6 Commits

Author SHA1 Message Date
Jeff Brown
cc4f7db698 Fix input channel leak.
Bug: 5156144

Input channels could leak or simply live longer than they should
in some cases.

1. Monitor channels (used by the pointer location overlay) are never
   unregistered, so they would leak.

   Added code to handle failures in the receive callback by closing
   the input channel.

2. The DragState held onto its input window and application handles
   even after the input channel was disposed.

   Added code to null these handles out when they are no longer needed.

3. Input channels previously used as input event targets would stick
   around until the targets were cleared (usually on the next
   event).

   Added code to detect when the input dispatcher is in
   an idle state and to proactively clear the targets then
   to ensure that resources are released promptly.

4. Native input window handles held onto the input channel even
   after the input window was removed from the input dispatcher.
   Consequently, the input channel would not be disposed until
   the input window handle itself was freed.  Since the input
   window handle is held from managed code, this meant that the
   window's input channel could stick around until the next GC.

   Refactored the input window handle to separate the properties
   (info) and identify (handle) state into different objects.
   Then modified the dispatcher to release the properties (info)
   when no longer needed, including the input channel.

7. The pointer location overlay does not actually use its
   standard input channel, only the monitor input channel.

   Added INPUT_FEATURE_NO_INPUT_CHANNEL to allow windows to
   request that they not be provided with an input channel
   at all.

Improved some of the error handling logic to emit the status
code as part of the exception message.

Change-Id: I01988d4391a70c6678c8b0e936ca051af680b1a5
2011-08-30 21:40:28 -07:00
Jeff Brown
9302c8796f Refactor input dispatcher use of window/app handles.
This change moves the cached window and application input state
into the handle objects themselves.  It simplifies the dispatcher
somewhat because it no longer needs to fix up references to
transient InputWindow objects each time the window list is updated.

This change will also make it easier to optimize setInputWindows
to avoid doing a lot of redundant data copying.  In principle, only
the modified fields need to be updated.  However, for now we
continue to update all fields in unison as before.

It turns out that the input dispatcher was inappropriately retaining
pointers to InputWindow objects within the mWindows InputWindow
vector.  This vector is copy-on-write so it is possible and the
item pointers to change if an editing operation is performed on
the vector when it does not exclusively own the underlying
SharedBuffer.  This bug was uncovered by a previous change that
replaced calls to clear() and appendVector() with a simple use
of operator= which caused the buffer to be shared.  Consequently
after editItemAt was called (which it shouldn't have, actually)
the buffer was copied and the cached InputWindow pointers became
invalid.  Oops.  This change fixes the problem.

Change-Id: I0a259339a6015fcf9113dc4081a6875e047fd425
2011-07-14 04:11:21 -07:00
Jeff Brown
4fb76253f2 Fix bounds check.
Bug: 4413945

Also remove dead code from header file.

Change-Id: I4e9fc9d7f8612fedb436e62649d308bd737ed138
2011-05-13 12:51:12 -07:00
Dianne Hackborn
e2515eebf4 Better compat mode part one: start scaling windows.
First step of improving app screen size compatibility mode.  When
running in compat mode, an application's windows are scaled up on
the screen rather than being small with 1:1 pixels.

Currently we scale the application to fill the entire screen, so
don't use an even pixel scaling.  Though this may have some
negative impact on the appearance (it looks okay to me), it has a
big benefit of allowing us to now treat these apps as normal
full-screens apps and do the normal transition animations as you
move in and out and around in them.

This introduces fun stuff in the input system to take care of
modifying pointer coordinates to account for the app window
surface scaling.  The input dispatcher is told about the scale
that is being applied to each window and, when there is one,
adjusts pointer events appropriately as they are being sent
to the transport.

Also modified is CompatibilityInfo, which has been greatly
simplified to not be so insane and incomprehendible.  It is
now simple -- when constructed it determines if the given app
is compatible with the current screen size and density, and
that is that.

There are new APIs on ActivityManagerService to put applications
that we would traditionally consider compatible with larger screens
in compatibility mode.  This is the start of a facility to have
a UI affordance for a user to switch apps in and out of
compatibility.

To test switching of modes, there is a new variation of the "am"
command to do this: am screen-compat [on|off] [package]

This mode switching has the fundamentals of restarting activities
when it is changed, though the state still needs to be persisted
and the overall mode switch cleaned up.

For the few small apps I have tested, things mostly seem to be
working well.  I know of one problem with the text selection
handles being drawn at the wrong position because at some point
the window offset is being scaled incorrectly.  There are
probably other similar issues around the interaction between
two windows because the different window coordinate spaces are
done in a hacky way instead of being formally integrated into
the window manager layout process.

Change-Id: Ie038e3746b448135117bd860859d74e360938557
2011-05-09 17:03:24 -07:00
Jeff Brown
fbf0977321 Support non-rectangular input regions.
This enables the system bar to carve out a region through which
events will be sent to the IME behind it.

Bug: 3238092
Change-Id: I69b855a8d9b5b3ee525266c0861826e53e5b5028
2011-01-16 18:58:49 -08:00
Jeff Brown
928e054931 Prevent events from getting backlogged.
This change implements two heuristics.

1. When events are older than 10 seconds, they are dropped.

2. If the application is currently busy processing an event and
   the user touches a window belonging to a different application
   then we drop the currently queued events so the other application
   can start processing the gesture immediately.

Note that the system takes care of synthesizing cancelation events
automatically for any events that it drops.

Added some new handle types to allow the native dispatcher to
indirectly refer to the WindowManager's window state and app window
token.  This was done to enable the dispatcher to identify the
application to which each window belongs but it also eliminates
some lookup tables and linear searches through the window list
on each key press.

Bug: 3224911
Change-Id: I9dae8dfe23d195d76865f97011fe2f1d351e2940
2011-01-10 17:23:05 -08:00