Naomi Luis d8588e1c12 overlay: Close correct channels during state change events.
When there is a state change, open/close only the required channels
instead of blindly closing all the channels. When HDMI is connected,
open the overlay for queueing the frames on the external, and when
hdmi is disconnected, only close the external overlay channel.

CRs-fixed: 333324
(cherry picked from commit a6b5ac4c385e03b3c798bb07d0d3ba6140aebbe8)

Change-Id: I7333a6fe106cd0aa99471542daf3c84c04c5d056
2012-03-16 23:13:48 -05:00

1493 lines
51 KiB
C++

/*
* Copyright (C) 2010 The Android Open Source Project
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include <stdlib.h>
#include <stdint.h>
#include <string.h>
#include <unistd.h>
#include <hardware/hardware.h>
#include <fcntl.h>
#include <errno.h>
#include <cutils/log.h>
#include <cutils/atomic.h>
#include <cutils/properties.h>
#include <hardware/hwcomposer.h>
#include <overlayLib.h>
#include <overlayLibUI.h>
#include <copybit.h>
#include <EGL/egl.h>
#include <EGL/eglext.h>
#include <ui/android_native_buffer.h>
#include <gralloc_priv.h>
#include <genlock.h>
#include <qcom_ui.h>
#include <gr.h>
/*****************************************************************************/
#define ALIGN(x, align) (((x) + ((align)-1)) & ~((align)-1))
#define LIKELY( exp ) (__builtin_expect( (exp) != 0, true ))
#define UNLIKELY( exp ) (__builtin_expect( (exp) != 0, false ))
enum HWCLayerType{
HWC_SINGLE_VIDEO = 0x1,
HWC_ORIG_RESOLUTION = 0x2,
HWC_S3D_LAYER = 0x4,
HWC_STOP_UI_MIRRORING_MASK = 0xF
};
#ifdef COMPOSITION_BYPASS
enum BypassState {
BYPASS_ON,
BYPASS_OFF,
BYPASS_OFF_PENDING,
};
enum {
MAX_BYPASS_LAYERS = 2,
ANIM_FRAME_COUNT = 30,
};
enum BypassBufferLockState {
BYPASS_BUFFER_UNLOCKED,
BYPASS_BUFFER_LOCKED,
};
#endif
enum eHWCOverlayStatus {
HWC_OVERLAY_OPEN,
HWC_OVERLAY_PREPARE_TO_CLOSE,
HWC_OVERLAY_CLOSED
};
struct hwc_context_t {
hwc_composer_device_t device;
/* our private state goes below here */
overlay::Overlay* mOverlayLibObject;
native_handle_t *previousOverlayHandle;
#ifdef COMPOSITION_BYPASS
overlay::OverlayUI* mOvUI[MAX_BYPASS_LAYERS];
native_handle_t* previousBypassHandle[MAX_BYPASS_LAYERS];
BypassBufferLockState bypassBufferLockState[MAX_BYPASS_LAYERS];
int animCount;
BypassState bypassState;
#endif
#if defined HDMI_DUAL_DISPLAY
external_display mHDMIEnabled; // Type of external display
bool pendingHDMI;
#endif
int previousLayerCount;
eHWCOverlayStatus hwcOverlayStatus;
};
static int hwc_device_open(const struct hw_module_t* module, const char* name,
struct hw_device_t** device);
static struct hw_module_methods_t hwc_module_methods = {
open: hwc_device_open
};
struct private_hwc_module_t {
hwc_module_t base;
copybit_device_t *copybitEngine;
framebuffer_device_t *fbDevice;
int compositionType;
bool isBypassEnabled; //from build.prop ro.sf.compbypass.enable
};
struct private_hwc_module_t HAL_MODULE_INFO_SYM = {
base: {
common: {
tag: HARDWARE_MODULE_TAG,
version_major: 1,
version_minor: 0,
id: HWC_HARDWARE_MODULE_ID,
name: "Hardware Composer Module",
author: "The Android Open Source Project",
methods: &hwc_module_methods,
}
},
copybitEngine: NULL,
fbDevice: NULL,
compositionType: 0,
isBypassEnabled: false,
};
/*****************************************************************************/
static void dump_layer(hwc_layer_t const* l) {
LOGD("\ttype=%d, flags=%08x, handle=%p, tr=%02x, blend=%04x, {%d,%d,%d,%d}, {%d,%d,%d,%d}",
l->compositionType, l->flags, l->handle, l->transform, l->blending,
l->sourceCrop.left,
l->sourceCrop.top,
l->sourceCrop.right,
l->sourceCrop.bottom,
l->displayFrame.left,
l->displayFrame.top,
l->displayFrame.right,
l->displayFrame.bottom);
}
static inline int min(const int& a, const int& b) {
return (a < b) ? a : b;
}
static inline int max(const int& a, const int& b) {
return (a > b) ? a : b;
}
static int setVideoOverlayStatusInGralloc(hwc_context_t* ctx, const bool enable) {
#if defined HDMI_DUAL_DISPLAY
private_hwc_module_t* hwcModule = reinterpret_cast<private_hwc_module_t*>(
ctx->device.common.module);
if(!hwcModule) {
LOGE("%s: invalid params", __FUNCTION__);
return -1;
}
framebuffer_device_t *fbDev = hwcModule->fbDevice;
if (!fbDev) {
LOGE("%s: fbDev is NULL", __FUNCTION__);
return -1;
}
// Inform the gralloc to stop or start UI mirroring
fbDev->videoOverlayStarted(fbDev, enable);
#endif
return 0;
}
static void setHWCOverlayStatus(hwc_context_t *ctx, bool isVideoPresent) {
switch (ctx->hwcOverlayStatus) {
case HWC_OVERLAY_OPEN:
ctx->hwcOverlayStatus =
isVideoPresent ? HWC_OVERLAY_OPEN : HWC_OVERLAY_PREPARE_TO_CLOSE;
break;
case HWC_OVERLAY_PREPARE_TO_CLOSE:
ctx->hwcOverlayStatus =
isVideoPresent ? HWC_OVERLAY_OPEN : HWC_OVERLAY_CLOSED;
break;
case HWC_OVERLAY_CLOSED:
ctx->hwcOverlayStatus =
isVideoPresent ? HWC_OVERLAY_OPEN : HWC_OVERLAY_CLOSED;
break;
default:
LOGE("%s: Invalid hwcOverlayStatus (status =%d)", __FUNCTION__,
ctx->hwcOverlayStatus);
break;
}
}
static int hwc_closeOverlayChannels(hwc_context_t* ctx) {
#ifdef USE_OVERLAY
overlay::Overlay *ovLibObject = ctx->mOverlayLibObject;
if(!ovLibObject) {
LOGE("%s: invalid params", __FUNCTION__);
return -1;
}
if (HWC_OVERLAY_PREPARE_TO_CLOSE == ctx->hwcOverlayStatus) {
// Video mirroring is going on, and we do not have any layers to
// mirror directly. Close the current video channel and inform the
// gralloc to start UI mirroring
ovLibObject->closeChannel();
// Inform the gralloc that video overlay has stopped.
