OSDN Git Service

android-x86/frameworks-native.git
10 years agomerge in klp-release history after reset to klp-dev
The Android Automerger [Mon, 26 Aug 2013 13:01:12 +0000 (06:01 -0700)]
merge in klp-release history after reset to klp-dev

10 years agoMerge changes I34935d21,If423cd29 into klp-dev
Mathias Agopian [Sat, 24 Aug 2013 00:12:41 +0000 (00:12 +0000)]
Merge changes I34935d21,If423cd29 into klp-dev

* changes:
  handle several vsync signal correctly
  only use format compatible with CPU consumers

10 years agoMerge "make sure SF initialization is ran at target priority and correct thread group...
Mathias Agopian [Sat, 24 Aug 2013 00:12:05 +0000 (00:12 +0000)]
Merge "make sure SF initialization is ran at target priority and correct thread group" into klp-dev

10 years agomake sure SF initialization is ran at target priority and correct thread group
Mathias Agopian [Fri, 23 Aug 2013 22:56:38 +0000 (15:56 -0700)]
make sure SF initialization is ran at target priority and correct thread group

SF can spawn threads (indirectly) during initialization
and we want those to be spawned at URGENT_DISPLAY_PRIORITY
(in theory they should set their own, but some code lives
in vendor libraries and doesn't).

Bug: 10430209
Change-Id: I5b3a8f979297de287614c8eafd8267bef1176e4b

10 years agofix DimLayer in GLES2.0 SF
Mathias Agopian [Fri, 23 Aug 2013 23:12:34 +0000 (16:12 -0700)]
fix DimLayer in GLES2.0 SF

DimLayer state wasn't initialized properly.

Bug: 10442967
Change-Id: I384c1d85c58c96ae29d5daa1fabee9f4ce003501

10 years agoRemove input flinger stubs. (DO NOT MERGE)
Jeff Brown [Fri, 23 Aug 2013 01:34:33 +0000 (18:34 -0700)]
Remove input flinger stubs.  (DO NOT MERGE)

Bug: 10446930
Change-Id: I4dd46f539cd2a2e324dfc83b45ba2787f7404a4a

10 years agoMerge "Add ability to test supported content types to MediaDrm" into klp-dev
Jeff Tinker [Thu, 22 Aug 2013 23:08:09 +0000 (23:08 +0000)]
Merge "Add ability to test supported content types to MediaDrm" into klp-dev

10 years agoMerge "Surface: Fix controlledByApp parameter" into klp-dev
Eino-Ville Talvala [Thu, 22 Aug 2013 19:23:27 +0000 (19:23 +0000)]
Merge "Surface: Fix controlledByApp parameter" into klp-dev

10 years agomerge in klp-release history after reset to klp-dev
The Android Automerger [Thu, 22 Aug 2013 13:01:19 +0000 (06:01 -0700)]
merge in klp-release history after reset to klp-dev

10 years agohandle several vsync signal correctly
Mathias Agopian [Thu, 22 Aug 2013 00:45:46 +0000 (17:45 -0700)]
handle several vsync signal correctly

Change-Id: I34935d2197ce8e914fef2f110896e47b44225ad2

10 years agoonly use format compatible with CPU consumers
Mathias Agopian [Wed, 21 Aug 2013 23:36:34 +0000 (16:36 -0700)]
only use format compatible with CPU consumers

EGL now picks the buffer format out of a small set
of formats compatible with CPU consumers instead of
using the EGL_NATIVE_VISUAL_ID.

Bug: 10194508
Change-Id: If423cd29601b7a3ace8670f4c73004132cfc4b31

10 years agoAdd ability to test supported content types to MediaDrm
Jeff Tinker [Wed, 21 Aug 2013 18:57:40 +0000 (11:57 -0700)]
Add ability to test supported content types to MediaDrm

bug: 10244066
Change-Id: I6666d0dbfdc6e791deb9a82c918255af4fa19816

10 years agoscreenshot layers wouldn't work in some cases
Mathias Agopian [Wed, 21 Aug 2013 22:23:15 +0000 (15:23 -0700)]
screenshot layers wouldn't work in some cases

specifically when the display size and the screenshot window
size didn't match, the buffer would be rejected.
We simply fix this by setting the scalling mode to
"SCALE_TO_WINDOW".

