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
|
|
|
|
//
|
|
|
|
// Provides a shared memory transport for input events.
|
|
|
|
//
|
|
|
|
#define LOG_TAG "InputTransport"
|
|
|
|
|
|
|
|
//#define LOG_NDEBUG 0
|
|
|
|
|
|
|
|
// Log debug messages about channel signalling (send signal, receive signal)
|
2010-06-16 08:53:36 +00:00
|
|
|
#define DEBUG_CHANNEL_SIGNALS 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 whenever InputChannel objects are created/destroyed
|
2010-06-16 08:53:36 +00:00
|
|
|
#define DEBUG_CHANNEL_LIFECYCLE 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 transport actions (initialize, reset, publish, ...)
|
2010-06-16 08:53:36 +00:00
|
|
|
#define DEBUG_TRANSPORT_ACTIONS 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/ashmem.h>
|
|
|
|
#include <cutils/log.h>
|
|
|
|
#include <errno.h>
|
|
|
|
#include <fcntl.h>
|
|
|
|
#include <sys/mman.h>
|
|
|
|
#include <ui/InputTransport.h>
|
|
|
|
#include <unistd.h>
|
|
|
|
|
|
|
|
namespace android {
|
|
|
|
|
|
|
|
// Must be at least sizeof(InputMessage) + sufficient space for pointer data
|
|
|
|
static const int DEFAULT_MESSAGE_BUFFER_SIZE = 16384;
|
|
|
|
|
|
|
|
// Signal sent by the producer to the consumer to inform it that a new message is
|
|
|
|
// available to be consumed in the shared memory buffer.
|
|
|
|
static const char INPUT_SIGNAL_DISPATCH = 'D';
|
|
|
|
|
|
|
|
// Signal sent by the consumer to the producer to inform it that it has finished
|
|
|
|
// consuming the most recent message.
|
|
|
|
static const char INPUT_SIGNAL_FINISHED = 'f';
|
|
|
|
|
|
|
|
|
|
|
|
// --- InputChannel ---
|
|
|
|
|
|
|
|
InputChannel::InputChannel(const String8& name, int32_t ashmemFd, int32_t receivePipeFd,
|
|
|
|
int32_t sendPipeFd) :
|
|
|
|
mName(name), mAshmemFd(ashmemFd), mReceivePipeFd(receivePipeFd), mSendPipeFd(sendPipeFd) {
|
|
|
|
#if DEBUG_CHANNEL_LIFECYCLE
|
|
|
|
LOGD("Input channel constructed: name='%s', ashmemFd=%d, receivePipeFd=%d, sendPipeFd=%d",
|
|
|
|
mName.string(), ashmemFd, receivePipeFd, sendPipeFd);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
int result = fcntl(mReceivePipeFd, F_SETFL, O_NONBLOCK);
|
|
|
|
LOG_ALWAYS_FATAL_IF(result != 0, "channel '%s' ~ Could not make receive pipe "
|
|
|
|
"non-blocking. errno=%d", mName.string(), errno);
|
|
|
|
|
|
|
|
result = fcntl(mSendPipeFd, F_SETFL, O_NONBLOCK);
|
|
|
|
LOG_ALWAYS_FATAL_IF(result != 0, "channel '%s' ~ Could not make send pipe "
|
|
|
|
"non-blocking. errno=%d", mName.string(), errno);
|
|
|
|
}
|
|
|
|
|
|
|
|
InputChannel::~InputChannel() {
|
|
|
|
#if DEBUG_CHANNEL_LIFECYCLE
|
|
|
|
LOGD("Input channel destroyed: name='%s', ashmemFd=%d, receivePipeFd=%d, sendPipeFd=%d",
|
|
|
|
mName.string(), mAshmemFd, mReceivePipeFd, mSendPipeFd);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
::close(mAshmemFd);
|
|
|
|
::close(mReceivePipeFd);
|
|
|
|
::close(mSendPipeFd);
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputChannel::openInputChannelPair(const String8& name,
|
2010-06-16 08:53:36 +00:00
|
|
|
sp<InputChannel>& outServerChannel, sp<InputChannel>& outClientChannel) {
|
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
|
|
|
status_t result;
|
|
|
|
|
|
|
|
int serverAshmemFd = ashmem_create_region(name.string(), DEFAULT_MESSAGE_BUFFER_SIZE);
|
|
|
|
if (serverAshmemFd < 0) {
|
|
|
|
result = -errno;
|
|
|
|
LOGE("channel '%s' ~ Could not create shared memory region. errno=%d",
|
|
|
|
name.string(), errno);
|
|
|
|
} else {
|
|
|
|
result = ashmem_set_prot_region(serverAshmemFd, PROT_READ | PROT_WRITE);
|
|
|
|
if (result < 0) {
|
|
|
|
LOGE("channel '%s' ~ Error %d trying to set protection of ashmem fd %d.",
|
|
|
|
name.string(), result, serverAshmemFd);
|
|
|
|
} else {
|
|
|
|
// Dup the file descriptor because the server and client input channel objects that
|
|
|
|
// are returned may have different lifetimes but they share the same shared memory region.