setVideoOverlayStatusInGralloc(ctx, false);
}
#endif
return 0;
}
#ifdef COMPOSITION_BYPASS
// To-do: Merge this with other blocks & move them to a separate file.
void unlockPreviousBypassBuffers(hwc_context_t* ctx) {
// Unlock the previous bypass buffers. We can blindly unlock the buffers here,
// because buffers will be in this list only if the lock was successfully acquired.
for(int i = 0; i < MAX_BYPASS_LAYERS; i++) {
if (ctx->previousBypassHandle[i]) {
private_handle_t *hnd = (private_handle_t*) ctx->previousBypassHandle[i];
// Validate the handle to make sure it hasn't been deallocated.
if (private_handle_t::validate(ctx->previousBypassHandle[i])) {
continue;
}
// Check if the handle was locked previously
if (private_handle_t::PRIV_FLAGS_HWC_LOCK & hnd->flags) {
if (GENLOCK_FAILURE == genlock_unlock_buffer(ctx->previousBypassHandle[i])) {
LOGE("%s: genlock_unlock_buffer failed", __FUNCTION__);
} else {
ctx->previousBypassHandle[i] = NULL;
// Reset the lock flag
hnd->flags &= ~private_handle_t::PRIV_FLAGS_HWC_LOCK;
}
}
}
}
}
void closeBypass(hwc_context_t* ctx) {
unlockPreviousBypassBuffers(ctx);
for (int index = 0 ; index < MAX_BYPASS_LAYERS; index++) {
ctx->mOvUI[index]->closeChannel();
}
#ifdef DEBUG
LOGE("%s", __FUNCTION__);
#endif
}
#endif
/*
* Configures mdp pipes
*/
static int prepareOverlay(hwc_context_t *ctx, hwc_layer_t *layer, const bool waitForVsync) {
int ret = 0;
#ifdef COMPOSITION_BYPASS
if(ctx && (ctx->bypassState != BYPASS_OFF)) {
closeBypass(ctx);
ctx->bypassState = BYPASS_OFF;
}
#endif
if (LIKELY(ctx && ctx->mOverlayLibObject)) {
private_hwc_module_t* hwcModule =
reinterpret_cast<private_hwc_module_t*>(ctx->device.common.module);
if (UNLIKELY(!hwcModule)) {
LOGE("prepareOverlay null module ");
return -1;
}
private_handle_t *hnd = (private_handle_t *)layer->handle;
overlay::Overlay *ovLibObject = ctx->mOverlayLibObject;
overlay_buffer_info info;
info.width = hnd->width;
info.height = hnd->height;
info.format = hnd->format;
info.size = hnd->size;
info.secure = (hnd->flags &
private_handle_t::PRIV_FLAGS_SECURE_BUFFER)? true:false;
int hdmiConnected = 0;
#if defined HDMI_DUAL_DISPLAY
hdmiConnected = (int)ctx->mHDMIEnabled;
#endif
ret = ovLibObject->setSource(info, layer->transform,
hdmiConnected, waitForVsync);
if (!ret) {
LOGE("prepareOverlay setSource failed");
return -1;
}
ret = ovLibObject->setTransform(layer->transform);
if (!ret) {
LOGE("prepareOverlay setTransform failed transform %x",
layer->transform);
return -1;
}
hwc_rect_t sourceCrop = layer->sourceCrop;
ret = ovLibObject->setCrop(sourceCrop.left, sourceCrop.top,
(sourceCrop.right - sourceCrop.left),
(sourceCrop.bottom - sourceCrop.top));
if (!ret) {
LOGE("prepareOverlay setCrop failed");
return -1;
}
#if defined HDMI_DUAL_DISPLAY
// Send the device orientation to overlayLib
if(hwcModule) {
framebuffer_device_t *fbDev = reinterpret_cast<framebuffer_device_t*>
(hwcModule->fbDevice);
if(fbDev) {
private_module_t* m = reinterpret_cast<private_module_t*>(
fbDev->common.module);
if(m)
ovLibObject->setDeviceOrientation(m->orientation);
}
}
#endif
if (layer->flags & HWC_USE_ORIGINAL_RESOLUTION) {
framebuffer_device_t* fbDev = hwcModule->fbDevice;
ret = ovLibObject->setPosition(0, 0,
fbDev->width, fbDev->height);
} else {
hwc_rect_t displayFrame = layer->displayFrame;
ret = ovLibObject->setPosition(displayFrame.left, displayFrame.top,
(displayFrame.right - displayFrame.left),
(displayFrame.bottom - displayFrame.top));
}
if (!ret) {
LOGE("prepareOverlay setPosition failed");
return -1;
}
}
return 0;
}
void unlockPreviousOverlayBuffer(hwc_context_t* ctx)
{
if (ctx->previousOverlayHandle) {
// Validate the handle before attempting to use it.
if (!private_handle_t::validate(ctx->previousOverlayHandle)) {
private_handle_t *hnd = (private_handle_t*)ctx->previousOverlayHandle;
// Unlock any previously locked buffers
if (private_handle_t::PRIV_FLAGS_HWC_LOCK & hnd->flags) {
if (GENLOCK_NO_ERROR == genlock_unlock_buffer(ctx->previousOverlayHandle)) {
ctx->previousOverlayHandle = NULL;
hnd->flags &= ~private_handle_t::PRIV_FLAGS_HWC_LOCK;
} else {
LOGE("%s: genlock_unlock_buffer failed", __FUNCTION__);
}
}
}
}
}
bool canSkipComposition(hwc_context_t* ctx, int yuvBufferCount, int currentLayerCount,
int numLayersNotUpdating)
{
if (!ctx) {
LOGE("canSkipComposition invalid context");
return false;
}
bool compCountChanged = false;
if (yuvBufferCount == 1) {
if (currentLayerCount != ctx->previousLayerCount) {
compCountChanged = true;
ctx->previousLayerCount = currentLayerCount;
}
if (!compCountChanged) {
if ((currentLayerCount == 1) ||
((currentLayerCount-1) == numLayersNotUpdating)) {
// We either have only one overlay layer or we have
// all the non-UI layers not updating. In this case
// we can skip the composition of the UI layers.
return true;
}
}
} else {
ctx->previousLayerCount = -1;
}
return false;
}
inline void getLayerResolution(const hwc_layer_t* layer, int& width, int& height)
{
hwc_rect_t displayFrame = layer->displayFrame;
width = displayFrame.right - displayFrame.left;
height = displayFrame.bottom - displayFrame.top;
}
static bool canUseCopybit(const framebuffer_device_t* fbDev, const hwc_layer_list_t* list) {
if(!fbDev) {
LOGE("ERROR: %s : fb device is invalid",__func__);
return false;
}
int fb_w = fbDev->width;
int fb_h = fbDev->height;
/*
* We can use copybit when
* 1. We have 1 layer to compose
* 2. We have 2 layers to compose
* a. Sum of both layers covers full screen
* b. One of the layers is full screen and the
* other is less than full screen (includes
* pop ups, volume bar etc.)