Bug: 9992306
Change-Id: Ib821767899af330bb70d3cbbfa7d41b02794a075

10 years agoMerge "dumpstate: Fix wifi list of networks dump" into klp-dev
Dmitry Shmidt [Wed, 21 Aug 2013 19:35:25 +0000 (19:35 +0000)]
Merge "dumpstate: Fix wifi list of networks dump" into klp-dev

10 years agoSurface: Fix controlledByApp parameter
Eino-Ville Talvala [Wed, 21 Aug 2013 18:53:37 +0000 (11:53 -0700)]
Surface: Fix controlledByApp parameter

Bug: 10425641
Change-Id: I1a47e2ce4ecbdb8e67030a5d9ddf6604c5e2497d

10 years agoMerge "Fix a bug where non-cropped layer could be scaled incorrectly" into klp-dev
Mathias Agopian [Wed, 21 Aug 2013 18:55:19 +0000 (18:55 +0000)]
Merge "Fix a bug where non-cropped layer could be scaled incorrectly" into klp-dev

10 years agodumpstate: Fix wifi list of networks dump
Dmitry Shmidt [Wed, 21 Aug 2013 17:58:29 +0000 (10:58 -0700)]
dumpstate: Fix wifi list of networks dump

Bug: 9056601

Change-Id: Ia0adb67afdf3d478033d0ee34bbb95e621e2fe51
Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>
10 years agoMerge "Provide HWC prepare with a valid output buffer" into klp-dev
Jesse Hall [Wed, 21 Aug 2013 14:38:51 +0000 (14:38 +0000)]
Merge "Provide HWC prepare with a valid output buffer" into klp-dev

10 years agoMerge "Fix HWComposer dumping of float source crops" into klp-dev
Jesse Hall [Wed, 21 Aug 2013 14:37:50 +0000 (14:37 +0000)]
Merge "Fix HWComposer dumping of float source crops" into klp-dev

10 years agoMerge "Fix virtual displays for HWC 1.0" into klp-dev
Andy McFadden [Wed, 21 Aug 2013 14:31:34 +0000 (14:31 +0000)]
Merge "Fix virtual displays for HWC 1.0" into klp-dev

10 years agomerge in klp-release history after reset to klp-dev
The Android Automerger [Wed, 21 Aug 2013 13:02:26 +0000 (06:02 -0700)]
merge in klp-release history after reset to klp-dev

10 years agoFix a bug where non-cropped layer could be scaled incorrectly
Mathias Agopian [Wed, 21 Aug 2013 04:41:07 +0000 (21:41 -0700)]
Fix a bug where non-cropped layer could be scaled incorrectly

If a layer is not cropped but its bounds are outside of the
viewport (i.e.: clipped), the crop rectangle passed to
hw composer would be invalid because it started invalid
in the first place (to indicate "no crop").

Bug: 10410944
Change-Id: I4ae4d49a1adef0be7fa4304ecf84b1a5b7d03fe0

10 years agoProvide HWC prepare with a valid output buffer
Jesse Hall [Tue, 20 Aug 2013 23:35:32 +0000 (16:35 -0700)]
Provide HWC prepare with a valid output buffer

We weren't dequeing and setting the output buffer until just before
set(). This didn't allow HWC to make decisions in prepare() based on
the output buffer format, dimensions, etc.

Now we dequeue the output buffer at the beginning of the composition
loop and provide it to HWC in prepare. In GLES-only rendering, we may
have to cancel the buffer and acquire a new one if GLES requests a
buffer with properties different than the one we already dequeued.

Bug: 10365313
Change-Id: I96b4b0a851920e4334ef05080d58097d46467ab8

10 years agoFix HWComposer dumping of float source crops
Jesse Hall [Tue, 20 Aug 2013 23:11:50 +0000 (16:11 -0700)]
Fix HWComposer dumping of float source crops

Change-Id: I45a9344b5fab17ccb54bebd01382d738a03860e6

10 years agoFix virtual displays for HWC 1.0
Andy McFadden [Tue, 20 Aug 2013 23:04:47 +0000 (16:04 -0700)]
Fix virtual displays for HWC 1.0

We weren't swapping the buffers.

