exynos: reorganized and updated from insignal

Changes needed on exynos4210 devices:

libcsc -> libseccscapi
libswconverter -> remove

TARGET_HAL_PATH := hardware/samsung/exynos4/hal
TARGET_OMX_PATH := hardware/samsung/exynos/multimedia/openmax
$(call inherit-product, hardware/samsung/exynos4210.mk)

Change-Id: Ic59ef95b85ef37b3f38fb36cf6a364a5414685ee

9 files changed, 2233 insertions, 0 deletions

diff --git a/exynos/multimedia/utils/csc/exynos5/Android.mk b/exynos/multimedia/utils/csc/exynos5/Android.mk
new file mode 100644
index 0000000..955a3e2
--- /dev/null
+++ b/exynos/multimedia/utils/csc/exynos5/Android.mk
@@ -0,0 +1,35 @@
+LOCAL_PATH := $(call my-dir)
+
+include $(CLEAR_VARS)
+
+LOCAL_COPY_HEADERS_TO := libsecmm
+LOCAL_COPY_HEADERS := \
+	color_space_convertor.h \
+	csc_fimc.h
+
+LOCAL_MODULE_TAGS := optional
+
+LOCAL_SRC_FILES := \
+	color_space_convertor.c \
+	csc_fimc.cpp \
+	csc_tiled_to_linear_y_neon.s \
+	csc_tiled_to_linear_uv_neon.s \
+	csc_tiled_to_linear_uv_deinterleave_neon.s \
+	csc_interleave_memcpy_neon.s
+
+LOCAL_C_INCLUDES := \
+	$(TOP)/device/samsung/multimedia/openmax/include/khronos \
+	$(TOP)/device/samsung/$(TARGET_BOARD_PLATFORM)/include
+
+LOCAL_MODULE := libseccscapi
+
+LOCAL_PRELINK_MODULE := false
+
+LOCAL_CFLAGS :=
+
+LOCAL_ARM_MODE := arm
+
+LOCAL_STATIC_LIBRARIES :=
+LOCAL_SHARED_LIBRARIES := liblog libfimc
+
+include $(BUILD_STATIC_LIBRARY)
diff --git a/exynos/multimedia/utils/csc/exynos5/color_space_convertor.c b/exynos/multimedia/utils/csc/exynos5/color_space_convertor.c
new file mode 100644
index 0000000..a651d62
--- /dev/null
+++ b/exynos/multimedia/utils/csc/exynos5/color_space_convertor.c
@@ -0,0 +1,640 @@
+/*
+ *
+ * Copyright 2010 Samsung Electronics S.LSI Co. LTD
+ *
+ * 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.
+ */
+
+/*
+ * @file    color_space_convertor.c
+ *
+ * @brief   SEC_OMX specific define. It support MFC 6.x tiled.
+ *
+ * @author  ShinWon Lee (shinwon.lee@samsung.com)
+ *
+ * @version 1.0
+ *
+ * @history
+ *   2011.12.01 : Create
+ */
+
+#include "stdio.h"
+#include "stdlib.h"
+#include "color_space_convertor.h"
+
+/* 2D Configurable tiled memory access (TM)
+ * Return the linear address from tiled position (x, y) */
+unsigned int Tile2D_To_Linear(
+    unsigned int width,
+    unsigned int height,
+    unsigned int xpos,
+    unsigned int ypos,
+    int crFlag)
+{
+    int  tileNumX;
+    int  tileX, tileY;
+    int  tileAddr;
+    int  offset;
+    int  addr;
+
+    width = ((width + 15) / 16) * 16;
+    tileNumX = width / 16;
+
+    /* crFlag - 0: Y plane, 1: CbCr plane */
+    if (crFlag == 0) {
+        tileX = xpos / 16;
+        tileY = ypos / 16;
+        tileAddr = tileY * tileNumX + tileX;
+        offset = (ypos & 15) * 16 + (xpos & 15);
+        addr = (tileAddr << 8) | offset;
+    } else {
+        tileX = xpos / 16;
+        tileY = ypos / 8;
+        tileAddr = tileY * tileNumX + tileX;
+        offset = (ypos & 7) * 16 + (xpos & 15);
+        addr = (tileAddr << 7) | offset;
+    }
+
+    return addr;
+}
+
+/*
+ * De-interleaves src to dest1, dest2
+ *
+ * @param dest1
+ *   Address of de-interleaved data[out]
+ *
+ * @param dest2
+ *   Address of de-interleaved data[out]
+ *
+ * @param src
+ *   Address of interleaved data[in]
+ *
+ * @param src_size
+ *   Size of interleaved data[in]
+ */
+void csc_deinterleave_memcpy(
+    unsigned char *dest1,
+    unsigned char *dest2,
+    unsigned char *src,
+    unsigned int src_size)
+{
+    unsigned int i = 0;
+    for(i=0; i<src_size/2; i++) {
+        dest1[i] = src[i*2];
+        dest2[i] = src[i*2+1];
+    }
+}
+
+/*
+ * Interleaves src1, src2 to dest
+ *
+ * @param dest
+ *   Address of interleaved data[out]
+ *
+ * @param src1
+ *   Address of de-interleaved data[in]
+ *
+ * @param src2
+ *   Address of de-interleaved data[in]
+ *
+ * @param src_size
+ *   Size of de-interleaved data[in]
+ */
+void csc_interleave_memcpy(
+    unsigned char *dest,
+    unsigned char *src1,
+    unsigned char *src2,
+    unsigned int src_size)
+{
+    unsigned int i = 0;
+    for(i=0; i<src_size; i++) {
+        dest[i*2] = src1[i];
+        dest[i*2+1] = src2[i];
+    }
+}
+
+/*
+ * Converts tiled data to linear for mfc 6.x tiled
+ * 1. y of nv12t to y of yuv420p
+ * 2. y of nv12t to y of yuv420s
+ *
+ * @param dst
+ *   y address of yuv420[out]
+ *
+ * @param src
+ *   y address of nv12t[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420[in]
+ *   it should be even
+ *
+ * @param yuv420_height
+ *   real height of yuv420[in]
+ *   it should be even.
+ *
+ */
+void csc_tiled_to_linear_y(
+    unsigned char *y_dst,
+    unsigned char *y_src,
+    unsigned int width,
+    unsigned int height)
+{
+    unsigned int i, j, k;
+    unsigned int aligned_width, aligned_height;
+    unsigned int tiled_width;
+    unsigned int src_offset, dst_offset;
+
+    aligned_height = height & (~0xF);
+    aligned_width = width & (~0xF);
+    tiled_width = ((width + 15) >> 4) << 4;
+
+    for (i = 0; i < aligned_height; i = i + 16) {
+        for (j = 0; j<aligned_width; j = j + 16) {
+            src_offset = (tiled_width * i) + (j << 4);
+            dst_offset = width * i + j;
+            for (k = 0; k < 8; k++) {
+                memcpy(y_dst + dst_offset, y_src + src_offset, 16);
+                src_offset += 16;
+                dst_offset += width;
+                memcpy(y_dst + dst_offset, y_src + src_offset, 16);
+                src_offset += 16;
+                dst_offset += width;
+            }
+        }
+        if (aligned_width != width) {
+            src_offset = (tiled_width * i) + (j << 4);
+            dst_offset = width * i + j;
+            for (k = 0; k < 8; k++) {
+                memcpy(y_dst + dst_offset, y_src + src_offset, width - j);
+                src_offset += 16;
+                dst_offset += width;
+                memcpy(y_dst + dst_offset, y_src + src_offset, width - j);
+                src_offset += 16;
+                dst_offset += width;
+            }
+        }
+    }
+
+    if (aligned_height != height) {
+        for (j = 0; j<aligned_width; j = j + 16) {
+            src_offset = (tiled_width * i) + (j << 4);
+            dst_offset = width * i + j;
+            for (k = 0; k < height - aligned_height; k = k + 2) {
+                memcpy(y_dst + dst_offset, y_src + src_offset, 16);
+                src_offset += 16;
+                dst_offset += width;
+                memcpy(y_dst + dst_offset, y_src + src_offset, 16);
+                src_offset += 16;
+                dst_offset += width;
+            }
+        }
+        if (aligned_width != width) {
+            src_offset = (tiled_width * i) + (j << 4);
+            dst_offset = width * i + j;
+            for (k = 0; k < height - aligned_height; k = k + 2) {
+                memcpy(y_dst + dst_offset, y_src + src_offset, width - j);
+                src_offset += 16;
+                dst_offset += width;
+                memcpy(y_dst + dst_offset, y_src + src_offset, width - j);
+                src_offset += 16;
+                dst_offset += width;
+            }
+        }
+    }
+}
+
+/*
+ * Converts tiled data to linear for mfc 6.x tiled
+ * 1. uv of nv12t to uv of yuv420s
+ *
+ * @param dst