* TODO: Need to revisit this logic to use copybit
* based on the total blitting region instead of total
* layers count
*/
bool use_copybit = (list->numHwLayers == 1);
if(list->numHwLayers == 2) {
int w1, h1;
int w2, h2;
getLayerResolution(&list->hwLayers[0], w1, h1);
getLayerResolution(&list->hwLayers[1], w2, h2);
use_copybit = ((fb_w >= w1) && (fb_w >= w2) && ((fb_h * 2) > (h1 + h2)));
}
return use_copybit;
}
#ifdef COMPOSITION_BYPASS
/*
* Configures pipe(s) for composition bypass
*/
static int prepareBypass(hwc_context_t *ctx, hwc_layer_t *layer, int index,
int lastLayerIndex) {
if (ctx && ctx->mOvUI[index]) {
private_hwc_module_t* hwcModule = reinterpret_cast<
private_hwc_module_t*>(ctx->device.common.module);
if (!hwcModule) {
LOGE("prepareBypass null module ");
return -1;
}
private_handle_t *hnd = (private_handle_t *)layer->handle;
if(!hnd) {
LOGE("prepareBypass handle null");
return -1;
}
hwc_rect_t sourceCrop = layer->sourceCrop;
if((sourceCrop.right - sourceCrop.left) > hwcModule->fbDevice->width ||
(sourceCrop.bottom - sourceCrop.top) > hwcModule->fbDevice->height) {
ctx->animCount = ANIM_FRAME_COUNT;
return -1;
}
overlay::OverlayUI *ovUI = ctx->mOvUI[index];
int ret = 0;
int orientation = layer->transform;
overlay_buffer_info info;
info.width = sourceCrop.right - sourceCrop.left;
info.height = sourceCrop.bottom - sourceCrop.top;
info.format = hnd->format;
info.size = hnd->size;
info.secure = (hnd->flags &
private_handle_t::PRIV_FLAGS_SECURE_BUFFER)? true:false;
const bool useVGPipe = true;
//only last layer should wait for vsync
const bool waitForVsync = (index == lastLayerIndex);
const int fbnum = 0;
const bool isFg = (index == 0);
//Just to differentiate zorders for different layers
const int zorder = index;
const bool isVGPipe = true;
ovUI->setSource(info, orientation);
ovUI->setDisplayParams(fbnum, waitForVsync, isFg, zorder, isVGPipe);
hwc_rect_t displayFrame = layer->displayFrame;
ovUI->setPosition(displayFrame.left, displayFrame.top,
(displayFrame.right - displayFrame.left),
(displayFrame.bottom - displayFrame.top));
if(ovUI->commit() != overlay::NO_ERROR) {
LOGE("%s: Bypass Overlay Commit failed", __FUNCTION__);
return -1;
}
}
return 0;
}
static int drawLayerUsingBypass(hwc_context_t *ctx, hwc_layer_t *layer,
int index) {
if (ctx && ctx->mOvUI[index]) {
overlay::OverlayUI *ovUI = ctx->mOvUI[index];
int ret = 0;
private_handle_t *hnd = (private_handle_t *)layer->handle;
ctx->bypassBufferLockState[index] = BYPASS_BUFFER_UNLOCKED;
if (GENLOCK_FAILURE == genlock_lock_buffer(hnd, GENLOCK_READ_LOCK,
GENLOCK_MAX_TIMEOUT)) {
LOGE("%s: genlock_lock_buffer(READ) failed", __FUNCTION__);
return -1;
}
ctx->bypassBufferLockState[index] = BYPASS_BUFFER_LOCKED;
ret = ovUI->queueBuffer(hnd);
if (ret) {
LOGE("drawLayerUsingBypass queueBuffer failed");
// Unlock the locked buffer
if (GENLOCK_FAILURE == genlock_unlock_buffer(hnd)) {
LOGE("%s: genlock_unlock_buffer failed", __FUNCTION__);
}
ctx->bypassBufferLockState[index] = BYPASS_BUFFER_UNLOCKED;
return -1;
}
}
return 0;
}
/* Checks if 2 layers intersect */
static bool isIntersect(const hwc_rect_t& one, const hwc_rect_t& two) {
hwc_rect_t result;
result.left = max(one.left, two.left);
result.top = max(one.top, two.top);
result.right = min(one.right, two.right);
result.bottom = min(one.bottom, two.bottom);
const int width = result.right - result.left;
const int height = result.bottom - result.top;
const bool isEmpty = width <= 0 || height <= 0;
return !isEmpty;
}
/* Check if layers are disjoint */
static bool isDisjoint(const hwc_layer_list_t* list) {
//Validate supported layer range
if(list->numHwLayers <= 0 || list->numHwLayers > MAX_BYPASS_LAYERS) {
return false;
}
for(int i = 0; i < (list->numHwLayers) - 1; i++) {
for(int j = i + 1; j < list->numHwLayers; j++) {
if(isIntersect(list->hwLayers[i].displayFrame,
list->hwLayers[j].displayFrame)) {
return false;
}
}
}
return true;
}
static bool usesContiguousMemory(const hwc_layer_list_t* list) {
for(int i = 0; i < list->numHwLayers; i++) {
const private_handle_t *hnd =
reinterpret_cast<const private_handle_t *>(list->hwLayers[i].handle);
if(hnd != NULL && (hnd->flags &
private_handle_t::PRIV_FLAGS_NONCONTIGUOUS_MEM
)) {
// Bypass cannot work for non contiguous buffers
return false;
}
}
return true;
}
/*
* Checks if doing comp. bypass is possible.
* It is possible if
* 1. If video is not on
* 2. There are 2 layers
* 3. The memory type is contiguous
*/
inline static bool isBypassDoable(hwc_composer_device_t *dev, const int yuvCount,
const hwc_layer_list_t* list) {
hwc_context_t* ctx = (hwc_context_t*)(dev);
private_hwc_module_t* hwcModule = reinterpret_cast<private_hwc_module_t*>(
dev->common.module);
//Check if enabled in build.prop
if(hwcModule->isBypassEnabled == false) {
return false;
}
// Check if memory type is contiguous
if(!usesContiguousMemory(list))
return false;
//Disable bypass during animation
if(UNLIKELY(ctx->animCount)) {
--(ctx->animCount);
return false;
}
#if defined HDMI_DUAL_DISPLAY
//Disable bypass when HDMI is enabled
if(ctx->mHDMIEnabled || ctx->pendingHDMI) {
return false;
}
#endif
return (yuvCount == 0) && (ctx->hwcOverlayStatus == HWC_OVERLAY_CLOSED) && isDisjoint(list);
}
/*
* Bypass is not efficient if area is greater than 1280x720
* AND rotation is necessary, since the rotator consumes
* time greater than 1 Vsync and is sequential.