Bug 10366404

Change-Id: I43eb7648231bedf24899da6fc0d9acc09e0b047e

10 years agoSurfaceFlinger now runs in the process's main thread
Mathias Agopian [Tue, 20 Aug 2013 00:26:18 +0000 (17:26 -0700)]
SurfaceFlinger now runs in the process's main thread

it used to spawn its own thread and return the main thread
to the binder thread pool -- this was confusing the naming
of things in the kernel.

Bug: 10331839

Change-Id: I2d13a6d73409a38109300fcbe6a04b4c41cb5d00

10 years agomerge in klp-release history after reset to klp-dev
The Android Automerger [Tue, 20 Aug 2013 13:01:38 +0000 (06:01 -0700)]
merge in klp-release history after reset to klp-dev

10 years agoFix build
Andy McFadden [Mon, 19 Aug 2013 15:56:07 +0000 (08:56 -0700)]
Fix build

Neglected the "checkbuild"-built tests.

Change-Id: I202692f1a6d1e306c88e820c959d07d16e7c6a00

10 years agoMerge "Re-enable frame dropping for non-auto timestamps" into klp-dev
Andy McFadden [Mon, 19 Aug 2013 14:27:28 +0000 (14:27 +0000)]
Merge "Re-enable frame dropping for non-auto timestamps" into klp-dev

10 years agomerge in klp-release history after reset to klp-dev
The Android Automerger [Mon, 19 Aug 2013 13:01:39 +0000 (06:01 -0700)]
merge in klp-release history after reset to klp-dev

10 years agoMerge "Use new HWC display type/count constants." into klp-dev
Jesse Hall [Sun, 18 Aug 2013 04:31:24 +0000 (04:31 +0000)]
Merge "Use new HWC display type/count constants." into klp-dev

10 years agoMerge "Encapsulate textures into their own Texture class" into klp-dev
Mathias Agopian [Sat, 17 Aug 2013 00:40:19 +0000 (00:40 +0000)]
Merge "Encapsulate textures into their own Texture class" into klp-dev

10 years agoEncapsulate textures into their own Texture class
Mathias Agopian [Thu, 15 Aug 2013 01:20:17 +0000 (18:20 -0700)]
Encapsulate textures into their own Texture class

the main reason for doing this is so that we can have
access to informations about a texture (like its dimension)
close to where we generate and use shaders in ES 2.0.
Previously, there wasn't any way to get to a texture's size
from a RenderEngine implementation.

Bug: 8679321

Change-Id: I388b338a70d07e3e8177dde248710ea1e4c82dff

10 years agoRe-enable frame dropping for non-auto timestamps
Andy McFadden [Fri, 16 Aug 2013 21:55:39 +0000 (14:55 -0700)]
Re-enable frame dropping for non-auto timestamps

This change adds an entire field to note whether the timestamp was
auto-generated by Surface or supplied by the application.

The value is used when deciding whether or not to drop frames based
on buffer presentation timestamps.  If a desired presentation time
was set explicitly, BufferQueue will use that value to decide if a
frame should be dropped.  If the timestamp was generated by Surface
at the time the buffer was queued, the timestamp is ignored.

Bug 10151804

Change-Id: Ibd571a7578351063b813cbdad2ddbeed70655ba5

10 years agoMerge "Replace sRGB_888 with sRGB_X_8888" into klp-dev
Jesse Hall [Fri, 16 Aug 2013 21:50:22 +0000 (21:50 +0000)]
Merge "Replace sRGB_888 with sRGB_X_8888" into klp-dev

10 years agoUse new HWC display type/count constants.
Jesse Hall [Fri, 16 Aug 2013 21:28:37 +0000 (14:28 -0700)]
Use new HWC display type/count constants.

