ADHOC/replicant-frameworks_native
3
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
b06cca6228
replicant-frameworks_native/libs/ui/InputReader.cpp

3412 lines
119 KiB
C++
Raw Normal View History

Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
//
// Copyright 2010 The Android Open Source Project
//
// The input reader.
//
#define LOG_TAG "InputReader"
//#define LOG_NDEBUG 0
// Log debug messages for each raw event received from the EventHub.
#define DEBUG_RAW_EVENTS 0
// Log debug messages about touch screen filtering hacks.
Native input event dispatching. Target identification is now fully native. Fixed a couple of minor issues related to input injection. Native input enabled by default, can be disabled by setting WindowManagerPolicy.ENABLE_NATIVE_INPUT_DISPATCH to false. Change-Id: I7edf66ed3e987cc9306ad4743ac57a116af452ff
2010-06-22 08:27:15 +00:00
#define DEBUG_HACKS 0
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
// Log debug messages about virtual key processing.
Native input event dispatching. Target identification is now fully native. Fixed a couple of minor issues related to input injection. Native input enabled by default, can be disabled by setting WindowManagerPolicy.ENABLE_NATIVE_INPUT_DISPATCH to false. Change-Id: I7edf66ed3e987cc9306ad4743ac57a116af452ff
2010-06-22 08:27:15 +00:00
#define DEBUG_VIRTUAL_KEYS 0
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
// Log debug messages about pointers.
Native input event dispatching. Target identification is now fully native. Fixed a couple of minor issues related to input injection. Native input enabled by default, can be disabled by setting WindowManagerPolicy.ENABLE_NATIVE_INPUT_DISPATCH to false. Change-Id: I7edf66ed3e987cc9306ad4743ac57a116af452ff
2010-06-22 08:27:15 +00:00
#define DEBUG_POINTERS 0
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Even more native input dispatch work in progress. Added more tests. Fixed a regression in Vector. Fixed bugs in pointer tracking. Fixed a starvation issue in PollLoop when setting or removing callbacks. Fixed a couple of policy nits. Modified the internal representation of MotionEvent to be more efficient and more consistent. Added code to skip/cancel virtual key processing when there are multiple pointers down. This helps to better disambiguate virtual key presses from stray touches (such as cheek presses). Change-Id: I2a7d2cce0195afb9125b23378baa94fd2fc6671c
2010-06-16 08:53:36 +00:00
// Log debug messages about pointer assignment calculations.
#define DEBUG_POINTER_ASSIGNMENT 0
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
#include <cutils/log.h>
#include <ui/InputReader.h>
#include <stddef.h>
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
#include <stdlib.h>
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
#include <unistd.h>
#include <errno.h>
#include <limits.h>
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
#include <math.h>
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
#define INDENT " "
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
namespace android {
// --- Static Functions ---
template<typename T>
inline static T abs(const T& value) {
return value < 0 ? - value : value;
}
template<typename T>
inline static T min(const T& a, const T& b) {
return a < b ? a : b;
}
Even more native input dispatch work in progress. Added more tests. Fixed a regression in Vector. Fixed bugs in pointer tracking. Fixed a starvation issue in PollLoop when setting or removing callbacks. Fixed a couple of policy nits. Modified the internal representation of MotionEvent to be more efficient and more consistent. Added code to skip/cancel virtual key processing when there are multiple pointers down. This helps to better disambiguate virtual key presses from stray touches (such as cheek presses). Change-Id: I2a7d2cce0195afb9125b23378baa94fd2fc6671c
2010-06-16 08:53:36 +00:00
template<typename T>
inline static void swap(T& a, T& b) {
T temp = a;
a = b;
b = temp;
}
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
inline static float avg(float x, float y) {
return (x + y) / 2;
}
inline static float pythag(float x, float y) {
return sqrtf(x * x + y * y);
}
Even more native input dispatch work in progress. Added more tests. Fixed a regression in Vector. Fixed bugs in pointer tracking. Fixed a starvation issue in PollLoop when setting or removing callbacks. Fixed a couple of policy nits. Modified the internal representation of MotionEvent to be more efficient and more consistent. Added code to skip/cancel virtual key processing when there are multiple pointers down. This helps to better disambiguate virtual key presses from stray touches (such as cheek presses). Change-Id: I2a7d2cce0195afb9125b23378baa94fd2fc6671c
2010-06-16 08:53:36 +00:00
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
int32_t updateMetaState(int32_t keyCode, bool down, int32_t oldMetaState) {
int32_t mask;
switch (keyCode) {
Add initial gamepad support. Change-Id: I0439648f6eb5405f200e4223c915eb3a418b32b9
2010-06-30 23:10:35 +00:00
case AKEYCODE_ALT_LEFT:
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
mask = AMETA_ALT_LEFT_ON;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
break;
Add initial gamepad support. Change-Id: I0439648f6eb5405f200e4223c915eb3a418b32b9
2010-06-30 23:10:35 +00:00
case AKEYCODE_ALT_RIGHT:
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
mask = AMETA_ALT_RIGHT_ON;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
break;
Add initial gamepad support. Change-Id: I0439648f6eb5405f200e4223c915eb3a418b32b9
2010-06-30 23:10:35 +00:00
case AKEYCODE_SHIFT_LEFT:
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
mask = AMETA_SHIFT_LEFT_ON;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
break;
Add initial gamepad support. Change-Id: I0439648f6eb5405f200e4223c915eb3a418b32b9
2010-06-30 23:10:35 +00:00
case AKEYCODE_SHIFT_RIGHT:
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
mask = AMETA_SHIFT_RIGHT_ON;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
break;
Add initial gamepad support. Change-Id: I0439648f6eb5405f200e4223c915eb3a418b32b9
2010-06-30 23:10:35 +00:00
case AKEYCODE_SYM:
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
mask = AMETA_SYM_ON;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
break;
default:
return oldMetaState;
}
int32_t newMetaState = down ? oldMetaState | mask : oldMetaState & ~ mask
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
& ~ (AMETA_ALT_ON | AMETA_SHIFT_ON);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
if (newMetaState & (AMETA_ALT_LEFT_ON | AMETA_ALT_RIGHT_ON)) {
newMetaState |= AMETA_ALT_ON;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
if (newMetaState & (AMETA_SHIFT_LEFT_ON | AMETA_SHIFT_RIGHT_ON)) {
newMetaState |= AMETA_SHIFT_ON;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
return newMetaState;
}
static const int32_t keyCodeRotationMap[][4] = {
// key codes enumerated counter-clockwise with the original (unrotated) key first
// no rotation, 90 degree rotation, 180 degree rotation, 270 degree rotation
Add initial gamepad support. Change-Id: I0439648f6eb5405f200e4223c915eb3a418b32b9
2010-06-30 23:10:35 +00:00
{ AKEYCODE_DPAD_DOWN, AKEYCODE_DPAD_RIGHT, AKEYCODE_DPAD_UP, AKEYCODE_DPAD_LEFT },
{ AKEYCODE_DPAD_RIGHT, AKEYCODE_DPAD_UP, AKEYCODE_DPAD_LEFT, AKEYCODE_DPAD_DOWN },
{ AKEYCODE_DPAD_UP, AKEYCODE_DPAD_LEFT, AKEYCODE_DPAD_DOWN, AKEYCODE_DPAD_RIGHT },
{ AKEYCODE_DPAD_LEFT, AKEYCODE_DPAD_DOWN, AKEYCODE_DPAD_RIGHT, AKEYCODE_DPAD_UP },
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
};
static const int keyCodeRotationMapSize =
sizeof(keyCodeRotationMap) / sizeof(keyCodeRotationMap[0]);
int32_t rotateKeyCode(int32_t keyCode, int32_t orientation) {
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
if (orientation != InputReaderPolicyInterface::ROTATION_0) {
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
for (int i = 0; i < keyCodeRotationMapSize; i++) {
if (keyCode == keyCodeRotationMap[i][0]) {
return keyCodeRotationMap[i][orientation];
}
}
}
return keyCode;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
static inline bool sourcesMatchMask(uint32_t sources, uint32_t sourceMask) {
return (sources & sourceMask & ~ AINPUT_SOURCE_CLASS_MASK) != 0;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// --- InputDeviceCalibration ---
InputDeviceCalibration::InputDeviceCalibration() {
}
void InputDeviceCalibration::clear() {
mProperties.clear();
}
void InputDeviceCalibration::addProperty(const String8& key, const String8& value) {
mProperties.add(key, value);
}
bool InputDeviceCalibration::tryGetProperty(const String8& key, String8& outValue) const {
ssize_t index = mProperties.indexOfKey(key);
if (index < 0) {
return false;
}
outValue = mProperties.valueAt(index);
return true;
}
bool InputDeviceCalibration::tryGetProperty(const String8& key, int32_t& outValue) const {
String8 stringValue;
if (! tryGetProperty(key, stringValue) || stringValue.length() == 0) {
return false;
}
char* end;
int value = strtol(stringValue.string(), & end, 10);
if (*end != '\0') {
LOGW("Input device calibration key '%s' has invalid value '%s'. Expected an integer.",
key.string(), stringValue.string());
return false;
}
outValue = value;
return true;
}
bool InputDeviceCalibration::tryGetProperty(const String8& key, float& outValue) const {
String8 stringValue;
if (! tryGetProperty(key, stringValue) || stringValue.length() == 0) {
return false;
}
char* end;
float value = strtof(stringValue.string(), & end);
if (*end != '\0') {
LOGW("Input device calibration key '%s' has invalid value '%s'. Expected a float.",
key.string(), stringValue.string());
return false;
}
outValue = value;
return true;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
// --- InputReader ---
InputReader::InputReader(const sp<EventHubInterface>& eventHub,
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
const sp<InputReaderPolicyInterface>& policy,
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
const sp<InputDispatcherInterface>& dispatcher) :
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mEventHub(eventHub), mPolicy(policy), mDispatcher(dispatcher),
mGlobalMetaState(0) {
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
configureExcludedDevices();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
updateGlobalMetaState();
updateInputConfiguration();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
InputReader::~InputReader() {
for (size_t i = 0; i < mDevices.size(); i++) {
delete mDevices.valueAt(i);
}
}
void InputReader::loopOnce() {
RawEvent rawEvent;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mEventHub->getEvent(& rawEvent);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
#if DEBUG_RAW_EVENTS
LOGD("Input event: device=0x%x type=0x%x scancode=%d keycode=%d value=%d",
rawEvent.deviceId, rawEvent.type, rawEvent.scanCode, rawEvent.keyCode,
rawEvent.value);
#endif
process(& rawEvent);
}
void InputReader::process(const RawEvent* rawEvent) {
switch (rawEvent->type) {
case EventHubInterface::DEVICE_ADDED:
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
addDevice(rawEvent->when, rawEvent->deviceId);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
break;
case EventHubInterface::DEVICE_REMOVED:
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
removeDevice(rawEvent->when, rawEvent->deviceId);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
break;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
default:
consumeEvent(rawEvent);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
break;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputReader::addDevice(nsecs_t when, int32_t deviceId) {
String8 name = mEventHub->getDeviceName(deviceId);
uint32_t classes = mEventHub->getDeviceClasses(deviceId);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Write a log message about the added device as a heading for subsequent log messages.
LOGI("Device added: id=0x%x, name=%s", deviceId, name.string());
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputDevice* device = createDevice(deviceId, name, classes);
device->configure();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
if (device->isIgnored()) {
LOGI(INDENT "Sources: none (device is ignored)");
} else {
LOGI(INDENT "Sources: 0x%08x", device->getSources());
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
bool added = false;
{ // acquire device registry writer lock
RWLock::AutoWLock _wl(mDeviceRegistryLock);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
ssize_t deviceIndex = mDevices.indexOfKey(deviceId);
if (deviceIndex < 0) {
mDevices.add(deviceId, device);
added = true;
}
} // release device registry writer lock
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (! added) {
LOGW("Ignoring spurious device added event for deviceId %d.", deviceId);
delete device;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
return;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
handleConfigurationChanged(when);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputReader::removeDevice(nsecs_t when, int32_t deviceId) {
bool removed = false;
InputDevice* device = NULL;
{ // acquire device registry writer lock
RWLock::AutoWLock _wl(mDeviceRegistryLock);
ssize_t deviceIndex = mDevices.indexOfKey(deviceId);
if (deviceIndex >= 0) {
device = mDevices.valueAt(deviceIndex);
mDevices.removeItemsAt(deviceIndex, 1);
removed = true;
}
} // release device registry writer lock
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (! removed) {
LOGW("Ignoring spurious device removed event for deviceId %d.", deviceId);
return;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Write a log message about the removed device as a heading for subsequent log messages.
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (device->isIgnored()) {
LOGI("Device removed: id=0x%x, name=%s (ignored non-input device)",
device->getId(), device->getName().string());
} else {
LOGI("Device removed: id=0x%x, name=%s, sources=%08x",
device->getId(), device->getName().string(), device->getSources());
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
device->reset();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
delete device;
handleConfigurationChanged(when);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputDevice* InputReader::createDevice(int32_t deviceId, const String8& name, uint32_t classes) {
InputDevice* device = new InputDevice(this, deviceId, name);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
const int32_t associatedDisplayId = 0; // FIXME: hardcoded for current single-display devices
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Switch-like devices.
if (classes & INPUT_DEVICE_CLASS_SWITCH) {
device->addMapper(new SwitchInputMapper(device));
Add initial gamepad support. Change-Id: I0439648f6eb5405f200e4223c915eb3a418b32b9
2010-06-30 23:10:35 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Keyboard-like devices.
uint32_t keyboardSources = 0;
int32_t keyboardType = AINPUT_KEYBOARD_TYPE_NON_ALPHABETIC;
if (classes & INPUT_DEVICE_CLASS_KEYBOARD) {
keyboardSources |= AINPUT_SOURCE_KEYBOARD;
}
if (classes & INPUT_DEVICE_CLASS_ALPHAKEY) {
keyboardType = AINPUT_KEYBOARD_TYPE_ALPHABETIC;
}
if (classes & INPUT_DEVICE_CLASS_DPAD) {
keyboardSources |= AINPUT_SOURCE_DPAD;
}
if (classes & INPUT_DEVICE_CLASS_GAMEPAD) {
keyboardSources |= AINPUT_SOURCE_GAMEPAD;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Add initial gamepad support. Change-Id: I0439648f6eb5405f200e4223c915eb3a418b32b9
2010-06-30 23:10:35 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (keyboardSources != 0) {
device->addMapper(new KeyboardInputMapper(device,
associatedDisplayId, keyboardSources, keyboardType));
Add initial gamepad support. Change-Id: I0439648f6eb5405f200e4223c915eb3a418b32b9
2010-06-30 23:10:35 +00:00
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Trackball-like devices.
if (classes & INPUT_DEVICE_CLASS_TRACKBALL) {
device->addMapper(new TrackballInputMapper(device, associatedDisplayId));
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Touchscreen-like devices.
if (classes & INPUT_DEVICE_CLASS_TOUCHSCREEN_MT) {
device->addMapper(new MultiTouchInputMapper(device, associatedDisplayId));
} else if (classes & INPUT_DEVICE_CLASS_TOUCHSCREEN) {
device->addMapper(new SingleTouchInputMapper(device, associatedDisplayId));
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return device;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputReader::consumeEvent(const RawEvent* rawEvent) {
int32_t deviceId = rawEvent->deviceId;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
{ // acquire device registry reader lock
RWLock::AutoRLock _rl(mDeviceRegistryLock);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
ssize_t deviceIndex = mDevices.indexOfKey(deviceId);
if (deviceIndex < 0) {
LOGW("Discarding event for unknown deviceId %d.", deviceId);
return;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputDevice* device = mDevices.valueAt(deviceIndex);
if (device->isIgnored()) {
//LOGD("Discarding event for ignored deviceId %d.", deviceId);
return;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
device->process(rawEvent);
} // release device registry reader lock
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputReader::handleConfigurationChanged(nsecs_t when) {
// Reset global meta state because it depends on the list of all configured devices.
updateGlobalMetaState();
// Update input configuration.
updateInputConfiguration();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Enqueue configuration changed.