+ *   uv address of yuv420s[out]
+ *
+ * @param src
+ *   uv address of nv12t[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420s[in]
+ *
+ * @param yuv420_height
+ *   real height of yuv420s[in]
+ *
+ */
+void csc_tiled_to_linear_uv(
+    unsigned char *uv_dst,
+    unsigned char *uv_src,
+    unsigned int width,
+    unsigned int height)
+{
+    unsigned int i, j, k;
+    unsigned int aligned_width, aligned_height;
+    unsigned int tiled_width;
+    unsigned int src_offset, dst_offset;
+
+    aligned_height = height & (~0x7);
+    aligned_width = width & (~0xF);
+    tiled_width = ((width + 15) >> 4) << 4;
+
+    for (i = 0; i < aligned_height; i = i + 8) {
+        for (j = 0; j<aligned_width; j = j + 16) {
+            src_offset = (tiled_width * i) + (j << 3);
+            dst_offset = width * i + j;
+            for (k = 0; k < 4; k++) {
+                memcpy(uv_dst + dst_offset, uv_src + src_offset, 16);
+                src_offset += 16;
+                dst_offset += width;
+                memcpy(uv_dst + dst_offset, uv_src + src_offset, 16);
+                src_offset += 16;
+                dst_offset += width;
+            }
+        }
+        if (aligned_width != width) {
+            src_offset = (tiled_width * i) + (j << 3);
+            dst_offset = width * i + j;
+            for (k = 0; k < 4; k++) {
+                memcpy(uv_dst + dst_offset, uv_src + src_offset, width - j);
+                src_offset += 16;
+                dst_offset += width;
+                memcpy(uv_dst + dst_offset, uv_src + src_offset, width - j);
+                src_offset += 16;
+                dst_offset += width;
+            }
+        }
+    }
+
+    if (aligned_height != height) {
+        for (j = 0; j<aligned_width; j = j + 16) {
+            src_offset = (tiled_width * i) + (j << 3);
+            dst_offset = width * i + j;
+            for (k = 0; k < height - aligned_height; k = k + 1) {
+                memcpy(uv_dst + dst_offset, uv_src + src_offset, 16);
+                src_offset += 16;
+                dst_offset += width;
+            }
+        }
+        if (aligned_width != width) {
+            src_offset = (tiled_width * i) + (j << 3);
+            dst_offset = width * i + j;
+            for (k = 0; k < height - aligned_height; k = k + 1) {
+                memcpy(uv_dst + dst_offset, uv_src + src_offset, width - j);
+                src_offset += 16;
+                dst_offset += width;
+            }
+        }
+    }
+}
+
+/*
+ * Converts tiled data to linear for mfc 6.x tiled
+ * 1. uv of nt12t to uv of yuv420p
+ *
+ * @param u_dst
+ *   u address of yuv420p[out]
+ *
+ * @param v_dst
+ *   v address of yuv420p[out]
+ *
+ * @param uv_src
+ *   uv address of nt12t[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420p[in]
+ *
+ * @param yuv420_height
+ *   real height of yuv420p[in]
+ */
+void csc_tiled_to_linear_uv_deinterleave(
+    unsigned char *u_dst,
+    unsigned char *v_dst,
+    unsigned char *uv_src,
+    unsigned int width,
+    unsigned int height)
+{
+    unsigned int i, j, k;
+    unsigned int aligned_width, aligned_height;
+    unsigned int tiled_width;
+    unsigned int src_offset, dst_offset;
+
+    aligned_height = height & (~0x7);
+    aligned_width = width & (~0xF);
+    tiled_width = ((width + 15) >> 4) << 4;
+
+    for (i = 0; i < aligned_height; i = i + 8) {
+        for (j = 0; j<aligned_width; j = j + 16) {
+            src_offset = (tiled_width * i) + (j << 3);
+            dst_offset = (width >> 1) * i + (j >> 1);
+            for (k = 0; k < 4; k++) {
+                csc_deinterleave_memcpy(u_dst + dst_offset, v_dst + dst_offset,
+                                        uv_src + src_offset, 16);
+                src_offset += 16;
+                dst_offset += width >> 1;
+                csc_deinterleave_memcpy(u_dst + dst_offset, v_dst + dst_offset,
+                                        uv_src + src_offset, 16);
+                src_offset += 16;
+                dst_offset += width >> 1;
+            }
+        }
+        if (aligned_width != width) {
+            src_offset = (tiled_width * i) + (j << 3);
+            dst_offset = (width >> 1) * i + (j >> 1);
+            for (k = 0; k < 4; k++) {
+                csc_deinterleave_memcpy(u_dst + dst_offset, v_dst + dst_offset,
+                                        uv_src + src_offset, width - j);
+                src_offset += 16;
+                dst_offset += width >> 1;
+                csc_deinterleave_memcpy(u_dst + dst_offset, v_dst + dst_offset,
+                                        uv_src + src_offset, width - j);
+                src_offset += 16;
+                dst_offset += width >> 1;
+            }
+        }
+    }
+    if (aligned_height != height) {
+        for (j = 0; j<aligned_width; j = j + 16) {
+            src_offset = (tiled_width * i) + (j << 3);
+            dst_offset = (width >> 1) * i + (j >> 1);
+            for (k = 0; k < height - aligned_height; k = k + 1) {
+                csc_deinterleave_memcpy(u_dst + dst_offset, v_dst + dst_offset,
+                                        uv_src + src_offset, 16);
+                src_offset += 16;
+                dst_offset += width >> 1;
+            }
+        }
+        if (aligned_width != width) {
+            src_offset = (tiled_width * i) + (j << 3);
+            dst_offset = (width >> 1) * i + (j >> 1);
+            for (k = 0; k < height - aligned_height; k = k + 1) {
+                csc_deinterleave_memcpy(u_dst + dst_offset, v_dst + dst_offset,
+                                        uv_src + src_offset, width - j);
+                src_offset += 16;
+                dst_offset += width >> 1;
+            }
+        }
+    }
+}
+
+/*
+ * Converts linear data to tiled
+ * It supports mfc 6.x tiled
+ * 1. y of yuv420 to y of nv12t
+ *
+ * @param dst
+ *   y address of nv12t[out]
+ *
+ * @param src
+ *   y address of yuv420[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420[in]
+ *   it should be even
+ *
+ * @param yuv420_height
+ *   real height of yuv420[in]
+ *   it should be even.
+ *
+ */
+void csc_linear_to_tiled_y(
+    unsigned char *y_dst,
+    unsigned char *y_src,
+    unsigned int width,
+    unsigned int height)
+{
+
+}
+
+/*
+ * Converts and interleaves linear data to tiled
+ * It supports mfc 6.x tiled
+ * 1. uv of nv12t to uv of yuv420
+ *
+ * @param dst
+ *   uv address of nv12t[out]
+ *
+ * @param src
+ *   u address of yuv420[in]
+ *
+ * @param src
+ *   v address of yuv420[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420[in]
+ *
+ * @param yuv420_height
+ *   real height of yuv420[in]
+ *
+ */
+void csc_linear_to_tiled_uv(
+    unsigned char *uv_dst,
+    unsigned char *u_src,
+    unsigned char *v_src,
+    unsigned int width,
+    unsigned int height)
+{
+
+}
+
+void Tile2D_To_YUV420(unsigned char *Y_plane, unsigned char *Cb_plane, unsigned char *Cr_plane,
+                        unsigned int y_addr, unsigned int c_addr, unsigned int width, unsigned int height)
+{
+    int x, y, j, k, l;
+    int out_of_width, actual_width;
+    unsigned int base_addr, data;
+
+    // y: 0, 16, 32, ...
+    for (y = 0; y < height; y += 16) {
+        // x: 0, 16, 32, ...