|
|
|
|
int clientAshmemFd;
|
|
|
|
clientAshmemFd = dup(serverAshmemFd);
|
|
|
|
if (clientAshmemFd < 0) {
|
|
|
|
result = -errno;
|
|
|
|
LOGE("channel '%s' ~ Could not dup() shared memory region fd. errno=%d",
|
|
|
|
name.string(), errno);
|
|
|
|
} else {
|
|
|
|
int forward[2];
|
|
|
|
if (pipe(forward)) {
|
|
|
|
result = -errno;
|
|
|
|
LOGE("channel '%s' ~ Could not create forward pipe. errno=%d",
|
|
|
|
name.string(), errno);
|
|
|
|
} else {
|
|
|
|
int reverse[2];
|
|
|
|
if (pipe(reverse)) {
|
|
|
|
result = -errno;
|
|
|
|
LOGE("channel '%s' ~ Could not create reverse pipe. errno=%d",
|
|
|
|
name.string(), errno);
|
|
|
|
} else {
|
|
|
|
String8 serverChannelName = name;
|
|
|
|
serverChannelName.append(" (server)");
|
2010-06-16 08:53:36 +00:00
|
|
|
outServerChannel = new InputChannel(serverChannelName,
|
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
|
|
|
serverAshmemFd, reverse[0], forward[1]);
|
|
|
|
|
|
|
|
String8 clientChannelName = name;
|
|
|
|
clientChannelName.append(" (client)");
|
2010-06-16 08:53:36 +00:00
|
|
|
outClientChannel = new InputChannel(clientChannelName,
|
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
|
|
|
clientAshmemFd, forward[0], reverse[1]);
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
::close(forward[0]);
|
|
|
|
::close(forward[1]);
|
|
|
|
}
|
|
|
|
::close(clientAshmemFd);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
::close(serverAshmemFd);
|
|
|
|
}
|
|
|
|
|
2010-06-16 08:53:36 +00:00
|
|
|
outServerChannel.clear();
|
|
|
|
outClientChannel.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
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputChannel::sendSignal(char signal) {
|
|
|
|
ssize_t nWrite = ::write(mSendPipeFd, & signal, 1);
|
|
|
|
|
|
|
|
if (nWrite == 1) {
|
|
|
|
#if DEBUG_CHANNEL_SIGNALS
|
|
|
|
LOGD("channel '%s' ~ sent signal '%c'", mName.string(), signal);
|
|
|
|
#endif
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
#if DEBUG_CHANNEL_SIGNALS
|
|
|
|
LOGD("channel '%s' ~ error sending signal '%c', errno=%d", mName.string(), signal, errno);
|
|
|
|
#endif
|
|
|
|
return -errno;
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputChannel::receiveSignal(char* outSignal) {
|
|
|
|
ssize_t nRead = ::read(mReceivePipeFd, outSignal, 1);
|
|
|
|
if (nRead == 1) {
|
|
|
|
#if DEBUG_CHANNEL_SIGNALS
|
|
|
|
LOGD("channel '%s' ~ received signal '%c'", mName.string(), *outSignal);
|
|
|
|
#endif
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
|
2010-06-16 08:53:36 +00:00
|
|
|
if (nRead == 0) { // check for EOF
|
|
|
|
#if DEBUG_CHANNEL_SIGNALS
|
|
|
|
LOGD("channel '%s' ~ receive signal failed because peer was closed", mName.string());
|
|
|
|
#endif
|
|
|
|
return DEAD_OBJECT;
|
|
|
|
}
|
|
|
|
|
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 (errno == EAGAIN) {
|
|
|
|
#if DEBUG_CHANNEL_SIGNALS
|
|
|
|