Change-Id: I774d0c68906ac6dc69268f708c30a6b0868b8816

10 years agoMerge "gui: CpuConsumer::lockNextBuffer change return code when too many bufs acquire...
Igor Murashkin [Fri, 16 Aug 2013 21:02:01 +0000 (21:02 +0000)]
Merge "gui: CpuConsumer::lockNextBuffer change return code when too many bufs acquired" into klp-dev

10 years agoReplace sRGB_888 with sRGB_X_8888
Jesse Hall [Fri, 16 Aug 2013 14:50:11 +0000 (07:50 -0700)]
Replace sRGB_888 with sRGB_X_8888

Bug: 10357459
Change-Id: I3a1d92cce2c9d3e930408129fb20887276fffddc

10 years agoMerge "Add support for EGL_KHR_gl_colorspace" into klp-dev
Jesse Hall [Thu, 15 Aug 2013 23:17:49 +0000 (23:17 +0000)]
Merge "Add support for EGL_KHR_gl_colorspace" into klp-dev

10 years agoMerge "Fix ES2 composition on some devices" into klp-dev
Mathias Agopian [Thu, 15 Aug 2013 23:12:26 +0000 (23:12 +0000)]
Merge "Fix ES2 composition on some devices" into klp-dev

10 years agoSimplify handling of opaque layers in shader generation
Mathias Agopian [Wed, 14 Aug 2013 23:33:27 +0000 (16:33 -0700)]
Simplify handling of opaque layers in shader generation

just ensure the alpha value is 1.0 in the opaque case
when reading the color from the texture or the
global color.

Bug: 8679321

Change-Id: Ia38b30e97c3bce5a2d534a40c0d66e0bfc3ea40d

10 years agoFix ES2 composition on some devices
Mathias Agopian [Thu, 15 Aug 2013 21:56:51 +0000 (14:56 -0700)]
Fix ES2 composition on some devices

- turns out fragment shaders don't have default precision by default
- GLES 1.x extensions that became core in GLES 2.0 don't always work
  as extensions in GLES 2.0 (!)

Bug: 8679321
Change-Id: I5a4a93e158247910399325a965af5d2e3bbece9b

10 years agoDisable GLES 2.x rendering
Andy McFadden [Thu, 15 Aug 2013 17:39:11 +0000 (10:39 -0700)]
Disable GLES 2.x rendering

Bug 10340705

Change-Id: I91f3baf0d79655d8d031c100d0b116f745c2fbef

10 years agoInitialize mIndent
Andy McFadden [Thu, 15 Aug 2013 17:05:01 +0000 (10:05 -0700)]
Initialize mIndent

Bug 10334731

Change-Id: Ia3aa118a1516cc480b0d85179b78c4447f982dd6

10 years agoDisable GLES 2.x rendering
Andy McFadden [Thu, 15 Aug 2013 17:39:11 +0000 (10:39 -0700)]
Disable GLES 2.x rendering

Bug 10340705

Change-Id: I91f3baf0d79655d8d031c100d0b116f745c2fbef

10 years agoInitialize mIndent
Andy McFadden [Thu, 15 Aug 2013 17:05:01 +0000 (10:05 -0700)]
Initialize mIndent

Bug 10334731

Change-Id: Ia3aa118a1516cc480b0d85179b78c4447f982dd6

10 years agomerge in klp-release history after reset to klp-dev
The Android Automerger [Thu, 15 Aug 2013 13:01:06 +0000 (06:01 -0700)]
merge in klp-release history after reset to klp-dev

10 years agogui: CpuConsumer::lockNextBuffer change return code when too many bufs acquired
Igor Murashkin [Thu, 15 Aug 2013 01:49:12 +0000 (18:49 -0700)]
gui: CpuConsumer::lockNextBuffer change return code when too many bufs acquired

- Return NOT_ENOUGH_DATA instead of INVALID_OPERATION when too many
  buffers have already been locked.
- INVALID_OPERATION is nominally used when something irrecoverable happens,
  but in this case the client just needs to call unlockBuffer to go back into a
  good state.

Bug: 10333400
Change-Id: I3a034d77de85741429f832a90eedd670afa1dc94

10 years agofix build. optional test broke.
Mathias Agopian [Thu, 15 Aug 2013 01:22:02 +0000 (18:22 -0700)]
fix build. optional test broke.