*/
inline static bool isBypassEfficient(const framebuffer_device_t* fbDev,
const hwc_layer_list_t* list, hwc_context_t* ctx) {
bool rotationNeeded = false;
for(int i = 0; i < list->numHwLayers; ++i) {
if(list->hwLayers[i].transform) {
rotationNeeded = true;
break;
}
}
return !(rotationNeeded);
}
bool setupBypass(hwc_context_t* ctx, hwc_layer_list_t* list) {
for (int index = 0 ; index < list->numHwLayers; index++) {
if(prepareBypass(ctx, &(list->hwLayers[index]), index,
list->numHwLayers - 1) != 0) {
return false;
}
}
return true;
}
void setBypassLayerFlags(hwc_context_t* ctx, hwc_layer_list_t* list) {
for (int index = 0 ; index < list->numHwLayers; index++) {
list->hwLayers[index].flags |= HWC_COMP_BYPASS;
list->hwLayers[index].compositionType = HWC_USE_OVERLAY;
#ifdef DEBUG
LOGE("%s: layer = %d", __FUNCTION__, index);
#endif
}
}
void unsetBypassLayerFlags(hwc_layer_list_t* list) {
for (int index = 0 ; index < list->numHwLayers; index++) {
if(list->hwLayers[index].flags & HWC_COMP_BYPASS) {
list->hwLayers[index].flags = 0;
}
}
}
void unsetBypassBufferLockState(hwc_context_t* ctx) {
for (int i=0; i< MAX_BYPASS_LAYERS; i++) {
ctx->bypassBufferLockState[i] = BYPASS_BUFFER_UNLOCKED;
}
}
void storeLockedBypassHandle(hwc_layer_list_t* list, hwc_context_t* ctx) {
for (int index = 0 ; index < list->numHwLayers; index++) {
// Store the current bypass handle.
if (list->hwLayers[index].flags & HWC_COMP_BYPASS) {
private_handle_t *hnd = (private_handle_t*)list->hwLayers[index].handle;
if (ctx->bypassBufferLockState[index] == BYPASS_BUFFER_LOCKED) {
ctx->previousBypassHandle[index] = (native_handle_t*)list->hwLayers[index].handle;
hnd->flags |= private_handle_t::PRIV_FLAGS_HWC_LOCK;
} else
ctx->previousBypassHandle[index] = NULL;
}
}
}
#endif //COMPOSITION_BYPASS
static void handleHDMIStateChange(hwc_composer_device_t *dev, int externaltype) {
#if defined HDMI_DUAL_DISPLAY
hwc_context_t* ctx = (hwc_context_t*)(dev);
private_hwc_module_t* hwcModule = reinterpret_cast<private_hwc_module_t*>(
dev->common.module);
framebuffer_device_t *fbDev = hwcModule->fbDevice;
if (fbDev) {
fbDev->enableHDMIOutput(fbDev, externaltype);
}
if(ctx && ctx->mOverlayLibObject) {
overlay::Overlay *ovLibObject = ctx->mOverlayLibObject;
if (!externaltype) {
// Close the external overlay channels if HDMI is disconnected
ovLibObject->closeExternalChannel();
}
}
#endif
}
/*
* function to set the status of external display in hwc
* Just mark flags and do stuff after eglSwapBuffers
* externaltype - can be HDMI, WIFI or OFF
*/
static void hwc_enableHDMIOutput(hwc_composer_device_t *dev, int externaltype) {
#if defined HDMI_DUAL_DISPLAY
hwc_context_t* ctx = (hwc_context_t*)(dev);
private_hwc_module_t* hwcModule = reinterpret_cast<private_hwc_module_t*>(
dev->common.module);
framebuffer_device_t *fbDev = hwcModule->fbDevice;
overlay::Overlay *ovLibObject = ctx->mOverlayLibObject;
if(externaltype && (externaltype != ctx->mHDMIEnabled)) {
// Close the current external display - as the SF will
// prioritize and send the correct external display
handleHDMIStateChange(dev, 0);
}
// Store the external display
ctx->mHDMIEnabled = (external_display)externaltype;
if(ctx->mHDMIEnabled) { //On connect, allow bypass to draw once to FB
ctx->pendingHDMI = true;
} else { //On disconnect, close immediately (there will be no bypass)
handleHDMIStateChange(dev, ctx->mHDMIEnabled);
}
#endif
}
static bool isValidDestination(const framebuffer_device_t* fbDev, const hwc_rect_t& rect)
{
if (!fbDev) {
LOGE("%s: fbDev is null", __FUNCTION__);
return false;
}
int dest_width = (rect.right - rect.left);
int dest_height = (rect.bottom - rect.top);
if (rect.left < 0 || rect.right < 0 || rect.top < 0 || rect.bottom < 0
|| dest_width <= 0 || dest_height <= 0) {
LOGE("%s: destination: left=%d right=%d top=%d bottom=%d width=%d"
"height=%d", __FUNCTION__, rect.left, rect.right, rect.top,
rect.bottom, dest_width, dest_height);
return false;
}
if ((rect.left+dest_width) > fbDev->width || (rect.top+dest_height) > fbDev->height) {
LOGE("%s: destination out of bound params", __FUNCTION__);
return false;
}
return true;
}
static int getYUVBufferCount (const hwc_layer_list_t* list) {
int yuvBufferCount = 0;
if (list) {
for (size_t i=0 ; i<list->numHwLayers; i++) {
private_handle_t *hnd = (private_handle_t *)list->hwLayers[i].handle;
if (hnd && (hnd->bufferType == BUFFER_TYPE_VIDEO) &&
!(list->hwLayers[i].flags & HWC_DO_NOT_USE_OVERLAY)) {
yuvBufferCount++;
if (yuvBufferCount > 1) {
break;
}
}
}
}
return yuvBufferCount;
}
static int getS3DVideoFormat (const hwc_layer_list_t* list) {
int s3dFormat = 0;
if (list) {
for (size_t i=0; i<list->numHwLayers; i++) {
private_handle_t *hnd = (private_handle_t *)list->hwLayers[i].handle;
if (hnd && (hnd->bufferType == BUFFER_TYPE_VIDEO))
s3dFormat = FORMAT_3D_INPUT(hnd->format);
if (s3dFormat)
break;
}
}
return s3dFormat;
}
static int getS3DFormat (const hwc_layer_list_t* list) {
int s3dFormat = 0;
if (list) {
for (size_t i=0; i<list->numHwLayers; i++) {
private_handle_t *hnd = (private_handle_t *)list->hwLayers[i].handle;
if (hnd)