+        for (x = 0; x < width; x += 16) {
+            out_of_width = (x + 16) > width ? 1 : 0;
+            base_addr = y_addr + Tile2D_To_Linear(width, height, x, y, 0);
+
+            for (k = 0; (k < 16) && ((y + k) < height); k++) {
+                actual_width = out_of_width ? ((width%4)?((width%16) / 4 + 1) : ((width%16) / 4)) : 4;
+                for (l = 0; l < actual_width; l++) {
+                    data = *((unsigned int*)(base_addr + 16*k + l*4));
+                    for (j = 0; (j < 4) && (x + l*4 + j) < width; j++) {
+                        Y_plane[(y+k)*width + x + l*4 +j] = (data>>(8*j))&0xff;
+                    }
+                }
+            }
+        }
+    }
+
+    for (y = 0; y < height/2; y += 8) {
+        for (x = 0; x < width; x += 16) {
+            out_of_width = (x + 16) > width ? 1 : 0;
+            base_addr = c_addr + Tile2D_To_Linear(width, height/2, x, y, 1);
+            for (k = 0; (k < 8) && ((y+k) < height/2); k++) {
+                actual_width = out_of_width ? ((width%4) ? ((width%16) / 4 + 1) : ((width%16) / 4)) : 4;
+                for (l = 0; l < actual_width; l++) {
+                    data = *((unsigned int*)(base_addr + 16*k + l*4));
+                    for (j = 0; (j < 2) && (x/2 + l*2 +j) < width/2; j++) {
+                        Cb_plane[(y+k)*width/2 + x/2 + l*2 +j] = (data>> (8*2*j))&0xff;
+                        Cr_plane[(y+k)*width/2 + x/2 + l*2 +j] = (data>>(8*2*j+8))&0xff;
+                    }
+                }
+            }
+        }
+    }
+}
+
+/*
+ * Converts RGB565 to YUV420P
+ *
+ * @param y_dst
+ *   Y plane address of YUV420P[out]
+ *
+ * @param u_dst
+ *   U plane address of YUV420P[out]
+ *
+ * @param v_dst
+ *   V plane address of YUV420P[out]
+ *
+ * @param rgb_src
+ *   Address of RGB565[in]
+ *
+ * @param width
+ *   Width of RGB565[in]
+ *
+ * @param height
+ *   Height of RGB565[in]
+ */
+void csc_RGB565_to_YUV420P(
+    unsigned char *y_dst,
+    unsigned char *u_dst,
+    unsigned char *v_dst,
+    unsigned char *rgb_src,
+    int width,
+    int height)
+{
+    unsigned int i, j;
+    unsigned int tmp;
+
+    unsigned int R, G, B;
+    unsigned int Y, U, V;
+
+    unsigned int offset1 = width * height;
+    unsigned int offset2 = width/2 * height/2;
+
+    unsigned short int *pSrc = (unsigned short int *)rgb_src;
+
+    unsigned char *pDstY = (unsigned char *)y_dst;
+    unsigned char *pDstU = (unsigned char *)u_dst;
+    unsigned char *pDstV = (unsigned char *)v_dst;
+
+    unsigned int yIndex = 0;
+    unsigned int uIndex = 0;
+    unsigned int vIndex = 0;
+
+    for (j = 0; j < height; j++) {
+        for (i = 0; i < width; i++) {
+            tmp = pSrc[j * width + i];
+
+            R = (tmp & 0x0000F800) >> 8;
+            G = (tmp & 0x000007E0) >> 3;
+            B = (tmp & 0x0000001F);
+            B = B << 3;
+
+            Y = ((66 * R) + (129 * G) + (25 * B) + 128);
+            Y = Y >> 8;
+            Y += 16;
+
+            pDstY[yIndex++] = (unsigned char)Y;
+
+            if ((j % 2) == 0 && (i % 2) == 0) {
+                U = ((-38 * R) - (74 * G) + (112 * B) + 128);
+                U = U >> 8;
+                U += 128;
+                V = ((112 * R) - (94 * G) - (18 * B) + 128);
+                V = V >> 8;
+                V += 128;
+
+                pDstU[uIndex++] = (unsigned char)U;
+                pDstV[vIndex++] = (unsigned char)V;
+            }
+        }
+    }
+}
+
+/*
+ * Converts RGB565 to YUV420SP
+ *
+ * @param y_dst
+ *   Y plane address of YUV420SP[out]
+ *
+ * @param uv_dst
+ *   UV plane address of YUV420SP[out]
+ *
+ * @param rgb_src
+ *   Address of RGB565[in]
+ *
+ * @param width
+ *   Width of RGB565[in]
+ *
+ * @param height
+ *   Height of RGB565[in]
+ */
+void csc_RGB565_to_YUV420SP(
+    unsigned char *y_dst,
+    unsigned char *uv_dst,
+    unsigned char *rgb_src,
+    int width,
+    int height)
+{
+    unsigned int i, j;
+    unsigned int tmp;
+
+    unsigned int R, G, B;
+    unsigned int Y, U, V;
+
+    unsigned int offset = width * height;
+
+    unsigned short int *pSrc = (unsigned short int *)rgb_src;
+
+    unsigned char *pDstY = (unsigned char *)y_dst;
+    unsigned char *pDstUV = (unsigned char *)uv_dst;
+
+    unsigned int yIndex = 0;
+    unsigned int uvIndex = 0;
+
+    for (j = 0; j < height; j++) {
+        for (i = 0; i < width; i++) {
+            tmp = pSrc[j * width + i];
+
+            R = (tmp & 0x0000F800) >> 11;
+            R = R * 8;
+            G = (tmp & 0x000007E0) >> 5;
+            G = G * 4;
+            B = (tmp & 0x0000001F);
+            B = B * 8;
+
+            Y = ((66 * R) + (129 * G) + (25 * B) + 128);
+            Y = Y >> 8;
+            Y += 16;
+
+            pDstY[yIndex++] = (unsigned char)Y;
+
+            if ((j % 2) == 0 && (i % 2) == 0) {
+                U = ((-38 * R) - (74 * G) + (112 * B) + 128);
+                U = U >> 8;
+                U += 128;
+                V = ((112 * R) - (94 * G) - (18 * B) + 128);
+                V = V >> 8;
+                V += 128;
+
+                pDstUV[uvIndex++] = (unsigned char)U;
+                pDstUV[uvIndex++] = (unsigned char)V;
+            }
+        }
+    }
+}
diff --git a/exynos/multimedia/utils/csc/exynos5/color_space_convertor.h b/exynos/multimedia/utils/csc/exynos5/color_space_convertor.h
new file mode 100644
index 0000000..202011e
--- /dev/null
+++ b/exynos/multimedia/utils/csc/exynos5/color_space_convertor.h
@@ -0,0 +1,411 @@
+/*
+ *
+ * Copyright 2010 Samsung Electronics S.LSI Co. LTD
+ *
+ * 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.
+ */
+
+/*
+ * @file    color_space_convertor.h
+ * @brief   SEC_OMX specific define. It support MFC 6.x tiled.
+ *   NV12T(tiled) layout:
+ *   Each element is not pixel. It is 64x32 pixel block.
+ *   uv pixel block is interleaved as u v u v u v ...
+ *   y1    y2    y7    y8    y9    y10   y15   y16
+ *   y3    y4    y5    y6    y11   y12   y13   y14
+ *   y17   y18   y23   y24   y25   y26   y31   y32
+ *   y19   y20   y21   y22   y27   y28   y29   y30
+ *   uv1   uv2   uv7   uv8   uv9   uv10  uv15  uv16
+ *   uv3   uv4   uv5   uv6   uv11  uv12  uv13  uv14
+ *   YUV420Planar(linear) layout:
+ *   Each element is not pixel. It is 64x32 pixel block.
+ *   y1    y2    y3    y4    y5    y6    y7    y8
+ *   y9    y10   y11   y12   y13   y14   y15   y16
+ *   y17   y18   y19   y20   y21   y22   y23   y24
+ *   y25   y26   y27   y28   y29   y30   y31   y32
+ *   u1    u2    u3    u4    u5    u6    u7    u8
+ *   v1    v2    v3    v4    v5    v6    v7    v8
+ *   YUV420Semiplanar(linear) layout:
+ *   Each element is not pixel. It is 64x32 pixel block.
+ *   uv pixel block is interleaved as u v u v u v ...
+ *   y1    y2    y3    y4    y5    y6    y7    y8
+ *   y9    y10   y11   y12   y13   y14   y15   y16
+ *   y17   y18   y19   y20   y21   y22   y23   y24
+ *   y25   y26   y27   y28   y29   y30   y31   y32
+ *   uv1   uv2   uv3   uv4   uv5   uv6   uv7   uv8
+ *   uv9   uv10  uv11  uv12  uv13  uv14  uv15  uv16
+ * @author  ShinWon Lee (shinwon.lee@samsung.com)
+ * @version 1.0
+ * @history
+ *   2011.12.01 : Create
+ */
+
+#ifndef COLOR_SPACE_CONVERTOR_H_
+#define COLOR_SPACE_CONVERTOR_H_
+
+/*--------------------------------------------------------------------------------*/
+/* Format Conversion API                                                          */
+/*--------------------------------------------------------------------------------*/
+/* C Code */
+/*
+ * De-interleaves src to dest1, dest2
+ *
+ * @param dest1
+ *   Address of de-interleaved data[out]
+ *
+ * @param dest2
+ *   Address of de-interleaved data[out]
+ *
+ * @param src
+ *   Address of interleaved data[in]
+ *
+ * @param src_size
+ *   Size of interleaved data[in]
+ */
+void csc_deinterleave_memcpy(
+    unsigned char *dest1,
+    unsigned char *dest2,
+    unsigned char *src,
+    unsigned int src_size);
+
+/*
+ * Interleaves src1, src2 to dest
+ *
+ * @param dest
+ *   Address of interleaved data[out]
+ *
+ * @param src1
+ *   Address of de-interleaved data[in]
+ *
+ * @param src2
+ *   Address of de-interleaved data[in]
+ *
+ * @param src_size
+ *   Size of de-interleaved data[in]
+ */
+void csc_interleave_memcpy(
+    unsigned char *dest,
+    unsigned char *src1,
+    unsigned char *src2,
+    unsigned int src_size);
+
+/*
+ * Converts tiled data to linear
+ * It supports mfc 6.x tiled
+ * 1. y of nv12t to y of yuv420p
+ * 2. y of nv12t to y of yuv420s
+ *
+ * @param dst
+ *   y address of yuv420[out]
+ *
+ * @param src
+ *   y address of nv12t[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420[in]
+ *   it should be even
+ *
+ * @param yuv420_height
+ *   real height of yuv420[in]
+ *   it should be even.
