Cleaned up the implementation of Surface and SurfaceSession
to use more consistent naming and structure.
Added JNI for all of the new surface flinger display API calls.
Enforced the requirement that all Surfaces created by
the window manager be named.
Updated the display manager service to use the new methods.
Change-Id: I2a658f1bfd0437e1c6f9d22df8d4ffcce7284ca2
There is no graceful way to kill Android application processes.
They typically have many threads running doing various things
When System.exit() is called, those threads just keep going
while the cleanup actions run until the process finally.
Performing shutdown actions can easily cause more harm than good.
For example, closing the Binder driver's file descriptor may
cause other threads waiting on Binder to wake up and then crash
in nasty ways after receiving EBADF.
So when an Android application exits, skip the cleanup and just
call _exit() to end it all.
Bug: 6168809
Change-Id: I29790c064426a0bf7dae7cdf444eea3eef1d5275
Some native activities experienced ANRs when the input consumer
deferred an input event due to client-side batching. If the
input channel was fully emptied then the client had no way of
knowing that it should wake up to handle the deferred input event.
This patch also fixes some lock issues in the native activity
input queue implementation. In at least one error case, it
was possible for a function to exit without releasing the lock.
Bug: 6051176
Change-Id: I4d9d843237e69b9834f8d8b360031b677fcab8c3
To support this feature, the input dispatcher now allows input
events to be acknowledged out-of-order. As a result, the
consumer can choose to defer handling an input event from one
device (because it is building a big batch) while continuing
to handle input events from other devices.
The InputEventReceiver now sends a notification when a batch
is pending. The ViewRoot handles this notification by scheduling
a draw on the next sync. When the draw happens, the InputEventReceiver
is instructed to consume all pending batched input events, the
input event queue is fully processed (as much as possible),
and then the ViewRoot performs traversals as usual.
With these changes in place, the input dispatch latency is
consistently less than one frame as long as the application itself
isn't stalled. Input events are delivered to the application
as soon as possible and are handled as soon as possible. In practice,
it is no longer possible for an application to build up a huge
backlog of touch events.
This is part of a series of changes to improve input system pipelining.
Bug: 5963420
Change-Id: I42c01117eca78f12d66d49a736c1c122346ccd1d
Bug: 5332296
The code is functionally equivalent, but a little more efficient
and much easier to maintain.
Change-Id: I90670a13799df05831843a5137ab234929281b7c
This removes the ParcelSurfaceTexture class since that functionality has been
folded into Surface.java. The change also updates the MediaPlayer to get rid
of setParcelSurfaceTexture() and modifies setTexture() to use the new Surface
functionality in order to simplify the code.
Change-Id: Iafa75ea3188263928128325d8a726786971b4de4
This adds support for setting a SurfaceTexture as the MediaPlayer video
sink by using a ParcelSurfaceTexture object. The goal is to enable a
SurfaceTexture to pass through Binder (via ParcelSurfaceTexture) and then
be set on the MediaPlayer.
Change-Id: Ife5689ce673eb4bee1c377019db761685217b71d
This adds a new ParcelSurfaceTexture.java class that can be instantiated with
a SurfaceTexture and used to send the corresponding ISurfaceTexture interface
to another process via Binder. The ParcelSurfaceTexture java object can then
be used to create an ANativeWindow based on the SurfaceTextureClient interface.
Change-Id: Ie38ea948b866e52f36a6d0f6cde19b54a8546817
This seems simpler and more contained, and I think the comment explaining
why hoop-jumping is necessary is a bit clearer now.
Change-Id: Ief4afd7cbb42188ed835fce23e497520bdb753a8
We can help you with that.
Note also that getParcelFileDescriptorFD did no such thing. All its callers
were passing in a regular java.io.FileDescriptor and expecting the int. No
ParcelFileDescriptors involved.
Change-Id: Idc233626f20c092e719f152562601f406cc1b64a
Refactored ViewRoot, NativeActivity and related classes to tell the
dispatcher whether an input event was actually handled by the application.
This will be used to move more of the global default key processing
into the system server instead of the application.