Change-Id: I0ed5e2cd4c8945ee95673ed0eec5820d9def5097

10 years agoFix build. optional tests were broken.
Mathias Agopian [Wed, 14 Aug 2013 23:26:43 +0000 (16:26 -0700)]
Fix build. optional tests were broken.

Change-Id: Iba9f5cb57f134559e1f9c55e293578d9f5d8cb96

10 years agoMerge "don't reallocate the mesh each time we use it" into klp-dev
Mathias Agopian [Wed, 14 Aug 2013 22:43:25 +0000 (22:43 +0000)]
Merge "don't reallocate the mesh each time we use it" into klp-dev

10 years agoMerge "SurfaceFlinger now uses GLES 2.x when available" into klp-dev
Mathias Agopian [Wed, 14 Aug 2013 22:42:51 +0000 (22:42 +0000)]
Merge "SurfaceFlinger now uses GLES 2.x when available" into klp-dev

10 years agomerge in klp-release history after reset to klp-dev
The Android Automerger [Wed, 14 Aug 2013 13:06:17 +0000 (06:06 -0700)]
merge in klp-release history after reset to klp-dev

10 years agodon't reallocate the mesh each time we use it
Mathias Agopian [Wed, 14 Aug 2013 03:51:23 +0000 (20:51 -0700)]
don't reallocate the mesh each time we use it

the Mesh object can be part of each Layer (at least currently).
also reworked the Mesh code a bit to make it easier to access
the vertex data.

Change-Id: I0490851ba898f0aa2e55b62958dcd8bdb535e98b

10 years agoMerge "dumpstate: switch to HAL static libraries" into klp-dev
Todd Poynor [Wed, 14 Aug 2013 00:22:41 +0000 (00:22 +0000)]
Merge "dumpstate: switch to HAL static libraries" into klp-dev

10 years agodumpstate: switch to HAL static libraries
Todd Poynor [Tue, 13 Aug 2013 23:30:01 +0000 (16:30 -0700)]
dumpstate: switch to HAL static libraries

Change-Id: I865a0b4b4a84a2d54a239e7b09874287f0c1ad56

10 years agoSurfaceFlinger now uses GLES 2.x when available
Mathias Agopian [Thu, 8 Aug 2013 04:24:32 +0000 (21:24 -0700)]
SurfaceFlinger now uses GLES 2.x when available

Bug: 8679321

Change-Id: I2b152d01fb4e2de2ea9fe87f1ddbd6826d7520d7

10 years agoallow "system" screenshots even when the secure flag is set
Mathias Agopian [Tue, 13 Aug 2013 22:55:43 +0000 (15:55 -0700)]
allow "system" screenshots even when the secure flag is set

we need too allow this case so that things like the rotation
animation can work.
with this change we only permit these screenshot if the
destination is SurfaceFlinger itself.

Bug: 10235036
Change-Id: I66fea5391e52b0d7f17f25827572b236f2d9eb71

10 years agoAdd support for EGL_KHR_gl_colorspace
Jesse Hall [Thu, 8 Aug 2013 20:40:22 +0000 (13:40 -0700)]
Add support for EGL_KHR_gl_colorspace

Change-Id: I684d0b8556cd6c84ee4b4d67e1bb95c3b96fccfb

10 years agoMerge "Don't swallow eglCreateContext errors" into klp-dev
Jesse Hall [Tue, 13 Aug 2013 18:33:07 +0000 (18:33 +0000)]
Merge "Don't swallow eglCreateContext errors" into klp-dev

10 years agomerge in klp-release history after reset to klp-dev
The Android Automerger [Tue, 13 Aug 2013 13:02:27 +0000 (06:02 -0700)]
merge in klp-release history after reset to klp-dev

10 years agomake sure GLConsumer::releaseTexImage() works after detachFromContext()
Mathias Agopian [Fri, 9 Aug 2013 01:16:21 +0000 (18:16 -0700)]
make sure GLConsumer::releaseTexImage() works after detachFromContext()

Change-Id: I27e0bc57e927d47b2b98113ee37b5396bcc72019

10 years agoDon't swallow eglCreateContext errors
Jesse Hall [Mon, 12 Aug 2013 23:43:34 +0000 (16:43 -0700)]
Don't swallow eglCreateContext errors

If the implementation eglCreateContext failed, we called
eglGetError(), so we could check that it set the EGL error properly.
But since we'd already called this, when the app called eglGetError()
it would get EGL_SUCCESS!

