This adds a second createBufferQueue method that returns
sp<IGraphicBuffer*> interfaces instead of sp<BnGraphicBuffer*>, since
most clients don't actually need the Binderized versions (but some
might).
Change-Id: Iaf4f719c96ddb6f704afc75cf52be22588173e32
Sideband streams are essentially a device-specific buffer queue that
bypasses the BufferQueue system. They can be used for situations with
hard real-time requirements like high-quality TV and video playback
with A/V sync. A handle to the stream is provided by the source HAL,
and attached to a BufferQueue. The sink HAL can read buffers via the
stream handle rather than acquiring individual buffers from the
BufferQueue.
Change-Id: Ib3f262eddfc520f4bbe3d9b91753ed7dd09d3a9b
Adds detachBuffer and attachBuffer calls to both the producer and
consumer sides of BufferQueue. Buffers may be detached while dequeued
by the producer or acquired by the consumer, and when attached, enter
the dequeued and acquired states, respectively.
Bug: 13173343
Change-Id: Ic152692b0a94d99e0135b9bfa62747dab2a54220
* commit 'd1c87d37025c49f6a47fe43328572da495ff04c1':
Binder: Fix some valgrind errors.
Binder: Don't cast directly from a pointer to binder_uintptr_t
Binder: Disable attemptIncStrongHandle
* commit '58242fc29881cf29d56ee6e5fde6d73b16d0b67c':
ServiceManager: Implement PING_TRANSACTION
ServiceManager: Use 32/64 bit types from new binder header
Binder: Use 64 bit pointers in 32 processes if selected by the target
Add BINDER_IPC_32BIT to CFLAGS unless TARGET_USES_64_BIT_BINDER is true
Binder: Make binder portable
ServiceManager: Fix the binder interface
ServiceManager: Store handles in uint32_t instead of void *
ServiceManager: Generic Fixes
ServiceManager: Add extra error handling
ServiceManager: Fix Android.mk
ServiceManager: Make use of kernel exported structures
Now that BufferQueue has been split into core + producer + consumer,
rewrite BufferQueue to be a thin layer over a producer and consumer
interface. Eventually, this layer will be deprecated in favor of
only using either the producer or consumer interface, as applicable.
Change-Id: I340ae5f5b633b244fb594615ff52ba50b9e2f7e4
- This also fixes a hang in the camera service when trying to shut down
if the producer is infinitely blocked while trying to get a new buffer.
Bug: 13250382
Change-Id: I32ca82162bb8645b97dbe084e13e05ca05529a42
* commit 'c5294424386bedad2a240074d0845999f28e662d':
Binder: Fix some valgrind errors.
Binder: Don't cast directly from a pointer to binder_uintptr_t
Binder: Disable attemptIncStrongHandle
* commit 'd1c87d37025c49f6a47fe43328572da495ff04c1':
Binder: Fix some valgrind errors.
Binder: Don't cast directly from a pointer to binder_uintptr_t
Binder: Disable attemptIncStrongHandle
* commit '6f286114ec1d597f3dae987bba6e72dbbc404e32':
Revert "Fix "Binder: Make sure binder objects do not overlap" to work old binder kernel interface"
This branch has the new driver interface so it should not have this change.
This reverts commit 6790329358.
Change-Id: I9b6d0d653d98dbdb521ef964c5e7c2f14859d340
When using 64 bit binder pointers, only initializing the 32 bit
handle, in a stack allocated struct, will pass uninitialized stack
data to the kernel and other processes.
Change-Id: I3432d9d36bb251d8ddb0a863661aeb80aabb3d92
When using the 64 bit binder interface from a 32 bit process the
pointer may get sign extended and cause the kernel to fail to read
from it.
Change-Id: I90fcf53880e2aa92e230a9723f9b3f7696170e32
Moves the "opaque layer" from Layer to Layer::State. This allows
it to be updated as part of a transaction.
Bug 12387406
Change-Id: I0a114ce6adf77cd12fb08f96e0691b76c475768d
The driver does not support BC_ATTEMPT_ACQUIRE and will return an error.
