There are implmentations that can dead lock when calling certain
functions in certain states. It works fine before HIDL because
SurfaceFlinger calls those functions in another thread. We allow
for 4 hwbinder threads in this commit to simulate how SurfaceFlinger
calls hwcomposer.
Bug: 38183197
Test: camera, videos, multi windows, screencap, screenrecord
Change-Id: Ie05b2ca349b8c1ed9a3ac962981434f2efee92d9
android.hidl.base@1.0 and android.hidl.manager@1.0 are built into libhidltransport.
Test: links
Bug: 33276472
Merged-In: I08aaad80f7e2fc262aa3a8b66fe932e8133a928d
Change-Id: I08aaad80f7e2fc262aa3a8b66fe932e8133a928d
When setLayerBuffer() was called on a layer previously destroyed by
destroyLayer() the code would crash. Instead, return an error.
Bug: 37159844
Test: Ran vr flinger in a way that would trigger the crash, and
confirmed that I now get error logs instead of crashing. Unfortunately
the error code is consumed by the Composer wrapper class and not
propagated back to the caller, but that's a separate issue (b/37901601).
Change-Id: I75a5b954d47a1deac44d03851f60f347085eca89
We must use the mapper HAL instead of gralloc0/gralloc1 from the
composer.
Bug: 37540361
Test: boots on marlin, angler, and ryu
Change-Id: I5a3ff6a025bf51a3507a4f33fa77e9506a6f1ec9
By setting vendor_available, the following may become true:
* a prebuilt library from this release may be used at runtime by
in a later releasse (by vendor code compiled against this release).
so this library shouldn't depend on runtime state that may change
in the future.
* this library may be loaded twice into a single process (potentially
an old version and a newer version). The symbols will be isolated
using linker namespaces, but this may break assumptions about 1
library in 1 process (your singletons will run twice).
Background:
This means that these modules may be built and installed twice --
once for the system partition and once for the vendor partition. The
system version will build just like today, and will be used by the
framework components on /system. The vendor version will build
against a reduced set of exports and libraries -- similar to, but
separate from, the NDK. This means that all your dependencies must
also mark vendor_available.
At runtime, /system binaries will load libraries from /system/lib*,
while /vendor binaries will load libraries from /vendor/lib*. There
are some exceptions in both directions -- bionic(libc,etc) and liblog
are always loaded from /system. And SP-HALs (OpenGL, etc) may load
/vendor code into /system processes, but the dependencies of those
libraries will load from /vendor until it reaches a library that's
always on /system. In the SP-HAL case, if both framework and vendor
libraries depend on a library of the same name, both versions will be
loaded, but they will be isolated from each other.
It's possible to compile differently -- reducing your source files,
exporting different include directories, etc. For details see:
https://android-review.googlesource.com/368372
None of this is enabled unless the device opts into the system/vendor
split with BOARD_VNDK_VERSION := current.
Bug: 36426473
Bug: 36079834
Test: m -j libhwcomposer-client
Test: attempt to compile with BOARD_VNDK_VERSION := current
Change-Id: I735b861fcc25bc1048ce0ce3ad48432980248b06
Revise IAllocator and IMapper to reduce IPC and to support gralloc0
devices.
Specifically, IAllocator is trimmed down to have essentially only
allocate(BufferDescriptor descriptor, uint32_t count)
generates (Error error,
uint32_t stride,
vec<handle> buffers);
The ability to allocate buffers with shared backing store is
removed. ProducerUsage and ConsumerUsage are moved to the
graphics.common package and are merged and renamed to BufferUsage.
BufferUsage's bits follow gralloc0.
IMapper gains
typedef vec<uint32_t> BufferDescriptor;
createDescriptor(BufferDescriptorInfo descriptorInfo)
generates (Error error,
BufferDescriptor descriptor);
where BufferDescriptor is an implementation-defined blob. lockFlex
is replaced by lockYCbCr. All getters are removed.
Reference counting with retain/release is replaced by
importBuffer/freeBuffer.
Most if not all gralloc1 features are not used by the runtime yet.
There is also not too much test written for them. As such, they
tend to behave differently between implementations and cannot be
used reliably.
Bug: 36481301
Test: builds and boots on Pixel
Change-Id: I1d31105120517ea2c128c7a19297acf3bfd312bb
This has been copy pasted and isn't actually required.
Test: all hals manually tested to work on internal marlin (which have them binderized).
Note: tv hals not tested (and they never have been). Filed b/36562029
Note: for thermal hal, could only run VTS/check logs for errors.
Test: lshal outputs proper stuff
Fixes: 31928447
Change-Id: Id1be584dc0fa2d70e9189b922335146bf6d1382d
Remove undeclared dependency of libhwcomposer-client on Hwc.h which
created a circular dependency.
Test: Manual
Change-Id: I74d5c4e2db5d247f8b406d1ea42fece41e76659d
We cannot lookup _and_ update buffer cache entry in lookupBuffer.
The old buffer is still in use by hwcomposer2. Add updateBuffer to
do the update after the new buffer has replaced the old buffer in
hwcomposer2.
While at it, s/BufferClone/BufferCacheEntry/g.
Bug: 36064845
Test: manual
Change-Id: I59b61c0198ad528c40020fdebbe27a6cc359226f
Perform a final presentDisplay to clear the screen and get HWC2 out
of any special state (e.g., some implementations put themselves in
special states between validateDisplay and presentDisplay). Our
only portable choice really is to abort(), but that is not an option
because we have other clients such as VTS or VR.
Bug: 35872161
Test: manual
Change-Id: I9028705607d1b86d418a379c6e90e833d638b3f5
Add libVtsHalGraphicsComposerUtils which provides a wrapper to
IComposer. Port tests to be based on
libVtsHalGraphicsComposerUtils.
Test: manual
Change-Id: I400c347b54702c3d45954e6cdc101d3dc1241efd
The getService() and registerAsService() methods of interface objects
now have default parameters of "default" for the service name. HALs
will not have to use any service name unless they want to register
more than one service.
Test: builds
Bug: 33844934
Change-Id: I0c7987ffa7b22bb7386a4a1320242b3248e434e4
The getService() and registerAsService() methods of interface objects
now have default parameters of "default" for the service name. HALs
will not have to use any service name unless they want to register
more than one service.
Test: builds
Bug: 33844934
Change-Id: Ic3859ccf21e7ee5177f47cb06e3e0c93dd572bf6
The getService() and registerAsService() methods of interface objects
now have default parameters of "default" for the service name. HALs
will not have to use any service name unless they want to register
more than one service.
Test: builds
Bug: 33844934
Change-Id: I3d51429681d834f53dc5f6765e97675eafdceda4