s3dFormat = FORMAT_3D_INPUT(hnd->format);
if (s3dFormat)
break;
}
}
return s3dFormat;
}
static int getLayerS3DFormat (hwc_layer_t &layer) {
int s3dFormat = 0;
private_handle_t *hnd = (private_handle_t *)layer.handle;
if (hnd)
s3dFormat = FORMAT_3D_INPUT(hnd->format);
return s3dFormat;
}
static bool isS3DCompositionRequired() {
#ifdef HDMI_AS_PRIMARY
return overlay::is3DTV();
#endif
return false;
}
static void markUILayerForS3DComposition (hwc_layer_t &layer, int s3dVideoFormat) {
#ifdef HDMI_AS_PRIMARY
layer.compositionType = HWC_FRAMEBUFFER;
switch(s3dVideoFormat) {
case HAL_3D_IN_SIDE_BY_SIDE_L_R:
case HAL_3D_IN_SIDE_BY_SIDE_R_L:
layer.hints |= HWC_HINT_DRAW_S3D_SIDE_BY_SIDE;
break;
case HAL_3D_IN_TOP_BOTTOM:
layer.hints |= HWC_HINT_DRAW_S3D_TOP_BOTTOM;
break;
default:
LOGE("%s: Unknown S3D input format 0x%x", __FUNCTION__, s3dVideoFormat);
break;
}
#endif
return;
}
static int getLayersNotUpdatingCount(const hwc_layer_list_t* list) {
int numLayersNotUpdating = 0;
if (list) {
for (size_t i=0 ; i<list->numHwLayers; i++) {
private_handle_t *hnd = (private_handle_t *)list->hwLayers[i].handle;
if (hnd && (hnd->bufferType != BUFFER_TYPE_VIDEO) &&
list->hwLayers[i].flags & HWC_LAYER_NOT_UPDATING)
numLayersNotUpdating++;
}
}
return numLayersNotUpdating;
}
static int hwc_prepare(hwc_composer_device_t *dev, hwc_layer_list_t* list) {
hwc_context_t* ctx = (hwc_context_t*)(dev);
if(!ctx) {
LOGE("hwc_prepare invalid context");
return -1;
}
private_hwc_module_t* hwcModule = reinterpret_cast<private_hwc_module_t*>(
dev->common.module);
if (!list || !hwcModule) {
LOGE("hwc_prepare invalid list or module");
#ifdef COMPOSITION_BYPASS
unlockPreviousBypassBuffers(ctx);
unsetBypassBufferLockState(ctx);
#endif
unlockPreviousOverlayBuffer(ctx);
return -1;
}
int yuvBufferCount = 0;
int layerType = 0;
bool isS3DCompositionNeeded = false;
int s3dVideoFormat = 0;
int numLayersNotUpdating = 0;
bool useCopybit = false;
if (list) {
useCopybit = canUseCopybit(hwcModule->fbDevice, list);
yuvBufferCount = getYUVBufferCount(list);
bool skipComposition = false;
if (yuvBufferCount == 1) {
numLayersNotUpdating = getLayersNotUpdatingCount(list);
skipComposition = canSkipComposition(ctx, yuvBufferCount,
list->numHwLayers, numLayersNotUpdating);
s3dVideoFormat = getS3DVideoFormat(list);
if (s3dVideoFormat)
isS3DCompositionNeeded = isS3DCompositionRequired();
} else if((s3dVideoFormat = getS3DFormat(list))){
if (s3dVideoFormat)
isS3DCompositionNeeded = isS3DCompositionRequired();
} else {
unlockPreviousOverlayBuffer(ctx);
}
if (list->flags & HWC_GEOMETRY_CHANGED) {
if (yuvBufferCount == 1) {
// Inform the gralloc of the current video overlay status
setVideoOverlayStatusInGralloc(ctx, true);
}
}
for (size_t i=0 ; i<list->numHwLayers ; i++) {
private_handle_t *hnd = (private_handle_t *)list->hwLayers[i].handle;
// If there is a single Fullscreen layer, we can bypass it - TBD
// If there is only one video/camera buffer, we can bypass itn
if (list->hwLayers[i].flags & HWC_SKIP_LAYER) {
// During the animaton UI layers are marked as SKIP
// need to still mark the layer for S3D composition
if (isS3DCompositionNeeded)
markUILayerForS3DComposition(list->hwLayers[i], s3dVideoFormat);
ssize_t layer_countdown = ((ssize_t)i) - 1;
// Mark every layer below the SKIP layer to be composed by the GPU
while (layer_countdown >= 0)
{
private_handle_t *countdown_handle =
(private_handle_t *)list->hwLayers[layer_countdown].handle;
if (countdown_handle && (countdown_handle->bufferType == BUFFER_TYPE_VIDEO)
&& (yuvBufferCount == 1)) {
unlockPreviousOverlayBuffer(ctx);
skipComposition = false;
}
list->hwLayers[layer_countdown].compositionType = HWC_FRAMEBUFFER;
list->hwLayers[layer_countdown].hints &= ~HWC_HINT_CLEAR_FB;
layer_countdown--;
}
continue;
}
if (hnd && (hnd->bufferType == BUFFER_TYPE_VIDEO) && (yuvBufferCount == 1)) {
bool waitForVsync = true;
if (!isValidDestination(hwcModule->fbDevice, list->hwLayers[i].displayFrame)) {
list->hwLayers[i].compositionType = HWC_FRAMEBUFFER;
#ifdef USE_OVERLAY
} else if(prepareOverlay(ctx, &(list->hwLayers[i]), waitForVsync) == 0) {
list->hwLayers[i].compositionType = HWC_USE_OVERLAY;
list->hwLayers[i].hints |= HWC_HINT_CLEAR_FB;
// We've opened the channel. Set the state to open.
ctx->hwcOverlayStatus = HWC_OVERLAY_OPEN;
#endif
}
else if (hwcModule->compositionType & (COMPOSITION_TYPE_C2D|
COMPOSITION_TYPE_MDP)) {
//Fail safe path: If drawing with overlay fails,
//Use C2D if available.
list->hwLayers[i].compositionType = HWC_USE_COPYBIT;
}
else {
//If C2D is not enabled fall back to GPU.
list->hwLayers[i].compositionType = HWC_FRAMEBUFFER;
}
if (HWC_USE_OVERLAY != list->hwLayers[i].compositionType) {
unlockPreviousOverlayBuffer(ctx);
skipComposition = false;
}
} else if (getLayerS3DFormat(list->hwLayers[i])) {
int flags = WAIT_FOR_VSYNC;
flags |= (1 == list->numHwLayers) ? DISABLE_FRAMEBUFFER_FETCH : 0;
#ifdef USE_OVERLAY
if(prepareOverlay(ctx, &(list->hwLayers[i]), flags) == 0) {
list->hwLayers[i].compositionType = HWC_USE_OVERLAY;
list->hwLayers[i].hints |= HWC_HINT_CLEAR_FB;
// We've opened the channel. Set the state to open.