IPCThreadState does not handle driver errors, and will resend the failed
command blocking all other commands.
Change-Id: I643986037341821b27b62dc82df933844f4842b8
Add the ability to ignore layers' transformation matrices during
screenshot capture, which will allow the window manager to capture
unrotated images for recents during the device rotation animation.
Bug: 11805195
Change-Id: I854d87bc84ca06ef9a054a454af1c080ee66fbb8
Enhanced the input system to support concurrent dispatch of touch
events on multiple displays which is required for this to work.
Add method to apply offset to PointerCoords.
Change-Id: I55fe4a9a8785ae5a2d3341d214fa3c5107f3963a
Fix warnings related to casting pointers to ints, using %d or %ld to
print size_t/ssize_t, and unused parameters.
Change-Id: I7a13ba83d402952989c1f795cd9e880a95b98d9e
* commit 'c93865de80a16d3638936890fb42eb175284044b':
ServiceManager: Implement PING_TRANSACTION
ServiceManager: Use 32/64 bit types from new binder header
Binder: Use 64 bit pointers in 32 processes if selected by the target
Add BINDER_IPC_32BIT to CFLAGS unless TARGET_USES_64_BIT_BINDER is true
Binder: Make binder portable
ServiceManager: Fix the binder interface
ServiceManager: Store handles in uint32_t instead of void *
ServiceManager: Generic Fixes
ServiceManager: Add extra error handling
ServiceManager: Fix Android.mk
ServiceManager: Make use of kernel exported structures
Changes include
- Binder attempts to cast pointers to a int datatype
which is not sufficient on a 64-bit platform.
- This patch introduces new read/write functions into
Parcel that allow pointers to be written using the
uintptr_t datatype for compile-time data type size
selection.
- Change access specifier for the methods above.
- Binder uses the 64bit android_atomic_release_cas64
(aka cmpxchg)
Change-Id: I595280541e0ba1d19c94b2ca2127bf9d96efabf1
Signed-off-by: Matthew Leach <matthew.leach@arm.com>
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>
Moves the "opaque layer" from Layer to Layer::State. This allows
it to be updated as part of a transaction.
Bug 12387406
Change-Id: I0a114ce6adf77cd12fb08f96e0691b76c475768d
Cookies are really indices into vectors and arrays, so
they don't need to be void*. We choose int32_t instead
of size_t to allow their width to be well specified.
Change-Id: I319f271bd4f102dee7cb9f00df4c75ea697456f5
Add SRGBTest.RenderToSRGBSurface, which validates that passing the
SRGB colorspace attributes to eglCreateWindowSurface actually
results in an SRGB surface by comparing the output values to those
of a standard RGB surface.
Change-Id: I3d5fef8070ed8fa2357ddd1c5fcc849ae3fbd12a
Extract the many different test fixtures and suites in SurfaceTexture_test.cpp
into separate files. No real functional changes, just tweaking headers to keep
things building (and adding a few copyright headers I forgot last time).
Change-Id: Id801bd5d617f0cc61d22508fb9b71b41694bdecf
This lets us share zip archive processing code with both
the runtime (Art, dalvik) and critical java code
(StrictJarFile).
This change also moves several utility methods to ZipUtils
and dedups code across several zip inflation methods.
One of the side effects of this change is that several
processing loops are now O(n) instead of O(n^2).
bug: 10193060
(cherry picked from commit afd31e08299008fdc5c2813f21b2573f29dc53df)
Change-Id: Iae67e62f1dc6dfc3f43e29bc38e3ffd1cb14d191
The CpuConsumer error code was changed recently from INVALID_OPERATION to
NOT_ENOUGH_DATA, however the CpuConsumer_test was not updated. This change
fixes the return code checked by CpuConsumer_test so the test no longer
fails.
Change-Id: Ie548ddcb313fa34c130ac63355090cb44ac15ebf
* Basic tests only. Needs more complicated queue/dequeue tests.
* Also needs consumer-side tests to really be thorough.