Bug: 10181333
Change-Id: Ic45d3a8fcb8c4421e04844c6d2f52761790b5948

10 years agomerge in klp-release history after reset to klp-dev
The Android Automerger [Mon, 12 Aug 2013 13:02:07 +0000 (06:02 -0700)]
merge in klp-release history after reset to klp-dev

10 years agoMerge "make sure we have a context when creating the 1st surface" into klp-dev
Mathias Agopian [Fri, 9 Aug 2013 04:41:19 +0000 (04:41 +0000)]
Merge "make sure we have a context when creating the 1st surface" into klp-dev

10 years agomake GLConsumer's debug texture static
Mathias Agopian [Fri, 9 Aug 2013 00:46:48 +0000 (17:46 -0700)]
make GLConsumer's debug texture static

this is so that we don't burn a gralloc handle and a bit
of memory per GLConsumer.

Change-Id: I30c3e7ec45d7f1b5426aca8e06c7716870877706

10 years agomake sure we have a context when creating the 1st surface
Mathias Agopian [Thu, 8 Aug 2013 03:10:20 +0000 (20:10 -0700)]
make sure we have a context when creating the 1st surface

also add an option to dump the stack trace when
calling a GL function without a context.

Change-Id: I57b72bb8c322ac4253c3077bf150621bd9863b69

10 years agoAdd ISurfaceComposer::destroyDisplay
Jesse Hall [Thu, 8 Aug 2013 19:15:49 +0000 (12:15 -0700)]
Add ISurfaceComposer::destroyDisplay

Bug: 10191053
Change-Id: Ia89286f95421344a60ffedaaca5825c4e3cd7f9e

10 years agomerge in klp-release history after reset to klp-dev
The Android Automerger [Thu, 8 Aug 2013 14:54:13 +0000 (07:54 -0700)]
merge in klp-release history after reset to klp-dev

10 years agoMerge "don't hardcode debug texture dimensions"
Mathias Agopian [Wed, 7 Aug 2013 20:36:07 +0000 (20:36 +0000)]
Merge "don't hardcode debug texture dimensions"

10 years agodon't hardcode debug texture dimensions
Mathias Agopian [Wed, 7 Aug 2013 20:35:20 +0000 (13:35 -0700)]
don't hardcode debug texture dimensions

Change-Id: I356333801d246a8cafb3e977a77dc7325e8b822d

10 years agoam 046b72fc: Merge "fix a possible deadlock when removing a layer and destroying...
Jean-Baptiste Queru [Wed, 7 Aug 2013 20:18:51 +0000 (13:18 -0700)]
am 046b72fc: Merge "fix a possible deadlock when removing a layer and destroying a client"

* commit '046b72fc5d93aae3b47b88b9f6683512d01d75d0':
  fix a possible deadlock when removing a layer and destroying a client

10 years agobatteryservice: add battery current now and charge counter data
Todd Poynor [Wed, 31 Jul 2013 02:12:10 +0000 (19:12 -0700)]
batteryservice: add battery current now and charge counter data

Change-Id: I4b3ea25342d4a03a8e38bfee2ef3b1ce54f82582

10 years agoMerge "Add EGL_KHR_get_all_proc_addresses to extension string"
Jesse Hall [Wed, 7 Aug 2013 15:23:02 +0000 (15:23 +0000)]
Merge "Add EGL_KHR_get_all_proc_addresses to extension string"

10 years agoMerge "Set the swap interval of virtual display surfaces to 0"
Jesse Hall [Wed, 7 Aug 2013 15:23:00 +0000 (15:23 +0000)]
Merge "Set the swap interval of virtual display surfaces to 0"

10 years agomerge in klp-release history after reset to master
The Android Automerger [Wed, 7 Aug 2013 13:00:54 +0000 (06:00 -0700)]
merge in klp-release history after reset to master

10 years agoSet the swap interval of virtual display surfaces to 0
Jesse Hall [Wed, 7 Aug 2013 00:08:53 +0000 (17:08 -0700)]
Set the swap interval of virtual display surfaces to 0

This prevents slow/malicious virtual display consumers from
backpressuring SurfaceFlinger and preventing it from updating other
displays.

