blob: 9c9c7b481651d3b81e63525492531b2f9e66a980 [file] [log] [blame]
/*
* Copyright (C) 2010 ARM Limited. All rights reserved.
*
* Portions of this code have been modified from the original.
* These modifications are:
* * includes
* * gralloc_alloc_buffer()
* * gralloc_alloc_framebuffer_locked()
* * gralloc_alloc_framebuffer()
* * alloc_device_alloc()
* * alloc_device_free()
* * alloc_device_close()
* * alloc_device_open()
*
* Copyright (C) 2008 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 <string.h>
#include <errno.h>
#include <pthread.h>
#include <cutils/log.h>
#include <cutils/atomic.h>
#include <hardware/hardware.h>
#include <hardware/gralloc.h>
#include "sec_format.h"
#include "graphics.h"
#include "gralloc_priv.h"
#include "gralloc_helper.h"
#include "framebuffer_device.h"
#include "ump.h"
#include "ump_ref_drv.h"
#include "secion.h"
/*****************************************************************************/
#include <limits.h>
#include <unistd.h>
#include <fcntl.h>
#include <errno.h>
#include <pthread.h>
#include <stdlib.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/ioctl.h>
#include "format.h"
#include <linux/videodev2.h>
#include "s5p_fimc.h"
#ifdef SAMSUNG_EXYNOS4x12
#define PFX_NODE_FIMC1 "/dev/video3"
#endif
#ifdef SAMSUNG_EXYNOS4210
#define PFX_NODE_FIMC1 "/dev/video1"
#endif
#ifndef OMX_COLOR_FormatYUV420Planar
#define OMX_COLOR_FormatYUV420Planar 0x13
#endif
#ifndef OMX_COLOR_FormatYUV420SemiPlanar
#define OMX_COLOR_FormatYUV420SemiPlanar 0x15
#endif
#define PFX_NODE_MEM "/dev/exynos-mem"
static int gMemfd = 0;
bool ion_dev_open = true;
static pthread_mutex_t l_surface= PTHREAD_MUTEX_INITIALIZER;
static int buffer_offset = 0;
static int gfd = 0;
#ifdef USE_PARTIAL_FLUSH
extern struct private_handle_rect *rect_list;
extern private_handle_rect *find_rect(int secure_id);
extern private_handle_rect *find_last_rect(int secure_id);
extern int release_rect(int secure_id);
#endif
#define EXYNOS4_ALIGN( value, base ) (((value) + ((base) - 1)) & ~((base) - 1))
static int gralloc_alloc_buffer(alloc_device_t* dev, size_t size, int usage,
buffer_handle_t* pHandle, int w, int h,
int format, int bpp, int stride_raw, int stride)
{
ump_handle ump_mem_handle;
void *cpu_ptr;
ump_secure_id ump_id;
size = round_up_to_page_size(size);
#ifdef INSIGNAL_FIMC1
if (usage & GRALLOC_USAGE_HW_FIMC1) {
int dev_fd=0;
char node[20];
int ret;
int paddr=0;
int offset=0;
struct v4l2_control vc;
sprintf(node, "%s", PFX_NODE_FIMC1);
if (gfd == 0) {
gfd = open(node, O_RDWR);
if (gfd < 0) {
ALOGE("%s:: %s Post processor open error\n", __func__, node);
