoryBase constructor.

11 years agoam 20e154f1: Merge "Second try at adding a compatibility symbol for the MemoryBase...
Elliott Hughes [Thu, 14 Mar 2013 05:40:26 +0000 (22:40 -0700)]
am 20e154f1: Merge "Second try at adding a compatibility symbol for the MemoryBase constructor."

* commit '20e154f16f315d7ae0b3204db0004d19a8b0bc48':
  Second try at adding a compatibility symbol for the MemoryBase constructor.

11 years agoMerge "Second try at adding a compatibility symbol for the MemoryBase constructor."
Elliott Hughes [Thu, 14 Mar 2013 04:35:13 +0000 (04:35 +0000)]
Merge "Second try at adding a compatibility symbol for the MemoryBase constructor."

11 years agoSecond try at adding a compatibility symbol for the MemoryBase constructor.
Elliott Hughes [Thu, 14 Mar 2013 03:54:01 +0000 (20:54 -0700)]
Second try at adding a compatibility symbol for the MemoryBase constructor.

Third-party libraries are currently trying to use the
MemoryBase constructor but failing because we fixed the
definition of ssize_t. This is a stop-gap for users of
this private API until we can get them fixed.

Bug: 8253769
Change-Id: Ie7c86f45fa39cb53539cab0ffe8585a585656714

11 years agoam d8151306: Merge "Revert "Add a compatibility symbol for the MemoryBase constructor.""
Elliott Hughes [Thu, 14 Mar 2013 03:32:33 +0000 (20:32 -0700)]
am d8151306: Merge "Revert "Add a compatibility symbol for the MemoryBase constructor.""

* commit 'd8151306f539f8c2f3b66c1fe06be9566ba8aca3':
  Revert "Add a compatibility symbol for the MemoryBase constructor."

11 years agoMerge "Revert "Add a compatibility symbol for the MemoryBase constructor.""
Elliott Hughes [Thu, 14 Mar 2013 02:56:48 +0000 (02:56 +0000)]
Merge "Revert "Add a compatibility symbol for the MemoryBase constructor.""

11 years agoRevert "Add a compatibility symbol for the MemoryBase constructor."
Elliott Hughes [Thu, 14 Mar 2013 02:56:34 +0000 (02:56 +0000)]
Revert "Add a compatibility symbol for the MemoryBase constructor."

This reverts commit 7f974ff2f317ffbf9256579109f000ec876c3037

Change-Id: I28ebfc5ac88bd025471b613aa7958d23749e6991

11 years agoam 8facd1b2: Merge "Add a compatibility symbol for the MemoryBase constructor."
Elliott Hughes [Thu, 14 Mar 2013 02:28:17 +0000 (19:28 -0700)]
am 8facd1b2: Merge "Add a compatibility symbol for the MemoryBase constructor."

* commit '8facd1b2ea02796b575f123b420590d2b8354910':
  Add a compatibility symbol for the MemoryBase constructor.

11 years agoMerge "Add a compatibility symbol for the MemoryBase constructor."
Elliott Hughes [Thu, 14 Mar 2013 01:19:44 +0000 (01:19 +0000)]
Merge "Add a compatibility symbol for the MemoryBase constructor."

11 years agoAdd a compatibility symbol for the MemoryBase constructor.
Elliott Hughes [Wed, 13 Mar 2013 21:35:55 +0000 (14:35 -0700)]
Add a compatibility symbol for the MemoryBase constructor.

Third-party libraries are currently trying to use the
MemoryBase constructor but failing because we fixed the
definition of ssize_t. This is a stop-gap for users of
this private API until we can get them fixed.

Bug: 8253769
Change-Id: I8a19770f3252d88ee87023fde625cc6289924b0d

11 years agoSupport broadcast when dumpstate finishes.
Jeff Sharkey [Wed, 13 Mar 2013 22:45:50 +0000 (15:45 -0700)]
Support broadcast when dumpstate finishes.