ctx->hwcOverlayStatus = HWC_OVERLAY_OPEN;
}
#endif
} else if (isS3DCompositionNeeded) {
markUILayerForS3DComposition(list->hwLayers[i], s3dVideoFormat);
} else if (list->hwLayers[i].flags & HWC_USE_ORIGINAL_RESOLUTION) {
list->hwLayers[i].compositionType = HWC_USE_OVERLAY;
list->hwLayers[i].hints |= HWC_HINT_CLEAR_FB;
layerType |= HWC_ORIG_RESOLUTION;
}
else if (hnd && (hwcModule->compositionType &
(COMPOSITION_TYPE_C2D|COMPOSITION_TYPE_MDP))) {
list->hwLayers[i].compositionType = HWC_USE_COPYBIT;
} else if ((hwcModule->compositionType == COMPOSITION_TYPE_DYN)
&& useCopybit) {
list->hwLayers[i].compositionType = HWC_USE_COPYBIT;
}
else {
list->hwLayers[i].compositionType = HWC_FRAMEBUFFER;
}
}
if (skipComposition) {
list->flags |= HWC_SKIP_COMPOSITION;
} else {
list->flags &= ~HWC_SKIP_COMPOSITION;
}
#ifdef COMPOSITION_BYPASS
//Check if bypass is feasible
if(isBypassDoable(dev, yuvBufferCount, list) &&
isBypassEfficient(hwcModule->fbDevice, list, ctx)) {
//Setup bypass
if(setupBypass(ctx, list)) {
//Overwrite layer flags only if setup succeeds.
setBypassLayerFlags(ctx, list);
list->flags |= HWC_SKIP_COMPOSITION;
ctx->bypassState = BYPASS_ON;
}
} else {
unlockPreviousBypassBuffers(ctx);
unsetBypassLayerFlags(list);
unsetBypassBufferLockState(ctx);
if(ctx->bypassState == BYPASS_ON) {
ctx->bypassState = BYPASS_OFF_PENDING;
}
}
#endif
}
return 0;
}
// ---------------------------------------------------------------------------
struct range {
int current;
int end;
};
struct region_iterator : public copybit_region_t {
region_iterator(hwc_region_t region) {
mRegion = region;
r.end = region.numRects;
r.current = 0;
this->next = iterate;
}
private:
static int iterate(copybit_region_t const * self, copybit_rect_t* rect) {
if (!self || !rect) {
LOGE("iterate invalid parameters");
return 0;
}
region_iterator const* me = static_cast<region_iterator const*>(self);
if (me->r.current != me->r.end) {
rect->l = me->mRegion.rects[me->r.current].left;
rect->t = me->mRegion.rects[me->r.current].top;
rect->r = me->mRegion.rects[me->r.current].right;
rect->b = me->mRegion.rects[me->r.current].bottom;
me->r.current++;
return 1;
}
return 0;
}
hwc_region_t mRegion;
mutable range r;
};
static int drawLayerUsingCopybit(hwc_composer_device_t *dev, hwc_layer_t *layer, EGLDisplay dpy,
EGLSurface surface)
{
hwc_context_t* ctx = (hwc_context_t*)(dev);
if(!ctx) {
LOGE("drawLayerUsingCopybit null context ");
return -1;
}
private_hwc_module_t* hwcModule = reinterpret_cast<private_hwc_module_t*>(dev->common.module);
if(!hwcModule) {
LOGE("drawLayerUsingCopybit null module ");
return -1;
}
private_handle_t *hnd = (private_handle_t *)layer->handle;
if(!hnd) {
LOGE("drawLayerUsingCopybit invalid handle");
return -1;
}
// Lock this buffer for read.
genlock_lock_type lockType = GENLOCK_READ_LOCK;
int err = genlock_lock_buffer(hnd, lockType, GENLOCK_MAX_TIMEOUT);
if (GENLOCK_FAILURE == err) {
LOGE("%s: genlock_lock_buffer(READ) failed", __FUNCTION__);
return -1;
}
//render buffer
android_native_buffer_t *renderBuffer = (android_native_buffer_t *)eglGetRenderBufferANDROID(dpy, surface);
if (!renderBuffer) {
LOGE("eglGetRenderBufferANDROID returned NULL buffer");
genlock_unlock_buffer(hnd);
return -1;
}
private_handle_t *fbHandle = (private_handle_t *)renderBuffer->handle;
if(!fbHandle) {
LOGE("Framebuffer handle is NULL");
genlock_unlock_buffer(hnd);
return -1;
}
int alignment = 32;
if( HAL_PIXEL_FORMAT_RGB_565 == fbHandle->format )
alignment = 16;
// Set the copybit source:
copybit_image_t src;
src.w = ALIGN(hnd->width, alignment);
src.h = hnd->height;
src.format = hnd->format;
src.base = (void *)hnd->base;
src.handle = (native_handle_t *)layer->handle;
src.horiz_padding = src.w - hnd->width;
// Initialize vertical padding to zero for now,
// this needs to change to accomodate vertical stride
// if needed in the future
src.vert_padding = 0;
// Copybit source rect
hwc_rect_t sourceCrop = layer->sourceCrop;
copybit_rect_t srcRect = {sourceCrop.left, sourceCrop.top,
sourceCrop.right,
sourceCrop.bottom};
// Copybit destination rect
hwc_rect_t displayFrame = layer->displayFrame;
copybit_rect_t dstRect = {displayFrame.left, displayFrame.top,
displayFrame.right,
displayFrame.bottom};
// Copybit dst
copybit_image_t dst;
dst.w = ALIGN(fbHandle->width,alignment);
dst.h = fbHandle->height;
dst.format = fbHandle->format;
dst.base = (void *)fbHandle->base;
dst.handle = (native_handle_t *)renderBuffer->handle;
copybit_device_t *copybit = hwcModule->copybitEngine;
int32_t screen_w = displayFrame.right - displayFrame.left;
int32_t screen_h = displayFrame.bottom - displayFrame.top;
int32_t src_crop_width = sourceCrop.right - sourceCrop.left;
int32_t src_crop_height = sourceCrop.bottom -sourceCrop.top;
float copybitsMaxScale = (float)copybit->get(copybit,COPYBIT_MAGNIFICATION_LIMIT);
if((layer->transform == HWC_TRANSFORM_ROT_90) ||
(layer->transform == HWC_TRANSFORM_ROT_270)) {
//swap screen width and height
int tmp = screen_w;
screen_w = screen_h;
screen_h = tmp;
}
private_handle_t *tmpHnd = NULL;
if(screen_w <=0 || screen_h<=0 ||src_crop_width<=0 || src_crop_height<=0 ) {
LOGE("%s: wrong params for display screen_w=%d src_crop_width=%d screen_w=%d \
src_crop_width=%d", __FUNCTION__, screen_w,
src_crop_width,screen_w,src_crop_width);
genlock_unlock_buffer(hnd);
return -1;
}
float dsdx = (float)screen_w/src_crop_width;
float dtdy = (float)screen_h/src_crop_height;
int scaleLimit = copybitsMaxScale * copybitsMaxScale;
if(dsdx > scaleLimit || dtdy > scaleLimit) {
LOGE("%s: greater than max supported size ", __FUNCTION__ );
genlock_unlock_buffer(hnd);
return -1;
}
if(dsdx > copybitsMaxScale || dtdy > copybitsMaxScale){
// The requested scale is out of the range the hardware
// can support.