+ *
+ */
+void csc_tiled_to_linear_y(
+    unsigned char *y_dst,
+    unsigned char *y_src,
+    unsigned int width,
+    unsigned int height);
+
+/*
+ * Converts tiled data to linear
+ * It supports mfc 6.x tiled
+ * 1. uv of nv12t to y of yuv420s
+ *
+ * @param dst
+ *   uv address of yuv420s[out]
+ *
+ * @param src
+ *   uv address of nv12t[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420s[in]
+ *
+ * @param yuv420_height
+ *   real height of yuv420s[in]
+ *
+ */
+void csc_tiled_to_linear_uv(
+    unsigned char *uv_dst,
+    unsigned char *uv_src,
+    unsigned int width,
+    unsigned int height);
+
+/*
+ * Converts tiled data to linear
+ * It supports mfc 6.x tiled
+ * 1. uv of nt12t to uv of yuv420p
+ *
+ * @param u_dst
+ *   u address of yuv420p[out]
+ *
+ * @param v_dst
+ *   v address of yuv420p[out]
+ *
+ * @param uv_src
+ *   uv address of nt12t[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420p[in]
+ *
+ * @param yuv420_height
+ *   real height of yuv420p[in]
+ */
+void csc_tiled_to_linear_uv_deinterleave(
+    unsigned char *u_dst,
+    unsigned char *v_dst,
+    unsigned char *uv_src,
+    unsigned int width,
+    unsigned int height);
+
+/*
+ * Converts linear data to tiled
+ * It supports mfc 6.x tiled
+ * 1. y of yuv420 to y of nv12t
+ *
+ * @param dst
+ *   y address of nv12t[out]
+ *
+ * @param src
+ *   y address of yuv420[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420[in]
+ *   it should be even
+ *
+ * @param yuv420_height
+ *   real height of yuv420[in]
+ *   it should be even.
+ *
+ */
+void csc_linear_to_tiled_y(
+    unsigned char *y_dst,
+    unsigned char *y_src,
+    unsigned int width,
+    unsigned int height);
+
+/*
+ * Converts and interleaves linear data to tiled
+ * It supports mfc 6.x tiled
+ * 1. uv of nv12t to uv of yuv420
+ *
+ * @param dst
+ *   uv address of nv12t[out]
+ *
+ * @param src
+ *   u address of yuv420[in]
+ *
+ * @param src
+ *   v address of yuv420[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420[in]
+ *
+ * @param yuv420_height
+ *   real height of yuv420[in]
+ *
+ */
+void csc_linear_to_tiled_uv(
+    unsigned char *uv_dst,
+    unsigned char *u_src,
+    unsigned char *v_src,
+    unsigned int width,
+    unsigned int height);
+
+/*
+ * Converts RGB565 to YUV420P
+ *
+ * @param y_dst
+ *   Y plane address of YUV420P[out]
+ *
+ * @param u_dst
+ *   U plane address of YUV420P[out]
+ *
+ * @param v_dst
+ *   V plane address of YUV420P[out]
+ *
+ * @param rgb_src
+ *   Address of RGB565[in]
+ *
+ * @param width
+ *   Width of RGB565[in]
+ *
+ * @param height
+ *   Height of RGB565[in]
+ */
+void csc_RGB565_to_YUV420P(
+    unsigned char *y_dst,
+    unsigned char *u_dst,
+    unsigned char *v_dst,
+    unsigned char *rgb_src,
+    int width,
+    int height);
+
+/*
+ * Converts RGB565 to YUV420S
+ *
+ * @param y_dst
+ *   Y plane address of YUV420S[out]
+ *
+ * @param uv_dst
+ *   UV plane address of YUV420S[out]
+ *
+ * @param rgb_src
+ *   Address of RGB565[in]
+ *
+ * @param width
+ *   Width of RGB565[in]
+ *
+ * @param height
+ *   Height of RGB565[in]
+ */
+void csc_RGB565_to_YUV420SP(
+    unsigned char *y_dst,
+    unsigned char *uv_dst,
+    unsigned char *rgb_src,
+    int width,
+    int height);
+
+/*
+ * Converts tiled data to linear for mfc 6.x
+ * 1. Y of NV12T to Y of YUV420P
+ * 2. Y of NV12T to Y of YUV420S
+ *
+ * @param dst
+ *   Y address of YUV420[out]
+ *
+ * @param src
+ *   Y address of NV12T[in]
+ *
+ * @param yuv420_width
+ *   real width of YUV420[in]
+ *
+ * @param yuv420_height
+ *   Y: real height of YUV420[in]
+ *
+ */
+void csc_tiled_to_linear_y_neon(
+    unsigned char *y_dst,
+    unsigned char *y_src,
+    unsigned int width,
+    unsigned int height);
+
+/*
+ * Converts tiled data to linear for mfc 6.x
+ * 1. UV of NV12T to Y of YUV420S
+ *
+ * @param u_dst
+ *   UV plane address of YUV420P[out]
+ *
+ * @param nv12t_src
+ *   Y or UV plane address of NV12T[in]
+ *
+ * @param yuv420_width
+ *   real width of YUV420[in]
+ *
+ * @param yuv420_height
+ *   (real height)/2 of YUV420[in]
+ */
+void csc_tiled_to_linear_uv_neon(
+    unsigned char *uv_dst,
+    unsigned char *uv_src,
+    unsigned int width,
+    unsigned int height);
+
+/*
+ * Converts tiled data to linear for mfc 6.x
+ * Deinterleave src to u_dst, v_dst
+ * 1. UV of NV12T to Y of YUV420P
+ *
+ * @param u_dst
+ *   U plane address of YUV420P[out]
+ *
+ * @param v_dst
+ *   V plane address of YUV420P[out]
+ *
+ * @param nv12t_src
+ *   Y or UV plane address of NV12T[in]
+ *
+ * @param yuv420_width
+ *   real width of YUV420[in]
+ *
+ * @param yuv420_height
+ *   (real height)/2 of YUV420[in]
+ */
+void csc_tiled_to_linear_uv_deinterleave_neon(
+    unsigned char *u_dst,
+    unsigned char *v_dst,
+    unsigned char *uv_src,
+    unsigned int width,
+    unsigned int height);
+
+/*
+ * Converts linear data to tiled
+ * It supports mfc 6.x tiled
+ * 1. y of yuv420 to y of nv12t
+ *
+ * @param dst
+ *   y address of nv12t[out]
+ *
+ * @param src
+ *   y address of yuv420[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420[in]
+ *   it should be even
+ *
+ * @param yuv420_height
+ *   real height of yuv420[in]
+ *   it should be even.
+ *
+ */
+void csc_linear_to_tiled_y_neon(
+    unsigned char *y_dst,
+    unsigned char *y_src,
+    unsigned int width,
+    unsigned int height);
+
+/*
+ * Converts and interleave linear data to tiled
+ * It supports mfc 6.x tiled
+ * 1. uv of nv12t to uv of yuv420
+ *
+ * @param dst
+ *   uv address of yuv420[out]
+ *
+ * @param src
+ *   uv address of nv12t[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420[in]
+ *
+ * @param yuv420_height
+ *   real height of yuv420[in]
+ *
+ */
+void csc_linear_to_tiled_uv_neon(
+    unsigned char *uv_dst,
+    unsigned char *uv_src,
+    unsigned int width,
+    unsigned int height);
+
+#endif /*COLOR_SPACE_CONVERTOR_H_*/
diff --git a/exynos/multimedia/utils/csc/exynos5/csc_fimc.cpp b/exynos/multimedia/utils/csc/exynos5/csc_fimc.cpp
new file mode 100644
index 0000000..89d0f97
--- /dev/null
+++ b/exynos/multimedia/utils/csc/exynos5/csc_fimc.cpp
@@ -0,0 +1,225 @@
+/*
+ * Copyright (C) 2009 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.