LOGD("channel '%s' ~ receive signal failed because no signal available", mName.string());
|
|
|
|
#endif
|
|
|
|
return WOULD_BLOCK;
|
|
|
|
}
|
|
|
|
|
|
|
|
#if DEBUG_CHANNEL_SIGNALS
|
|
|
|
LOGD("channel '%s' ~ receive signal failed, errno=%d", mName.string(), errno);
|
|
|
|
#endif
|
|
|
|
return -errno;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
// --- InputPublisher ---
|
|
|
|
|
|
|
|
InputPublisher::InputPublisher(const sp<InputChannel>& channel) :
|
|
|
|
mChannel(channel), mSharedMessage(NULL),
|
|
|
|
mPinned(false), mSemaphoreInitialized(false), mWasDispatched(false),
|
|
|
|
mMotionEventSampleDataTail(NULL) {
|
|
|
|
}
|
|
|
|
|
|
|
|
InputPublisher::~InputPublisher() {
|
|
|
|
reset();
|
|
|
|
|
|
|
|
if (mSharedMessage) {
|
|
|
|
munmap(mSharedMessage, mAshmemSize);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputPublisher::initialize() {
|
|
|
|
#if DEBUG_TRANSPORT_ACTIONS
|
|
|
|
LOGD("channel '%s' publisher ~ initialize",
|
|
|
|
mChannel->getName().string());
|
|
|
|
#endif
|
|
|
|
|
|
|
|
int ashmemFd = mChannel->getAshmemFd();
|
|
|
|
int result = ashmem_get_size_region(ashmemFd);
|
|
|
|
if (result < 0) {
|
|
|
|
LOGE("channel '%s' publisher ~ Error %d getting size of ashmem fd %d.",
|
|
|
|
mChannel->getName().string(), result, ashmemFd);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
mAshmemSize = (size_t) result;
|
|
|
|
|
|
|
|
mSharedMessage = static_cast<InputMessage*>(mmap(NULL, mAshmemSize,
|
|
|
|
PROT_READ | PROT_WRITE, MAP_SHARED, ashmemFd, 0));
|
|
|
|
if (! mSharedMessage) {
|
|
|
|
LOGE("channel '%s' publisher ~ mmap failed on ashmem fd %d.",
|
|
|
|
mChannel->getName().string(), ashmemFd);
|
|
|
|
return NO_MEMORY;
|
|
|
|
}
|
|
|
|
|
|
|
|
mPinned = true;
|
|
|
|
mSharedMessage->consumed = false;
|
|
|
|
|
|
|
|
return reset();
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputPublisher::reset() {
|
|
|
|
#if DEBUG_TRANSPORT_ACTIONS
|
|
|
|
LOGD("channel '%s' publisher ~ reset",
|
|
|
|
mChannel->getName().string());
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (mPinned) {
|
|
|
|
// Destroy the semaphore since we are about to unpin the memory region that contains it.
|
|
|
|
int result;
|
|
|
|
if (mSemaphoreInitialized) {
|
|
|
|
if (mSharedMessage->consumed) {
|
|
|
|
result = sem_post(& mSharedMessage->semaphore);
|
|
|
|
if (result < 0) {
|
|
|
|
LOGE("channel '%s' publisher ~ Error %d in sem_post.",
|
|
|
|
mChannel->getName().string(), errno);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
result = sem_destroy(& mSharedMessage->semaphore);
|
|
|
|
if (result < 0) {
|
|
|
|
LOGE("channel '%s' publisher ~ Error %d in sem_destroy.",
|
|
|
|
mChannel->getName().string(), errno);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
|
|
|
|
mSemaphoreInitialized = false;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Unpin the region since we no longer care about its contents.