return false;
}
}
vc.id = V4L2_CID_RESERVED_MEM_BASE_ADDR;
vc.value = 0;
ret = ioctl(gfd, VIDIOC_G_CTRL, &vc);
if (ret < 0) {
ALOGE("Error in video VIDIOC_G_CTRL - V4L2_CID_RESERVED_MEM_BAES_ADDR (%d)\n", ret);
return false;
}
paddr = (unsigned int)vc.value;
if ((buffer_offset + size) >= FIMC1_RESERVED_SIZE * 1024)
buffer_offset = 0;
paddr += buffer_offset;
private_handle_t* hnd = new private_handle_t(private_handle_t::PRIV_FLAGS_USES_IOCTL, size, 0,
private_handle_t::LOCK_STATE_MAPPED, 0, 0);
*pHandle = hnd;
hnd->format = format;
hnd->usage = usage;
hnd->width = w;
hnd->height = h;
hnd->bpp = bpp;
hnd->paddr = paddr;
hnd->offset = buffer_offset;
hnd->stride = stride;
hnd->fd = gfd;
hnd->uoffset = (EXYNOS4_ALIGN((EXYNOS4_ALIGN(hnd->width, 16) * EXYNOS4_ALIGN(hnd->height, 16)), 4096));
hnd->voffset = (EXYNOS4_ALIGN((EXYNOS4_ALIGN((hnd->width >> 1), 16) * EXYNOS4_ALIGN((hnd->height >> 1), 16)), 4096));
buffer_offset += size;
if (gMemfd == 0) {
gMemfd = open(PFX_NODE_MEM, O_RDWR);
if (gMemfd < 0) {
ALOGE("%s:: %s exynos-mem open error\n", __func__, PFX_NODE_MEM);
return false;
}
}
size_t size = FIMC1_RESERVED_SIZE * 1024;
void *mappedAddress = mmap(0, size,
PROT_READ|PROT_WRITE, MAP_SHARED, gMemfd, (hnd->paddr - hnd->offset));
hnd->base = intptr_t(mappedAddress) + hnd->offset;
return 0;
} else {
#endif
ion_buffer ion_fd = 0;
unsigned int ion_flags = 0;
int priv_alloc_flag = private_handle_t::PRIV_FLAGS_USES_UMP;
#ifdef INSIGNAL_FIMC1
if (usage & GRALLOC_USAGE_HW_ION) {
#else
if (usage & GRALLOC_USAGE_HW_ION || usage & GRALLOC_USAGE_HW_FIMC1) {
#endif
if (!ion_dev_open) {
ALOGE("ERROR, failed to open ion");
return -1;
}
private_module_t* m = reinterpret_cast<private_module_t*>(dev->common.module);
ion_flags = ION_HEAP_EXYNOS_MASK;
ion_fd = ion_alloc(m->ion_client, size, 0, ion_flags);
if (ion_fd < 0) {
ALOGE("Failed to ion_alloc");
return -1;
}
cpu_ptr = ion_map(ion_fd, size, 0);
if (NULL == cpu_ptr) {
ALOGE("Failed to ion_map");
ion_free(ion_fd);
return -1;
}
ump_mem_handle = ump_ref_drv_ion_import(ion_fd, UMP_REF_DRV_CONSTRAINT_NONE);
if (UMP_INVALID_MEMORY_HANDLE != ump_mem_handle) {
priv_alloc_flag = private_handle_t::PRIV_FLAGS_USES_ION;
} else {
ALOGE("gralloc_alloc_buffer() failed to import ION memory");
ion_unmap(cpu_ptr, size);
ion_free(ion_fd);
return -1;
}
}
#ifdef SAMSUNG_EXYNOS_CACHE_UMP
else if ((usage&GRALLOC_USAGE_SW_READ_MASK) == GRALLOC_USAGE_SW_READ_OFTEN)
ump_mem_handle = ump_ref_drv_allocate(size, UMP_REF_DRV_CONSTRAINT_USE_CACHE);
else
ump_mem_handle = ump_ref_drv_allocate(size, UMP_REF_DRV_CONSTRAINT_NONE);
#else
else