+ */
+
+/*
+ * @file    csc_fimc.cpp
+ *
+ * @brief   csc_fimc use fimc1 to color space convertion
+ *
+ * @author  ShinWon Lee (shinwon.lee@samsung.com)
+ *
+ * @version 1.0
+ *
+ * @history
+ *   2011.11.01 : Create
+ */
+
+#include <utils/Log.h>
+#include <dlfcn.h>
+
+#include "SecFimc.h"
+#include "csc_fimc.h"
+
+#define ALIGN(value, base) (((value) + (base) - 1) & ~((base) - 1))
+
+unsigned int OMXtoHarPixelFomrat(OMX_COLOR_FORMATTYPE ColorFormat)
+{
+    unsigned int v4l2_format = 0;
+    switch (ColorFormat) {
+    case OMX_COLOR_FormatYUV420Planar:
+        v4l2_format = HAL_PIXEL_FORMAT_YCbCr_420_P;
+        break;
+    case OMX_COLOR_FormatYUV420SemiPlanar:
+        v4l2_format = HAL_PIXEL_FORMAT_YCbCr_420_SP;
+        break;
+    default:
+        v4l2_format = HAL_PIXEL_FORMAT_YCbCr_420_P;
+        break;
+    }
+    return v4l2_format;
+}
+
+/*
+ * create and open fimc handle
+ *
+ * @return
+ *   fimc handle
+ */
+void *csc_fimc_open()
+{
+    SecFimc *(*create_instance)();
+    void (*destroy_instance)(void *);
+    SecFimc *handle_fimc = NULL;
+    void* hdl = NULL;
+
+    hdl = dlopen("libfimc.so", RTLD_NOW);
+    if (hdl == NULL) {
+        LOGE("%s:: load libfimc.so failed", __func__);
+        return NULL;
+    }
+
+    create_instance = (SecFimc *(*)())dlsym(hdl, "create_instance");
+    handle_fimc = (SecFimc *)create_instance();
+    if (handle_fimc == NULL) {
+        LOGE("%s:: create handle_fimc failed", __func__);
+        return NULL;
+    }
+
+    if (!handle_fimc->create(SecFimc::DEV_1, SecFimc::MODE_MULTI_BUF, 1)) {
+        destroy_instance = (void (*)(void *))dlsym(hdl, "destroy_instance");
+        destroy_instance(handle_fimc);
+        LOGE("%s:: create() failed", __func__);
+        return NULL;
+    }
+    return (void *)handle_fimc;
+}
+
+/*
+ * close and destroy fimc handle
+ *
+ * @param handle
+ *   fimc handle[in]
+ *
+ * @return
+ *   pass or fail
+ */
+CSC_FIMC_ERROR_CODE csc_fimc_close(void *handle)
+{
+    void (*destroy_instance)(void *);
+    SecFimc *handle_fimc = (SecFimc *)handle;
+    void* hdl = NULL;
+
+    if (!handle_fimc->destroy()) {
+        LOGE("%s:: destroy() failed", __func__);
+        return CSC_FIMC_RET_FAIL;
+    }
+
+    hdl = dlopen("libfimc.so", RTLD_NOW);
+    if (hdl == NULL) {
+        LOGE("%s:: load libfimc.so failed", __func__);
+        return CSC_FIMC_RET_FAIL;
+    }
+
+    destroy_instance = (void (*)(void *))dlsym(hdl, "destroy_instance");
+    destroy_instance(handle);
+
+    return CSC_FIMC_RET_OK;
+}
+
+/*
+ * convert color space nv12t to omxformat
+ *
+ * @param handle
+ *   fimc handle[in]
+ *
+ * @param dst_addr
+ *   y,u,v address of dst_addr[out]
+ *
+ * @param src_addr
+ *   y,uv address of src_addr.Format is nv12t[in]
+ *
+ * @param width
+ *   width of dst image[in]
+ *
+ * @param height
+ *   height of dst image[in]
+ *
+ * @param omxformat
+ *   omxformat of dst image[in]
+ *
+ * @return
+ *   pass or fail
+ */
+CSC_FIMC_ERROR_CODE csc_fimc_convert_nv12t(
+    void *handle,
+    void **dst_addr,
+    void **src_addr,
+    unsigned int width,
+    unsigned int height,
+    OMX_COLOR_FORMATTYPE omxformat)
+{
+    int rotate_value = 0;
+
+    SecFimc *handle_fimc = (SecFimc *)handle;
+
+    unsigned int src_crop_x = 0;
+    unsigned int src_crop_y = 0;
+    unsigned int src_crop_width = width;
+    unsigned int src_crop_height = height;
+
+    unsigned int dst_crop_x = 0;
+    unsigned int dst_crop_y = 0;
+    unsigned int dst_crop_width = width;
+    unsigned int dst_crop_height = height;
+
+    unsigned int HarPixelformat = 0;
+    HarPixelformat = OMXtoHarPixelFomrat(omxformat);
+
+    // set post processor configuration
+    if (!handle_fimc->setSrcParams(width, height, src_crop_x, src_crop_y,
+                                   &src_crop_width, &src_crop_height,
+                                   HAL_PIXEL_FORMAT_CUSTOM_YCbCr_420_SP_TILED)) {
+        LOGE("%s:: setSrcParms() failed", __func__);
+        return CSC_FIMC_RET_FAIL;
+    }
+
+    if (!handle_fimc->setSrcAddr((unsigned int)src_addr[0],
+                                 (unsigned int)src_addr[1],
+                                 (unsigned int)src_addr[1],
+                                 HAL_PIXEL_FORMAT_CUSTOM_YCbCr_420_SP_TILED)) {
+        LOGE("%s:: setSrcPhyAddr() failed", __func__);
+        return CSC_FIMC_RET_FAIL;
+    }
+
+    if (!handle_fimc->setRotVal(rotate_value)) {
+        LOGE("%s:: setRotVal() failed", __func__);
+        return CSC_FIMC_RET_FAIL;
+    }
+
+    if (!handle_fimc->setDstParams(width, height, dst_crop_x, dst_crop_y,
+                                   &dst_crop_width, &dst_crop_height,
+                                   HarPixelformat)) {
+        LOGE("%s:: setDstParams() failed", __func__);
+        return CSC_FIMC_RET_FAIL;
+    }
+
+    switch (omxformat) {
+    case OMX_COLOR_FormatYUV420SemiPlanar:
+        if (!handle_fimc->setDstAddr((unsigned int)(dst_addr[0]),
+                                     (unsigned int)(dst_addr[1]),
+                                     (unsigned int)(dst_addr[1]))) {
+            LOGE("%s:: setDstPhyAddr() failed", __func__);
+            return CSC_FIMC_RET_FAIL;
+        }
+        break;
+    case OMX_COLOR_FormatYUV420Planar:
+    default:
+        if (!handle_fimc->setDstAddr((unsigned int)(dst_addr[0]),
+                                     (unsigned int)(dst_addr[1]),
+                                     (unsigned int)(dst_addr[2]))) {
+            LOGE("%s:: setDstPhyAddr() failed", __func__);
+            return CSC_FIMC_RET_FAIL;
+        }
+        break;
+    }
+
+    if (!handle_fimc->draw(0, 0)) {
+        LOGE("%s:: handleOneShot() failed", __func__);
+        return CSC_FIMC_RET_FAIL;
+    }
+
+    return CSC_FIMC_RET_OK;
+}
diff --git a/exynos/multimedia/utils/csc/exynos5/csc_fimc.h b/exynos/multimedia/utils/csc/exynos5/csc_fimc.h
new file mode 100644
index 0000000..3ae24ac
--- /dev/null
+++ b/exynos/multimedia/utils/csc/exynos5/csc_fimc.h
@@ -0,0 +1,106 @@
+/*
+ * Copyright (C) 2009 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.
+ */
+
+/*
+ * @file    csc_fimc.h
+ *
+ * @brief   csc_fimc use fimc1 to color space convertion
+ *
+ * @author  ShinWon Lee (shinwon.lee@samsung.com)
+ *
+ * @version 1.0
+ *
+ * @history
+ *   2011.11.01 : Create
+ */
+
+#ifndef CSC_FIMC_H
+
+#define CSC_FIMC_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <OMX_Video.h>
+
+/*--------------------------------------------------------------------------------*/
+/* Structure and Type                                                             */
+/*--------------------------------------------------------------------------------*/
+typedef enum {
+    CSC_FIMC_RET_OK                 = 0,
+    CSC_FIMC_RET_FAIL               = -1
+} CSC_FIMC_ERROR_CODE;
+
+/*--------------------------------------------------------------------------------*/
+/* CSC FIMC APIs                                                                  */
+/*--------------------------------------------------------------------------------*/
+/*
+ * create and open fimc handle
+ *
+ * @return
+ *   fimc handle
+ */
+void *csc_fimc_open();
+
+/*
+ * close and destroy fimc handle
+ *
+ * @param handle
+ *   fimc handle[in]
+ *
+ * @return
+ *   error code
+ */
+CSC_FIMC_ERROR_CODE csc_fimc_close(void *handle);
+
+/*
+ * convert color space nv12t to omxformat
+ *
+ * @param handle
+ *   fimc handle[in]
+ *
+ * @param dst_addr
+ *   y,u,v address of dst_addr[out]
+ *
+ * @param src_addr
+ *   y,uv address of src_addr.Format is nv12t[in]
+ *
+ * @param width
+ *   width of dst image[in]
+ *
+ * @param height
+ *   height of dst image[in]
+ *
+ * @param omxformat
+ *   omxformat of dst image[in]
+ *
+ * @return
+ *   error code
+ */
+CSC_FIMC_ERROR_CODE csc_fimc_convert_nv12t(
+    void *handle,
+    void **dst_addr,
+    void **src_addr,
+    unsigned int width,
+    unsigned int height,
+    OMX_COLOR_FORMATTYPE omxformat);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/exynos/multimedia/utils/csc/exynos5/csc_interleave_memcpy_neon.s b/exynos/multimedia/utils/csc/exynos5/csc_interleave_memcpy_neon.s
new file mode 100644
index 0000000..3d2b41f
--- /dev/null
+++ b/exynos/multimedia/utils/csc/exynos5/csc_interleave_memcpy_neon.s
@@ -0,0 +1,120 @@
+/*
+ *
+ * Copyright 2012 Samsung Electronics S.LSI Co. LTD
+ *
+ * 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.