|
|
|
|
int ashmemFd = mChannel->getAshmemFd();
|
|
|
|
result = ashmem_unpin_region(ashmemFd, 0, 0);
|
|
|
|
if (result < 0) {
|
|
|
|
LOGE("channel '%s' publisher ~ Error %d unpinning ashmem fd %d.",
|
|
|
|
mChannel->getName().string(), result, ashmemFd);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
|
|
|
|
mPinned = false;
|
|
|
|
}
|
|
|
|
|
|
|
|
mMotionEventSampleDataTail = NULL;
|
|
|
|
mWasDispatched = false;
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputPublisher::publishInputEvent(
|
|
|
|
int32_t type,
|
|
|
|
int32_t deviceId,
|
|
|
|
int32_t nature) {
|
|
|
|
if (mPinned) {
|
|
|
|
LOGE("channel '%s' publisher ~ Attempted to publish a new event but publisher has "
|
|
|
|
"not yet been reset.", mChannel->getName().string());
|
|
|
|
return INVALID_OPERATION;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Pin the region.
|
|
|
|
// We do not check for ASHMEM_NOT_PURGED because we don't care about the previous
|
|
|
|
// contents of the buffer so it does not matter whether it was purged in the meantime.
|
|
|
|
int ashmemFd = mChannel->getAshmemFd();
|
|
|
|
int result = ashmem_pin_region(ashmemFd, 0, 0);
|
|
|
|
if (result < 0) {
|
|
|
|
LOGE("channel '%s' publisher ~ Error %d pinning ashmem fd %d.",
|
|
|
|
mChannel->getName().string(), result, ashmemFd);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
|
|
|
|
mPinned = true;
|
|
|
|
|
|
|
|
result = sem_init(& mSharedMessage->semaphore, 1, 1);
|
|
|
|
if (result < 0) {
|
|
|
|
LOGE("channel '%s' publisher ~ Error %d in sem_init.",
|
|
|
|
mChannel->getName().string(), errno);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
|
|
|
|
mSemaphoreInitialized = true;
|
|
|
|
|
|
|
|
mSharedMessage->consumed = false;
|
|
|
|
mSharedMessage->type = type;
|
|
|
|
mSharedMessage->deviceId = deviceId;
|
|
|
|
mSharedMessage->nature = nature;
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputPublisher::publishKeyEvent(
|
|
|
|
int32_t deviceId,
|
|
|
|
int32_t nature,
|
|
|
|
int32_t action,
|
|
|
|
int32_t flags,
|
|
|
|
int32_t keyCode,
|
|
|
|
int32_t scanCode,
|
|
|
|
int32_t metaState,
|
|
|
|
int32_t repeatCount,
|
|
|
|
nsecs_t downTime,
|
|
|
|
nsecs_t eventTime) {
|
|
|
|
#if DEBUG_TRANSPORT_ACTIONS
|
|
|
|
LOGD("channel '%s' publisher ~ publishKeyEvent: deviceId=%d, nature=%d, "
|
|
|
|
"action=%d, flags=%d, keyCode=%d, scanCode=%d, metaState=%d, repeatCount=%d,"
|
|
|
|
"downTime=%lld, eventTime=%lld",
|
|
|
|
mChannel->getName().string(),
|
|
|
|
deviceId, nature, action, flags, keyCode, scanCode, metaState, repeatCount,
|
|
|
|
downTime, eventTime);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
status_t result = publishInputEvent(INPUT_EVENT_TYPE_KEY, deviceId, nature);
|
|
|
|
if (result < 0) {
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
|
|
|
mSharedMessage->key.action = action;
|
|
|
|
mSharedMessage->key.flags = flags;
|
|
|
|
mSharedMessage->key.keyCode = keyCode;
|
|
|
|
mSharedMessage->key.scanCode = scanCode;
|
|
|
|
mSharedMessage->key.metaState = metaState;
|
|
|
|
mSharedMessage->key.repeatCount = repeatCount;
|
|
|
|
mSharedMessage->key.downTime = downTime;
|
|
|
|
mSharedMessage->key.eventTime = eventTime;