Add flag that sends BUGREPORT_FINISHED broadcast when dumpstate
is finished, including the resulting filesystem paths.  Also reduces
directory creation to 0770.

Bug: 7005318
Change-Id: Id4c6b699a56f8acd859b7ab73368500e1a8f3c67

11 years agosize IMemoryHeap properly for screenshots
Mathias Agopian [Wed, 13 Mar 2013 22:22:11 +0000 (15:22 -0700)]
size IMemoryHeap properly for screenshots

since we're using glReadPixels(), we only need to use
the width (as opposed to the stride) of the source
screenshot.

Bug: 8374664
Change-Id: I145c80f4fff5444df7c77c4f52e70a7203caddbd

11 years agoUpdated MediaDrm API based on review input.
Jeff Tinker [Fri, 8 Feb 2013 01:46:18 +0000 (17:46 -0800)]
Updated MediaDrm API based on review input.

Change-Id: I06716d2616bbe7efaea6cbf9ba7bf303dfae1650

11 years agoSurface can now be created only from an IGraphicBufferProducer
Mathias Agopian [Wed, 13 Mar 2013 01:45:09 +0000 (18:45 -0700)]
Surface can now be created only from an IGraphicBufferProducer

it can't write itself to a parcel, nor can it be created from a
parcel.

Change-Id: I69165d5c54d6024b3e546e80d8b57e3dedda7893

11 years agoget rid of ISurface
Mathias Agopian [Wed, 13 Mar 2013 00:11:48 +0000 (17:11 -0700)]
get rid of ISurface

ISurface was only used to get the IGraphicBufferProducer from
a Layer. It's now replaced by a BBinder subclass / IBinder and
is only used as a handle to the surface, to both refer to it
and manage its life-time.

Also cleaned-up a bit the ISurfaceComposer interface and
"create layer" code path.

Change-Id: I68d0e02d57b862cffb31d5168c3bc10cea0906eb

11 years agoMerge "atrace: add the webview category" into jb-mr2-dev
Jamie Gennis [Tue, 12 Mar 2013 19:21:29 +0000 (19:21 +0000)]
Merge "atrace: add the webview category" into jb-mr2-dev

11 years agotemporary: enable ASSERTs in RefBase
Mathias Agopian [Tue, 12 Mar 2013 04:27:16 +0000 (21:27 -0700)]
temporary: enable ASSERTs in RefBase

this is in an attempt to get more data on
bug 8328715.

Change-Id: I9333a67c2d7f67f4d9b2fc5eb1ad8a7b2d1c6dcb

11 years agoFix missing recents screenshots
Mathias Agopian [Tue, 12 Mar 2013 03:47:24 +0000 (20:47 -0700)]
Fix missing recents screenshots

We were using the "visible layer list" when taking screenshots,
which doesn't work when a layer is behind other opaque layers
and therefore hidden.

We fix this by using the full layer list, filtered by the
layerstack of the display we're looking at.

Bug: 7552304
Change-Id: I4b6f77e5511aea94f8d218975b6e22738e7e5d5b

11 years agoatrace: add the webview category
Jamie Gennis [Tue, 12 Mar 2013 02:37:53 +0000 (19:37 -0700)]
atrace: add the webview category

Change-Id: Iefb552cd275bb5784903e8bf328890ecd233e45b

11 years agoMerge "Add TOUCH_NAVIGATION input source" into jb-mr2-dev
Michael Wright [Tue, 12 Mar 2013 01:15:48 +0000 (01:15 +0000)]
Merge "Add TOUCH_NAVIGATION input source" into jb-mr2-dev

11 years agoworkaround a deadlock when taking screenshots into a surface
Mathias Agopian [Tue, 12 Mar 2013 00:59:59 +0000 (17:59 -0700)]
workaround a deadlock when taking screenshots into a surface

When disconnecting from BufferQueue, we now drain the queue
except the head (which means in the screenshot case we won't
have to block, but we might not have a buffer to show, this
will appear as an error in the log).