mDispatcher->notifyConfigurationChanged(when);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputReader::configureExcludedDevices() {
Vector<String8> excludedDeviceNames;
mPolicy->getExcludedDeviceNames(excludedDeviceNames);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
for (size_t i = 0; i < excludedDeviceNames.size(); i++) {
mEventHub->addExcludedDevice(excludedDeviceNames[i]);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputReader::updateGlobalMetaState() {
{ // acquire state lock
AutoMutex _l(mStateLock);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mGlobalMetaState = 0;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
{ // acquire device registry reader lock
RWLock::AutoRLock _rl(mDeviceRegistryLock);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
for (size_t i = 0; i < mDevices.size(); i++) {
InputDevice* device = mDevices.valueAt(i);
mGlobalMetaState |= device->getMetaState();
}
} // release device registry reader lock
} // release state lock
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputReader::getGlobalMetaState() {
{ // acquire state lock
AutoMutex _l(mStateLock);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return mGlobalMetaState;
} // release state lock
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputReader::updateInputConfiguration() {
{ // acquire state lock
AutoMutex _l(mStateLock);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t touchScreenConfig = InputConfiguration::TOUCHSCREEN_NOTOUCH;
int32_t keyboardConfig = InputConfiguration::KEYBOARD_NOKEYS;
int32_t navigationConfig = InputConfiguration::NAVIGATION_NONAV;
{ // acquire device registry reader lock
RWLock::AutoRLock _rl(mDeviceRegistryLock);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputDeviceInfo deviceInfo;
for (size_t i = 0; i < mDevices.size(); i++) {
InputDevice* device = mDevices.valueAt(i);
device->getDeviceInfo(& deviceInfo);
uint32_t sources = deviceInfo.getSources();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if ((sources & AINPUT_SOURCE_TOUCHSCREEN) == AINPUT_SOURCE_TOUCHSCREEN) {
touchScreenConfig = InputConfiguration::TOUCHSCREEN_FINGER;
}
if ((sources & AINPUT_SOURCE_TRACKBALL) == AINPUT_SOURCE_TRACKBALL) {
navigationConfig = InputConfiguration::NAVIGATION_TRACKBALL;
} else if ((sources & AINPUT_SOURCE_DPAD) == AINPUT_SOURCE_DPAD) {
navigationConfig = InputConfiguration::NAVIGATION_DPAD;
}
if (deviceInfo.getKeyboardType() == AINPUT_KEYBOARD_TYPE_ALPHABETIC) {
keyboardConfig = InputConfiguration::KEYBOARD_QWERTY;
}
}
} // release device registry reader lock
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mInputConfiguration.touchScreen = touchScreenConfig;
mInputConfiguration.keyboard = keyboardConfig;
mInputConfiguration.navigation = navigationConfig;
} // release state lock
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputReader::getInputConfiguration(InputConfiguration* outConfiguration) {
{ // acquire state lock
AutoMutex _l(mStateLock);
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
*outConfiguration = mInputConfiguration;
} // release state lock
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
status_t InputReader::getInputDeviceInfo(int32_t deviceId, InputDeviceInfo* outDeviceInfo) {
{ // acquire device registry reader lock
RWLock::AutoRLock _rl(mDeviceRegistryLock);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
ssize_t deviceIndex = mDevices.indexOfKey(deviceId);
if (deviceIndex < 0) {
return NAME_NOT_FOUND;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputDevice* device = mDevices.valueAt(deviceIndex);
if (device->isIgnored()) {
return NAME_NOT_FOUND;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
device->getDeviceInfo(outDeviceInfo);
return OK;
} // release device registy reader lock
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputReader::getInputDeviceIds(Vector<int32_t>& outDeviceIds) {
outDeviceIds.clear();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
{ // acquire device registry reader lock
RWLock::AutoRLock _rl(mDeviceRegistryLock);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
size_t numDevices = mDevices.size();
for (size_t i = 0; i < numDevices; i++) {
InputDevice* device = mDevices.valueAt(i);
if (! device->isIgnored()) {
outDeviceIds.add(device->getId());
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
} // release device registy reader lock
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputReader::getKeyCodeState(int32_t deviceId, uint32_t sourceMask,
int32_t keyCode) {
return getState(deviceId, sourceMask, keyCode, & InputDevice::getKeyCodeState);
}
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputReader::getScanCodeState(int32_t deviceId, uint32_t sourceMask,
int32_t scanCode) {
return getState(deviceId, sourceMask, scanCode, & InputDevice::getScanCodeState);
}
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputReader::getSwitchState(int32_t deviceId, uint32_t sourceMask, int32_t switchCode) {
return getState(deviceId, sourceMask, switchCode, & InputDevice::getSwitchState);
}
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputReader::getState(int32_t deviceId, uint32_t sourceMask, int32_t code,
GetStateFunc getStateFunc) {
{ // acquire device registry reader lock
RWLock::AutoRLock _rl(mDeviceRegistryLock);
int32_t result = AKEY_STATE_UNKNOWN;
if (deviceId >= 0) {
ssize_t deviceIndex = mDevices.indexOfKey(deviceId);
if (deviceIndex >= 0) {
InputDevice* device = mDevices.valueAt(deviceIndex);
if (! device->isIgnored() && sourcesMatchMask(device->getSources(), sourceMask)) {
result = (device->*getStateFunc)(sourceMask, code);
}
}
} else {
size_t numDevices = mDevices.size();
for (size_t i = 0; i < numDevices; i++) {
InputDevice* device = mDevices.valueAt(i);
if (! device->isIgnored() && sourcesMatchMask(device->getSources(), sourceMask)) {
result = (device->*getStateFunc)(sourceMask, code);
if (result >= AKEY_STATE_DOWN) {
return result;
}
}
}
}
return result;
} // release device registy reader lock
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
bool InputReader::hasKeys(int32_t deviceId, uint32_t sourceMask,
size_t numCodes, const int32_t* keyCodes, uint8_t* outFlags) {
memset(outFlags, 0, numCodes);
return markSupportedKeyCodes(deviceId, sourceMask, numCodes, keyCodes, outFlags);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
bool InputReader::markSupportedKeyCodes(int32_t deviceId, uint32_t sourceMask, size_t numCodes,
const int32_t* keyCodes, uint8_t* outFlags) {
{ // acquire device registry reader lock
RWLock::AutoRLock _rl(mDeviceRegistryLock);
bool result = false;
if (deviceId >= 0) {
ssize_t deviceIndex = mDevices.indexOfKey(deviceId);
if (deviceIndex >= 0) {
InputDevice* device = mDevices.valueAt(deviceIndex);
if (! device->isIgnored() && sourcesMatchMask(device->getSources(), sourceMask)) {
result = device->markSupportedKeyCodes(sourceMask,
numCodes, keyCodes, outFlags);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
} else {
size_t numDevices = mDevices.size();
for (size_t i = 0; i < numDevices; i++) {
InputDevice* device = mDevices.valueAt(i);
if (! device->isIgnored() && sourcesMatchMask(device->getSources(), sourceMask)) {
result |= device->markSupportedKeyCodes(sourceMask,
numCodes, keyCodes, outFlags);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
}
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return result;
} // release device registy reader lock
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
void InputReader::dump(String8& dump) {
dumpDeviceInfo(dump);
}
static void dumpMotionRange(String8& dump,
const char* name, const InputDeviceInfo::MotionRange* range) {
if (range) {
dump.appendFormat(" %s = { min: %0.3f, max: %0.3f, flat: %0.3f, fuzz: %0.3f }\n",
name, range->min, range->max, range->flat, range->fuzz);
}
}
#define DUMP_MOTION_RANGE(range) \
dumpMotionRange(dump, #range, deviceInfo.getMotionRange(AINPUT_MOTION_RANGE_##range));
void InputReader::dumpDeviceInfo(String8& dump) {
Vector<int32_t> deviceIds;
getInputDeviceIds(deviceIds);
InputDeviceInfo deviceInfo;
for (size_t i = 0; i < deviceIds.size(); i++) {
int32_t deviceId = deviceIds[i];
status_t result = getInputDeviceInfo(deviceId, & deviceInfo);
if (result == NAME_NOT_FOUND) {
continue;
} else if (result != OK) {
dump.appendFormat(" ** Unexpected error %d getting information about input devices.\n",
result);
continue;
}
dump.appendFormat(" Device %d: '%s'\n",
deviceInfo.getId(), deviceInfo.getName().string());
dump.appendFormat(" sources = 0x%08x\n",
deviceInfo.getSources());
dump.appendFormat(" keyboardType = %d\n",
deviceInfo.getKeyboardType());
dump.append(" motion ranges:\n");
DUMP_MOTION_RANGE(X);
DUMP_MOTION_RANGE(Y);
DUMP_MOTION_RANGE(PRESSURE);
DUMP_MOTION_RANGE(SIZE);
DUMP_MOTION_RANGE(TOUCH_MAJOR);
DUMP_MOTION_RANGE(TOUCH_MINOR);
DUMP_MOTION_RANGE(TOOL_MAJOR);
DUMP_MOTION_RANGE(TOOL_MINOR);
DUMP_MOTION_RANGE(ORIENTATION);
}
}
#undef DUMP_MOTION_RANGE
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// --- InputReaderThread ---
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputReaderThread::InputReaderThread(const sp<InputReaderInterface>& reader) :
Thread(/*canCallJava*/ true), mReader(reader) {
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputReaderThread::~InputReaderThread() {
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
bool InputReaderThread::threadLoop() {
mReader->loopOnce();
return true;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// --- InputDevice ---
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputDevice::InputDevice(InputReaderContext* context, int32_t id, const String8& name) :
mContext(context), mId(id), mName(name), mSources(0) {
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputDevice::~InputDevice() {
size_t numMappers = mMappers.size();
for (size_t i = 0; i < numMappers; i++) {
delete mMappers[i];
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mMappers.clear();
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputDevice::addMapper(InputMapper* mapper) {
mMappers.add(mapper);
}
Native input event dispatching. Target identification is now fully native. Fixed a couple of minor issues related to input injection. Native input enabled by default, can be disabled by setting WindowManagerPolicy.ENABLE_NATIVE_INPUT_DISPATCH to false. Change-Id: I7edf66ed3e987cc9306ad4743ac57a116af452ff
2010-06-22 08:27:15 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputDevice::configure() {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
if (! isIgnored()) {
mContext->getPolicy()->getInputDeviceCalibration(mName, mCalibration);
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mSources = 0;
size_t numMappers = mMappers.size();
for (size_t i = 0; i < numMappers; i++) {
InputMapper* mapper = mMappers[i];
mapper->configure();
mSources |= mapper->getSources();
Native input event dispatching. Target identification is now fully native. Fixed a couple of minor issues related to input injection. Native input enabled by default, can be disabled by setting WindowManagerPolicy.ENABLE_NATIVE_INPUT_DISPATCH to false. Change-Id: I7edf66ed3e987cc9306ad4743ac57a116af452ff
2010-06-22 08:27:15 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Native input event dispatching. Target identification is now fully native. Fixed a couple of minor issues related to input injection. Native input enabled by default, can be disabled by setting WindowManagerPolicy.ENABLE_NATIVE_INPUT_DISPATCH to false. Change-Id: I7edf66ed3e987cc9306ad4743ac57a116af452ff
2010-06-22 08:27:15 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputDevice::reset() {
size_t numMappers = mMappers.size();
for (size_t i = 0; i < numMappers; i++) {
InputMapper* mapper = mMappers[i];
mapper->reset();
Native input event dispatching. Target identification is now fully native. Fixed a couple of minor issues related to input injection. Native input enabled by default, can be disabled by setting WindowManagerPolicy.ENABLE_NATIVE_INPUT_DISPATCH to false. Change-Id: I7edf66ed3e987cc9306ad4743ac57a116af452ff
2010-06-22 08:27:15 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Native input event dispatching. Target identification is now fully native. Fixed a couple of minor issues related to input injection. Native input enabled by default, can be disabled by setting WindowManagerPolicy.ENABLE_NATIVE_INPUT_DISPATCH to false. Change-Id: I7edf66ed3e987cc9306ad4743ac57a116af452ff
2010-06-22 08:27:15 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputDevice::process(const RawEvent* rawEvent) {
size_t numMappers = mMappers.size();
for (size_t i = 0; i < numMappers; i++) {
InputMapper* mapper = mMappers[i];
mapper->process(rawEvent);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputDevice::getDeviceInfo(InputDeviceInfo* outDeviceInfo) {
outDeviceInfo->initialize(mId, mName);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
size_t numMappers = mMappers.size();
for (size_t i = 0; i < numMappers; i++) {
InputMapper* mapper = mMappers[i];
mapper->populateDeviceInfo(outDeviceInfo);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputDevice::getKeyCodeState(uint32_t sourceMask, int32_t keyCode) {
return getState(sourceMask, keyCode, & InputMapper::getKeyCodeState);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputDevice::getScanCodeState(uint32_t sourceMask, int32_t scanCode) {
return getState(sourceMask, scanCode, & InputMapper::getScanCodeState);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputDevice::getSwitchState(uint32_t sourceMask, int32_t switchCode) {
return getState(sourceMask, switchCode, & InputMapper::getSwitchState);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputDevice::getState(uint32_t sourceMask, int32_t code, GetStateFunc getStateFunc) {
int32_t result = AKEY_STATE_UNKNOWN;
size_t numMappers = mMappers.size();
for (size_t i = 0; i < numMappers; i++) {
InputMapper* mapper = mMappers[i];
if (sourcesMatchMask(mapper->getSources(), sourceMask)) {
result = (mapper->*getStateFunc)(sourceMask, code);
if (result >= AKEY_STATE_DOWN) {
return result;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return result;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
bool InputDevice::markSupportedKeyCodes(uint32_t sourceMask, size_t numCodes,
const int32_t* keyCodes, uint8_t* outFlags) {
bool result = false;
size_t numMappers = mMappers.size();
for (size_t i = 0; i < numMappers; i++) {
InputMapper* mapper = mMappers[i];
if (sourcesMatchMask(mapper->getSources(), sourceMask)) {
result |= mapper->markSupportedKeyCodes(sourceMask, numCodes, keyCodes, outFlags);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return result;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputDevice::getMetaState() {
int32_t result = 0;
size_t numMappers = mMappers.size();
for (size_t i = 0; i < numMappers; i++) {
InputMapper* mapper = mMappers[i];
result |= mapper->getMetaState();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return result;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// --- InputMapper ---
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputMapper::InputMapper(InputDevice* device) :
mDevice(device), mContext(device->getContext()) {
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputMapper::~InputMapper() {
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputMapper::populateDeviceInfo(InputDeviceInfo* info) {
info->addSource(getSources());
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputMapper::configure() {
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void InputMapper::reset() {
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputMapper::getKeyCodeState(uint32_t sourceMask, int32_t keyCode) {
return AKEY_STATE_UNKNOWN;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputMapper::getScanCodeState(uint32_t sourceMask, int32_t scanCode) {
return AKEY_STATE_UNKNOWN;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputMapper::getSwitchState(uint32_t sourceMask, int32_t switchCode) {
return AKEY_STATE_UNKNOWN;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
bool InputMapper::markSupportedKeyCodes(uint32_t sourceMask, size_t numCodes,
const int32_t* keyCodes, uint8_t* outFlags) {
return false;
}
Native input event dispatching. Target identification is now fully native. Fixed a couple of minor issues related to input injection. Native input enabled by default, can be disabled by setting WindowManagerPolicy.ENABLE_NATIVE_INPUT_DISPATCH to false. Change-Id: I7edf66ed3e987cc9306ad4743ac57a116af452ff
2010-06-22 08:27:15 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t InputMapper::getMetaState() {
return 0;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
bool InputMapper::applyStandardPolicyActions(nsecs_t when, int32_t policyActions) {
return policyActions & InputReaderPolicyInterface::ACTION_DISPATCH;
}
Native input event dispatching. Target identification is now fully native. Fixed a couple of minor issues related to input injection. Native input enabled by default, can be disabled by setting WindowManagerPolicy.ENABLE_NATIVE_INPUT_DISPATCH to false. Change-Id: I7edf66ed3e987cc9306ad4743ac57a116af452ff
2010-06-22 08:27:15 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// --- SwitchInputMapper ---
SwitchInputMapper::SwitchInputMapper(InputDevice* device) :
InputMapper(device) {
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
SwitchInputMapper::~SwitchInputMapper() {
}
uint32_t SwitchInputMapper::getSources() {
return 0;
}
void SwitchInputMapper::process(const RawEvent* rawEvent) {
switch (rawEvent->type) {
case EV_SW:
processSwitch(rawEvent->when, rawEvent->scanCode, rawEvent->value);
break;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void SwitchInputMapper::processSwitch(nsecs_t when, int32_t switchCode, int32_t switchValue) {
uint32_t policyFlags = 0;
int32_t policyActions = getPolicy()->interceptSwitch(
when, switchCode, switchValue, policyFlags);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
applyStandardPolicyActions(when, policyActions);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t SwitchInputMapper::getSwitchState(uint32_t sourceMask, int32_t switchCode) {
return getEventHub()->getSwitchState(getDeviceId(), switchCode);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// --- KeyboardInputMapper ---
KeyboardInputMapper::KeyboardInputMapper(InputDevice* device, int32_t associatedDisplayId,
uint32_t sources, int32_t keyboardType) :
InputMapper(device), mAssociatedDisplayId(associatedDisplayId), mSources(sources),
mKeyboardType(keyboardType) {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
initializeLocked();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
KeyboardInputMapper::~KeyboardInputMapper() {
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
void KeyboardInputMapper::initializeLocked() {
mLocked.metaState = AMETA_NONE;
mLocked.downTime = 0;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
uint32_t KeyboardInputMapper::getSources() {
return mSources;
}
void KeyboardInputMapper::populateDeviceInfo(InputDeviceInfo* info) {
InputMapper::populateDeviceInfo(info);
info->setKeyboardType(mKeyboardType);
}
void KeyboardInputMapper::reset() {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
for (;;) {
int32_t keyCode, scanCode;
{ // acquire lock
AutoMutex _l(mLock);
// Synthesize key up event on reset if keys are currently down.
if (mLocked.keyDowns.isEmpty()) {
initializeLocked();
break; // done
}
const KeyDown& keyDown = mLocked.keyDowns.top();
keyCode = keyDown.keyCode;
scanCode = keyDown.scanCode;
} // release lock
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
nsecs_t when = systemTime(SYSTEM_TIME_MONOTONIC);
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
processKey(when, false, keyCode, scanCode, 0);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
InputMapper::reset();
getContext()->updateGlobalMetaState();
}
void KeyboardInputMapper::process(const RawEvent* rawEvent) {
switch (rawEvent->type) {
case EV_KEY: {
int32_t scanCode = rawEvent->scanCode;
if (isKeyboardOrGamepadKey(scanCode)) {
processKey(rawEvent->when, rawEvent->value != 0, rawEvent->keyCode, scanCode,
rawEvent->flags);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
break;
}
}
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
bool KeyboardInputMapper::isKeyboardOrGamepadKey(int32_t scanCode) {
return scanCode < BTN_MOUSE
|| scanCode >= KEY_OK
|| (scanCode >= BTN_GAMEPAD && scanCode < BTN_DIGI);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
void KeyboardInputMapper::processKey(nsecs_t when, bool down, int32_t keyCode,
int32_t scanCode, uint32_t policyFlags) {
int32_t newMetaState;
nsecs_t downTime;
bool metaStateChanged = false;
{ // acquire lock
AutoMutex _l(mLock);
if (down) {
// Rotate key codes according to orientation if needed.