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputPublisher::publishMotionEvent(
|
|
|
|
int32_t deviceId,
|
|
|
|
int32_t nature,
|
|
|
|
int32_t action,
|
|
|
|
int32_t edgeFlags,
|
|
|
|
int32_t metaState,
|
|
|
|
float xOffset,
|
|
|
|
float yOffset,
|
|
|
|
float xPrecision,
|
|
|
|
float yPrecision,
|
|
|
|
nsecs_t downTime,
|
|
|
|
nsecs_t eventTime,
|
|
|
|
size_t pointerCount,
|
|
|
|
const int32_t* pointerIds,
|
|
|
|
const PointerCoords* pointerCoords) {
|
|
|
|
#if DEBUG_TRANSPORT_ACTIONS
|
|
|
|
LOGD("channel '%s' publisher ~ publishMotionEvent: deviceId=%d, nature=%d, "
|
|
|
|
"action=%d, edgeFlags=%d, metaState=%d, xOffset=%f, yOffset=%f, "
|
|
|
|
"xPrecision=%f, yPrecision=%f, downTime=%lld, eventTime=%lld, "
|
|
|
|
"pointerCount=%d",
|
|
|
|
mChannel->getName().string(),
|
|
|
|
deviceId, nature, action, edgeFlags, metaState, xOffset, yOffset,
|
|
|
|
xPrecision, yPrecision, downTime, eventTime, pointerCount);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (pointerCount > MAX_POINTERS || pointerCount < 1) {
|
|
|
|
LOGE("channel '%s' publisher ~ Invalid number of pointers provided: %d.",
|
|
|
|
mChannel->getName().string(), pointerCount);
|
|
|
|
return BAD_VALUE;
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t result = publishInputEvent(INPUT_EVENT_TYPE_MOTION, deviceId, nature);
|
|
|
|
if (result < 0) {
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
|
|
|
mSharedMessage->motion.action = action;
|
|
|
|
mSharedMessage->motion.edgeFlags = edgeFlags;
|
|
|
|
mSharedMessage->motion.metaState = metaState;
|
|
|
|
mSharedMessage->motion.xOffset = xOffset;
|
|
|
|
mSharedMessage->motion.yOffset = yOffset;
|
|
|
|
mSharedMessage->motion.xPrecision = xPrecision;
|
|
|
|
mSharedMessage->motion.yPrecision = yPrecision;
|
|
|
|
mSharedMessage->motion.downTime = downTime;
|
|
|
|
mSharedMessage->motion.pointerCount = pointerCount;
|
|
|
|
|
|
|
|
mSharedMessage->motion.sampleCount = 1;
|
|
|
|
mSharedMessage->motion.sampleData[0].eventTime = eventTime;
|
|
|
|
|
|
|
|
for (size_t i = 0; i < pointerCount; i++) {
|
|
|
|
mSharedMessage->motion.pointerIds[i] = pointerIds[i];
|
|
|
|
mSharedMessage->motion.sampleData[0].coords[i] = pointerCoords[i];
|
|
|
|
}
|
|
|
|
|
|
|
|
// Cache essential information about the motion event to ensure that a malicious consumer
|
|
|
|
// cannot confuse the publisher by modifying the contents of the shared memory buffer while
|
|
|
|
// it is being updated.
|
|
|
|
if (action == MOTION_EVENT_ACTION_MOVE) {
|
|
|
|
mMotionEventPointerCount = pointerCount;
|
|
|
|
mMotionEventSampleDataStride = InputMessage::sampleDataStride(pointerCount);
|
|
|
|
mMotionEventSampleDataTail = InputMessage::sampleDataPtrIncrement(
|
|
|
|
mSharedMessage->motion.sampleData, mMotionEventSampleDataStride);
|
|
|
|
} else {
|
|
|
|
mMotionEventSampleDataTail = NULL;
|
|
|
|
}
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputPublisher::appendMotionSample(
|
|
|
|
nsecs_t eventTime,
|
|
|
|
const PointerCoords* pointerCoords) {
|
|
|
|
#if DEBUG_TRANSPORT_ACTIONS
|
|
|
|
LOGD("channel '%s' publisher ~ appendMotionSample: eventTime=%lld",
|
|
|
|
mChannel->getName().string(), eventTime);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (! mPinned || ! mMotionEventSampleDataTail) {