ump_mem_handle = ump_ref_drv_allocate(size, UMP_REF_DRV_CONSTRAINT_NONE);
#endif
if (UMP_INVALID_MEMORY_HANDLE != ump_mem_handle) {
if (!(usage & GRALLOC_USAGE_HW_ION || usage & GRALLOC_USAGE_HW_FIMC1))
cpu_ptr = ump_mapped_pointer_get(ump_mem_handle);
if (NULL != cpu_ptr) {
ump_id = ump_secure_id_get(ump_mem_handle);
if (UMP_INVALID_SECURE_ID != ump_id) {
private_handle_t* hnd;
hnd = new private_handle_t(priv_alloc_flag, size, (int)cpu_ptr,
private_handle_t::LOCK_STATE_MAPPED, ump_id, ump_mem_handle, ion_fd, 0, 0);
if (NULL != hnd) {
*pHandle = hnd;
#ifdef USE_PARTIAL_FLUSH
if (hnd->flags & private_handle_t::PRIV_FLAGS_USES_UMP) {
private_handle_rect *psRect;
private_handle_rect *psFRect;
psRect = (private_handle_rect *)calloc(1, sizeof(private_handle_rect));
psRect->handle = (int)hnd->ump_id;
psRect->stride = stride_raw;
psFRect = find_last_rect((int)hnd->ump_id);
psFRect->next = psRect;
}
#endif
hnd->format = format;
hnd->usage = usage;
hnd->width = w;
hnd->height = h;
hnd->bpp = bpp;
hnd->stride = stride;
if(hnd->format == HAL_PIXEL_FORMAT_YV12) {
hnd->uoffset = ((EXYNOS4_ALIGN(hnd->width, 16) * hnd->height));
hnd->voffset = ((EXYNOS4_ALIGN((hnd->width >> 1), 16) * (hnd->height >> 1)));
} else {
hnd->uoffset = ((EXYNOS4_ALIGN(hnd->width, 16) * EXYNOS4_ALIGN(hnd->height, 16)));
hnd->voffset = ((EXYNOS4_ALIGN((hnd->width >> 1), 16) * EXYNOS4_ALIGN((hnd->height >> 1), 16)));
}
return 0;
} else {
ALOGE("gralloc_alloc_buffer() failed to allocate handle");
}
} else {
ALOGE("gralloc_alloc_buffer() failed to retrieve valid secure id");
}
ump_mapped_pointer_release(ump_mem_handle);
} else {
ALOGE("gralloc_alloc_buffer() failed to map UMP memory");
}
ump_reference_release(ump_mem_handle);
} else {
ALOGE("gralloc_alloc_buffer() failed to allcoate UMP memory");
}
#ifdef INSIGNAL_FIMC1
}
#endif
return -1;
}
static int gralloc_alloc_framebuffer_locked(alloc_device_t* dev, size_t size, int usage,
buffer_handle_t* pHandle, int w, int h,
int format, int bpp)
{
private_module_t* m = reinterpret_cast<private_module_t*>(dev->common.module);
/* allocate the framebuffer */
if (m->framebuffer == NULL) {
/* initialize the framebuffer, the framebuffer is mapped once and forever. */
int err = init_frame_buffer_locked(m);
if (err < 0)
return err;
}
const uint32_t bufferMask = m->bufferMask;
const uint32_t numBuffers = m->numBuffers;
const size_t bufferSize = m->finfo.line_length * m->info.yres;
if (numBuffers == 1) {
/*
* If we have only one buffer, we never use page-flipping. Instead,
* we return a regular buffer which will be memcpy'ed to the main
* screen when post is called.