// Note: getDisplayInfo is non-reentrant so we can continue holding the lock.
if (mAssociatedDisplayId >= 0) {
int32_t orientation;
if (! getPolicy()->getDisplayInfo(mAssociatedDisplayId, NULL, NULL, & orientation)) {
return;
}
keyCode = rotateKeyCode(keyCode, orientation);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
// Add key down.
ssize_t keyDownIndex = findKeyDownLocked(scanCode);
if (keyDownIndex >= 0) {
// key repeat, be sure to use same keycode as before in case of rotation
keyCode = mLocked.keyDowns.top().keyCode;
} else {
// key down
mLocked.keyDowns.push();
KeyDown& keyDown = mLocked.keyDowns.editTop();
keyDown.keyCode = keyCode;
keyDown.scanCode = scanCode;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.downTime = when;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
} else {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
// Remove key down.
ssize_t keyDownIndex = findKeyDownLocked(scanCode);
if (keyDownIndex >= 0) {
// key up, be sure to use same keycode as before in case of rotation
keyCode = mLocked.keyDowns.top().keyCode;
mLocked.keyDowns.removeAt(size_t(keyDownIndex));
} else {
// key was not actually down
LOGI("Dropping key up from device %s because the key was not down. "
"keyCode=%d, scanCode=%d",
getDeviceName().string(), keyCode, scanCode);
return;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
int32_t oldMetaState = mLocked.metaState;
newMetaState = updateMetaState(keyCode, down, oldMetaState);
if (oldMetaState != newMetaState) {
mLocked.metaState = newMetaState;
metaStateChanged = true;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
downTime = mLocked.downTime;
} // release lock
if (metaStateChanged) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
getContext()->updateGlobalMetaState();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
applyPolicyAndDispatch(when, policyFlags, down, keyCode, scanCode, newMetaState, downTime);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
void KeyboardInputMapper::applyPolicyAndDispatch(nsecs_t when, uint32_t policyFlags, bool down,
int32_t keyCode, int32_t scanCode, int32_t metaState, nsecs_t downTime) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t policyActions = getPolicy()->interceptKey(when,
getDeviceId(), down, keyCode, scanCode, policyFlags);
Fix native input dispatch in the emulator. Set a default orientation of ROTATION_0. Added some more careful checks based on whether we have valid absolute axis information from the driver. Reset key repeating during configuration changes since the keyboard device may have been removed. Change-Id: I685960828acffcb17595fc5683309e8064a76714
2010-06-29 23:52:21 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (! applyStandardPolicyActions(when, policyActions)) {
return; // event dropped
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
int32_t keyEventAction = down ? AKEY_EVENT_ACTION_DOWN : AKEY_EVENT_ACTION_UP;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t keyEventFlags = AKEY_EVENT_FLAG_FROM_SYSTEM;
if (policyFlags & POLICY_FLAG_WOKE_HERE) {
keyEventFlags = keyEventFlags | AKEY_EVENT_FLAG_WOKE_HERE;
}
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
getDispatcher()->notifyKey(when, getDeviceId(), AINPUT_SOURCE_KEYBOARD, policyFlags,
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
keyEventAction, keyEventFlags, keyCode, scanCode, metaState, downTime);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
ssize_t KeyboardInputMapper::findKeyDownLocked(int32_t scanCode) {
size_t n = mLocked.keyDowns.size();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
for (size_t i = 0; i < n; i++) {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
if (mLocked.keyDowns[i].scanCode == scanCode) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return i;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
return -1;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t KeyboardInputMapper::getKeyCodeState(uint32_t sourceMask, int32_t keyCode) {
return getEventHub()->getKeyCodeState(getDeviceId(), keyCode);
}
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t KeyboardInputMapper::getScanCodeState(uint32_t sourceMask, int32_t scanCode) {
return getEventHub()->getScanCodeState(getDeviceId(), scanCode);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
bool KeyboardInputMapper::markSupportedKeyCodes(uint32_t sourceMask, size_t numCodes,
const int32_t* keyCodes, uint8_t* outFlags) {
return getEventHub()->markSupportedKeyCodes(getDeviceId(), numCodes, keyCodes, outFlags);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t KeyboardInputMapper::getMetaState() {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
{ // acquire lock
AutoMutex _l(mLock);
return mLocked.metaState;
} // release lock
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Fix individual pointer id up/down reporting. Fix a minor threading bug in InputManager dump. Change-Id: Ic2eecf7df5a8dc9f40561fcb03ebe58a2c073778
2010-07-16 22:01:56 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// --- TrackballInputMapper ---
TrackballInputMapper::TrackballInputMapper(InputDevice* device, int32_t associatedDisplayId) :
InputMapper(device), mAssociatedDisplayId(associatedDisplayId) {
mXPrecision = TRACKBALL_MOVEMENT_THRESHOLD;
mYPrecision = TRACKBALL_MOVEMENT_THRESHOLD;
mXScale = 1.0f / TRACKBALL_MOVEMENT_THRESHOLD;
mYScale = 1.0f / TRACKBALL_MOVEMENT_THRESHOLD;
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
initializeLocked();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
TrackballInputMapper::~TrackballInputMapper() {
}
uint32_t TrackballInputMapper::getSources() {
return AINPUT_SOURCE_TRACKBALL;
}
void TrackballInputMapper::populateDeviceInfo(InputDeviceInfo* info) {
InputMapper::populateDeviceInfo(info);
info->addMotionRange(AINPUT_MOTION_RANGE_X, -1.0f, 1.0f, 0.0f, mXScale);
info->addMotionRange(AINPUT_MOTION_RANGE_Y, -1.0f, 1.0f, 0.0f, mYScale);
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
void TrackballInputMapper::initializeLocked() {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mAccumulator.clear();
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.down = false;
mLocked.downTime = 0;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
void TrackballInputMapper::reset() {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
for (;;) {
{ // acquire lock
AutoMutex _l(mLock);
if (! mLocked.down) {
initializeLocked();
break; // done
}
} // release lock
// Synthesize trackball button up event on reset.
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
nsecs_t when = systemTime(SYSTEM_TIME_MONOTONIC);
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mAccumulator.fields = Accumulator::FIELD_BTN_MOUSE;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mAccumulator.btnMouse = false;
sync(when);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
InputMapper::reset();
}
void TrackballInputMapper::process(const RawEvent* rawEvent) {
switch (rawEvent->type) {
case EV_KEY:
switch (rawEvent->scanCode) {
case BTN_MOUSE:
mAccumulator.fields |= Accumulator::FIELD_BTN_MOUSE;
mAccumulator.btnMouse = rawEvent->value != 0;
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
// Sync now since BTN_MOUSE is not necessarily followed by SYN_REPORT and
// we need to ensure that we report the up/down promptly.
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
sync(rawEvent->when);
break;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
break;
case EV_REL:
switch (rawEvent->scanCode) {
case REL_X:
mAccumulator.fields |= Accumulator::FIELD_REL_X;
mAccumulator.relX = rawEvent->value;
break;
case REL_Y:
mAccumulator.fields |= Accumulator::FIELD_REL_Y;
mAccumulator.relY = rawEvent->value;
break;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
break;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
case EV_SYN:
switch (rawEvent->scanCode) {
case SYN_REPORT:
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
sync(rawEvent->when);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
break;
}
break;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void TrackballInputMapper::sync(nsecs_t when) {
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
uint32_t fields = mAccumulator.fields;
if (fields == 0) {
return; // no new state changes, so nothing to do
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
int motionEventAction;
PointerCoords pointerCoords;
nsecs_t downTime;
{ // acquire lock
AutoMutex _l(mLock);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
bool downChanged = fields & Accumulator::FIELD_BTN_MOUSE;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
if (downChanged) {
if (mAccumulator.btnMouse) {
mLocked.down = true;
mLocked.downTime = when;
} else {
mLocked.down = false;
}
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
downTime = mLocked.downTime;
float x = fields & Accumulator::FIELD_REL_X ? mAccumulator.relX * mXScale : 0.0f;
float y = fields & Accumulator::FIELD_REL_Y ? mAccumulator.relY * mYScale : 0.0f;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
if (downChanged) {
motionEventAction = mLocked.down ? AMOTION_EVENT_ACTION_DOWN : AMOTION_EVENT_ACTION_UP;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
} else {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
motionEventAction = AMOTION_EVENT_ACTION_MOVE;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
pointerCoords.x = x;
pointerCoords.y = y;
pointerCoords.pressure = mLocked.down ? 1.0f : 0.0f;
pointerCoords.size = 0;
pointerCoords.touchMajor = 0;
pointerCoords.touchMinor = 0;
pointerCoords.toolMajor = 0;
pointerCoords.toolMinor = 0;
pointerCoords.orientation = 0;
if (mAssociatedDisplayId >= 0 && (x != 0.0f || y != 0.0f)) {
// Rotate motion based on display orientation if needed.
// Note: getDisplayInfo is non-reentrant so we can continue holding the lock.
int32_t orientation;
if (! getPolicy()->getDisplayInfo(mAssociatedDisplayId, NULL, NULL, & orientation)) {
return;
}
float temp;
switch (orientation) {
case InputReaderPolicyInterface::ROTATION_90:
temp = pointerCoords.x;
pointerCoords.x = pointerCoords.y;
pointerCoords.y = - temp;
break;
case InputReaderPolicyInterface::ROTATION_180:
pointerCoords.x = - pointerCoords.x;
pointerCoords.y = - pointerCoords.y;
break;
case InputReaderPolicyInterface::ROTATION_270:
temp = pointerCoords.x;
pointerCoords.x = - pointerCoords.y;
pointerCoords.y = temp;
break;
}
}
} // release lock
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
applyPolicyAndDispatch(when, motionEventAction, & pointerCoords, downTime);
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
mAccumulator.clear();
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
}
void TrackballInputMapper::applyPolicyAndDispatch(nsecs_t when, int32_t motionEventAction,
PointerCoords* pointerCoords, nsecs_t downTime) {
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
uint32_t policyFlags = 0;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t policyActions = getPolicy()->interceptGeneric(when, policyFlags);
if (! applyStandardPolicyActions(when, policyActions)) {
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
return; // event dropped
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
int32_t metaState = mContext->getGlobalMetaState();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
int32_t pointerId = 0;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
getDispatcher()->notifyMotion(when, getDeviceId(), AINPUT_SOURCE_TRACKBALL, policyFlags,
Add support for secure views. Added the MotionEvent.FLAG_WINDOW_IS_OBSCURED flag which is set by the input manager whenever another visible window is partly or wholly obscured the target of a touch event so that applications can filter touches accordingly. Added a "filterTouchesWhenObscured" attribute to View which can be used to enable filtering of touches when the view's window is obscured. Change-Id: I936d9c85013fd2d77fb296a600528d30a29027d2
2010-09-02 00:01:00 +00:00
motionEventAction, 0, metaState, AMOTION_EVENT_EDGE_FLAG_NONE,
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
1, & pointerId, pointerCoords, mXPrecision, mYPrecision, downTime);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Fix safe mode and KeyEvent.getMaxKeyCode(). Bug: 2901731 Change-Id: I78617c1b9dee3790fc590e5af4b5083368873184
2010-08-10 22:47:53 +00:00
int32_t TrackballInputMapper::getScanCodeState(uint32_t sourceMask, int32_t scanCode) {
if (scanCode >= BTN_MOUSE && scanCode < BTN_JOYSTICK) {
return getEventHub()->getScanCodeState(getDeviceId(), scanCode);
} else {
return AKEY_STATE_UNKNOWN;
}
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// --- TouchInputMapper ---
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
TouchInputMapper::TouchInputMapper(InputDevice* device, int32_t associatedDisplayId) :
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
InputMapper(device), mAssociatedDisplayId(associatedDisplayId) {
mLocked.surfaceOrientation = -1;
mLocked.surfaceWidth = -1;
mLocked.surfaceHeight = -1;
initializeLocked();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
TouchInputMapper::~TouchInputMapper() {
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
uint32_t TouchInputMapper::getSources() {
return mAssociatedDisplayId >= 0 ? AINPUT_SOURCE_TOUCHSCREEN : AINPUT_SOURCE_TOUCHPAD;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void TouchInputMapper::populateDeviceInfo(InputDeviceInfo* info) {
InputMapper::populateDeviceInfo(info);
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
{ // acquire lock
AutoMutex _l(mLock);
// Ensure surface information is up to date so that orientation changes are
// noticed immediately.
configureSurfaceLocked();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
info->addMotionRange(AINPUT_MOTION_RANGE_X, mLocked.orientedRanges.x);
info->addMotionRange(AINPUT_MOTION_RANGE_Y, mLocked.orientedRanges.y);
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
if (mLocked.orientedRanges.havePressure) {
info->addMotionRange(AINPUT_MOTION_RANGE_PRESSURE,
mLocked.orientedRanges.pressure);
}
if (mLocked.orientedRanges.haveSize) {
info->addMotionRange(AINPUT_MOTION_RANGE_SIZE,
mLocked.orientedRanges.size);
}
if (mLocked.orientedRanges.haveTouchArea) {
info->addMotionRange(AINPUT_MOTION_RANGE_TOUCH_MAJOR,
mLocked.orientedRanges.touchMajor);
info->addMotionRange(AINPUT_MOTION_RANGE_TOUCH_MINOR,
mLocked.orientedRanges.touchMinor);
}
if (mLocked.orientedRanges.haveToolArea) {
info->addMotionRange(AINPUT_MOTION_RANGE_TOOL_MAJOR,
mLocked.orientedRanges.toolMajor);
info->addMotionRange(AINPUT_MOTION_RANGE_TOOL_MINOR,
mLocked.orientedRanges.toolMinor);
}
if (mLocked.orientedRanges.haveOrientation) {
info->addMotionRange(AINPUT_MOTION_RANGE_ORIENTATION,
mLocked.orientedRanges.orientation);
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
} // release lock
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
void TouchInputMapper::initializeLocked() {
mCurrentTouch.clear();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mLastTouch.clear();
mDownTime = 0;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
for (uint32_t i = 0; i < MAX_POINTERS; i++) {
mAveragingTouchFilter.historyStart[i] = 0;
mAveragingTouchFilter.historyEnd[i] = 0;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mJumpyTouchFilter.jumpyPointsDropped = 0;
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.currentVirtualKey.down = false;
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
mLocked.orientedRanges.havePressure = false;
mLocked.orientedRanges.haveSize = false;
mLocked.orientedRanges.haveTouchArea = false;
mLocked.orientedRanges.haveToolArea = false;
mLocked.orientedRanges.haveOrientation = false;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void TouchInputMapper::configure() {
InputMapper::configure();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Configure basic parameters.
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
configureParameters();
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
logParameters();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Configure absolute axis information.
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
configureRawAxes();
logRawAxes();
// Prepare input device calibration.
parseCalibration();
resolveCalibration();
logCalibration();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
{ // acquire lock
AutoMutex _l(mLock);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Configure surface dimensions and orientation.