Change-Id: I1099dd56d65b6e9dfa15377726d6054ce657c0ca
1. When alloc or realloc failed in the function SharedBuffer::editResize,
it would return a NULL pointer, then mStorage would update to be 1 by
SharedBuffer::data() if no pointer check here, which is an obviously
wrong address, and would cause corruption when used it e.g. in capacity().
So add the pointer check here for the return value of SharedBuffer::editResize,
if it's NULL do not use it to update mStorage, to avoid the value of mStorage
polluted.
2. when alloc or realloc falied in _grow & _shrink function, mStorage keep
the original value, so mCount should not be updated here.
Otherwise, mStorage might be 0 but mCount>0, so a corruption would happend
when it try to delete items from the Vector since mCount>0.
Change-Id: I7c3814e843c459834ca5eed392e8d63d1cb7d2d8
Signed-off-by: Shuo Gao <shuo.gao@intel.com>
Signed-off-by: Jian Luo <jian.luo@intel.com>
Signed-off-by: Bruce Beare <bruce.j.beare@intel.com>
Signed-off-by: Jack Ren <jack.ren@intel.com>
Author-tracking-BZ: 139626
When resampling is disabled, it's currently possible to consume
touches after the current frame time. This breaks some guarantees and
could cause unexpected behaviors.
Change-Id: I99908a2cac2df9f795dd4a07188b4451213cf3e4
When a BufferQueue producer disconnects and reconnects, we retain
the previously-queued buffers but empty the slots. This allows
the number of queued buffers to grow without limit. The low-memory
killer does not approve.
Bug 11069934
Change-Id: Ia2eaa954c7a3904b54209a3701dba01689e204d8
* commit '1f1e576840d5b36a091368abb6fe6610384b6449':
Change API from flush(handle) to flush(). Call flush on all active sensors in the given SensorEventConnection.
* commit '5d6aa95129a13aae37a60b3e4246a0592dab396d':
Change API from flush(handle) to flush(). Call flush on all active sensors in the given SensorEventConnection.
Since the latency was introduced to avoid egregious errors during
resampling, remove it when we aren't resampling.
Change-Id: Ia8e9a9afb1e16212fec81a8c9c661c598e1b0ebf
This change makes GLConsumer use the EGL_ANDROID_image_crop extension when
available on a device. The crop rectangle is passed to the EGL driver when
creating EGLImages, allowing the crop to be performed by the driver rather than
using the texture transform matrix.
Bug: 10897141
Change-Id: I63e9a5d5c85067376abc420e3639154468346311
we add a flag to ANativeWindow::setBufferTransform that means
"apply the inverse rotation of the display this buffer is displayed
onto to".
Bug: 10804238
Change-Id: Id2447676271950463e8dbcef1b95935c5c3f32b2
The UniquePtr.h file is no longer in libcore/... so there is no need to
have this extra copy of it anymore.
Change-Id: I4ec4f86d2844e524b50441c3a9100a50a6a45b1a
BitTube used to send objects one at a time and didn't
handle errors properly.
We now send all the objects in one call, which means they
have to be read as a single batch as well. This changes the
BitTube API.
Update SensorService to the new API.
Also added an API to set the size of the send buffer.
Bug: 10641596
Change-Id: I77c70d35e351fdba0416fae4b7ca3b1d56272251
Initialize local variable to avoid syscall ioctl warning.
Change-Id: I33a90917856018a8527305bb362948ef03bd734c
Origin-Change-Id: I022ebce40b3774a815639a5af34bc7aeb2489936
Signed-off-by: Jin Wei <wei.a.jin@intel.com>
Signed-off-by: Jian Luo <jian.luo@intel.com>
Signed-off-by: Bruce Beare <bruce.j.beare@intel.com>
Signed-off-by: Jack Ren <jack.ren@intel.com>
Author-tracking-BZ: 62797
- this implements vec2, vec3, vec4, which are float vectors
of size 2, 3 and 4 respectively.
the code allows easy instantiation of vectors of a different
type via the tvec{2|3|4}<T> template classes.