*/
int newUsage = (usage & ~GRALLOC_USAGE_HW_FB) | GRALLOC_USAGE_HW_2D;
ALOGE("fallback to single buffering");
return gralloc_alloc_buffer(dev, bufferSize, newUsage, pHandle, w, h, format, bpp, 0, 0);
}
if (bufferMask >= ((1LU<<numBuffers)-1))
return -ENOMEM;
int vaddr = m->framebuffer->base;
/* find a free slot */
for (uint32_t i = 0; i < numBuffers; i++) {
if ((bufferMask & (1LU<<i)) == 0) {
m->bufferMask |= (1LU<<i);
break;
}
vaddr += bufferSize;
}
/*
* The entire framebuffer memory is already mapped,
* now create a buffer object for parts of this memory
*/
private_handle_t* hnd = new private_handle_t
(private_handle_t::PRIV_FLAGS_FRAMEBUFFER, size, vaddr,
0, dup(m->framebuffer->fd), vaddr - m->framebuffer->base);
hnd->format = format;
hnd->usage = usage;
hnd->width = w;
hnd->height = h;
hnd->bpp = bpp;
*pHandle = hnd;
return 0;
}
static int gralloc_alloc_framebuffer(alloc_device_t* dev, size_t size, int usage,
buffer_handle_t* pHandle, int w, int h,
int format, int bpp)
{
private_module_t* m = reinterpret_cast<private_module_t*>(dev->common.module);
pthread_mutex_lock(&m->lock);
int err = gralloc_alloc_framebuffer_locked(dev, size, usage, pHandle, w, h, format, bpp);
pthread_mutex_unlock(&m->lock);
return err;
}
static int alloc_device_alloc(alloc_device_t* dev, int w, int h, int format,
int usage, buffer_handle_t* pHandle, int* pStride)
{
if (!pHandle || !pStride)
return -EINVAL;
size_t size = 0;
size_t stride = 0;
size_t stride_raw = 0;
if (format == HAL_PIXEL_FORMAT_YCbCr_420_SP ||
format == HAL_PIXEL_FORMAT_YCbCr_420_SP_TILED ||
format == HAL_PIXEL_FORMAT_YCrCb_420_SP ||
format == HAL_PIXEL_FORMAT_YCbCr_422_SP ||
format == HAL_PIXEL_FORMAT_YCbCr_420_P ||
format == HAL_PIXEL_FORMAT_YV12 ||
format == HAL_PIXEL_FORMAT_CUSTOM_YCrCb_420_SP ||
format == HAL_PIXEL_FORMAT_CUSTOM_YCbCr_420_SP_TILED ||
format == GGL_PIXEL_FORMAT_L_8 ||
format == OMX_COLOR_FormatYUV420Planar ||
format == OMX_COLOR_FormatYUV420SemiPlanar) {
/* FIXME: there is no way to return the vstride */
int vstride;
stride = EXYNOS4_ALIGN(w, 16);
vstride = EXYNOS4_ALIGN(h, 16);
switch (format) {
case HAL_PIXEL_FORMAT_YCbCr_420_SP:
case HAL_PIXEL_FORMAT_YCbCr_420_SP_TILED:
case HAL_PIXEL_FORMAT_YCrCb_420_SP:
case HAL_PIXEL_FORMAT_YCbCr_420_P:
case HAL_PIXEL_FORMAT_YV12:
case HAL_PIXEL_FORMAT_CUSTOM_YCrCb_420_SP:
case HAL_PIXEL_FORMAT_CUSTOM_YCbCr_420_SP_TILED:
case OMX_COLOR_FormatYUV420Planar:
case OMX_COLOR_FormatYUV420SemiPlanar:
size = stride * vstride + EXYNOS4_ALIGN((w / 2), 16) * EXYNOS4_ALIGN((h / 2), 16) * 2;
#ifdef INSIGNAL_FIMC1
if (usage & GRALLOC_USAGE_HW_FIMC1) {
#else
if (usage & GRALLOC_USAGE_HW_ION || usage & GRALLOC_USAGE_HW_FIMC1) {
#endif
size += PAGE_SIZE * 2;
}
break;
case HAL_PIXEL_FORMAT_YCbCr_422_SP:
size = (stride * vstride) + (w/2 * h/2) * 2;
break;
case GGL_PIXEL_FORMAT_L_8:
size = (stride * vstride);
break;
default:
return -EINVAL;
}
} else {
int bpp = 0;
switch (format) {
case HAL_PIXEL_FORMAT_RGBA_8888:
case HAL_PIXEL_FORMAT_RGBX_8888:
case HAL_PIXEL_FORMAT_BGRA_8888:
bpp = 4;
break;
case HAL_PIXEL_FORMAT_RGB_888:
bpp = 3;
break;
case HAL_PIXEL_FORMAT_RGB_565:
case HAL_PIXEL_FORMAT_RGBA_5551:
case HAL_PIXEL_FORMAT_RGBA_4444:
bpp = 2;
break;
default:
return -EINVAL;