|
|
|
|
LOGE("channel '%s' publisher ~ Cannot append motion sample because there is no current "
|
|
|
|
"MOTION_EVENT_ACTION_MOVE event.", mChannel->getName().string());
|
|
|
|
return INVALID_OPERATION;
|
|
|
|
}
|
|
|
|
|
|
|
|
InputMessage::SampleData* newTail = InputMessage::sampleDataPtrIncrement(
|
|
|
|
mMotionEventSampleDataTail, mMotionEventSampleDataStride);
|
|
|
|
size_t newBytesUsed = reinterpret_cast<char*>(newTail) -
|
|
|
|
reinterpret_cast<char*>(mSharedMessage);
|
|
|
|
|
|
|
|
if (newBytesUsed > mAshmemSize) {
|
2010-06-22 08:27:15 +00:00
|
|
|
#if DEBUG_TRANSPORT_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
|
|
|
LOGD("channel '%s' publisher ~ Cannot append motion sample because the shared memory "
|
|
|
|
"buffer is full. Buffer size: %d bytes, pointers: %d, samples: %d",
|
|
|
|
mChannel->getName().string(),
|
|
|
|
mAshmemSize, mMotionEventPointerCount, mSharedMessage->motion.sampleCount);
|
2010-06-22 08:27:15 +00:00
|
|
|
#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
|
|
|
return NO_MEMORY;
|
|
|
|
}
|
|
|
|
|
|
|
|
int result;
|
|
|
|
if (mWasDispatched) {
|
|
|
|
result = sem_trywait(& mSharedMessage->semaphore);
|
|
|
|
if (result < 0) {
|
|
|
|
if (errno == EAGAIN) {
|
|
|
|
// Only possible source of contention is the consumer having consumed (or being in the
|
|
|
|
// process of consuming) the message and left the semaphore count at 0.
|
2010-06-22 08:27:15 +00:00
|
|
|
#if DEBUG_TRANSPORT_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
|
|
|
LOGD("channel '%s' publisher ~ Cannot append motion sample because the message has "
|
|
|
|
"already been consumed.", mChannel->getName().string());
|
2010-06-22 08:27:15 +00:00
|
|
|
#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
|
|
|
return FAILED_TRANSACTION;
|
|
|
|
} else {
|
|
|
|
LOGE("channel '%s' publisher ~ Error %d in sem_trywait.",
|
|
|
|
mChannel->getName().string(), errno);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
mMotionEventSampleDataTail->eventTime = eventTime;
|
|
|
|
for (size_t i = 0; i < mMotionEventPointerCount; i++) {
|
|
|
|
mMotionEventSampleDataTail->coords[i] = pointerCoords[i];
|
|
|
|
}
|
|
|
|
mMotionEventSampleDataTail = newTail;
|
|
|
|
|
|
|
|
mSharedMessage->motion.sampleCount += 1;
|
|
|
|
|
|
|
|
if (mWasDispatched) {
|
|
|
|
result = sem_post(& mSharedMessage->semaphore);
|
|
|
|
if (result < 0) {
|
|
|
|
LOGE("channel '%s' publisher ~ Error %d in sem_post.",
|
|
|
|
mChannel->getName().string(), errno);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputPublisher::sendDispatchSignal() {
|
|
|
|
#if DEBUG_TRANSPORT_ACTIONS
|
|
|
|
LOGD("channel '%s' publisher ~ sendDispatchSignal",
|
|
|
|
mChannel->getName().string());
|
|
|
|
#endif
|
|
|
|
|
|
|
|
mWasDispatched = true;
|
|
|
|
return mChannel->sendSignal(INPUT_SIGNAL_DISPATCH);
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputPublisher::receiveFinishedSignal() {
|
|
|
|
#if DEBUG_TRANSPORT_ACTIONS
|
|
|
|
LOGD("channel '%s' publisher ~ receiveFinishedSignal",
|
|
|
|
mChannel->getName().string());
|
|
|
|
#endif
|
|
|
|
|
|
|
|
char signal;
|
|
|
|