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
configureSurfaceLocked();
} // release lock
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
void TouchInputMapper::configureParameters() {
mParameters.useBadTouchFilter = getPolicy()->filterTouchEvents();
mParameters.useAveragingTouchFilter = getPolicy()->filterTouchEvents();
mParameters.useJumpyTouchFilter = getPolicy()->filterJumpyTouchEvents();
}
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
void TouchInputMapper::logParameters() {
if (mParameters.useBadTouchFilter) {
LOGI(INDENT "Bad touch filter enabled.");
}
if (mParameters.useAveragingTouchFilter) {
LOGI(INDENT "Averaging touch filter enabled.");
}
if (mParameters.useJumpyTouchFilter) {
LOGI(INDENT "Jumpy touch filter enabled.");
}
}
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
void TouchInputMapper::configureRawAxes() {
mRawAxes.x.clear();
mRawAxes.y.clear();
mRawAxes.pressure.clear();
mRawAxes.touchMajor.clear();
mRawAxes.touchMinor.clear();
mRawAxes.toolMajor.clear();
mRawAxes.toolMinor.clear();
mRawAxes.orientation.clear();
}
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
static void logAxisInfo(RawAbsoluteAxisInfo axis, const char* name) {
if (axis.valid) {
LOGI(INDENT "Raw %s axis: min=%d, max=%d, flat=%d, fuzz=%d",
name, axis.minValue, axis.maxValue, axis.flat, axis.fuzz);
} else {
LOGI(INDENT "Raw %s axis: unknown range", name);
}
}
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
void TouchInputMapper::logRawAxes() {
logAxisInfo(mRawAxes.x, "x");
logAxisInfo(mRawAxes.y, "y");
logAxisInfo(mRawAxes.pressure, "pressure");
logAxisInfo(mRawAxes.touchMajor, "touchMajor");
logAxisInfo(mRawAxes.touchMinor, "touchMinor");
logAxisInfo(mRawAxes.toolMajor, "toolMajor");
logAxisInfo(mRawAxes.toolMinor, "toolMinor");
logAxisInfo(mRawAxes.orientation, "orientation");
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
bool TouchInputMapper::configureSurfaceLocked() {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Update orientation and dimensions if needed.
int32_t orientation;
int32_t width, height;
if (mAssociatedDisplayId >= 0) {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
// Note: getDisplayInfo is non-reentrant so we can continue holding the lock.
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (! getPolicy()->getDisplayInfo(mAssociatedDisplayId, & width, & height, & orientation)) {
return false;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
} else {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
orientation = InputReaderPolicyInterface::ROTATION_0;
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
width = mRawAxes.x.getRange();
height = mRawAxes.y.getRange();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
bool orientationChanged = mLocked.surfaceOrientation != orientation;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (orientationChanged) {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.surfaceOrientation = orientation;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
bool sizeChanged = mLocked.surfaceWidth != width || mLocked.surfaceHeight != height;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (sizeChanged) {
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI("Device reconfigured (display size changed): id=0x%x, name=%s",
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
getDeviceId(), getDeviceName().string());
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT "Width: %dpx", width);
LOGI(INDENT "Height: %dpx", height);
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.surfaceWidth = width;
mLocked.surfaceHeight = height;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Configure X and Y factors.
if (mRawAxes.x.valid && mRawAxes.y.valid) {
mLocked.xOrigin = mRawAxes.x.minValue;
mLocked.yOrigin = mRawAxes.y.minValue;
mLocked.xScale = float(width) / mRawAxes.x.getRange();
mLocked.yScale = float(height) / mRawAxes.y.getRange();
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.xPrecision = 1.0f / mLocked.xScale;
mLocked.yPrecision = 1.0f / mLocked.yScale;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
configureVirtualKeysLocked();
Fix native input dispatch in the emulator. Set a default orientation of ROTATION_0. Added some more careful checks based on whether we have valid absolute axis information from the driver. Reset key repeating during configuration changes since the keyboard device may have been removed. Change-Id: I685960828acffcb17595fc5683309e8064a76714
2010-06-29 23:52:21 +00:00
} else {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
LOGW(INDENT "Touch device did not report support for X or Y axis!");
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.xOrigin = 0;
mLocked.yOrigin = 0;
mLocked.xScale = 1.0f;
mLocked.yScale = 1.0f;
mLocked.xPrecision = 1.0f;
mLocked.yPrecision = 1.0f;
Fix native input dispatch in the emulator. Set a default orientation of ROTATION_0. Added some more careful checks based on whether we have valid absolute axis information from the driver. Reset key repeating during configuration changes since the keyboard device may have been removed. Change-Id: I685960828acffcb17595fc5683309e8064a76714
2010-06-29 23:52:21 +00:00
}
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Scale factor for terms that are not oriented in a particular axis.
// If the pixels are square then xScale == yScale otherwise we fake it
// by choosing an average.
mLocked.geometricScale = avg(mLocked.xScale, mLocked.yScale);
// Size of diagonal axis.
float diagonalSize = pythag(width, height);
// TouchMajor and TouchMinor factors.
if (mCalibration.touchAreaCalibration != Calibration::TOUCH_AREA_CALIBRATION_NONE) {
mLocked.orientedRanges.haveTouchArea = true;
mLocked.orientedRanges.touchMajor.min = 0;
mLocked.orientedRanges.touchMajor.max = diagonalSize;
mLocked.orientedRanges.touchMajor.flat = 0;
mLocked.orientedRanges.touchMajor.fuzz = 0;
mLocked.orientedRanges.touchMinor = mLocked.orientedRanges.touchMajor;
}
// ToolMajor and ToolMinor factors.
if (mCalibration.toolAreaCalibration != Calibration::TOOL_AREA_CALIBRATION_NONE) {
mLocked.toolAreaLinearScale = 0;
mLocked.toolAreaLinearBias = 0;
if (mCalibration.toolAreaCalibration == Calibration::TOOL_AREA_CALIBRATION_LINEAR) {
if (mCalibration.haveToolAreaLinearScale) {
mLocked.toolAreaLinearScale = mCalibration.toolAreaLinearScale;
} else if (mRawAxes.toolMajor.valid && mRawAxes.toolMajor.maxValue != 0) {
mLocked.toolAreaLinearScale = float(min(width, height))
/ mRawAxes.toolMajor.maxValue;
}
if (mCalibration.haveToolAreaLinearBias) {
mLocked.toolAreaLinearBias = mCalibration.toolAreaLinearBias;
}
}
mLocked.orientedRanges.haveToolArea = true;
mLocked.orientedRanges.toolMajor.min = 0;
mLocked.orientedRanges.toolMajor.max = diagonalSize;
mLocked.orientedRanges.toolMajor.flat = 0;
mLocked.orientedRanges.toolMajor.fuzz = 0;
mLocked.orientedRanges.toolMinor = mLocked.orientedRanges.toolMajor;
}
// Pressure factors.
if (mCalibration.pressureCalibration != Calibration::PRESSURE_CALIBRATION_NONE) {
RawAbsoluteAxisInfo rawPressureAxis;
switch (mCalibration.pressureSource) {
case Calibration::PRESSURE_SOURCE_PRESSURE:
rawPressureAxis = mRawAxes.pressure;
break;
case Calibration::PRESSURE_SOURCE_TOUCH:
rawPressureAxis = mRawAxes.touchMajor;
break;
default:
rawPressureAxis.clear();
}
mLocked.pressureScale = 0;
if (mCalibration.pressureCalibration == Calibration::PRESSURE_CALIBRATION_PHYSICAL
|| mCalibration.pressureCalibration
== Calibration::PRESSURE_CALIBRATION_AMPLITUDE) {
if (mCalibration.havePressureScale) {
mLocked.pressureScale = mCalibration.pressureScale;
} else if (rawPressureAxis.valid && rawPressureAxis.maxValue != 0) {
mLocked.pressureScale = 1.0f / rawPressureAxis.maxValue;
}
}
mLocked.orientedRanges.havePressure = true;
mLocked.orientedRanges.pressure.min = 0;
mLocked.orientedRanges.pressure.max = 1.0;
mLocked.orientedRanges.pressure.flat = 0;
mLocked.orientedRanges.pressure.fuzz = 0;
}
// Size factors.
if (mCalibration.sizeCalibration != Calibration::SIZE_CALIBRATION_NONE) {
mLocked.sizeScale = 0;
if (mCalibration.sizeCalibration == Calibration::SIZE_CALIBRATION_NORMALIZED) {
if (mRawAxes.toolMajor.valid && mRawAxes.toolMajor.maxValue != 0) {
mLocked.sizeScale = 1.0f / mRawAxes.toolMajor.maxValue;
}
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
mLocked.orientedRanges.haveSize = true;
mLocked.orientedRanges.size.min = 0;
mLocked.orientedRanges.size.max = 1.0;
mLocked.orientedRanges.size.flat = 0;
mLocked.orientedRanges.size.fuzz = 0;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Orientation
if (mCalibration.orientationCalibration != Calibration::ORIENTATION_CALIBRATION_NONE) {
mLocked.orientationScale = 0;
if (mCalibration.orientationCalibration
== Calibration::ORIENTATION_CALIBRATION_INTERPOLATED) {
if (mRawAxes.orientation.valid && mRawAxes.orientation.maxValue != 0) {
mLocked.orientationScale = float(M_PI_2) / mRawAxes.orientation.maxValue;
}
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
mLocked.orientedRanges.orientation.min = - M_PI_2;
mLocked.orientedRanges.orientation.max = M_PI_2;
mLocked.orientedRanges.orientation.flat = 0;
mLocked.orientedRanges.orientation.fuzz = 0;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
if (orientationChanged || sizeChanged) {
// Compute oriented surface dimensions, precision, and scales.
float orientedXScale, orientedYScale;
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
switch (mLocked.surfaceOrientation) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
case InputReaderPolicyInterface::ROTATION_90:
case InputReaderPolicyInterface::ROTATION_270:
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.orientedSurfaceWidth = mLocked.surfaceHeight;
mLocked.orientedSurfaceHeight = mLocked.surfaceWidth;
mLocked.orientedXPrecision = mLocked.yPrecision;
mLocked.orientedYPrecision = mLocked.xPrecision;
orientedXScale = mLocked.yScale;
orientedYScale = mLocked.xScale;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
break;
default:
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.orientedSurfaceWidth = mLocked.surfaceWidth;
mLocked.orientedSurfaceHeight = mLocked.surfaceHeight;
mLocked.orientedXPrecision = mLocked.xPrecision;
mLocked.orientedYPrecision = mLocked.yPrecision;
orientedXScale = mLocked.xScale;
orientedYScale = mLocked.yScale;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
break;
Fix native input dispatch in the emulator. Set a default orientation of ROTATION_0. Added some more careful checks based on whether we have valid absolute axis information from the driver. Reset key repeating during configuration changes since the keyboard device may have been removed. Change-Id: I685960828acffcb17595fc5683309e8064a76714
2010-06-29 23:52:21 +00:00
}
Add support for new input sources. Added several new coordinate values to MotionEvents to capture touch major/minor area, tool major/minor area and orientation. Renamed NDK input constants per convention. Added InputDevice class in Java which will eventually provide useful information about available input devices. Added APIs for manufacturing new MotionEvent objects with multiple pointers and all necessary coordinate data. Fixed a bug in the input dispatcher where it could get stuck with a pointer down forever. Fixed a bug in the WindowManager where the input window list could end up containing stale removed windows. Fixed a bug in the WindowManager where the input channel was being removed only after the final animation transition had taken place which caused spurious WINDOW DIED log messages to be printed. Change-Id: Ie55084da319b20aad29b28a0499b8dd98bb5da68
2010-07-15 01:48:53 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Configure position ranges.
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.orientedRanges.x.min = 0;
mLocked.orientedRanges.x.max = mLocked.orientedSurfaceWidth;
mLocked.orientedRanges.x.flat = 0;
mLocked.orientedRanges.x.fuzz = orientedXScale;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.orientedRanges.y.min = 0;
mLocked.orientedRanges.y.max = mLocked.orientedSurfaceHeight;
mLocked.orientedRanges.y.flat = 0;
mLocked.orientedRanges.y.fuzz = orientedYScale;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
if (sizeChanged) {
logMotionRangesLocked();
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return true;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
static void logMotionRangeInfo(InputDeviceInfo::MotionRange* range, const char* name) {
if (range) {
LOGI(INDENT "Output %s range: min=%f, max=%f, flat=%f, fuzz=%f",
name, range->min, range->max, range->flat, range->fuzz);
} else {
LOGI(INDENT "Output %s range: unsupported", name);
}
}
void TouchInputMapper::logMotionRangesLocked() {
logMotionRangeInfo(& mLocked.orientedRanges.x, "x");
logMotionRangeInfo(& mLocked.orientedRanges.y, "y");
logMotionRangeInfo(mLocked.orientedRanges.havePressure
? & mLocked.orientedRanges.pressure : NULL, "pressure");
logMotionRangeInfo(mLocked.orientedRanges.haveSize
? & mLocked.orientedRanges.size : NULL, "size");
logMotionRangeInfo(mLocked.orientedRanges.haveTouchArea
? & mLocked.orientedRanges.touchMajor : NULL, "touchMajor");
logMotionRangeInfo(mLocked.orientedRanges.haveTouchArea
? & mLocked.orientedRanges.touchMinor : NULL, "touchMinor");
logMotionRangeInfo(mLocked.orientedRanges.haveToolArea
? & mLocked.orientedRanges.toolMajor : NULL, "toolMajor");
logMotionRangeInfo(mLocked.orientedRanges.haveToolArea
? & mLocked.orientedRanges.toolMinor : NULL, "toolMinor");
logMotionRangeInfo(mLocked.orientedRanges.haveOrientation
? & mLocked.orientedRanges.orientation : NULL, "orientation");
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
void TouchInputMapper::configureVirtualKeysLocked() {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
assert(mRawAxes.x.valid && mRawAxes.y.valid);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
// Note: getVirtualKeyDefinitions is non-reentrant so we can continue holding the lock.