- this also implements mat4 which is a float 4x4 matrix. the
tmat44<T> template class allows easy instantiation of a
4x4 matrix of a different value_type.
The vector types have some minimal support for the
glsl style swizzled access; for instance:
vec4 u;
vec3 v = u.xyz;
only .x, .xy, .xyz and their .stpq / .rgba equivalent are
supported.
most operators are supported on both vector and matrices:
arithmetic, unary, compound assignment and comparison
(bit-wise operators NOT supported).
- operations available on vectors include:
dot, length, distance, normalize and cross
- operations available on matrices include:
transpose, inverse, trace
- and a few utilities to create matrices:
ortho, frustum, lookAt
Change-Id: I64add89ae90fa78d3f2f59985b63495575378635
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
- 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
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
Disable dropping of frames based on timestamp. Resume auto-
generating timestamps in Surface.
Bug 10151804
Change-Id: I15de26158e1d7ef22a5b150e685a126dc48ae2b4
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
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
libutils is being moved from frameworks/native/ to system/core/ in order
to facilitate native C++ platform (non-frameworks) code.
Change-Id: I44089fb960591a40b8a9c30faabb10459d107d71
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
Making an object Flattenable doesn't force it to
become virtual anymore. For instance, Fence and GraphicBuffer
are now non-virtual classes.
Also change Flatennable protocol a bit so that it updates
its parameters (pointers, sizes) to make it easier
to implement a flattenable in terms of other flattenables.
Change-Id: Ie81dc7637180b3c2cfcbaf644f8987ca804eb891
BufferQueue::dequeueBuffer() could incorrectly return
WOULD_BLOCK while in "cannot block" mode if it happened
while a consumer acquired the last allowed buffer
before releasing the old one (which is a valid thing
to do).
Change-Id: I318e5408871ba85e068ea9ef4dc9b578f1bb1043
* changes:
make sure to reset the framenumber when a buffer is marked FREE
Make ANW.setSwapInterval(0) work again
BuffferQueue disconnect is now always asynchrnous
BufferQueue improvements and APIs changes
we can now queue/dequeue a buffer in asynchrnous mode by using the
async parameter to these calls. async mode is only specified
with those calls (it is not modal anymore).
as a consequence it can only be specified when the buffer count
is not overidden, as error is returned otherwise.
Change-Id: Ic63f4f96f671cb9d65c4cecbcc192615e09a8b6b
we tag queued buffers with the "bufferqueue cannot block" flag
and use that bit to discard a buffer in the queue by new ones
comming in. this allows us to remove the buffer queue drain in
disconnect while maintaining the right behaviour if it gets
connected again (since each buffer remembers how it was enqueued).
Change-Id: I1e703d363a687b70b19ba49cef32213116e8bd3f
this is the first step of a series of improvements to
BufferQueue. A few things happen in this change:
- setSynchronousMode() goes away as well as the SynchronousModeAllowed flag
- BufferQueue now defaults to (what used to be) synchronous mode
- a new "controlled by app" flag is passed when creating consumers and producers
those flags are used to put the BufferQueue in a mode where it
will never block if both flags are set. This is achieved by:
- returning an error from dequeueBuffer() if it would block
- making sure a buffer is always available by replacing
the previous buffer with the new one in queueBuffer()
(note: this is similar to what asynchrnous mode used to be)
Note: in this change EGL's swap-interval 0 is broken; this will be
fixed in another change.
Change-Id: I691f9507d6e2e158287e3039f2a79a4d4434211d
When acquiring a buffer, SurfaceFlinger now computes the expected
presentation time and passes it to the BufferQueue acquireBuffer()
method. If it's not yet time to display the buffer, acquireBuffer()
returns PRESENT_LATER instead of a buffer.
The current implementation of the expected-present-time computation
uses approximations and guesswork.
Bug 7900302
Change-Id: If9345611c5983a11a811935aaf27d6388a5036f1
This is currently safe to do only for processes that disallow any binder
threads to be created: setThreadPoolMaxThreadCount(0).
Change-Id: I8a27f3cf26f4d51edb7f222af487ac256cbcab65