Bug: 8362363
Change-Id: If80989aac3c917beea2ebddf3cbb502849d394da

11 years agoMerge "Add Thread::isRunning and Condition::signal(WakeUpType)" into jb-mr2-dev
Romain Guy [Mon, 11 Mar 2013 21:38:26 +0000 (21:38 +0000)]
Merge "Add Thread::isRunning and Condition::signal(WakeUpType)" into jb-mr2-dev

11 years agoAdd Thread::isRunning and Condition::signal(WakeUpType)
Romain Guy [Mon, 11 Mar 2013 21:34:56 +0000 (14:34 -0700)]
Add Thread::isRunning and Condition::signal(WakeUpType)

The signal() method is useful to choose whether to wake up one or
all threads.

Change-Id: I062ab6d3ddd306a9fb735549ea140e2a76eed75a

11 years agoAdd TOUCH_NAVIGATION input source
Michael Wright [Mon, 11 Mar 2013 21:20:14 +0000 (14:20 -0700)]
Add TOUCH_NAVIGATION input source

Change-Id: Id76e2e85246269ba7b16bb9635e41593b45cac1c

11 years agoam 1d6e52d1: am 50d36d58: am 75bcd737: am 4ddcb758: (-s ours) am 9ef024ec: (-s ours...
Jean-Baptiste Queru [Mon, 11 Mar 2013 20:44:33 +0000 (13:44 -0700)]
am 1d6e52d1: am 50d36d58: am 75bcd737: am 4ddcb758: (-s ours) am 9ef024ec: (-s ours) Reconcile with jb-mr1-release - do not merge

* commit '1d6e52d17e2df913e10925c5e3e652d3421ff320':

11 years agoam 50d36d58: am 75bcd737: am 4ddcb758: (-s ours) am 9ef024ec: (-s ours) Reconcile...
Jean-Baptiste Queru [Mon, 11 Mar 2013 20:13:37 +0000 (13:13 -0700)]
am 50d36d58: am 75bcd737: am 4ddcb758: (-s ours) am 9ef024ec: (-s ours) Reconcile with jb-mr1-release - do not merge

* commit '50d36d58369cdf48bc6fcdba96b36c846be8587a':

11 years agoam 75bcd737: am 4ddcb758: (-s ours) am 9ef024ec: (-s ours) Reconcile with jb-mr1...
Jean-Baptiste Queru [Mon, 11 Mar 2013 20:00:44 +0000 (13:00 -0700)]
am 75bcd737: am 4ddcb758: (-s ours) am 9ef024ec: (-s ours) Reconcile with jb-mr1-release - do not merge

* commit '75bcd737770f85114673c4411b8e2ecc0dd32bdf':

11 years agoam 4ddcb758: (-s ours) am 9ef024ec: (-s ours) Reconcile with jb-mr1-release - do...
Jean-Baptiste Queru [Mon, 11 Mar 2013 19:41:50 +0000 (12:41 -0700)]
am 4ddcb758: (-s ours) am 9ef024ec: (-s ours) Reconcile with jb-mr1-release - do not merge

* commit '4ddcb75861cfb2835fcc20f2071222d2f55450ba':

11 years agoam ba8abb51: Merge "Ensure dumpstate children die"
jp abgrall [Mon, 11 Mar 2013 19:28:06 +0000 (12:28 -0700)]
am ba8abb51: Merge "Ensure dumpstate children die"

* commit 'ba8abb517d1eddebeda62bf907fa446e3c3b81f6':
  Ensure dumpstate children die

11 years agoam 9ef024ec: (-s ours) Reconcile with jb-mr1-release - do not merge
Jean-Baptiste Queru [Mon, 11 Mar 2013 19:23:09 +0000 (12:23 -0700)]
am 9ef024ec: (-s ours) Reconcile with jb-mr1-release - do not merge