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
Vector<VirtualKeyDefinition> virtualKeyDefinitions;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
getPolicy()->getVirtualKeyDefinitions(getDeviceName(), virtualKeyDefinitions);
Fix native input dispatch in the emulator. Set a default orientation of ROTATION_0. Added some more careful checks based on whether we have valid absolute axis information from the driver. Reset key repeating during configuration changes since the keyboard device may have been removed. Change-Id: I685960828acffcb17595fc5683309e8064a76714
2010-06-29 23:52:21 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.virtualKeys.clear();
Fix native input dispatch in the emulator. Set a default orientation of ROTATION_0. Added some more careful checks based on whether we have valid absolute axis information from the driver. Reset key repeating during configuration changes since the keyboard device may have been removed. Change-Id: I685960828acffcb17595fc5683309e8064a76714
2010-06-29 23:52:21 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
if (virtualKeyDefinitions.size() == 0) {
return;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.virtualKeys.setCapacity(virtualKeyDefinitions.size());
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
int32_t touchScreenLeft = mRawAxes.x.minValue;
int32_t touchScreenTop = mRawAxes.y.minValue;
int32_t touchScreenWidth = mRawAxes.x.getRange();
int32_t touchScreenHeight = mRawAxes.y.getRange();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
for (size_t i = 0; i < virtualKeyDefinitions.size(); i++) {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
const VirtualKeyDefinition& virtualKeyDefinition =
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
virtualKeyDefinitions[i];
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.virtualKeys.add();
VirtualKey& virtualKey = mLocked.virtualKeys.editTop();
Fix native input dispatch in the emulator. Set a default orientation of ROTATION_0. Added some more careful checks based on whether we have valid absolute axis information from the driver. Reset key repeating during configuration changes since the keyboard device may have been removed. Change-Id: I685960828acffcb17595fc5683309e8064a76714
2010-06-29 23:52:21 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
virtualKey.scanCode = virtualKeyDefinition.scanCode;
int32_t keyCode;
uint32_t flags;
if (getEventHub()->scancodeToKeycode(getDeviceId(), virtualKey.scanCode,
& keyCode, & flags)) {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
LOGW(INDENT "VirtualKey %d: could not obtain key code, ignoring",
virtualKey.scanCode);
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.virtualKeys.pop(); // drop the key
continue;
}
Fix native input dispatch in the emulator. Set a default orientation of ROTATION_0. Added some more careful checks based on whether we have valid absolute axis information from the driver. Reset key repeating during configuration changes since the keyboard device may have been removed. Change-Id: I685960828acffcb17595fc5683309e8064a76714
2010-06-29 23:52:21 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
virtualKey.keyCode = keyCode;
virtualKey.flags = flags;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
// convert the key definition's display coordinates into touch coordinates for a hit box
int32_t halfWidth = virtualKeyDefinition.width / 2;
int32_t halfHeight = virtualKeyDefinition.height / 2;
virtualKey.hitLeft = (virtualKeyDefinition.centerX - halfWidth)
* touchScreenWidth / mLocked.surfaceWidth + touchScreenLeft;
virtualKey.hitRight= (virtualKeyDefinition.centerX + halfWidth)
* touchScreenWidth / mLocked.surfaceWidth + touchScreenLeft;
virtualKey.hitTop = (virtualKeyDefinition.centerY - halfHeight)
* touchScreenHeight / mLocked.surfaceHeight + touchScreenTop;
virtualKey.hitBottom = (virtualKeyDefinition.centerY + halfHeight)
* touchScreenHeight / mLocked.surfaceHeight + touchScreenTop;
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
LOGI(INDENT "VirtualKey %d: keyCode=%d hitLeft=%d hitRight=%d hitTop=%d hitBottom=%d",
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
virtualKey.scanCode, virtualKey.keyCode,
virtualKey.hitLeft, virtualKey.hitRight, virtualKey.hitTop, virtualKey.hitBottom);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
void TouchInputMapper::parseCalibration() {
const InputDeviceCalibration& in = getDevice()->getCalibration();
Calibration& out = mCalibration;
// Touch Area
out.touchAreaCalibration = Calibration::TOUCH_AREA_CALIBRATION_DEFAULT;
String8 touchAreaCalibrationString;
if (in.tryGetProperty(String8("touch.touchArea.calibration"), touchAreaCalibrationString)) {
if (touchAreaCalibrationString == "none") {
out.touchAreaCalibration = Calibration::TOUCH_AREA_CALIBRATION_NONE;
} else if (touchAreaCalibrationString == "geometric") {
out.touchAreaCalibration = Calibration::TOUCH_AREA_CALIBRATION_GEOMETRIC;
} else if (touchAreaCalibrationString == "pressure") {
out.touchAreaCalibration = Calibration::TOUCH_AREA_CALIBRATION_PRESSURE;
} else if (touchAreaCalibrationString != "default") {
LOGW("Invalid value for touch.touchArea.calibration: '%s'",
touchAreaCalibrationString.string());
}
}
// Tool Area
out.toolAreaCalibration = Calibration::TOOL_AREA_CALIBRATION_DEFAULT;
String8 toolAreaCalibrationString;
if (in.tryGetProperty(String8("tool.toolArea.calibration"), toolAreaCalibrationString)) {
if (toolAreaCalibrationString == "none") {
out.toolAreaCalibration = Calibration::TOOL_AREA_CALIBRATION_NONE;
} else if (toolAreaCalibrationString == "geometric") {
out.toolAreaCalibration = Calibration::TOOL_AREA_CALIBRATION_GEOMETRIC;
} else if (toolAreaCalibrationString == "linear") {
out.toolAreaCalibration = Calibration::TOOL_AREA_CALIBRATION_LINEAR;
} else if (toolAreaCalibrationString != "default") {
LOGW("Invalid value for tool.toolArea.calibration: '%s'",
toolAreaCalibrationString.string());
}
}
out.haveToolAreaLinearScale = in.tryGetProperty(String8("touch.toolArea.linearScale"),
out.toolAreaLinearScale);
out.haveToolAreaLinearBias = in.tryGetProperty(String8("touch.toolArea.linearBias"),
out.toolAreaLinearBias);
out.haveToolAreaIsSummed = in.tryGetProperty(String8("touch.toolArea.isSummed"),
out.toolAreaIsSummed);
// Pressure
out.pressureCalibration = Calibration::PRESSURE_CALIBRATION_DEFAULT;
String8 pressureCalibrationString;
if (in.tryGetProperty(String8("tool.pressure.calibration"), pressureCalibrationString)) {
if (pressureCalibrationString == "none") {
out.pressureCalibration = Calibration::PRESSURE_CALIBRATION_NONE;
} else if (pressureCalibrationString == "physical") {
out.pressureCalibration = Calibration::PRESSURE_CALIBRATION_PHYSICAL;
} else if (pressureCalibrationString == "amplitude") {
out.pressureCalibration = Calibration::PRESSURE_CALIBRATION_AMPLITUDE;
} else if (pressureCalibrationString != "default") {
LOGW("Invalid value for tool.pressure.calibration: '%s'",
pressureCalibrationString.string());
}
}
out.pressureSource = Calibration::PRESSURE_SOURCE_DEFAULT;
String8 pressureSourceString;
if (in.tryGetProperty(String8("touch.pressure.source"), pressureSourceString)) {
if (pressureSourceString == "pressure") {
out.pressureSource = Calibration::PRESSURE_SOURCE_PRESSURE;
} else if (pressureSourceString == "touch") {
out.pressureSource = Calibration::PRESSURE_SOURCE_TOUCH;
} else if (pressureSourceString != "default") {
LOGW("Invalid value for touch.pressure.source: '%s'",
pressureSourceString.string());
}
}
out.havePressureScale = in.tryGetProperty(String8("touch.pressure.scale"),
out.pressureScale);
// Size
out.sizeCalibration = Calibration::SIZE_CALIBRATION_DEFAULT;
String8 sizeCalibrationString;
if (in.tryGetProperty(String8("tool.size.calibration"), sizeCalibrationString)) {
if (sizeCalibrationString == "none") {
out.sizeCalibration = Calibration::SIZE_CALIBRATION_NONE;
} else if (sizeCalibrationString == "normalized") {
out.sizeCalibration = Calibration::SIZE_CALIBRATION_NORMALIZED;
} else if (sizeCalibrationString != "default") {
LOGW("Invalid value for tool.size.calibration: '%s'",
sizeCalibrationString.string());
}
}
// Orientation
out.orientationCalibration = Calibration::ORIENTATION_CALIBRATION_DEFAULT;
String8 orientationCalibrationString;
if (in.tryGetProperty(String8("tool.orientation.calibration"), orientationCalibrationString)) {
if (orientationCalibrationString == "none") {
out.orientationCalibration = Calibration::ORIENTATION_CALIBRATION_NONE;
} else if (orientationCalibrationString == "interpolated") {
out.orientationCalibration = Calibration::ORIENTATION_CALIBRATION_INTERPOLATED;
} else if (orientationCalibrationString != "default") {
LOGW("Invalid value for tool.orientation.calibration: '%s'",
orientationCalibrationString.string());
}
}
}
void TouchInputMapper::resolveCalibration() {
// Pressure
switch (mCalibration.pressureSource) {
case Calibration::PRESSURE_SOURCE_DEFAULT:
if (mRawAxes.pressure.valid) {
mCalibration.pressureSource = Calibration::PRESSURE_SOURCE_PRESSURE;
} else if (mRawAxes.touchMajor.valid) {
mCalibration.pressureSource = Calibration::PRESSURE_SOURCE_TOUCH;
}
break;
case Calibration::PRESSURE_SOURCE_PRESSURE:
if (! mRawAxes.pressure.valid) {
LOGW("Calibration property touch.pressure.source is 'pressure' but "
"the pressure axis is not available.");
}
break;
case Calibration::PRESSURE_SOURCE_TOUCH:
if (! mRawAxes.touchMajor.valid) {
LOGW("Calibration property touch.pressure.source is 'touch' but "
"the touchMajor axis is not available.");
}
break;
default:
break;
}
switch (mCalibration.pressureCalibration) {
case Calibration::PRESSURE_CALIBRATION_DEFAULT:
if (mCalibration.pressureSource != Calibration::PRESSURE_SOURCE_DEFAULT) {
mCalibration.pressureCalibration = Calibration::PRESSURE_CALIBRATION_AMPLITUDE;
} else {
mCalibration.pressureCalibration = Calibration::PRESSURE_CALIBRATION_NONE;
}
break;
default:
break;
}
// Tool Area
switch (mCalibration.toolAreaCalibration) {
case Calibration::TOOL_AREA_CALIBRATION_DEFAULT:
if (mRawAxes.toolMajor.valid) {
mCalibration.toolAreaCalibration = Calibration::TOOL_AREA_CALIBRATION_LINEAR;
} else {
mCalibration.toolAreaCalibration = Calibration::TOOL_AREA_CALIBRATION_NONE;
}
break;
default:
break;
}
// Touch Area
switch (mCalibration.touchAreaCalibration) {
case Calibration::TOUCH_AREA_CALIBRATION_DEFAULT:
if (mCalibration.pressureCalibration != Calibration::PRESSURE_CALIBRATION_NONE
&& mCalibration.toolAreaCalibration != Calibration::TOOL_AREA_CALIBRATION_NONE) {
mCalibration.touchAreaCalibration = Calibration::TOUCH_AREA_CALIBRATION_PRESSURE;
} else {
mCalibration.touchAreaCalibration = Calibration::TOUCH_AREA_CALIBRATION_NONE;
}
break;
default:
break;
}
// Size
switch (mCalibration.sizeCalibration) {
case Calibration::SIZE_CALIBRATION_DEFAULT:
if (mRawAxes.toolMajor.valid) {
mCalibration.sizeCalibration = Calibration::SIZE_CALIBRATION_NORMALIZED;
} else {
mCalibration.sizeCalibration = Calibration::SIZE_CALIBRATION_NONE;
}
break;
default:
break;
}
// Orientation
switch (mCalibration.orientationCalibration) {
case Calibration::ORIENTATION_CALIBRATION_DEFAULT:
if (mRawAxes.orientation.valid) {
mCalibration.orientationCalibration = Calibration::ORIENTATION_CALIBRATION_INTERPOLATED;
} else {
mCalibration.orientationCalibration = Calibration::ORIENTATION_CALIBRATION_NONE;
}
break;
default:
break;
}
}
void TouchInputMapper::logCalibration() {
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT "Calibration:");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Touch Area
switch (mCalibration.touchAreaCalibration) {
case Calibration::TOUCH_AREA_CALIBRATION_NONE:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.touchArea.calibration: none");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
case Calibration::TOUCH_AREA_CALIBRATION_GEOMETRIC:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.touchArea.calibration: geometric");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
case Calibration::TOUCH_AREA_CALIBRATION_PRESSURE:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.touchArea.calibration: pressure");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
default:
assert(false);
}
// Tool Area
switch (mCalibration.toolAreaCalibration) {
case Calibration::TOOL_AREA_CALIBRATION_NONE:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.toolArea.calibration: none");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
case Calibration::TOOL_AREA_CALIBRATION_GEOMETRIC:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.toolArea.calibration: geometric");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
case Calibration::TOOL_AREA_CALIBRATION_LINEAR:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.toolArea.calibration: linear");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
default:
assert(false);
}
if (mCalibration.haveToolAreaLinearScale) {
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.toolArea.linearScale: %f", mCalibration.toolAreaLinearScale);
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
}
if (mCalibration.haveToolAreaLinearBias) {
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.toolArea.linearBias: %f", mCalibration.toolAreaLinearBias);
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
}
if (mCalibration.haveToolAreaIsSummed) {
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.toolArea.isSummed: %d", mCalibration.toolAreaIsSummed);
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
}
// Pressure
switch (mCalibration.pressureCalibration) {
case Calibration::PRESSURE_CALIBRATION_NONE:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.pressure.calibration: none");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
case Calibration::PRESSURE_CALIBRATION_PHYSICAL:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.pressure.calibration: physical");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
case Calibration::PRESSURE_CALIBRATION_AMPLITUDE:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.pressure.calibration: amplitude");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
default:
assert(false);
}
switch (mCalibration.pressureSource) {
case Calibration::PRESSURE_SOURCE_PRESSURE:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.pressure.source: pressure");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
case Calibration::PRESSURE_SOURCE_TOUCH:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.pressure.source: touch");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
case Calibration::PRESSURE_SOURCE_DEFAULT:
break;
default:
assert(false);
}
if (mCalibration.havePressureScale) {
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.pressure.scale: %f", mCalibration.pressureScale);
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
}
// Size
switch (mCalibration.sizeCalibration) {
case Calibration::SIZE_CALIBRATION_NONE:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.size.calibration: none");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
case Calibration::SIZE_CALIBRATION_NORMALIZED:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.size.calibration: normalized");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
default:
assert(false);
}
// Orientation
switch (mCalibration.orientationCalibration) {
case Calibration::ORIENTATION_CALIBRATION_NONE:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.orientation.calibration: none");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
case Calibration::ORIENTATION_CALIBRATION_INTERPOLATED:
Input dispatcher ANR handling enhancements. This change is essentially a rewrite of the main input dispatcher loop with the target identification folded in. Since the input dispatcher now has all of the window state, it can make better decisions about when to ANR. Added a .5 second deadline for processing app switch keys. This behavior predates Gingerbread but had not previously been ported. Fixed some timing inaccuracies in the ANR accounting that could cause applications to ANR sooner than they should have. Added a mechanism for tracking key and motion events that have been dispatched to a window so that appropriate cancelation events can be synthesized when recovering from ANR. This change helps to keep applications in sync so they don't end up with stuck buttons upon recovery from ANRs. Added more comments to describe the tricky parts of PollLoop. Change-Id: I13dffca27acb436fc383980db536abc4d8b9e6f1
2010-09-08 18:49:43 +00:00
LOGI(INDENT INDENT "touch.orientation.calibration: interpolated");
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
break;
default:
assert(false);
}
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void TouchInputMapper::reset() {
// Synthesize touch up event if touch is currently down.
// This will also take care of finishing virtual key processing if needed.
if (mLastTouch.pointerCount != 0) {
nsecs_t when = systemTime(SYSTEM_TIME_MONOTONIC);
mCurrentTouch.clear();
syncTouch(when, true);
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
{ // acquire lock
AutoMutex _l(mLock);
initializeLocked();
} // release lock
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
InputMapper::reset();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Fix native input dispatch in the emulator. Set a default orientation of ROTATION_0. Added some more careful checks based on whether we have valid absolute axis information from the driver. Reset key repeating during configuration changes since the keyboard device may have been removed. Change-Id: I685960828acffcb17595fc5683309e8064a76714
2010-06-29 23:52:21 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void TouchInputMapper::syncTouch(nsecs_t when, bool havePointerIds) {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
// Apply generic policy actions.
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
uint32_t policyFlags = 0;
int32_t policyActions = getPolicy()->interceptGeneric(when, policyFlags);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (! applyStandardPolicyActions(when, policyActions)) {
mLastTouch.clear();
return; // event dropped
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
// Preprocess pointer data.
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (mParameters.useBadTouchFilter) {
if (applyBadTouchFilter()) {
havePointerIds = false;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (mParameters.useJumpyTouchFilter) {
if (applyJumpyTouchFilter()) {
havePointerIds = false;
}
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (! havePointerIds) {
calculatePointerIds();
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
TouchData temp;
TouchData* savedTouch;
if (mParameters.useAveragingTouchFilter) {
temp.copyFrom(mCurrentTouch);
savedTouch = & temp;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
applyAveragingTouchFilter();
} else {
savedTouch = & mCurrentTouch;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
// Process touches and virtual keys.
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
TouchResult touchResult = consumeOffScreenTouches(when, policyFlags);
if (touchResult == DISPATCH_TOUCH) {
dispatchTouches(when, policyFlags);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
// Copy current touch to last touch in preparation for the next cycle.
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (touchResult == DROP_STROKE) {
mLastTouch.clear();
} else {
mLastTouch.copyFrom(*savedTouch);
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
}
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
TouchInputMapper::TouchResult TouchInputMapper::consumeOffScreenTouches(
nsecs_t when, uint32_t policyFlags) {
int32_t keyEventAction, keyEventFlags;
int32_t keyCode, scanCode, downTime;
TouchResult touchResult;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
{ // acquire lock
AutoMutex _l(mLock);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
// Update surface size and orientation, including virtual key positions.
if (! configureSurfaceLocked()) {
return DROP_STROKE;
}
// Check for virtual key press.
if (mLocked.currentVirtualKey.down) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (mCurrentTouch.pointerCount == 0) {
// Pointer went up while virtual key was down.
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.currentVirtualKey.down = false;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
#if DEBUG_VIRTUAL_KEYS
LOGD("VirtualKeys: Generating key up: keyCode=%d, scanCode=%d",
mCurrentVirtualKey.keyCode, mCurrentVirtualKey.scanCode);
#endif
keyEventAction = AKEY_EVENT_ACTION_UP;
keyEventFlags = AKEY_EVENT_FLAG_FROM_SYSTEM | AKEY_EVENT_FLAG_VIRTUAL_HARD_KEY;
touchResult = SKIP_TOUCH;
goto DispatchVirtualKey;
}
if (mCurrentTouch.pointerCount == 1) {
int32_t x = mCurrentTouch.pointers[0].x;
int32_t y = mCurrentTouch.pointers[0].y;
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
const VirtualKey* virtualKey = findVirtualKeyHitLocked(x, y);
if (virtualKey && virtualKey->keyCode == mLocked.currentVirtualKey.keyCode) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Pointer is still within the space of the virtual key.
return SKIP_TOUCH;
}
}
// Pointer left virtual key area or another pointer also went down.
// Send key cancellation and drop the stroke so subsequent motions will be
// considered fresh downs. This is useful when the user swipes away from the
// virtual key area into the main display surface.