+ */
+
+/*
+ * @file    csc_linear_to_tiled_crop_neon.s
+ * @brief   SEC_OMX specific define
+ * @author  ShinWon Lee (shinwon.lee@samsung.com)
+ * @version 1.0
+ * @history
+ *   2012.01.04 : Create
+ */
+
+/*
+ * Interleave src1, src2 to dst
+ *
+ * @param dest
+ *   dst address[out]
+ *
+ * @param src1
+ *   src1 address[in]
+ *
+ * @param src2
+ *   src2 address[in]
+ *
+ * @param src_size
+ *   src_size or src1
+ */
+
+    .arch armv7-a
+    .text
+    .global csc_interleave_memcpy_neon
+    .type   csc_interleave_memcpy_neon, %function
+csc_interleave_memcpy_neon:
+    .fnstart
+
+    @r0     dest
+    @r1     src1
+    @r2     src2
+    @r3     src_size
+    @r4
+    @r5
+    @r6
+    @r7
+    @r8     temp1
+    @r9     temp2
+    @r10    dest_addr
+    @r11    src1_addr
+    @r12    src2_addr
+    @r14    i
+
+    stmfd       sp!, {r8-r12,r14}       @ backup registers
+
+    mov         r10, r0
+    mov         r11, r1
+    mov         r12, r2
+    mov         r14, r3
+
+    cmp         r14, #128
+    blt         LESS_THAN_128
+
+LOOP_128:
+    vld1.8      {q0}, [r11]!
+    vld1.8      {q2}, [r11]!
+    vld1.8      {q4}, [r11]!
+    vld1.8      {q6}, [r11]!
+    vld1.8      {q8}, [r11]!
+    vld1.8      {q10}, [r11]!
+    vld1.8      {q12}, [r11]!
+    vld1.8      {q14}, [r11]!
+    vld1.8      {q1}, [r12]!
+    vld1.8      {q3}, [r12]!
+    vld1.8      {q5}, [r12]!
+    vld1.8      {q7}, [r12]!
+    vld1.8      {q9}, [r12]!
+    vld1.8      {q11}, [r12]!
+    vld1.8      {q13}, [r12]!
+    vld1.8      {q15}, [r12]!
+
+    vst2.8      {q0, q1}, [r10]!
+    vst2.8      {q2, q3}, [r10]!
+    vst2.8      {q4, q5}, [r10]!
+    vst2.8      {q6, q7}, [r10]!
+    vst2.8      {q8, q9}, [r10]!
+    vst2.8      {q10, q11}, [r10]!
+    vst2.8      {q12, q13}, [r10]!
+    vst2.8      {q14, q15}, [r10]!
+
+    sub         r14, #128
+    cmp         r14, #128
+    bgt         LOOP_128
+
+LESS_THAN_128:
+    cmp         r14, #0
+    beq         RESTORE_REG
+
+LOOP_1:
+    ldrb        r8, [r11], #1
+    ldrb        r9, [r12], #1
+    strb        r8, [r10], #1
+    strb        r9, [r10], #1
+    subs        r14, #1
+    bne         LOOP_1
+
+RESTORE_REG:
+    ldmfd       sp!, {r8-r12,r15}       @ restore registers
+    .fnend
diff --git a/exynos/multimedia/utils/csc/exynos5/csc_tiled_to_linear_uv_deinterleave_neon.s b/exynos/multimedia/utils/csc/exynos5/csc_tiled_to_linear_uv_deinterleave_neon.s
new file mode 100644
index 0000000..1e76a2e
--- /dev/null
+++ b/exynos/multimedia/utils/csc/exynos5/csc_tiled_to_linear_uv_deinterleave_neon.s
@@ -0,0 +1,249 @@
+/*
+ *
+ * Copyright 2010 Samsung Electronics S.LSI Co. LTD
+ *
+ * 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.
+ */
+
+/*
+ * @file    csc_tiled_to_linear_uv_deinterleave_neon.s
+ * @brief   SEC_OMX specific define. It support MFC 6.x tiled.
+ * @author  ShinWon Lee (shinwon.lee@samsung.com)
+ * @version 1.0
+ * @history
+ *   2011.12.01 : Create
+ */
+
+/*
+ * Converts and Deinterleave tiled data to linear for mfc 6.x
+ * 1. UV of NV12T to Y of YUV420P
+ *
+ * @param u_dst
+ *   U address of YUV420[out]
+ *
+ * @param v_dst
+ *   V address of YUV420[out]
+ *
+ * @param uv_src
+ *   UV address of NV12T[in]
+ *
+ * @param yuv420_width
+ *   real width of YUV420[in]. It should be even.
+ *
+ * @param yuv420_height
+ *   real height of YUV420[in] It should be even.
+ */
+
+    .arch armv7-a
+    .text
+    .global csc_tiled_to_linear_uv_deinterleave_neon
+    .type   csc_tiled_to_linear_uv_deinterleave_neon, %function
+csc_tiled_to_linear_uv_deinterleave_neon:
+    .fnstart
+
+    .equ CACHE_LINE_SIZE, 64
+
+    @r0     u_dst
+    @r1     v_dst
+    @r2     uv_src
+    @r3     width
+    @r4     height
+    @r5     i
+    @r6     j
+    @r7     dst_offset
+    @r8     src_offset
+    @r9     aligned_height
+    @r10    aligned_width
+    @r11    tiled_width
+    @r12    temp1
+    @r14    temp2
+
+    stmfd       sp!, {r4-r12,r14}       @ backup registers
+    ldr         r4, [sp, #40]           @ r4 = height
+
+    bic         r9, r4, #0x7            @ aligned_height = height & (~0x7)
+    bic         r10, r3, #0xF           @ aligned_width = width & (~0xF)
+    add         r11, r3, #15            @ tiled_width = ((width + 15) >> 4) << 4
+    mov         r11, r11, asr #4
+    mov         r11, r11, lsl #4
+
+    mov         r5, #0
+LOOP_MAIN_ALIGNED_HEIGHT:
+    mul         r8, r11, r5             @ src_offset = tiled_width * i
+    mov         r6, #0
+    add         r8, r2, r8              @ src_offset = uv_src + src_offset
+LOOP_MAIN_ALIGNED_WIDTH:
+    mov         r12, r3, asr #1         @ temp1 = (width >> 1) * i + (j >> 1)
+    mul         r12, r12, r5
+
+    pld         [r8, #CACHE_LINE_SIZE]
+    vld2.8      {q0, q1}, [r8]!
+    add         r12, r12, r6, asr #1
+    vld2.8      {q2, q3}, [r8]!
+    add         r7, r0, r12             @ dst_offset = u_dst + temp1
+    pld         [r8, #CACHE_LINE_SIZE]
+    vld2.8      {q4, q5}, [r8]!
+    mov         r14, r3, asr #1         @ temp2 = width / 2
+    vld2.8      {q6, q7}, [r8]!
+
+    vst1.8      {d0}, [r7], r14
+    vst1.8      {d1}, [r7], r14
+    vst1.8      {d4}, [r7], r14
+    vst1.8      {d5}, [r7], r14
+    vst1.8      {d8}, [r7], r14
+    vst1.8      {d9}, [r7], r14
+    vst1.8      {d12}, [r7], r14
+    vst1.8      {d13}, [r7], r14
+
+    add         r7, r1, r12             @ dst_offset = v_dst + temp1
+
+    vst1.8      {d2}, [r7], r14
+    vst1.8      {d3}, [r7], r14
+    vst1.8      {d6}, [r7], r14
+    vst1.8      {d7}, [r7], r14
+    vst1.8      {d10}, [r7], r14
+    vst1.8      {d11}, [r7], r14
+    add         r6, r6, #16
+    vst1.8      {d14}, [r7], r14
+    cmp         r6, r10
+    vst1.8      {d15}, [r7], r14
+    blt         LOOP_MAIN_ALIGNED_WIDTH
+
+MAIN_REMAIN_WIDTH_START:
+    cmp         r10, r3                 @ if (aligned_width != width) {
+    beq         MAIN_REMAIN_WIDTH_END
+    stmfd       sp!, {r0-r2,r4}         @ backup registers
+    mul         r8, r11, r5             @ src_offset = (tiled_width * i) + (j << 3)
+    add         r8, r8, r6, lsl #3
+    add         r8, r2, r8              @ r8 = uv_src + src_offset
+    mov         r12, r3, asr #1         @ temp1 = (width >> 1) * i + (j >> 1)
+    mul         r12, r12, r5
+    add         r12, r12, r6, asr #1
+    add         r7, r0, r12             @ r7 = u_dst + temp1
+    add         r12, r1, r12            @ r12 = v_dst + temp1
+    sub         r14, r3, r6             @ r14 = (width - j) / 2
+    mov         r14, r14, asr #1
+
+    mov         r4, #0
+LOOP_MAIN_REMAIN_HEIGHT:
+    mov         r0, #0                  @ r0 is index in de-interleave
+LOOP_MAIN_REMAIN_WIDTH:
+    ldrb        r1, [r8], #1
+    ldrb        r2, [r8], #1
+    strb        r1, [r7], #1
+    strb        r2, [r12], #1
+    add         r0, #1
+    cmp         r0, r14
+    blt         LOOP_MAIN_REMAIN_WIDTH
+
+    sub         r8, r8, r14, lsl #1
+    sub         r7, r7, r14
+    sub         r12, r12, r14
+    add         r8, r8, #16
+    add         r7, r7, r3, asr #1
+    add         r12, r12, r3, asr #1
+
+    add         r4, #1
+    cmp         r4, #8
+    blt         LOOP_MAIN_REMAIN_HEIGHT
+    ldmfd       sp!, {r0-r2,r4}         @ restore registers
+MAIN_REMAIN_WIDTH_END:
+
+    add         r5, r5, #8
+    cmp         r5, r9
+    blt         LOOP_MAIN_ALIGNED_HEIGHT
+
+REMAIN_HEIGHT_START:
+    cmp         r9, r4                  @ if (aligned_height != height) {
+    beq         REMAIN_HEIGHT_END
+
+    mov         r6, #0
+LOOP_REMAIN_HEIGHT_WIDTH16:
+    mul         r8, r11, r5             @ src_offset = (tiled_width * i) + (j << 3)
+    add         r8, r8, r6, lsl #3
+    add         r8, r2, r8              @ src_offset = uv_src + src_offset
+
+    mov         r12, r3, asr #1         @ temp1 = (width >> 1) * i + (j >> 1)
+    mul         r12, r12, r5
+    add         r12, r12, r6, asr #1
+    add         r7, r0, r12             @ r7 = u_dst + temp1
+    add         r12, r1, r12            @ r12 = v_dst + temp1
+    mov         r14, r3, asr #1         @ temp2 = width / 2
+
+    stmfd       sp!, {r0-r1}            @ backup registers
+    mov         r0, #0
+    sub         r1, r4, r9
+LOOP_REMAIN_HEIGHT_WIDTH16_HEIGHT1:
+    vld2.8      {d0, d1}, [r8]!