}
size_t bpr = EXYNOS4_ALIGN((w*bpp), 8);
size = bpr * h;
stride = bpr / bpp;
stride_raw = bpr;
}
int err;
pthread_mutex_lock(&l_surface);
if (usage & GRALLOC_USAGE_HW_FB)
err = gralloc_alloc_framebuffer(dev, size, usage, pHandle, w, h, format, 32);
else
err = gralloc_alloc_buffer(dev, size, usage, pHandle, w, h, format, 0, (int)stride_raw, (int)stride);
pthread_mutex_unlock(&l_surface);
if (err < 0)
return err;
*pStride = stride;
return 0;
}
static int alloc_device_free(alloc_device_t* dev, buffer_handle_t handle)
{
if (private_handle_t::validate(handle) < 0)
return -EINVAL;
private_handle_t const* hnd = reinterpret_cast<private_handle_t const*>(handle);
private_module_t* m = reinterpret_cast<private_module_t*>(dev->common.module);
pthread_mutex_lock(&l_surface);
if (hnd->flags & private_handle_t::PRIV_FLAGS_FRAMEBUFFER) {
/* free this buffer */
const size_t bufferSize = m->finfo.line_length * m->info.yres;
int index = (hnd->base - m->framebuffer->base) / bufferSize;
m->bufferMask &= ~(1<<index);
close(hnd->fd);
} else if (hnd->flags & private_handle_t::PRIV_FLAGS_USES_IOCTL) {
void* base = (void*)(intptr_t(hnd->base) - hnd->offset);
size_t size = FIMC1_RESERVED_SIZE * 1024;
if (munmap(base, size) < 0)
ALOGE("Could not unmap %s", strerror(errno));
if (0 < gMemfd) {
close(gMemfd);
gMemfd = 0;
}
} else if (hnd->flags & private_handle_t::PRIV_FLAGS_USES_UMP) {
#ifdef USE_PARTIAL_FLUSH
if (!release_rect((int)hnd->ump_id))
ALOGE("secure id: 0x%x, release error",(int)hnd->ump_id);
#endif
ump_mapped_pointer_release((ump_handle)hnd->ump_mem_handle);
ump_reference_release((ump_handle)hnd->ump_mem_handle);
} else if (hnd->flags & private_handle_t::PRIV_FLAGS_USES_ION) {
#ifdef USE_PARTIAL_FLUSH
if (!release_rect((int)hnd->ump_id))
ALOGE("secure id: 0x%x, release error",(int)hnd->ump_id);
#endif
ump_mapped_pointer_release((ump_handle)hnd->ump_mem_handle);
ump_reference_release((ump_handle)hnd->ump_mem_handle);
ion_unmap((void*)hnd->base, hnd->size);
ion_free(hnd->fd);
}
pthread_mutex_unlock(&l_surface);
delete hnd;
return 0;
}
static int alloc_device_close(struct hw_device_t *device)
{
alloc_device_t* dev = reinterpret_cast<alloc_device_t*>(device);
if (dev) {
private_module_t* m = reinterpret_cast<private_module_t*>(dev->common.module);
if (ion_dev_open)
ion_client_destroy(m->ion_client);
delete dev;
ump_close();
}
return 0;
}
int alloc_device_open(hw_module_t const* module, const char* name, hw_device_t** device)
{
alloc_device_t *dev;
dev = new alloc_device_t;
if (NULL == dev)
return -1;
dev->common.module = const_cast<hw_module_t*>(module);
private_module_t* m = reinterpret_cast<private_module_t*>(dev->common.module);
m->ion_client = ion_client_create();
ump_result ump_res = ump_open();
if (0 > m->ion_client)
ion_dev_open = false;
if (UMP_OK != ump_res) {
ALOGE("UMP open failed ump_res %d", ump_res);
delete dev;
return -1;
}
/* initialize our state here */
memset(dev, 0, sizeof(*dev));
/* initialize the procs */
dev->common.tag = HARDWARE_DEVICE_TAG;
dev->common.version = 0;
dev->common.module = const_cast<hw_module_t*>(module);
dev->common.close = alloc_device_close;
dev->alloc = alloc_device_alloc;
dev->free = alloc_device_free;
*device = &dev->common;
return 0;
}