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.currentVirtualKey.down = false;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
#if DEBUG_VIRTUAL_KEYS
LOGD("VirtualKeys: Canceling key: keyCode=%d, scanCode=%d",
mCurrentVirtualKey.keyCode, mCurrentVirtualKey.scanCode);
#endif
keyEventAction = AKEY_EVENT_ACTION_UP;
keyEventFlags = AKEY_EVENT_FLAG_FROM_SYSTEM | AKEY_EVENT_FLAG_VIRTUAL_HARD_KEY
| AKEY_EVENT_FLAG_CANCELED;
touchResult = DROP_STROKE;
goto DispatchVirtualKey;
} else {
if (mCurrentTouch.pointerCount >= 1 && mLastTouch.pointerCount == 0) {
// Pointer just went down. Handle off-screen touches, if needed.
int32_t x = mCurrentTouch.pointers[0].x;
int32_t y = mCurrentTouch.pointers[0].y;
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
if (! isPointInsideSurfaceLocked(x, y)) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// If exactly one pointer went down, check for virtual key hit.
// Otherwise we will drop the entire stroke.
if (mCurrentTouch.pointerCount == 1) {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
const VirtualKey* virtualKey = findVirtualKeyHitLocked(x, y);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (virtualKey) {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
mLocked.currentVirtualKey.down = true;
mLocked.currentVirtualKey.downTime = when;
mLocked.currentVirtualKey.keyCode = virtualKey->keyCode;
mLocked.currentVirtualKey.scanCode = virtualKey->scanCode;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
#if DEBUG_VIRTUAL_KEYS
LOGD("VirtualKeys: Generating key down: keyCode=%d, scanCode=%d",
mCurrentVirtualKey.keyCode, mCurrentVirtualKey.scanCode);
#endif
keyEventAction = AKEY_EVENT_ACTION_DOWN;
keyEventFlags = AKEY_EVENT_FLAG_FROM_SYSTEM
| AKEY_EVENT_FLAG_VIRTUAL_HARD_KEY;
touchResult = SKIP_TOUCH;
goto DispatchVirtualKey;
}
}
return DROP_STROKE;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return DISPATCH_TOUCH;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
DispatchVirtualKey:
// Collect remaining state needed to dispatch virtual key.
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
keyCode = mLocked.currentVirtualKey.keyCode;
scanCode = mLocked.currentVirtualKey.scanCode;
downTime = mLocked.currentVirtualKey.downTime;
} // release lock
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Dispatch virtual key.
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
applyPolicyAndDispatchVirtualKey(when, policyFlags, keyEventAction, keyEventFlags,
keyCode, scanCode, downTime);
return touchResult;
}
void TouchInputMapper::applyPolicyAndDispatchVirtualKey(nsecs_t when, uint32_t policyFlags,
int32_t keyEventAction, int32_t keyEventFlags,
int32_t keyCode, int32_t scanCode, nsecs_t downTime) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t metaState = mContext->getGlobalMetaState();
if (keyEventAction == AKEY_EVENT_ACTION_DOWN) {
getPolicy()->virtualKeyDownFeedback();
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t policyActions = getPolicy()->interceptKey(when, getDeviceId(),
keyEventAction == AKEY_EVENT_ACTION_DOWN, keyCode, scanCode, policyFlags);
if (applyStandardPolicyActions(when, policyActions)) {
getDispatcher()->notifyKey(when, getDeviceId(), AINPUT_SOURCE_KEYBOARD, policyFlags,
keyEventAction, keyEventFlags, keyCode, scanCode, metaState, downTime);
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void TouchInputMapper::dispatchTouches(nsecs_t when, uint32_t policyFlags) {
uint32_t currentPointerCount = mCurrentTouch.pointerCount;
uint32_t lastPointerCount = mLastTouch.pointerCount;
if (currentPointerCount == 0 && lastPointerCount == 0) {
return; // nothing to do!
}
BitSet32 currentIdBits = mCurrentTouch.idBits;
BitSet32 lastIdBits = mLastTouch.idBits;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (currentIdBits == lastIdBits) {
// No pointer id changes so this is a move event.
// The dispatcher takes care of batching moves so we don't have to deal with that here.
int32_t motionEventAction = AMOTION_EVENT_ACTION_MOVE;
dispatchTouch(when, policyFlags, & mCurrentTouch,
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
currentIdBits, -1, currentPointerCount, motionEventAction);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
} else {
// There may be pointers going up and pointers going down at the same time when pointer
// ids are reported by the device driver.
BitSet32 upIdBits(lastIdBits.value & ~ currentIdBits.value);
BitSet32 downIdBits(currentIdBits.value & ~ lastIdBits.value);
BitSet32 activeIdBits(lastIdBits.value);
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
uint32_t pointerCount = lastPointerCount;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
while (! upIdBits.isEmpty()) {
uint32_t upId = upIdBits.firstMarkedBit();
upIdBits.clearBit(upId);
BitSet32 oldActiveIdBits = activeIdBits;
activeIdBits.clearBit(upId);
int32_t motionEventAction;
if (activeIdBits.isEmpty()) {
motionEventAction = AMOTION_EVENT_ACTION_UP;
} else {
motionEventAction = AMOTION_EVENT_ACTION_POINTER_UP;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
dispatchTouch(when, policyFlags, & mLastTouch,
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
oldActiveIdBits, upId, pointerCount, motionEventAction);
pointerCount -= 1;
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
while (! downIdBits.isEmpty()) {
uint32_t downId = downIdBits.firstMarkedBit();
downIdBits.clearBit(downId);
BitSet32 oldActiveIdBits = activeIdBits;
activeIdBits.markBit(downId);
int32_t motionEventAction;
if (oldActiveIdBits.isEmpty()) {
motionEventAction = AMOTION_EVENT_ACTION_DOWN;
mDownTime = when;
} else {
motionEventAction = AMOTION_EVENT_ACTION_POINTER_DOWN;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
pointerCount += 1;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
dispatchTouch(when, policyFlags, & mCurrentTouch,
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
activeIdBits, downId, pointerCount, motionEventAction);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void TouchInputMapper::dispatchTouch(nsecs_t when, uint32_t policyFlags,
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
TouchData* touch, BitSet32 idBits, uint32_t changedId, uint32_t pointerCount,
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
int32_t motionEventAction) {
int32_t pointerIds[MAX_POINTERS];
PointerCoords pointerCoords[MAX_POINTERS];
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
int32_t motionEventEdgeFlags = 0;
float xPrecision, yPrecision;
{ // acquire lock
AutoMutex _l(mLock);
// Walk through the the active pointers and map touch screen coordinates (TouchData) into
// display coordinates (PointerCoords) and adjust for display orientation.
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
for (uint32_t outIndex = 0; ! idBits.isEmpty(); outIndex++) {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
uint32_t id = idBits.firstMarkedBit();
idBits.clearBit(id);
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
uint32_t inIndex = touch->idToIndex[id];
const PointerData& in = touch->pointers[inIndex];
// X and Y
float x = float(in.x - mLocked.xOrigin) * mLocked.xScale;
float y = float(in.y - mLocked.yOrigin) * mLocked.yScale;
// ToolMajor and ToolMinor
float toolMajor, toolMinor;
switch (mCalibration.toolAreaCalibration) {
case Calibration::TOOL_AREA_CALIBRATION_GEOMETRIC:
toolMajor = in.toolMajor * mLocked.geometricScale;
if (mRawAxes.toolMinor.valid) {
toolMinor = in.toolMinor * mLocked.geometricScale;
} else {
toolMinor = toolMajor;
}
break;
case Calibration::TOOL_AREA_CALIBRATION_LINEAR:
toolMajor = in.toolMajor != 0
? in.toolMajor * mLocked.toolAreaLinearScale + mLocked.toolAreaLinearBias
: 0;
if (mRawAxes.toolMinor.valid) {
toolMinor = in.toolMinor != 0
? in.toolMinor * mLocked.toolAreaLinearScale
+ mLocked.toolAreaLinearBias
: 0;
} else {
toolMinor = toolMajor;
}
break;
default:
toolMajor = 0;
toolMinor = 0;
break;
}
if (mCalibration.haveToolAreaIsSummed && mCalibration.toolAreaIsSummed) {
toolMajor /= pointerCount;
toolMinor /= pointerCount;
}
// Pressure
float rawPressure;
switch (mCalibration.pressureSource) {
case Calibration::PRESSURE_SOURCE_PRESSURE:
rawPressure = in.pressure;
break;
case Calibration::PRESSURE_SOURCE_TOUCH:
rawPressure = in.touchMajor;
break;
default:
rawPressure = 0;
}
float pressure;
switch (mCalibration.pressureCalibration) {
case Calibration::PRESSURE_CALIBRATION_PHYSICAL:
case Calibration::PRESSURE_CALIBRATION_AMPLITUDE:
pressure = rawPressure * mLocked.pressureScale;
break;
default:
pressure = 1;
break;
}
// TouchMajor and TouchMinor
float touchMajor, touchMinor;
switch (mCalibration.touchAreaCalibration) {
case Calibration::TOUCH_AREA_CALIBRATION_GEOMETRIC:
touchMajor = in.touchMajor * mLocked.geometricScale;
if (mRawAxes.touchMinor.valid) {
touchMinor = in.touchMinor * mLocked.geometricScale;
} else {
touchMinor = touchMajor;
}
break;
case Calibration::TOUCH_AREA_CALIBRATION_PRESSURE:
touchMajor = toolMajor * pressure;
touchMinor = toolMinor * pressure;
break;
default:
touchMajor = 0;
touchMinor = 0;
break;
}
if (touchMajor > toolMajor) {
touchMajor = toolMajor;
}
if (touchMinor > toolMinor) {
touchMinor = toolMinor;
}
// Size
float size;
switch (mCalibration.sizeCalibration) {
case Calibration::SIZE_CALIBRATION_NORMALIZED: {
float rawSize = mRawAxes.toolMinor.valid
? avg(in.toolMajor, in.toolMinor)
: in.toolMajor;
size = rawSize * mLocked.sizeScale;
break;
}
default:
size = 0;
break;
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
}
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Orientation
float orientation;
switch (mCalibration.orientationCalibration) {
case Calibration::ORIENTATION_CALIBRATION_INTERPOLATED:
orientation = in.orientation * mLocked.orientationScale;
break;
default:
orientation = 0;
}
// Adjust coords for orientation.
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
switch (mLocked.surfaceOrientation) {
case InputReaderPolicyInterface::ROTATION_90: {
float xTemp = x;
x = y;
y = mLocked.surfaceWidth - xTemp;
orientation -= M_PI_2;
if (orientation < - M_PI_2) {
orientation += M_PI;
}
break;
}
case InputReaderPolicyInterface::ROTATION_180: {
x = mLocked.surfaceWidth - x;
y = mLocked.surfaceHeight - y;
orientation = - orientation;
break;
}
case InputReaderPolicyInterface::ROTATION_270: {
float xTemp = x;
x = mLocked.surfaceHeight - y;
y = xTemp;
orientation += M_PI_2;
if (orientation > M_PI_2) {
orientation -= M_PI;
}
break;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
}
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Write output coords.
PointerCoords& out = pointerCoords[outIndex];
out.x = x;
out.y = y;
out.pressure = pressure;
out.size = size;
out.touchMajor = touchMajor;
out.touchMinor = touchMinor;
out.toolMajor = toolMajor;
out.toolMinor = toolMinor;
out.orientation = orientation;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
pointerIds[outIndex] = int32_t(id);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
if (id == changedId) {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
motionEventAction |= outIndex << AMOTION_EVENT_ACTION_POINTER_INDEX_SHIFT;
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
// Check edge flags by looking only at the first pointer since the flags are
// global to the event.
if (motionEventAction == AMOTION_EVENT_ACTION_DOWN) {
if (pointerCoords[0].x <= 0) {
motionEventEdgeFlags |= AMOTION_EVENT_EDGE_FLAG_LEFT;
} else if (pointerCoords[0].x >= mLocked.orientedSurfaceWidth) {
motionEventEdgeFlags |= AMOTION_EVENT_EDGE_FLAG_RIGHT;
}
if (pointerCoords[0].y <= 0) {
motionEventEdgeFlags |= AMOTION_EVENT_EDGE_FLAG_TOP;
} else if (pointerCoords[0].y >= mLocked.orientedSurfaceHeight) {
motionEventEdgeFlags |= AMOTION_EVENT_EDGE_FLAG_BOTTOM;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
xPrecision = mLocked.orientedXPrecision;
yPrecision = mLocked.orientedYPrecision;
} // release lock
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
getDispatcher()->notifyMotion(when, getDeviceId(), AINPUT_SOURCE_TOUCHSCREEN, policyFlags,
Add support for secure views. Added the MotionEvent.FLAG_WINDOW_IS_OBSCURED flag which is set by the input manager whenever another visible window is partly or wholly obscured the target of a touch event so that applications can filter touches accordingly. Added a "filterTouchesWhenObscured" attribute to View which can be used to enable filtering of touches when the view's window is obscured. Change-Id: I936d9c85013fd2d77fb296a600528d30a29027d2
2010-09-02 00:01:00 +00:00
motionEventAction, 0, getContext()->getGlobalMetaState(), motionEventEdgeFlags,
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
pointerCount, pointerIds, pointerCoords,
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
xPrecision, yPrecision, mDownTime);
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
bool TouchInputMapper::isPointInsideSurfaceLocked(int32_t x, int32_t y) {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
if (mRawAxes.x.valid && mRawAxes.y.valid) {
return x >= mRawAxes.x.minValue && x <= mRawAxes.x.maxValue
&& y >= mRawAxes.y.minValue && y <= mRawAxes.y.maxValue;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
return true;
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
const TouchInputMapper::VirtualKey* TouchInputMapper::findVirtualKeyHitLocked(
int32_t x, int32_t y) {
size_t numVirtualKeys = mLocked.virtualKeys.size();
for (size_t i = 0; i < numVirtualKeys; i++) {
const VirtualKey& virtualKey = mLocked.virtualKeys[i];
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
#if DEBUG_VIRTUAL_KEYS
LOGD("VirtualKeys: Hit test (%d, %d): keyCode=%d, scanCode=%d, "
"left=%d, top=%d, right=%d, bottom=%d",
x, y,
virtualKey.keyCode, virtualKey.scanCode,
virtualKey.hitLeft, virtualKey.hitTop,
virtualKey.hitRight, virtualKey.hitBottom);
#endif
if (virtualKey.isHit(x, y)) {
return & virtualKey;
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return NULL;
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
void TouchInputMapper::calculatePointerIds() {
uint32_t currentPointerCount = mCurrentTouch.pointerCount;
uint32_t lastPointerCount = mLastTouch.pointerCount;
if (currentPointerCount == 0) {
// No pointers to assign.
mCurrentTouch.idBits.clear();
} else if (lastPointerCount == 0) {
// All pointers are new.
mCurrentTouch.idBits.clear();
for (uint32_t i = 0; i < currentPointerCount; i++) {
mCurrentTouch.pointers[i].id = i;
mCurrentTouch.idToIndex[i] = i;
mCurrentTouch.idBits.markBit(i);
}
} else if (currentPointerCount == 1 && lastPointerCount == 1) {
// Only one pointer and no change in count so it must have the same id as before.
uint32_t id = mLastTouch.pointers[0].id;
mCurrentTouch.pointers[0].id = id;
mCurrentTouch.idToIndex[id] = 0;
mCurrentTouch.idBits.value = BitSet32::valueForBit(id);
} else {
// General case.
// We build a heap of squared euclidean distances between current and last pointers
// associated with the current and last pointer indices. Then, we find the best
// match (by distance) for each current pointer.
PointerDistanceHeapElement heap[MAX_POINTERS * MAX_POINTERS];
uint32_t heapSize = 0;
for (uint32_t currentPointerIndex = 0; currentPointerIndex < currentPointerCount;
currentPointerIndex++) {
for (uint32_t lastPointerIndex = 0; lastPointerIndex < lastPointerCount;
lastPointerIndex++) {
int64_t deltaX = mCurrentTouch.pointers[currentPointerIndex].x
- mLastTouch.pointers[lastPointerIndex].x;
int64_t deltaY = mCurrentTouch.pointers[currentPointerIndex].y
- mLastTouch.pointers[lastPointerIndex].y;
uint64_t distance = uint64_t(deltaX * deltaX + deltaY * deltaY);
// Insert new element into the heap (sift up).
heap[heapSize].currentPointerIndex = currentPointerIndex;
heap[heapSize].lastPointerIndex = lastPointerIndex;
heap[heapSize].distance = distance;
heapSize += 1;
}
}
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Heapify
for (uint32_t startIndex = heapSize / 2; startIndex != 0; ) {
startIndex -= 1;
for (uint32_t parentIndex = startIndex; ;) {
uint32_t childIndex = parentIndex * 2 + 1;
if (childIndex >= heapSize) {
break;
}
if (childIndex + 1 < heapSize
&& heap[childIndex + 1].distance < heap[childIndex].distance) {
childIndex += 1;
}
if (heap[parentIndex].distance <= heap[childIndex].distance) {
break;
}
swap(heap[parentIndex], heap[childIndex]);
parentIndex = childIndex;
}
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
#if DEBUG_POINTER_ASSIGNMENT
LOGD("calculatePointerIds - initial distance min-heap: size=%d", heapSize);
for (size_t i = 0; i < heapSize; i++) {
LOGD(" heap[%d]: cur=%d, last=%d, distance=%lld",
i, heap[i].currentPointerIndex, heap[i].lastPointerIndex,
heap[i].distance);
}
#endif
More work in progress on native events. Refactored the code to eliminate potential deadlocks due to re-entrant calls from the policy into the dispatcher. Also added some plumbing that will be used to notify the framework about ANRs. Change-Id: Iba7a10de0cb3c56cd7520d6ce716db52fdcc94ff
2010-06-15 08:31:58 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Pull matches out by increasing order of distance.