LOGD("%s:%d::Need to scale twice dsdx=%f, dtdy=%f,maxScaleInv=%f,screen_w=%d,screen_h=%d \
src_crop_width=%d src_crop_height=%d",__FUNCTION__,__LINE__,
dsdx,dtdy,copybitsMaxScale,screen_w,screen_h,src_crop_width,src_crop_height);
//Driver makes width and height as even
//that may cause wrong calculation of the ratio
//in display and crop.Hence we make
//crop width and height as even.
src_crop_width = (src_crop_width/2)*2;
src_crop_height = (src_crop_height/2)*2;
int tmp_w = src_crop_width*copybitsMaxScale;
int tmp_h = src_crop_height*copybitsMaxScale;
LOGD("%s:%d::tmp_w = %d,tmp_h = %d",__FUNCTION__,__LINE__,tmp_w,tmp_h);
int usage = GRALLOC_USAGE_PRIVATE_ADSP_HEAP |
GRALLOC_USAGE_PRIVATE_MM_HEAP;
if (0 == alloc_buffer(&tmpHnd, tmp_w, tmp_h, fbHandle->format, usage)){
copybit_image_t tmp_dst;
copybit_rect_t tmp_rect;
tmp_dst.w = tmp_w;
tmp_dst.h = tmp_h;
tmp_dst.format = tmpHnd->format;
tmp_dst.handle = tmpHnd;
tmp_dst.horiz_padding = src.horiz_padding;
tmp_dst.vert_padding = src.vert_padding;
tmp_rect.l = 0;
tmp_rect.t = 0;
tmp_rect.r = tmp_dst.w;
tmp_rect.b = tmp_dst.h;
//create one clip region
hwc_rect tmp_hwc_rect = {0,0,tmp_rect.r,tmp_rect.b};
hwc_region_t tmp_hwc_reg = {1,(hwc_rect_t const*)&tmp_hwc_rect};
region_iterator tmp_it(tmp_hwc_reg);
copybit->set_parameter(copybit,COPYBIT_TRANSFORM,0);
copybit->set_parameter(copybit, COPYBIT_PLANE_ALPHA,
(layer->blending == HWC_BLENDING_NONE) ? -1 : layer->alpha);
err = copybit->stretch(copybit,&tmp_dst, &src, &tmp_rect, &srcRect, &tmp_it);
if(err < 0){
LOGE("%s:%d::tmp copybit stretch failed",__FUNCTION__,__LINE__);
if(tmpHnd)
free_buffer(tmpHnd);
genlock_unlock_buffer(hnd);
return err;
}
// copy new src and src rect crop
src = tmp_dst;
srcRect = tmp_rect;
}
}
// Copybit region
hwc_region_t region = layer->visibleRegionScreen;
region_iterator copybitRegion(region);
copybit->set_parameter(copybit, COPYBIT_FRAMEBUFFER_WIDTH, renderBuffer->width);
copybit->set_parameter(copybit, COPYBIT_FRAMEBUFFER_HEIGHT, renderBuffer->height);
copybit->set_parameter(copybit, COPYBIT_TRANSFORM, layer->transform);
copybit->set_parameter(copybit, COPYBIT_PLANE_ALPHA,
(layer->blending == HWC_BLENDING_NONE) ? -1 : layer->alpha);
copybit->set_parameter(copybit, COPYBIT_PREMULTIPLIED_ALPHA,
(layer->blending == HWC_BLENDING_PREMULT)? COPYBIT_ENABLE : COPYBIT_DISABLE);
copybit->set_parameter(copybit, COPYBIT_DITHER,
(dst.format == HAL_PIXEL_FORMAT_RGB_565)? COPYBIT_ENABLE : COPYBIT_DISABLE);
err = copybit->stretch(copybit, &dst, &src, &dstRect, &srcRect, &copybitRegion);
if(tmpHnd)
free_buffer(tmpHnd);
if(err < 0)
LOGE("copybit stretch failed");
// Unlock this buffer since copybit is done with it.
err = genlock_unlock_buffer(hnd);
if (GENLOCK_FAILURE == err) {
LOGE("%s: genlock_unlock_buffer failed", __FUNCTION__);
}
return err;
}
static int drawLayerUsingOverlay(hwc_context_t *ctx, hwc_layer_t *layer)
{
if (ctx && ctx->mOverlayLibObject) {
private_hwc_module_t* hwcModule = reinterpret_cast<private_hwc_module_t*>(ctx->device.common.module);
if (!hwcModule) {
LOGE("drawLayerUsingLayer null module ");
return -1;
}
private_handle_t *hnd = (private_handle_t *)layer->handle;
overlay::Overlay *ovLibObject = ctx->mOverlayLibObject;
int ret = 0;
// Lock this buffer for read.
if (GENLOCK_NO_ERROR != genlock_lock_buffer(hnd, GENLOCK_READ_LOCK,
GENLOCK_MAX_TIMEOUT)) {
LOGE("%s: genlock_lock_buffer(READ) failed", __FUNCTION__);
return -1;
}
ret = ovLibObject->queueBuffer(hnd);
// Unlock the previously locked buffer, since the overlay has completed reading the buffer
unlockPreviousOverlayBuffer(ctx);
if (!ret) {
LOGE("drawLayerUsingOverlay queueBuffer failed");
// Unlock the buffer handle
genlock_unlock_buffer(hnd);
ctx->previousOverlayHandle = NULL;
} else {
// Store the current buffer handle as the one that is to be unlocked after
// the next overlay play call.