Bug: 10193714
Change-Id: I3e877d97202628d2d9abea24b66576f38299c14d

10 years agofix tests more
Mathias Agopian [Tue, 6 Aug 2013 23:47:04 +0000 (16:47 -0700)]
fix tests more

Change-Id: I05c3dd6a22440844f8bd071f4c2b620b2bd8c720

10 years agofix tests
Mathias Agopian [Tue, 6 Aug 2013 23:42:05 +0000 (16:42 -0700)]
fix tests

Change-Id: I0f4d613481eda2c44a2d631a81d75de7e52e7a62

10 years agoremove support for glReadPixels screenshot path
Mathias Agopian [Tue, 6 Aug 2013 23:07:33 +0000 (16:07 -0700)]
remove support for glReadPixels screenshot path

this was only needed on some chipset we're not
supporting in KLP.

Change-Id: I2e8fc07f135030bd61d2e241a8e426f1a4316425

10 years agoAdd EGL_KHR_get_all_proc_addresses to extension string
Jesse Hall [Tue, 6 Aug 2013 22:31:22 +0000 (15:31 -0700)]
Add EGL_KHR_get_all_proc_addresses to extension string

We've had the extended behavior since Android 4.3, but the extension
hadn't been ratified by Khronos yet so we couldn't add it to the
extension string.

Bug: 9681677
Change-Id: I78842316a3ab7a3f66a0ac1a4742d837de664c4c

10 years agoFix build. 3rd attempt
Mathias Agopian [Tue, 6 Aug 2013 22:19:15 +0000 (15:19 -0700)]
Fix build. 3rd attempt

Change-Id: I4317b34fd07890ea7cf9db51be42d72ef21aca89

10 years agoAll consumers now take an IGraphicBufferConsumer instead of a BufferQueue
Mathias Agopian [Fri, 2 Aug 2013 08:40:18 +0000 (01:40 -0700)]
All consumers now take an IGraphicBufferConsumer instead of a BufferQueue

this means they only have access to the consumer end of
the interface. we had a lot of code that assumed consumers
where holding a BufferQueue (i.e.: both ends), so most of
this change is untangling in fix that

Bug: 9265647
Change-Id: Ic2e2596ee14c7535f51bf26d9a897a0fc036d22c

10 years agoMerge "fix a possible deadlock when removing a layer and destroying a client"
Jean-Baptiste Queru [Tue, 6 Aug 2013 18:37:12 +0000 (18:37 +0000)]
Merge "fix a possible deadlock when removing a layer and destroying a client"

10 years agomerge in klp-release history after reset to master
The Android Automerger [Tue, 6 Aug 2013 15:17:35 +0000 (08:17 -0700)]
merge in klp-release history after reset to master

10 years agofix a possible deadlock when removing a layer and destroying a client
Mathias Agopian [Wed, 5 Jun 2013 21:30:54 +0000 (14:30 -0700)]
fix a possible deadlock when removing a layer and destroying a client

generally the last reference to a Layer is released in commitTransaction()
with mStateLock held. Layer itself only holds weak references to Client,
however, ~Layer() briefly promotes this weak reference -- during that time
the all other strong references to that Client go away, ~Layer is left with
the last one... then hell breaks loose as ~Client is called, which in turn
needs to acquire mStateLock.

We fix this by holding a temporary copy of the drawing state during
the transaction so that the side-effects of copying the current
state into the drawing state are seen only after mStateLock has
been released.

Bug: 9106453
Change-Id: Ic5348ac12283500ead87286a37565e8da35f1db2

10 years agoMerge "BufferQueue: Support query of consumer usage bits"
Eino-Ville Talvala [Mon, 5 Aug 2013 21:10:45 +0000 (21:10 +0000)]
Merge "BufferQueue: Support query of consumer usage bits"

10 years agomerge in klp-release history after reset to master
The Android Automerger [Mon, 5 Aug 2013 13:01:34 +0000 (06:01 -0700)]
merge in klp-release history after reset to master

10 years agoMerge "Roll back recent change"
Andy McFadden [Fri, 2 Aug 2013 22:34:13 +0000 (22:34 +0000)]
Merge "Roll back recent change"

10 years agoRoll back recent change
Andy McFadden [Fri, 2 Aug 2013 22:31:45 +0000 (15:31 -0700)]
Roll back recent change

Disable dropping of frames based on timestamp.  Resume auto-
generating timestamps in Surface.