status_t result = mChannel->receiveSignal(& signal);
|
|
|
|
if (result) {
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
if (signal != INPUT_SIGNAL_FINISHED) {
|
|
|
|
LOGE("channel '%s' publisher ~ Received unexpected signal '%c' from consumer",
|
|
|
|
mChannel->getName().string(), signal);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
// --- InputConsumer ---
|
|
|
|
|
|
|
|
InputConsumer::InputConsumer(const sp<InputChannel>& channel) :
|
|
|
|
mChannel(channel), mSharedMessage(NULL) {
|
|
|
|
}
|
|
|
|
|
|
|
|
InputConsumer::~InputConsumer() {
|
|
|
|
if (mSharedMessage) {
|
|
|
|
munmap(mSharedMessage, mAshmemSize);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputConsumer::initialize() {
|
|
|
|
#if DEBUG_TRANSPORT_ACTIONS
|
|
|
|
LOGD("channel '%s' consumer ~ initialize",
|
|
|
|
mChannel->getName().string());
|
|
|
|
#endif
|
|
|
|
|
|
|
|
int ashmemFd = mChannel->getAshmemFd();
|
|
|
|
int result = ashmem_get_size_region(ashmemFd);
|
|
|
|
if (result < 0) {
|
|
|
|
LOGE("channel '%s' consumer ~ Error %d getting size of ashmem fd %d.",
|
|
|
|
mChannel->getName().string(), result, ashmemFd);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
|
|
|
|
mAshmemSize = (size_t) result;
|
|
|
|
|
|
|
|
mSharedMessage = static_cast<InputMessage*>(mmap(NULL, mAshmemSize,
|
|
|
|
PROT_READ | PROT_WRITE, MAP_SHARED, ashmemFd, 0));
|
|
|
|
if (! mSharedMessage) {
|
|
|
|
LOGE("channel '%s' consumer ~ mmap failed on ashmem fd %d.",
|
|
|
|
mChannel->getName().string(), ashmemFd);
|
|
|
|
return NO_MEMORY;
|
|
|
|
}
|
|
|
|
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
|
2010-06-16 08:53:36 +00:00
|
|
|
status_t InputConsumer::consume(InputEventFactoryInterface* factory, InputEvent** outEvent) {
|
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_TRANSPORT_ACTIONS
|
|
|
|
LOGD("channel '%s' consumer ~ consume",
|
|
|
|
mChannel->getName().string());
|
|
|
|
#endif
|
|
|
|
|
2010-06-16 08:53:36 +00:00
|
|
|
*outEvent = NULL;
|
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
|
|
|
|
|
|
|
int ashmemFd = mChannel->getAshmemFd();
|
|
|
|
int result = ashmem_pin_region(ashmemFd, 0, 0);
|
|
|
|
if (result != ASHMEM_NOT_PURGED) {
|
|
|
|
if (result == ASHMEM_WAS_PURGED) {
|
|
|
|
LOGE("channel '%s' consumer ~ Error %d pinning ashmem fd %d because it was purged "
|
|
|
|
"which probably indicates that the publisher and consumer are out of sync.",
|
|
|
|
mChannel->getName().string(), result, ashmemFd);
|
|
|
|
return INVALID_OPERATION;
|
|
|
|
}
|
|
|
|
|
|
|
|
LOGE("channel '%s' consumer ~ Error %d pinning ashmem fd %d.",
|
|
|
|
mChannel->getName().string(), result, ashmemFd);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (mSharedMessage->consumed) {
|
|
|
|
LOGE("channel '%s' consumer ~ The current message has already been consumed.",
|
|
|
|
mChannel->getName().string());
|
|
|
|
return INVALID_OPERATION;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Acquire but *never release* the semaphore. Contention on the semaphore is used to signal
|
|
|
|
// to the publisher that the message has been consumed (or is in the process of being
|
|
|
|
// consumed). Eventually the publisher will reinitialize the semaphore for the next message.