* commit '9ef024ec8ae0318c89cc65dc0591aa40695393fc':

11 years agoReconcile with jb-mr1-release - do not merge
Jean-Baptiste Queru [Mon, 11 Mar 2013 19:19:30 +0000 (12:19 -0700)]
Reconcile with jb-mr1-release - do not merge

11 years agoMerge "Ensure dumpstate children die"
jp abgrall [Mon, 11 Mar 2013 18:09:35 +0000 (18:09 +0000)]
Merge "Ensure dumpstate children die"

11 years agoEnsure dumpstate children die
John Michelau [Thu, 7 Mar 2013 21:35:35 +0000 (15:35 -0600)]
Ensure dumpstate children die

Use prctl(PR_SET_PDEATHSIG, SIGKILL) in each forked child to ensure
it dies when dumpstate dies.

This is important for two cases:
- dumpstate runs a timer for each child process.  On expiration, it
  sends SIGTERM to kill the process.  Sometimes SIGTERM isn't enough
  to kill a hung process, so the child lives on.
- When dumpstate is killed by the user before completing, outstanding
  children continue to run and generate output.

Change-Id: I96e0dc918c26d56c9fee53611980ba2edd238712

11 years agomake sure to call compositionComplete after taking a screenshot
Mathias Agopian [Fri, 8 Mar 2013 21:54:46 +0000 (13:54 -0800)]
make sure to call compositionComplete after taking a screenshot

older drivers which are doing implicit synchronization need this
or they could deadlock.

Bug: 8341885
Change-Id: Icd980a6be16071678d6151e34725b3c1c547d7ee

11 years agoadd two libgui tests
Mathias Agopian [Fri, 8 Mar 2013 21:18:52 +0000 (13:18 -0800)]
add two libgui tests

we check that the order in which we destroy GLConsumer wrt.
releasing the corresponding EGLSurface via eglMake(Un)Current
doesn't leak a buffer.

On at least 2 devices this test doesn't pass.

Change-Id: I63ab83951b4b0a977f38571158f948cbd9dc7cec

11 years agoMerge changes Ibb3b6ff0,I2341e20c into jb-mr2-dev
Mathias Agopian [Fri, 8 Mar 2013 20:28:42 +0000 (20:28 +0000)]
Merge changes Ibb3b6ff0,I2341e20c into jb-mr2-dev

* changes:
  rename binder services main thread to Binder_*
  limit number of extra binder threads in SF to 4

11 years agoMerge "When disconnecting a surface, cancel don't queue the buffer" into jb-mr2-dev
Jesse Hall [Thu, 7 Mar 2013 23:43:56 +0000 (23:43 +0000)]
Merge "When disconnecting a surface, cancel don't queue the buffer" into jb-mr2-dev

11 years agoMerge "Defer destroying surfaces until not current" into jb-mr2-dev
Jesse Hall [Thu, 7 Mar 2013 23:41:45 +0000 (23:41 +0000)]
Merge "Defer destroying surfaces until not current" into jb-mr2-dev

11 years agorename binder services main thread to Binder_*
Mathias Agopian [Thu, 7 Mar 2013 23:34:28 +0000 (15:34 -0800)]
rename binder services main thread to Binder_*

When a binder service's main thread joins the thread pool
it retains its name (whatever the exec name was), which is
very confusing in systrace.

we now rename that thread just like its friends in the
thread pool.

Change-Id: Ibb3b6ff07304b247cfc6fb1694e72350c579513e

11 years agolimit number of extra binder threads in SF to 4
Mathias Agopian [Thu, 7 Mar 2013 23:32:47 +0000 (15:32 -0800)]
limit number of extra binder threads in SF to 4

it should be enough and it makes debugging easier.

Bug: 6609290
Change-Id: I2341e20cfcfde184fe1ed9b925215f8fa3481ced