+    vst1.8      {d0}, [r7], r14
+    vst1.8      {d1}, [r12], r14
+
+    add         r0, r0, #1
+    cmp         r0, r1
+    blt         LOOP_REMAIN_HEIGHT_WIDTH16_HEIGHT1
+    ldmfd       sp!, {r0-r1}            @ restore registers
+
+    add         r6, r6, #16
+    cmp         r6, r10
+    blt         LOOP_REMAIN_HEIGHT_WIDTH16
+
+REMAIN_HEIGHT_REMAIN_WIDTH_START:
+    cmp         r10, r3
+    beq         REMAIN_HEIGHT_REMAIN_WIDTH_END
+    mul         r8, r11, r5             @ src_offset = (tiled_width * i) + (j << 3)
+    add         r8, r8, r6, lsl #3
+    add         r8, r2, r8              @ src_offset = uv_src + src_offset
+
+    mov         r12, r3, asr #1         @ temp1 = (width >> 1) * i + (j >> 1)
+    mul         r12, r12, r5
+    add         r12, r12, r6, asr #1
+    add         r7, r0, r12             @ r7 = u_dst + temp1
+    add         r12, r1, r12            @ r12 = v_dst + temp1
+    sub         r14, r3, r6             @ r14 = (width - j) /2
+    mov         r14, r14, asr #1
+
+    stmfd       sp!, {r0-r2,r4-r5}            @ backup registers
+    mov         r0, #0
+    sub         r1, r4, r9
+LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1:
+
+    mov         r4, #0
+LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1_WIDTHx:
+    ldrb        r2, [r8], #1
+    ldrb        r5, [r8], #1
+    strb        r2, [r7], #1
+    strb        r5, [r12], #1
+    add         r4, #1
+    cmp         r4, r14
+    blt         LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1_WIDTHx
+
+    sub         r8, r8, r14, lsl #1
+    sub         r7, r7, r14
+    sub         r12, r12, r14
+    add         r8, r8, #16
+    add         r7, r7, r3, asr #1
+    add         r12, r12, r3, asr #1
+
+    add         r0, r0, #1
+    cmp         r0, r1
+    blt         LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1
+    ldmfd       sp!, {r0-r2,r4-r5}            @ restore registers
+
+REMAIN_HEIGHT_REMAIN_WIDTH_END:
+
+REMAIN_HEIGHT_END:
+
+RESTORE_REG:
+    ldmfd       sp!, {r4-r12,r15}       @ restore registers
+
+    .fnend
diff --git a/exynos/multimedia/utils/csc/exynos5/csc_tiled_to_linear_uv_neon.s b/exynos/multimedia/utils/csc/exynos5/csc_tiled_to_linear_uv_neon.s
new file mode 100644
index 0000000..ad58419
--- /dev/null
+++ b/exynos/multimedia/utils/csc/exynos5/csc_tiled_to_linear_uv_neon.s
@@ -0,0 +1,216 @@
+/*
+ *
+ * Copyright 2010 Samsung Electronics S.LSI Co. LTD
+ *
+ * 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.
+ */
+
+/*
+ * @file    csc_tiled_to_linear_uv.s
+ * @brief   SEC_OMX specific define. It support MFC 6.x tiled.
+ * @author  ShinWon Lee (shinwon.lee@samsung.com)
+ * @version 1.0
+ * @history
+ *   2011.12.01 : Create
+ */
+
+/*
+ * Converts tiled data to linear for mfc 6.x tiled
+ * 1. uv of nv12t to y of yuv420s
+ *
+ * @param dst
+ *   uv address of yuv420s[out]
+ *
+ * @param src
+ *   uv address of nv12t[in]
+ *
+ * @param yuv420_width
+ *   real width of yuv420s[in]
+ *
+ * @param yuv420_height
+ *   real height of yuv420s[in]
+ *
+ */
+    .arch armv7-a
+    .text
+    .global csc_tiled_to_linear_uv_neon
+    .type   csc_tiled_to_linear_uv_neon, %function
+csc_tiled_to_linear_uv_neon:
+    .fnstart
+
+    .equ CACHE_LINE_SIZE, 64
+
+    @r0     y_dst
+    @r1     y_src
+    @r2     width
+    @r3     height
+    @r4     temp3
+    @r5     i
+    @r6     j
+    @r7     dst_offset
+    @r8     src_offset
+    @r9     aligned_height
+    @r10    aligned_width
+    @r11    tiled_width
+    @r12    temp1
+    @r14    temp2
+
+    stmfd       sp!, {r4-r12,r14}       @ backup registers
+    ldr         r4, [sp, #40]           @ r4 = height
+
+    bic         r9, r3, #0x7            @ aligned_height = height & (~0xF)
+    bic         r10, r2, #0xF           @ aligned_width = width & (~0xF)
+    add         r11, r2, #15            @ tiled_width = ((width + 15) >> 4) << 4
+    mov         r11, r11, asr #4
+    mov         r11, r11, lsl #4
+
+    mov         r5, #0
+LOOP_MAIN_ALIGNED_HEIGHT:
+    mul         r8, r11, r5             @ src_offset = tiled_width * i
+    mov         r6, #0
+    add         r8, r1, r8              @ src_offset = y_src + src_offset
+LOOP_MAIN_ALIGNED_WIDTH:
+    pld         [r8, #CACHE_LINE_SIZE]
+    vld1.8      {q0, q1}, [r8]!
+    mul         r12, r2, r5             @ temp1 = width * i + j;
+    vld1.8      {q2, q3}, [r8]!
+    add         r12, r12, r6
+    pld         [r8, #CACHE_LINE_SIZE]
+    vld1.8      {q4, q5}, [r8]!
+    add         r7, r0, r12             @ dst_offset = y_dst + temp1
+    vld1.8      {q6, q7}, [r8]!
+
+    vst1.8      {q0}, [r7], r2
+    vst1.8      {q1}, [r7], r2
+    vst1.8      {q2}, [r7], r2
+    vst1.8      {q3}, [r7], r2
+    vst1.8      {q4}, [r7], r2
+    vst1.8      {q5}, [r7], r2
+    vst1.8      {q6}, [r7], r2
+    vst1.8      {q7}, [r7], r2
+    add         r6, r6, #16
+    cmp         r6, r10
+    blt         LOOP_MAIN_ALIGNED_WIDTH
+
+MAIN_REMAIN_WIDTH_START:
+    cmp         r10, r2                 @ if (aligned_width != width) {
+    beq         MAIN_REMAIN_WIDTH_END
+
+    mul         r8, r11, r5             @ src_offset = (tiled_width * i) + (j << 3);
+    add         r8, r8, r6, lsl #3
+    add         r8, r1, r8              @ r8 = y_src + src_offset
+
+    mul         r12, r2, r5             @ temp1 = width * i + j;
+    add         r12, r12, r6
+    add         r7, r0, r12             @ r7 = y_dst + temp1
+    sub         r14, r2, r6             @ r14 = width - j
+
+    stmfd       sp!, {r0-r1}            @ backup registers
+    mov         r1, #0
+LOOP_MAIN_REMAIN_HEIGHT:
+    mov         r0, #0                  @ r0 is index in memcpy
+LOOP_MAIN_REMAIN_WIDTH:
+    ldrh        r4, [r8], #2
+    strh        r4, [r7], #2
+    add         r0, #2
+    cmp         r0, r14
+    blt         LOOP_MAIN_REMAIN_WIDTH
+
+    sub         r8, r8, r14
+    sub         r7, r7, r14
+    add         r8, r8, #16
+    add         r7, r7, r2
+
+    add         r1, #1
+    cmp         r1, #8
+    blt         LOOP_MAIN_REMAIN_HEIGHT
+    ldmfd       sp!, {r0-r1}            @ restore registers
+MAIN_REMAIN_WIDTH_END:
+
+    add         r5, r5, #8
+    cmp         r5, r9
+    blt         LOOP_MAIN_ALIGNED_HEIGHT
+
+REMAIN_HEIGHT_START:
+    cmp         r9, r3                  @ if (aligned_height != height) {
+    beq         REMAIN_HEIGHT_END
+
+    mov         r6, #0
+LOOP_REMAIN_HEIGHT_WIDTH16:
+    mul         r8, r11, r5             @ src_offset = (tiled_width * i) + (j << 3)
+    add         r8, r8, r6, lsl #3
+    add         r8, r1, r8              @ src_offset = y_src + src_offset
+
+    mul         r12, r2, r5             @ temp1 = width * i + j;
+    add         r12, r12, r6
+    add         r7, r0, r12             @ r7 = y_dst + temp1
+
+    sub         r12, r3, r9
+    mov         r14, #0
+LOOP_REMAIN_HEIGHT_WIDTH16_HEIGHT1:
+    vld1.8      {q0}, [r8]!