// To avoid reassigning pointers that have already been matched, the loop keeps track
// of which last and current pointers have been matched using the matchedXXXBits variables.
// It also tracks the used pointer id bits.
BitSet32 matchedLastBits(0);
BitSet32 matchedCurrentBits(0);
BitSet32 usedIdBits(0);
bool first = true;
for (uint32_t i = min(currentPointerCount, lastPointerCount); i > 0; i--) {
for (;;) {
if (first) {
// The first time through the loop, we just consume the root element of
// the heap (the one with smallest distance).
first = false;
} else {
// Previous iterations consumed the root element of the heap.
// Pop root element off of the heap (sift down).
heapSize -= 1;
assert(heapSize > 0);
// Sift down.
heap[0] = heap[heapSize];
for (uint32_t parentIndex = 0; ;) {
uint32_t childIndex = parentIndex * 2 + 1;
if (childIndex >= heapSize) {
break;
}
if (childIndex + 1 < heapSize
&& heap[childIndex + 1].distance < heap[childIndex].distance) {
childIndex += 1;
}
if (heap[parentIndex].distance <= heap[childIndex].distance) {
break;
}
swap(heap[parentIndex], heap[childIndex]);
parentIndex = childIndex;
}
#if DEBUG_POINTER_ASSIGNMENT
LOGD("calculatePointerIds - reduced distance min-heap: size=%d", heapSize);
for (size_t i = 0; i < heapSize; i++) {
LOGD(" heap[%d]: cur=%d, last=%d, distance=%lld",
i, heap[i].currentPointerIndex, heap[i].lastPointerIndex,
heap[i].distance);
}
#endif
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
uint32_t currentPointerIndex = heap[0].currentPointerIndex;
if (matchedCurrentBits.hasBit(currentPointerIndex)) continue; // already matched
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
uint32_t lastPointerIndex = heap[0].lastPointerIndex;
if (matchedLastBits.hasBit(lastPointerIndex)) continue; // already matched
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
matchedCurrentBits.markBit(currentPointerIndex);
matchedLastBits.markBit(lastPointerIndex);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
uint32_t id = mLastTouch.pointers[lastPointerIndex].id;
mCurrentTouch.pointers[currentPointerIndex].id = id;
mCurrentTouch.idToIndex[id] = currentPointerIndex;
usedIdBits.markBit(id);
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
#if DEBUG_POINTER_ASSIGNMENT
LOGD("calculatePointerIds - matched: cur=%d, last=%d, id=%d, distance=%lld",
lastPointerIndex, currentPointerIndex, id, heap[0].distance);
#endif
break;
}
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// Assign fresh ids to new pointers.
if (currentPointerCount > lastPointerCount) {
for (uint32_t i = currentPointerCount - lastPointerCount; ;) {
uint32_t currentPointerIndex = matchedCurrentBits.firstUnmarkedBit();
uint32_t id = usedIdBits.firstUnmarkedBit();
mCurrentTouch.pointers[currentPointerIndex].id = id;
mCurrentTouch.idToIndex[id] = currentPointerIndex;
usedIdBits.markBit(id);
#if DEBUG_POINTER_ASSIGNMENT
LOGD("calculatePointerIds - assigned: cur=%d, id=%d",
currentPointerIndex, id);
#endif
if (--i == 0) break; // done
matchedCurrentBits.markBit(currentPointerIndex);
}
}
// Fix id bits.
mCurrentTouch.idBits = usedIdBits;
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
/* Special hack for devices that have bad screen data: if one of the
* points has moved more than a screen height from the last position,
* then drop it. */
bool TouchInputMapper::applyBadTouchFilter() {
// This hack requires valid axis parameters.
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
if (! mRawAxes.y.valid) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return false;
}
uint32_t pointerCount = mCurrentTouch.pointerCount;
// Nothing to do if there are no points.
if (pointerCount == 0) {
return false;
}
// Don't do anything if a finger is going down or up. We run
// here before assigning pointer IDs, so there isn't a good
// way to do per-finger matching.
if (pointerCount != mLastTouch.pointerCount) {
return false;
}
// We consider a single movement across more than a 7/16 of
// the long size of the screen to be bad. This was a magic value
// determined by looking at the maximum distance it is feasible
// to actually move in one sample.
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
int32_t maxDeltaY = mRawAxes.y.getRange() * 7 / 16;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// XXX The original code in InputDevice.java included commented out
// code for testing the X axis. Note that when we drop a point
// we don't actually restore the old X either. Strange.
// The old code also tries to track when bad points were previously
// detected but it turns out that due to the placement of a "break"
// at the end of the loop, we never set mDroppedBadPoint to true
// so it is effectively dead code.
// Need to figure out if the old code is busted or just overcomplicated
// but working as intended.
// Look through all new points and see if any are farther than
// acceptable from all previous points.
for (uint32_t i = pointerCount; i-- > 0; ) {
int32_t y = mCurrentTouch.pointers[i].y;
int32_t closestY = INT_MAX;
int32_t closestDeltaY = 0;
#if DEBUG_HACKS
LOGD("BadTouchFilter: Looking at next point #%d: y=%d", i, y);
#endif
for (uint32_t j = pointerCount; j-- > 0; ) {
int32_t lastY = mLastTouch.pointers[j].y;
int32_t deltaY = abs(y - lastY);
#if DEBUG_HACKS
LOGD("BadTouchFilter: Comparing with last point #%d: y=%d deltaY=%d",
j, lastY, deltaY);
#endif
if (deltaY < maxDeltaY) {
goto SkipSufficientlyClosePoint;
}
if (deltaY < closestDeltaY) {
closestDeltaY = deltaY;
closestY = lastY;
}
}
// Must not have found a close enough match.
#if DEBUG_HACKS
LOGD("BadTouchFilter: Dropping bad point #%d: newY=%d oldY=%d deltaY=%d maxDeltaY=%d",
i, y, closestY, closestDeltaY, maxDeltaY);
#endif
mCurrentTouch.pointers[i].y = closestY;
return true; // XXX original code only corrects one point
SkipSufficientlyClosePoint: ;
}
// No change.
return false;
}
/* Special hack for devices that have bad screen data: drop points where
* the coordinate value for one axis has jumped to the other pointer's location.
*/
bool TouchInputMapper::applyJumpyTouchFilter() {
// This hack requires valid axis parameters.
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
if (! mRawAxes.y.valid) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return false;
}
uint32_t pointerCount = mCurrentTouch.pointerCount;
if (mLastTouch.pointerCount != pointerCount) {
#if DEBUG_HACKS
LOGD("JumpyTouchFilter: Different pointer count %d -> %d",
mLastTouch.pointerCount, pointerCount);
for (uint32_t i = 0; i < pointerCount; i++) {
LOGD(" Pointer %d (%d, %d)", i,
mCurrentTouch.pointers[i].x, mCurrentTouch.pointers[i].y);
}
#endif
if (mJumpyTouchFilter.jumpyPointsDropped < JUMPY_TRANSITION_DROPS) {
if (mLastTouch.pointerCount == 1 && pointerCount == 2) {
// Just drop the first few events going from 1 to 2 pointers.
// They're bad often enough that they're not worth considering.
mCurrentTouch.pointerCount = 1;
mJumpyTouchFilter.jumpyPointsDropped += 1;
#if DEBUG_HACKS
LOGD("JumpyTouchFilter: Pointer 2 dropped");
#endif
return true;
} else if (mLastTouch.pointerCount == 2 && pointerCount == 1) {
// The event when we go from 2 -> 1 tends to be messed up too
mCurrentTouch.pointerCount = 2;
mCurrentTouch.pointers[0] = mLastTouch.pointers[0];
mCurrentTouch.pointers[1] = mLastTouch.pointers[1];
mJumpyTouchFilter.jumpyPointsDropped += 1;
#if DEBUG_HACKS
for (int32_t i = 0; i < 2; i++) {
LOGD("JumpyTouchFilter: Pointer %d replaced (%d, %d)", i,
mCurrentTouch.pointers[i].x, mCurrentTouch.pointers[i].y);
}
#endif
return true;
}
}
// Reset jumpy points dropped on other transitions or if limit exceeded.
mJumpyTouchFilter.jumpyPointsDropped = 0;
#if DEBUG_HACKS
LOGD("JumpyTouchFilter: Transition - drop limit reset");
#endif
return false;
}
// We have the same number of pointers as last time.
// A 'jumpy' point is one where the coordinate value for one axis
// has jumped to the other pointer's location. No need to do anything
// else if we only have one pointer.
if (pointerCount < 2) {
return false;
}
if (mJumpyTouchFilter.jumpyPointsDropped < JUMPY_DROP_LIMIT) {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
int jumpyEpsilon = mRawAxes.y.getRange() / JUMPY_EPSILON_DIVISOR;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
// We only replace the single worst jumpy point as characterized by pointer distance
// in a single axis.
int32_t badPointerIndex = -1;
int32_t badPointerReplacementIndex = -1;
int32_t badPointerDistance = INT_MIN; // distance to be corrected
for (uint32_t i = pointerCount; i-- > 0; ) {
int32_t x = mCurrentTouch.pointers[i].x;
int32_t y = mCurrentTouch.pointers[i].y;
#if DEBUG_HACKS
LOGD("JumpyTouchFilter: Point %d (%d, %d)", i, x, y);
#endif
// Check if a touch point is too close to another's coordinates
bool dropX = false, dropY = false;
for (uint32_t j = 0; j < pointerCount; j++) {
if (i == j) {
continue;
}
if (abs(x - mCurrentTouch.pointers[j].x) <= jumpyEpsilon) {
dropX = true;
break;
}
if (abs(y - mCurrentTouch.pointers[j].y) <= jumpyEpsilon) {
dropY = true;
break;
}
}
if (! dropX && ! dropY) {
continue; // not jumpy
}
// Find a replacement candidate by comparing with older points on the
// complementary (non-jumpy) axis.
int32_t distance = INT_MIN; // distance to be corrected
int32_t replacementIndex = -1;
if (dropX) {
// X looks too close. Find an older replacement point with a close Y.
int32_t smallestDeltaY = INT_MAX;
for (uint32_t j = 0; j < pointerCount; j++) {
int32_t deltaY = abs(y - mLastTouch.pointers[j].y);
if (deltaY < smallestDeltaY) {
smallestDeltaY = deltaY;
replacementIndex = j;
}
}
distance = abs(x - mLastTouch.pointers[replacementIndex].x);
} else {
// Y looks too close. Find an older replacement point with a close X.
int32_t smallestDeltaX = INT_MAX;
for (uint32_t j = 0; j < pointerCount; j++) {
int32_t deltaX = abs(x - mLastTouch.pointers[j].x);
if (deltaX < smallestDeltaX) {
smallestDeltaX = deltaX;
replacementIndex = j;
}
}
distance = abs(y - mLastTouch.pointers[replacementIndex].y);
}
// If replacing this pointer would correct a worse error than the previous ones
// considered, then use this replacement instead.
if (distance > badPointerDistance) {
badPointerIndex = i;
badPointerReplacementIndex = replacementIndex;
badPointerDistance = distance;
}
}
// Correct the jumpy pointer if one was found.
if (badPointerIndex >= 0) {
#if DEBUG_HACKS
LOGD("JumpyTouchFilter: Replacing bad pointer %d with (%d, %d)",
badPointerIndex,
mLastTouch.pointers[badPointerReplacementIndex].x,
mLastTouch.pointers[badPointerReplacementIndex].y);
#endif
mCurrentTouch.pointers[badPointerIndex].x =
mLastTouch.pointers[badPointerReplacementIndex].x;
mCurrentTouch.pointers[badPointerIndex].y =
mLastTouch.pointers[badPointerReplacementIndex].y;
mJumpyTouchFilter.jumpyPointsDropped += 1;
return true;
}
}
mJumpyTouchFilter.jumpyPointsDropped = 0;
return false;
}
/* Special hack for devices that have bad screen data: aggregate and
* compute averages of the coordinate data, to reduce the amount of
* jitter seen by applications. */
void TouchInputMapper::applyAveragingTouchFilter() {
for (uint32_t currentIndex = 0; currentIndex < mCurrentTouch.pointerCount; currentIndex++) {
uint32_t id = mCurrentTouch.pointers[currentIndex].id;
int32_t x = mCurrentTouch.pointers[currentIndex].x;
int32_t y = mCurrentTouch.pointers[currentIndex].y;
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
int32_t pressure;
switch (mCalibration.pressureSource) {
case Calibration::PRESSURE_SOURCE_PRESSURE:
pressure = mCurrentTouch.pointers[currentIndex].pressure;
break;
case Calibration::PRESSURE_SOURCE_TOUCH:
pressure = mCurrentTouch.pointers[currentIndex].touchMajor;
break;
default:
pressure = 1;
break;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (mLastTouch.idBits.hasBit(id)) {
// Pointer was down before and is still down now.
// Compute average over history trace.
uint32_t start = mAveragingTouchFilter.historyStart[id];
uint32_t end = mAveragingTouchFilter.historyEnd[id];
int64_t deltaX = x - mAveragingTouchFilter.historyData[end].pointers[id].x;
int64_t deltaY = y - mAveragingTouchFilter.historyData[end].pointers[id].y;
uint64_t distance = uint64_t(deltaX * deltaX + deltaY * deltaY);
#if DEBUG_HACKS
LOGD("AveragingTouchFilter: Pointer id %d - Distance from last sample: %lld",
id, distance);
#endif
if (distance < AVERAGING_DISTANCE_LIMIT) {
// Increment end index in preparation for recording new historical data.
end += 1;
if (end > AVERAGING_HISTORY_SIZE) {
end = 0;
}
// If the end index has looped back to the start index then we have filled
// the historical trace up to the desired size so we drop the historical
// data at the start of the trace.
if (end == start) {
start += 1;
if (start > AVERAGING_HISTORY_SIZE) {
start = 0;
}
}
// Add the raw data to the historical trace.
mAveragingTouchFilter.historyStart[id] = start;
mAveragingTouchFilter.historyEnd[id] = end;
mAveragingTouchFilter.historyData[end].pointers[id].x = x;
mAveragingTouchFilter.historyData[end].pointers[id].y = y;
mAveragingTouchFilter.historyData[end].pointers[id].pressure = pressure;
// Average over all historical positions in the trace by total pressure.
int32_t averagedX = 0;
int32_t averagedY = 0;
int32_t totalPressure = 0;
for (;;) {
int32_t historicalX = mAveragingTouchFilter.historyData[start].pointers[id].x;
int32_t historicalY = mAveragingTouchFilter.historyData[start].pointers[id].y;
int32_t historicalPressure = mAveragingTouchFilter.historyData[start]
.pointers[id].pressure;
averagedX += historicalX * historicalPressure;
averagedY += historicalY * historicalPressure;
totalPressure += historicalPressure;
if (start == end) {
break;
}
start += 1;
if (start > AVERAGING_HISTORY_SIZE) {
start = 0;
}
}
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
if (totalPressure != 0) {
averagedX /= totalPressure;
averagedY /= totalPressure;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
#if DEBUG_HACKS
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
LOGD("AveragingTouchFilter: Pointer id %d - "
"totalPressure=%d, averagedX=%d, averagedY=%d", id, totalPressure,
averagedX, averagedY);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
#endif
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
mCurrentTouch.pointers[currentIndex].x = averagedX;
mCurrentTouch.pointers[currentIndex].y = averagedY;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
} else {
#if DEBUG_HACKS
LOGD("AveragingTouchFilter: Pointer id %d - Exceeded max distance", id);
#endif
}
} else {
#if DEBUG_HACKS
LOGD("AveragingTouchFilter: Pointer id %d - Pointer went up", id);
#endif
}
// Reset pointer history.