|
|
|
|
result = sem_wait(& mSharedMessage->semaphore);
|
|
|
|
if (result < 0) {
|
|
|
|
LOGE("channel '%s' consumer ~ Error %d in sem_wait.",
|
|
|
|
mChannel->getName().string(), errno);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
|
|
|
|
mSharedMessage->consumed = true;
|
|
|
|
|
|
|
|
switch (mSharedMessage->type) {
|
|
|
|
case INPUT_EVENT_TYPE_KEY: {
|
|
|
|
KeyEvent* keyEvent = factory->createKeyEvent();
|
|
|
|
if (! keyEvent) return NO_MEMORY;
|
|
|
|
|
|
|
|
populateKeyEvent(keyEvent);
|
|
|
|
|
2010-06-16 08:53:36 +00:00
|
|
|
*outEvent = keyEvent;
|
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 INPUT_EVENT_TYPE_MOTION: {
|
|
|
|
MotionEvent* motionEvent = factory->createMotionEvent();
|
|
|
|
if (! motionEvent) return NO_MEMORY;
|
|
|
|
|
|
|
|
populateMotionEvent(motionEvent);
|
|
|
|
|
2010-06-16 08:53:36 +00:00
|
|
|
*outEvent = motionEvent;
|
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:
|
|
|
|
LOGE("channel '%s' consumer ~ Received message of unknown type %d",
|
|
|
|
mChannel->getName().string(), mSharedMessage->type);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputConsumer::sendFinishedSignal() {
|
|
|
|
#if DEBUG_TRANSPORT_ACTIONS
|
|
|
|
LOGD("channel '%s' consumer ~ sendFinishedSignal",
|
|
|
|
mChannel->getName().string());
|
|
|
|
#endif
|
|
|
|
|
|
|
|
return mChannel->sendSignal(INPUT_SIGNAL_FINISHED);
|
|
|
|
}
|
|
|
|
|
|
|
|
status_t InputConsumer::receiveDispatchSignal() {
|
|
|
|
#if DEBUG_TRANSPORT_ACTIONS
|
|
|
|
LOGD("channel '%s' consumer ~ receiveDispatchSignal",
|
|
|
|
mChannel->getName().string());
|
|
|
|
#endif
|
|
|
|
|
|
|
|
char signal;
|
|
|
|
status_t result = mChannel->receiveSignal(& signal);
|
|
|
|
if (result) {
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
if (signal != INPUT_SIGNAL_DISPATCH) {
|
|
|
|
LOGE("channel '%s' consumer ~ Received unexpected signal '%c' from publisher",
|
|
|
|
mChannel->getName().string(), signal);
|
|
|
|
return UNKNOWN_ERROR;
|
|
|
|
}
|
|
|
|
return OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
void InputConsumer::populateKeyEvent(KeyEvent* keyEvent) const {
|
|
|
|
keyEvent->initialize(
|
|
|
|
mSharedMessage->deviceId,
|
|
|
|
mSharedMessage->nature,
|
|
|
|
mSharedMessage->key.action,
|
|
|
|
mSharedMessage->key.flags,
|
|
|
|
mSharedMessage->key.keyCode,
|
|
|
|
mSharedMessage->key.scanCode,
|
|
|
|
mSharedMessage->key.metaState,
|
|
|
|
mSharedMessage->key.repeatCount,
|
|
|
|
mSharedMessage->key.downTime,
|
|
|
|
mSharedMessage->key.eventTime);
|
|
|
|
}
|
|
|
|
|
|
|
|
void InputConsumer::populateMotionEvent(MotionEvent* motionEvent) const {
|
|
|
|
motionEvent->initialize(
|
|
|
|
mSharedMessage->deviceId,
|
|
|
|
mSharedMessage->nature,
|
|
|
|
mSharedMessage->motion.action,
|
|
|
|
mSharedMessage->motion.edgeFlags,
|
|
|
|
mSharedMessage->motion.metaState,
|
2010-06-16 08:53:36 +00:00
|
|
|
mSharedMessage->motion.xOffset,
|
|
|
|
mSharedMessage->motion.yOffset,
|
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
|
|
|
mSharedMessage->motion.xPrecision,
|
|
|
|
mSharedMessage->motion.yPrecision,
|
|
|
|
mSharedMessage->motion.downTime,
|
|
|
|
mSharedMessage->motion.sampleData[0].eventTime,
|
|
|
|
mSharedMessage->motion.pointerCount,
|
|
|
|
mSharedMessage->motion.pointerIds,
|
|
|
|
mSharedMessage->motion.sampleData[0].coords);
|
|
|
|
|
|
|
|
size_t sampleCount = mSharedMessage->motion.sampleCount;
|
|
|
|
if (sampleCount > 1) {
|
|
|
|
InputMessage::SampleData* sampleData = mSharedMessage->motion.sampleData;
|
|
|
|
size_t sampleDataStride = InputMessage::sampleDataStride(
|
|
|
|
mSharedMessage->motion.pointerCount);
|
|
|
|
|
|
|
|
while (--sampleCount > 0) {
|
|
|
|
sampleData = InputMessage::sampleDataPtrIncrement(sampleData, sampleDataStride);
|
|
|
|
motionEvent->addSample(sampleData->eventTime, sampleData->coords);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
} // namespace android
|