Bug 10151804

Change-Id: I15de26158e1d7ef22a5b150e685a126dc48ae2b4

10 years agoMerge "Binderize the consumer side of BufferQueue"
Mathias Agopian [Fri, 2 Aug 2013 20:35:26 +0000 (20:35 +0000)]
Merge "Binderize the consumer side of BufferQueue"

10 years agoMerge "Drop frames based on presentation timestamp"
Andy McFadden [Fri, 2 Aug 2013 16:53:11 +0000 (16:53 +0000)]
Merge "Drop frames based on presentation timestamp"

10 years agoBinderize the consumer side of BufferQueue
Mathias Agopian [Thu, 1 Aug 2013 03:09:53 +0000 (20:09 -0700)]
Binderize the consumer side of BufferQueue

While currently untested, this should allow to move the
BuffereQueue in the consumer process and have everything
work as usual.

Bug: 9265647

Change-Id: I9ca8f099f7c65b9a27b7e7a3643b46d1b58eacfc

10 years agoDrop frames based on presentation timestamp
Andy McFadden [Thu, 1 Aug 2013 20:37:42 +0000 (13:37 -0700)]
Drop frames based on presentation timestamp

If there are two or more buffers pending that are ready for
immediate presentation, drop all but the last one.

Any code that didn't explicitly specify timestamps for buffers
was using the default value (auto-generated "now").  As a result,
surfaceflinger would drop frames whenever more than one buffer
was queued.  We now use zero as the auto-generated timestamp,
and we don't set the timestamp in eglBeginFrame().

Change-Id: I187f42d33de227cd3411ff0dcd3b9ce1961457eb

10 years agolog an error when dequeueBuffer gets a NULL Fence
Mathias Agopian [Thu, 1 Aug 2013 22:48:40 +0000 (15:48 -0700)]
log an error when dequeueBuffer gets a NULL Fence

Bug: 9858745
Change-Id: If35d6a7d770bf0a42d55bc55880a9ddb721fcfd0

10 years agoremove unused code form GraphicBuffer
Mathias Agopian [Thu, 1 Aug 2013 02:18:22 +0000 (19:18 -0700)]
remove unused code form GraphicBuffer

Change-Id: I0dd67f1f93f654331e8f17b55730931dbc695ca7

10 years agoremove libutils
Alex Ray [Tue, 30 Jul 2013 19:48:57 +0000 (12:48 -0700)]
remove libutils

libutils is being moved from frameworks/native/ to system/core/ in order
to facilitate native C++ platform (non-frameworks) code.

Change-Id: I44089fb960591a40b8a9c30faabb10459d107d71

10 years agomerge in klp-release history after reset to master
The Android Automerger [Wed, 31 Jul 2013 13:01:38 +0000 (06:01 -0700)]
merge in klp-release history after reset to master

10 years agoFix BlobCache_test
Brian Carlstrom [Wed, 31 Jul 2013 08:08:36 +0000 (01:08 -0700)]
Fix BlobCache_test

Change-Id: I5b20aeb4dcff80f1e18e1165b8461d20034dc1ac

10 years agoMake Flattenable not virtual
Mathias Agopian [Tue, 30 Jul 2013 04:24:40 +0000 (21:24 -0700)]
Make Flattenable not virtual

Fallout from the Flattenable change, update all its uses.

Additionnaly, fix/tighten size checks when (un)flatten()ing
things.

Removed the assumption by some flattenables (e.g.: Fence)
that the size passed to them would be exact (it can
and will be larger in some cases)

The code in Parcel is a bit complicated so that we don't
have to expose the full implementation (and also to
keep the code smallish).

Change-Id: I0bf1c8aca2a3128491b4f45510bc46667e566dde