ctx->previousOverlayHandle = hnd;
hnd->flags |= private_handle_t::PRIV_FLAGS_HWC_LOCK;
}
return ret;
}
return -1;
}
static int hwc_set(hwc_composer_device_t *dev,
hwc_display_t dpy,
hwc_surface_t sur,
hwc_layer_list_t* list)
{
hwc_context_t* ctx = (hwc_context_t*)(dev);
if(!ctx) {
LOGE("hwc_set invalid context");
return -1;
}
private_hwc_module_t* hwcModule = reinterpret_cast<private_hwc_module_t*>(
dev->common.module);
if (!list || !hwcModule) {
LOGE("hwc_set invalid list or module");
#ifdef COMPOSITION_BYPASS
unlockPreviousBypassBuffers(ctx);
unsetBypassBufferLockState(ctx);
#endif
unlockPreviousOverlayBuffer(ctx);
return -1;
}
int ret = 0;
for (size_t i=0; i<list->numHwLayers; i++) {
if (list->hwLayers[i].flags & HWC_SKIP_LAYER) {
continue;
#ifdef COMPOSITION_BYPASS
} else if (list->hwLayers[i].flags & HWC_COMP_BYPASS) {
drawLayerUsingBypass(ctx, &(list->hwLayers[i]), i);
#endif
} else if (list->hwLayers[i].compositionType == HWC_USE_OVERLAY) {
drawLayerUsingOverlay(ctx, &(list->hwLayers[i]));
} else if (list->flags & HWC_SKIP_COMPOSITION) {
continue;
}
else if (list->hwLayers[i].compositionType == HWC_USE_COPYBIT) {
drawLayerUsingCopybit(dev, &(list->hwLayers[i]), (EGLDisplay)dpy, (EGLSurface)sur);
}
}
#ifdef COMPOSITION_BYPASS
unlockPreviousBypassBuffers(ctx);
storeLockedBypassHandle(list, ctx);
// We have stored the handles, unset the current lock states in the context.
unsetBypassBufferLockState(ctx);
#endif
// Do not call eglSwapBuffers if we the skip composition flag is set on the list.
if (!(list->flags & HWC_SKIP_COMPOSITION)) {
EGLBoolean sucess = eglSwapBuffers((EGLDisplay)dpy, (EGLSurface)sur);
if (!sucess) {
ret = HWC_EGL_ERROR;
}
}
else {
CALC_FPS();
}
#ifdef COMPOSITION_BYPASS
if(ctx->bypassState == BYPASS_OFF_PENDING) {
closeBypass(ctx);
ctx->bypassState = BYPASS_OFF;
}
#endif
#if defined HDMI_DUAL_DISPLAY
if(ctx->pendingHDMI) {
handleHDMIStateChange(dev, ctx->mHDMIEnabled);
ctx->pendingHDMI = false;
}
#endif
hwc_closeOverlayChannels(ctx);
int yuvBufferCount = getYUVBufferCount(list);
setHWCOverlayStatus(ctx, yuvBufferCount);
return ret;
}
static int hwc_device_close(struct hw_device_t *dev)
{
if(!dev) {
LOGE("hwc_device_close null device pointer");
return -1;
}
struct hwc_context_t* ctx = (struct hwc_context_t*)dev;
private_hwc_module_t* hwcModule = reinterpret_cast<private_hwc_module_t*>(
ctx->device.common.module);
// Close the overlay and copybit modules
if(hwcModule->copybitEngine) {
copybit_close(hwcModule->copybitEngine);
hwcModule->copybitEngine = NULL;
}
if(hwcModule->fbDevice) {
framebuffer_close(hwcModule->fbDevice);
hwcModule->fbDevice = NULL;
}
unlockPreviousOverlayBuffer(ctx);
if (ctx) {
delete ctx->mOverlayLibObject;
ctx->mOverlayLibObject = NULL;
#ifdef COMPOSITION_BYPASS
for(int i = 0; i < MAX_BYPASS_LAYERS; i++) {
delete ctx->mOvUI[i];
}
unlockPreviousBypassBuffers(ctx);
unsetBypassBufferLockState(ctx);
#endif
free(ctx);
}
return 0;
}
/*****************************************************************************/
static int hwc_module_initialize(struct private_hwc_module_t* hwcModule)
{
// Open the overlay and copybit modules
hw_module_t const *module;
if (hw_get_module(COPYBIT_HARDWARE_MODULE_ID, &module) == 0) {
copybit_open(module, &(hwcModule->copybitEngine));
}
if (hw_get_module(GRALLOC_HARDWARE_MODULE_ID, &module) == 0) {
framebuffer_open(module, &(hwcModule->fbDevice));
}
// get the current composition type
char property[PROPERTY_VALUE_MAX];
if (property_get("debug.sf.hw", property, NULL) > 0) {
if(atoi(property) == 0) {
//debug.sf.hw = 0
hwcModule->compositionType = COMPOSITION_TYPE_CPU;
} else { //debug.sf.hw = 1
// Get the composition type
property_get("debug.composition.type", property, NULL);
if (property == NULL) {
hwcModule->compositionType = COMPOSITION_TYPE_GPU;
} else if ((strncmp(property, "mdp", 3)) == 0) {
hwcModule->compositionType = COMPOSITION_TYPE_MDP;
} else if ((strncmp(property, "c2d", 3)) == 0) {
hwcModule->compositionType = COMPOSITION_TYPE_C2D;
} else if ((strncmp(property, "dyn", 3)) == 0) {
hwcModule->compositionType = COMPOSITION_TYPE_DYN;
} else {
hwcModule->compositionType = COMPOSITION_TYPE_GPU;
}
if(!hwcModule->copybitEngine)
hwcModule->compositionType = COMPOSITION_TYPE_GPU;
}
} else { //debug.sf.hw is not set. Use cpu composition
hwcModule->compositionType = COMPOSITION_TYPE_CPU;
}
//Check if composition bypass is enabled
if(property_get("ro.sf.compbypass.enable", property, NULL) > 0) {
if(atoi(property) == 1) {
hwcModule->isBypassEnabled = true;
}
}
CALC_INIT();
return 0;
}
static int hwc_device_open(const struct hw_module_t* module, const char* name,
struct hw_device_t** device)
{
int status = -EINVAL;
if (!strcmp(name, HWC_HARDWARE_COMPOSER)) {
private_hwc_module_t* hwcModule = reinterpret_cast<private_hwc_module_t*>
(const_cast<hw_module_t*>(module));
hwc_module_initialize(hwcModule);
struct hwc_context_t *dev;
dev = (hwc_context_t*)malloc(sizeof(*dev));
/* initialize our state here */
memset(dev, 0, sizeof(*dev));
dev->mOverlayLibObject = new overlay::Overlay();
#ifdef COMPOSITION_BYPASS
for(int i = 0; i < MAX_BYPASS_LAYERS; i++) {
dev->mOvUI[i] = new overlay::OverlayUI();
dev->previousBypassHandle[i] = NULL;
}
unsetBypassBufferLockState(dev);
dev->animCount = 0;
dev->bypassState = BYPASS_OFF;
#endif
#if defined HDMI_DUAL_DISPLAY
dev->mHDMIEnabled = EXT_DISPLAY_OFF;
dev->pendingHDMI = false;
#endif
dev->previousOverlayHandle = NULL;
dev->hwcOverlayStatus = HWC_OVERLAY_CLOSED;
/* initialize the procs */
dev->device.common.tag = HARDWARE_DEVICE_TAG;
dev->device.common.version = 0;
dev->device.common.module = const_cast<hw_module_t*>(module);
dev->device.common.close = hwc_device_close;
dev->device.prepare = hwc_prepare;
dev->device.set = hwc_set;
dev->device.enableHDMIOutput = hwc_enableHDMIOutput;
*device = &dev->device.common;
status = 0;
}
return status;
}