Change-Id: If06b98b6f45c543e5ac5b1eae2b3baf9371fba28
Merge commit '86f1b643d10e0b5b17ae01db3b81020db87b3295'
* commit '86f1b643d10e0b5b17ae01db3b81020db87b3295':
Fix issue #3117918: No way to finish a native activity
Merge commit '14bc6b5d0677e5c454a67775c852f90389bb4567'
* commit '14bc6b5d0677e5c454a67775c852f90389bb4567':
Replace epoll() with poll() and rename PollLoop to Looper.
As part of this change, consolidated and cleaned up the Looper API so
that there are fewer distinctions between the NDK and non-NDK declarations
(no need for two callback types, etc.).
Removed the dependence on specific constants from sys/poll.h such as
POLLIN. Instead looper.h defines events like LOOPER_EVENT_INPUT for
the events that it supports. That should help make any future
under-the-hood implementation changes easier.
Fixed a couple of compiler warnings along the way.
Change-Id: I449a7ec780bf061bdd325452f823673e2b39b6ae
Merge commit 'a1205f07a6f0c745e8f42f16fe38c06af04143c7'
* commit 'a1205f07a6f0c745e8f42f16fe38c06af04143c7':
Modify native ALooper to take an explicit ident.
The ALooper API now uses an explicit "identifier" for the integer
that is returned rather than implicitly using the fd. This allows
the APIs that had the fd to be a little more sane.
Change-Id: I8507f535ad484c0bdc4a1bd016d87bb09acd7ff0
The goal here is to avoid the dalvik.vm.* property explosion by
having a single property that takes an arbitrary collection of Dalvik
command-line options. This is intended for testing of various
configurations by the Dalvik team, not industrial use.
Options should be separated by spaces, e.g.
adb shell setprop dalvik.vm.extra-opts "-showversion -Xmx4m"
will print the version banner and set the heap max to 4MB, which won't
get you very far.
The extra-opts options will appear last, which allows them to override
values set earlier (like the heap max).
Bug 2838629.
(cherry-pick from dalvik-dev branch)
Change-Id: Ibcbb1b62367cf2152798583e8722ef7e461ad19a
This significantly re-works the native key dispatching code to
allow events to be pre-dispatched to the current IME before
being processed by native code. It introduces one new public
API, which must be called after retrieving an event if the app
wishes for it to be pre-dispatched.
Currently the native code will only do pre-dispatching of
system keys, to avoid significant overhead for gaming input.
This should be improved to be smarted, filtering for only
keys that the IME is interested in. Unfortunately IMEs don't
currently provide this information. :p
Change-Id: Ic1c7aeec8b348164957f2cd88119eb5bd85c2a9f
Merge commit 'f8d9379bd834573feca085284970cf686993c330'
* commit 'f8d9379bd834573feca085284970cf686993c330':
IME events are now dispatched to native applications.
And also:
- APIs to show and hide the IME, and control its interaction with the app.
- APIs to tell the app when its window resizes and needs to be redrawn.
- API to tell the app the content rectangle of its window (to layout
around the IME or status bar).
There is still a problem with IME interaction -- we need a way for the
app to deliver events to the IME before it handles them, so that for
example the back key will close the IME instead of finishing the app.
Change-Id: I37b75fc2ec533750ef36ca3aedd2f0cc0b5813cd
Merge commit 'b14f325a0118ffce286155caaabaae149653462c'
* commit 'b14f325a0118ffce286155caaabaae149653462c':
Add ANativeWindow API for directly drawing to the surface bits.
Also other cleanup and fixes:
- We now properly set the default window format to 565.
- New APIs to set the window format and flags from native code.
- Tweaked glue for simpler handling of the "destroy" message.
- Um, other stuff.
Change-Id: Id7790a21a2fa9a19b91854d225324a7c1e7c6ade
Rather than polling for events from the native code in an event thread,
we now require the GPS HAL libraries to call our callbacks from a thread
that is registered with the JVM to call directly into Java.
This eliminates a thread from our code and removes one step in the chain
of message passing from the GPS to the Location Manager client.
Change-Id: I2745a157690310ba9a699a8369f54a7366c6b1ba
Signed-off-by: Mike Lockwood <lockwood@android.com>