mAveragingTouchFilter.historyStart[id] = 0;
mAveragingTouchFilter.historyEnd[id] = 0;
mAveragingTouchFilter.historyData[0].pointers[id].x = x;
mAveragingTouchFilter.historyData[0].pointers[id].y = y;
mAveragingTouchFilter.historyData[0].pointers[id].pressure = pressure;
}
}
int32_t TouchInputMapper::getKeyCodeState(uint32_t sourceMask, int32_t keyCode) {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
{ // acquire lock
AutoMutex _l(mLock);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
if (mLocked.currentVirtualKey.down && mLocked.currentVirtualKey.keyCode == keyCode) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return AKEY_STATE_VIRTUAL;
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
size_t numVirtualKeys = mLocked.virtualKeys.size();
for (size_t i = 0; i < numVirtualKeys; i++) {
const VirtualKey& virtualKey = mLocked.virtualKeys[i];
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (virtualKey.keyCode == keyCode) {
return AKEY_STATE_UP;
}
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
} // release lock
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return AKEY_STATE_UNKNOWN;
}
int32_t TouchInputMapper::getScanCodeState(uint32_t sourceMask, int32_t scanCode) {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
{ // acquire lock
AutoMutex _l(mLock);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
if (mLocked.currentVirtualKey.down && mLocked.currentVirtualKey.scanCode == scanCode) {
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return AKEY_STATE_VIRTUAL;
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
size_t numVirtualKeys = mLocked.virtualKeys.size();
for (size_t i = 0; i < numVirtualKeys; i++) {
const VirtualKey& virtualKey = mLocked.virtualKeys[i];
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (virtualKey.scanCode == scanCode) {
return AKEY_STATE_UP;
}
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
} // release lock
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return AKEY_STATE_UNKNOWN;
}
bool TouchInputMapper::markSupportedKeyCodes(uint32_t sourceMask, size_t numCodes,
const int32_t* keyCodes, uint8_t* outFlags) {
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
{ // acquire lock
AutoMutex _l(mLock);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
size_t numVirtualKeys = mLocked.virtualKeys.size();
for (size_t i = 0; i < numVirtualKeys; i++) {
const VirtualKey& virtualKey = mLocked.virtualKeys[i];
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
for (size_t i = 0; i < numCodes; i++) {
if (virtualKey.keyCode == keyCodes[i]) {
outFlags[i] = 1;
}
}
}
Improve thread safety of input mappers. Also fixed bug where old touch screen size could be reported by getMotionRange if an orientation change occurred but the user has not yet touched the screen. Bug: 2877345 Change-Id: I7878f47458f310ed6ebe6a5d1b2c9bec2c598ab9
2010-07-30 01:18:33 +00:00
} // release lock
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
return true;
}
// --- SingleTouchInputMapper ---
SingleTouchInputMapper::SingleTouchInputMapper(InputDevice* device, int32_t associatedDisplayId) :
TouchInputMapper(device, associatedDisplayId) {
initialize();
}
SingleTouchInputMapper::~SingleTouchInputMapper() {
}
void SingleTouchInputMapper::initialize() {
mAccumulator.clear();
mDown = false;
mX = 0;
mY = 0;
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
mPressure = 0; // default to 0 for devices that don't report pressure
mToolWidth = 0; // default to 0 for devices that don't report tool width
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
void SingleTouchInputMapper::reset() {
TouchInputMapper::reset();
initialize();
}
void SingleTouchInputMapper::process(const RawEvent* rawEvent) {
switch (rawEvent->type) {
case EV_KEY:
switch (rawEvent->scanCode) {
case BTN_TOUCH:
mAccumulator.fields |= Accumulator::FIELD_BTN_TOUCH;
mAccumulator.btnTouch = rawEvent->value != 0;
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
// Don't sync immediately. Wait until the next SYN_REPORT since we might
// not have received valid position information yet. This logic assumes that
// BTN_TOUCH is always followed by SYN_REPORT as part of a complete packet.
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
break;
}
break;
case EV_ABS:
switch (rawEvent->scanCode) {
case ABS_X:
mAccumulator.fields |= Accumulator::FIELD_ABS_X;
mAccumulator.absX = rawEvent->value;
break;
case ABS_Y:
mAccumulator.fields |= Accumulator::FIELD_ABS_Y;
mAccumulator.absY = rawEvent->value;
break;
case ABS_PRESSURE:
mAccumulator.fields |= Accumulator::FIELD_ABS_PRESSURE;
mAccumulator.absPressure = rawEvent->value;
break;
case ABS_TOOL_WIDTH:
mAccumulator.fields |= Accumulator::FIELD_ABS_TOOL_WIDTH;
mAccumulator.absToolWidth = rawEvent->value;
break;
}
break;
case EV_SYN:
switch (rawEvent->scanCode) {
case SYN_REPORT:
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
sync(rawEvent->when);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
break;
}
break;
}
}
void SingleTouchInputMapper::sync(nsecs_t when) {
uint32_t fields = mAccumulator.fields;
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
if (fields == 0) {
return; // no new state changes, so nothing to do
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (fields & Accumulator::FIELD_BTN_TOUCH) {
mDown = mAccumulator.btnTouch;
}
if (fields & Accumulator::FIELD_ABS_X) {
mX = mAccumulator.absX;
}
if (fields & Accumulator::FIELD_ABS_Y) {
mY = mAccumulator.absY;
}
if (fields & Accumulator::FIELD_ABS_PRESSURE) {
mPressure = mAccumulator.absPressure;
}
if (fields & Accumulator::FIELD_ABS_TOOL_WIDTH) {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
mToolWidth = mAccumulator.absToolWidth;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
mCurrentTouch.clear();
if (mDown) {
mCurrentTouch.pointerCount = 1;
mCurrentTouch.pointers[0].id = 0;
mCurrentTouch.pointers[0].x = mX;
mCurrentTouch.pointers[0].y = mY;
mCurrentTouch.pointers[0].pressure = mPressure;
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
mCurrentTouch.pointers[0].touchMajor = 0;
mCurrentTouch.pointers[0].touchMinor = 0;
mCurrentTouch.pointers[0].toolMajor = mToolWidth;
mCurrentTouch.pointers[0].toolMinor = mToolWidth;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mCurrentTouch.pointers[0].orientation = 0;
mCurrentTouch.idToIndex[0] = 0;
mCurrentTouch.idBits.markBit(0);
}
syncTouch(when, true);
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
mAccumulator.clear();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
void SingleTouchInputMapper::configureRawAxes() {
TouchInputMapper::configureRawAxes();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_X, & mRawAxes.x);
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_Y, & mRawAxes.y);
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_PRESSURE, & mRawAxes.pressure);
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_TOOL_WIDTH, & mRawAxes.toolMajor);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
// --- MultiTouchInputMapper ---
MultiTouchInputMapper::MultiTouchInputMapper(InputDevice* device, int32_t associatedDisplayId) :
TouchInputMapper(device, associatedDisplayId) {
initialize();
}
MultiTouchInputMapper::~MultiTouchInputMapper() {
}
void MultiTouchInputMapper::initialize() {
mAccumulator.clear();
}
void MultiTouchInputMapper::reset() {
TouchInputMapper::reset();
initialize();
}
void MultiTouchInputMapper::process(const RawEvent* rawEvent) {
switch (rawEvent->type) {
case EV_ABS: {
uint32_t pointerIndex = mAccumulator.pointerCount;
Accumulator::Pointer* pointer = & mAccumulator.pointers[pointerIndex];
switch (rawEvent->scanCode) {
case ABS_MT_POSITION_X:
pointer->fields |= Accumulator::FIELD_ABS_MT_POSITION_X;
pointer->absMTPositionX = rawEvent->value;
break;
case ABS_MT_POSITION_Y:
pointer->fields |= Accumulator::FIELD_ABS_MT_POSITION_Y;
pointer->absMTPositionY = rawEvent->value;
break;
case ABS_MT_TOUCH_MAJOR:
pointer->fields |= Accumulator::FIELD_ABS_MT_TOUCH_MAJOR;
pointer->absMTTouchMajor = rawEvent->value;
break;
case ABS_MT_TOUCH_MINOR:
pointer->fields |= Accumulator::FIELD_ABS_MT_TOUCH_MINOR;
pointer->absMTTouchMinor = rawEvent->value;
break;
case ABS_MT_WIDTH_MAJOR:
pointer->fields |= Accumulator::FIELD_ABS_MT_WIDTH_MAJOR;
pointer->absMTWidthMajor = rawEvent->value;
break;
case ABS_MT_WIDTH_MINOR:
pointer->fields |= Accumulator::FIELD_ABS_MT_WIDTH_MINOR;
pointer->absMTWidthMinor = rawEvent->value;
break;
case ABS_MT_ORIENTATION:
pointer->fields |= Accumulator::FIELD_ABS_MT_ORIENTATION;
pointer->absMTOrientation = rawEvent->value;
break;
case ABS_MT_TRACKING_ID:
pointer->fields |= Accumulator::FIELD_ABS_MT_TRACKING_ID;
pointer->absMTTrackingId = rawEvent->value;
break;
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
case ABS_MT_PRESSURE:
pointer->fields |= Accumulator::FIELD_ABS_MT_PRESSURE;
pointer->absMTPressure = rawEvent->value;
break;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
break;
}
case EV_SYN:
switch (rawEvent->scanCode) {
case SYN_MT_REPORT: {
// MultiTouch Sync: The driver has returned all data for *one* of the pointers.
uint32_t pointerIndex = mAccumulator.pointerCount;
if (mAccumulator.pointers[pointerIndex].fields) {
if (pointerIndex == MAX_POINTERS) {
LOGW("MultiTouch device driver returned more than maximum of %d pointers.",
MAX_POINTERS);
} else {
pointerIndex += 1;
mAccumulator.pointerCount = pointerIndex;
}
}
mAccumulator.pointers[pointerIndex].clear();
break;
}
case SYN_REPORT:
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
sync(rawEvent->when);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
break;
}
break;
}
}
void MultiTouchInputMapper::sync(nsecs_t when) {
static const uint32_t REQUIRED_FIELDS =
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
Accumulator::FIELD_ABS_MT_POSITION_X | Accumulator::FIELD_ABS_MT_POSITION_Y;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
uint32_t inCount = mAccumulator.pointerCount;
uint32_t outCount = 0;
bool havePointerIds = true;
mCurrentTouch.clear();
for (uint32_t inIndex = 0; inIndex < inCount; inIndex++) {
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
const Accumulator::Pointer& inPointer = mAccumulator.pointers[inIndex];
uint32_t fields = inPointer.fields;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if ((fields & REQUIRED_FIELDS) != REQUIRED_FIELDS) {
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
// Some drivers send empty MT sync packets without X / Y to indicate a pointer up.
// Drop this finger.
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
continue;
}
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
PointerData& outPointer = mCurrentTouch.pointers[outCount];
outPointer.x = inPointer.absMTPositionX;
outPointer.y = inPointer.absMTPositionY;
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
if (fields & Accumulator::FIELD_ABS_MT_PRESSURE) {
if (inPointer.absMTPressure <= 0) {
// Some devices send sync packets with X / Y but with a 0 presure to indicate
// a pointer up. Drop this finger.
continue;
}
outPointer.pressure = inPointer.absMTPressure;
} else {
// Default pressure to 0 if absent.
outPointer.pressure = 0;
}
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
if (fields & Accumulator::FIELD_ABS_MT_TOUCH_MAJOR) {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
if (inPointer.absMTTouchMajor <= 0) {
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
// Some devices send sync packets with X / Y but with a 0 touch major to indicate
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// a pointer going up. Drop this finger.
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
continue;
}
outPointer.touchMajor = inPointer.absMTTouchMajor;
} else {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Default touch area to 0 if absent.
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
outPointer.touchMajor = 0;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
if (fields & Accumulator::FIELD_ABS_MT_TOUCH_MINOR) {
outPointer.touchMinor = inPointer.absMTTouchMinor;
} else {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Assume touch area is circular.
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
outPointer.touchMinor = outPointer.touchMajor;
}
if (fields & Accumulator::FIELD_ABS_MT_WIDTH_MAJOR) {
outPointer.toolMajor = inPointer.absMTWidthMajor;
} else {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Default tool area to 0 if absent.
outPointer.toolMajor = 0;
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
if (fields & Accumulator::FIELD_ABS_MT_WIDTH_MINOR) {
outPointer.toolMinor = inPointer.absMTWidthMinor;
} else {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Assume tool area is circular.
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
outPointer.toolMinor = outPointer.toolMajor;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
if (fields & Accumulator::FIELD_ABS_MT_ORIENTATION) {
outPointer.orientation = inPointer.absMTOrientation;
} else {
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Default orientation to vertical if absent.
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
outPointer.orientation = 0;
}
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
// Assign pointer id using tracking id if available.
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (havePointerIds) {
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
if (fields & Accumulator::FIELD_ABS_MT_TRACKING_ID) {
uint32_t id = uint32_t(inPointer.absMTTrackingId);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
if (id > MAX_POINTER_ID) {
#if DEBUG_POINTERS
LOGD("Pointers: Ignoring driver provided pointer id %d because "
"it is larger than max supported id %d for optimizations",
id, MAX_POINTER_ID);
#endif
havePointerIds = false;
}
else {
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
outPointer.id = id;
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
mCurrentTouch.idToIndex[id] = outCount;
mCurrentTouch.idBits.markBit(id);
}
} else {
havePointerIds = false;
}
}
outCount += 1;
}
mCurrentTouch.pointerCount = outCount;
syncTouch(when, havePointerIds);
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
mAccumulator.clear();
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
void MultiTouchInputMapper::configureRawAxes() {
TouchInputMapper::configureRawAxes();
Fix some input device mapping bugs with certain drivers. On single-touch devices, pointer up/down is signalled by a BTN_TOUCH key event. Previously we handled BTN_TOUCH immediately but some drivers may produce the sequence BTN_TOUCH, ABS_X, ABS_Y, SYN_REPORT on pointer down which caused us to emit a bad initial pointer down location. Now we wait for SYN_REPORT before reporting the up or down. On multi-touch devices, pointer up can be signalled by as little as the sequence SYN_MT_REPORT, SYN_REPORT. This change ensures that we handle this case. Added support for reading ABS_MT_PRESSURE when available. Corrected mapping of touchMajor/touchMinor on single touch devices. Minor code cleanup. Change-Id: Ic7ec4811241ed85a06e59b8a839ca05180d491d4
2010-08-18 03:38:35 +00:00
Input device calibration and capabilities. Finished the input device capability API. Added a mechanism for calibrating touch devices to obtain more accurate information about the touch contact area. Improved pointer location to show new coordinates and capabilities. Optimized pointer location display and formatting to avoid allocating large numbers of temporary objects. The GC churn was causing the application to stutter very badly when more than a couple of fingers were down). Added more diagnostics. Change-Id: Ie25380278ed6f16c5b04cd9df848015850383498
2010-08-30 10:02:23 +00:00
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_MT_POSITION_X, & mRawAxes.x);
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_MT_POSITION_Y, & mRawAxes.y);
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_MT_TOUCH_MAJOR, & mRawAxes.touchMajor);
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_MT_TOUCH_MINOR, & mRawAxes.touchMinor);
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_MT_WIDTH_MAJOR, & mRawAxes.toolMajor);
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_MT_WIDTH_MINOR, & mRawAxes.toolMinor);
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_MT_ORIENTATION, & mRawAxes.orientation);
getEventHub()->getAbsoluteAxisInfo(getDeviceId(), ABS_MT_PRESSURE, & mRawAxes.pressure);
Refactor input reader to support new device types more easily. Refactored the input reader so that each raw input protocol is handled by a separate subclass of the new InputMapper type. This way, behaviors pertaining to keyboard, trackballs, touchscreens, switches and other devices are clearly distinguished for improved maintainability. Added partial support for describing capabilities of input devices (incomplete and untested for now, will be fleshed out in later commits). Simplified EventHub interface somewhat since InputReader is taking over more of the work. Cleaned up some of the interactions between InputManager and WindowManagerService related to reading input state. Fixed swiping finger from screen edge into display area. Added logging of device information to 'dumpsys window'. Change-Id: I17faffc33e3aec3a0f33f0b37e81a70609378612
2010-07-24 04:28:06 +00:00
}
Native input dispatch rewrite work in progress. The old dispatch mechanism has been left in place and continues to be used by default for now. To enable native input dispatch, edit the ENABLE_NATIVE_DISPATCH constant in WindowManagerPolicy. Includes part of the new input event NDK API. Some details TBD. To wire up input dispatch, as the ViewRoot adds a window to the window session it receives an InputChannel object as an output argument. The InputChannel encapsulates the file descriptors for a shared memory region and two pipe end-points. The ViewRoot then provides the InputChannel to the InputQueue. Behind the scenes, InputQueue simply attaches handlers to the native PollLoop object that underlies the MessageQueue. This way MessageQueue doesn't need to know anything about input dispatch per-se, it just exposes (in native code) a PollLoop that other components can use to monitor file descriptor state changes. There can be zero or more targets for any given input event. Each input target is specified by its input channel and some parameters including flags, an X/Y coordinate offset, and the dispatch timeout. An input target can request either synchronous dispatch (for foreground apps) or asynchronous dispatch (fire-and-forget for wallpapers and "outside" targets). Currently, finding the appropriate input targets for an event requires a call back into the WindowManagerServer from native code. In the future this will be refactored to avoid most of these callbacks except as required to handle pending focus transitions. End-to-end event dispatch mostly works! To do: event injection, rate limiting, ANRs, testing, optimization, etc. Change-Id: I8c36b2b9e0a2d27392040ecda0f51b636456de25
2010-04-23 01:58:52 +00:00
} // namespace android
Reference in New Issue Copy Permalink