+    vld1.8      {q1}, [r8]!
+    vst1.8      {q0}, [r7], r2
+    vst1.8      {q1}, [r7], r2
+
+    add         r14, r14, #2
+    cmp         r14, r12
+    blt         LOOP_REMAIN_HEIGHT_WIDTH16_HEIGHT1
+
+    add         r6, r6, #16
+    cmp         r6, r10
+    blt         LOOP_REMAIN_HEIGHT_WIDTH16
+
+REMAIN_HEIGHT_REMAIN_WIDTH_START:
+    cmp         r10, r2
+    beq         REMAIN_HEIGHT_REMAIN_WIDTH_END
+    mul         r8, r11, r5             @ src_offset = (tiled_width * i) + (j << 3)
+    add         r8, r8, r6, lsl #3
+    add         r8, r1, r8              @ src_offset = y_src + src_offset
+
+    mul         r12, r2, r5             @ temp1 = width * i + j;
+    add         r12, r12, r6
+    add         r7, r0, r12             @ r7 = y_dst + temp1
+
+    stmfd       sp!, {r0-r1,r3}         @ backup registers
+    mov         r0, #0
+    sub         r1, r3, r9
+LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1:
+
+    sub         r14, r2, r6
+    mov         r4, #0
+LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1_WIDTHx:
+    ldrh        r3, [r8], #2
+    strh        r3, [r7], #2
+    add         r4, #2
+    cmp         r4, r14
+    blt         LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1_WIDTHx
+
+    sub         r8, r8, r14
+    sub         r7, r7, r14
+    add         r8, r8, #16
+    add         r7, r7, r2
+
+    add         r0, r0, #1
+    cmp         r0, r1
+    blt         LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1
+    ldmfd       sp!, {r0-r1,r3}            @ restore registers
+
+REMAIN_HEIGHT_REMAIN_WIDTH_END:
+
+REMAIN_HEIGHT_END:
+
+RESTORE_REG:
+    ldmfd       sp!, {r4-r12,r15}       @ restore registers
+
+    .fnend
diff --git a/exynos/multimedia/utils/csc/exynos5/csc_tiled_to_linear_y_neon.s b/exynos/multimedia/utils/csc/exynos5/csc_tiled_to_linear_y_neon.s
new file mode 100644
index 0000000..1aa3b13
--- /dev/null
+++ b/exynos/multimedia/utils/csc/exynos5/csc_tiled_to_linear_y_neon.s
@@ -0,0 +1,231 @@
+/*
+ *
+ * Copyright 2010 Samsung Electronics S.LSI Co. LTD
+ *
+ * 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.
+ */
+
+/*
+ * @file    csc_tiled_to_linear_y.s
+ * @brief   SEC_OMX specific define. It support MFC 6.x tiled.
+ * @author  ShinWon Lee (shinwon.lee@samsung.com)
+ * @version 1.0
+ * @history
+ *   2011.12.01 : Create
+ */
+
+/*
+ * Converts tiled data to linear for mfc 6.x
+ * 1. Y of NV12T to Y of YUV420P
+ * 2. Y of NV12T to Y of YUV420S
+ *
+ * @param dst
+ *   Y address of YUV420[out]
+ *
+ * @param src
+ *   Y address of NV12T[in]
+ *
+ * @param yuv420_width
+ *   real width of YUV420[in]. It should be even.
+ *
+ * @param yuv420_height
+ *   real height of YUV420[in] It should be even.
+ *
+ */
+    .arch armv7-a
+    .text
+    .global csc_tiled_to_linear_y_neon
+    .type   csc_tiled_to_linear_y_neon, %function
+csc_tiled_to_linear_y_neon:
+    .fnstart
+
+    .equ CACHE_LINE_SIZE, 64
+
+    @r0     y_dst
+    @r1     y_src
+    @r2     width
+    @r3     height
+    @r4     temp3
+    @r5     i
+    @r6     j
+    @r7     dst_offset
+    @r8     src_offset
+    @r9     aligned_height
+    @r10    aligned_width
+    @r11    tiled_width
+    @r12    temp1
+    @r14    temp2
+
+    stmfd       sp!, {r4-r12,r14}       @ backup registers
+    ldr         r4, [sp, #40]           @ r4 = height
+
+    bic         r9, r3, #0xF            @ aligned_height = height & (~0xF)
+    bic         r10, r2, #0xF           @ aligned_width = width & (~0xF)
+    add         r11, r2, #15            @ tiled_width = ((width + 15) >> 4) << 4
+    mov         r11, r11, asr #4
+    mov         r11, r11, lsl #4
+
+    mov         r5, #0
+LOOP_MAIN_ALIGNED_HEIGHT:
+    mul         r8, r11, r5             @ src_offset = tiled_width * i
+    mov         r6, #0
+    add         r8, r1, r8              @ src_offset = y_src + src_offset
+LOOP_MAIN_ALIGNED_WIDTH:
+    pld         [r8, #CACHE_LINE_SIZE]
+    vld1.8      {q0, q1}, [r8]!
+    vld1.8      {q2, q3}, [r8]!
+    pld         [r8, #CACHE_LINE_SIZE]
+    vld1.8      {q4, q5}, [r8]!
+    vld1.8      {q6, q7}, [r8]!
+    mul         r12, r2, r5             @ temp1 = width * i + j;
+    pld         [r8, #CACHE_LINE_SIZE]
+    vld1.8      {q8, q9}, [r8]!
+    add         r12, r12, r6
+    vld1.8      {q10, q11}, [r8]!
+    add         r7, r0, r12             @ dst_offset = y_dst + temp1
+    pld         [r8, #CACHE_LINE_SIZE]
+    vld1.8      {q12, q13}, [r8]!
+    vld1.8      {q14, q15}, [r8]!
+
+    vst1.8      {q0}, [r7], r2
+    vst1.8      {q1}, [r7], r2
+    vst1.8      {q2}, [r7], r2
+    vst1.8      {q3}, [r7], r2
+    vst1.8      {q4}, [r7], r2
+    vst1.8      {q5}, [r7], r2
+    vst1.8      {q6}, [r7], r2
+    vst1.8      {q7}, [r7], r2
+    vst1.8      {q8}, [r7], r2
+    vst1.8      {q9}, [r7], r2
+    vst1.8      {q10}, [r7], r2
+    vst1.8      {q11}, [r7], r2
+    vst1.8      {q12}, [r7], r2
+    vst1.8      {q13}, [r7], r2
+    add         r6, r6, #16
+    vst1.8      {q14}, [r7], r2
+    cmp         r6, r10
+    vst1.8      {q15}, [r7], r2
+    blt         LOOP_MAIN_ALIGNED_WIDTH
+
+MAIN_REMAIN_WIDTH_START:
+    cmp         r10, r2                 @ if (aligned_width != width) {
+    beq         MAIN_REMAIN_WIDTH_END
+
+    mul         r8, r11, r5             @ src_offset = (tiled_width * i) + (j << 4);
+    add         r8, r8, r6, lsl #4
+    add         r8, r1, r8              @ r8 = y_src + src_offset
+
+    mul         r12, r2, r5             @ temp1 = width * i + j;
+    add         r12, r12, r6
+    add         r7, r0, r12             @ r7 = y_dst + temp1
+    sub         r14, r2, r6             @ r14 = width - j
+
+    stmfd       sp!, {r0-r1}            @ backup registers
+    mov         r1, #0
+LOOP_MAIN_REMAIN_HEIGHT:
+    mov         r0, #0                  @ r0 is index in memcpy
+LOOP_MAIN_REMAIN_WIDTH:
+    ldrh        r4, [r8], #2
+    strh        r4, [r7], #2
+    add         r0, #2
+    cmp         r0, r14
+    blt         LOOP_MAIN_REMAIN_WIDTH
+
+    sub         r8, r8, r14
+    sub         r7, r7, r14
+    add         r8, r8, #16
+    add         r7, r7, r2
+
+    add         r1, #1
+    cmp         r1, #16
+    blt         LOOP_MAIN_REMAIN_HEIGHT
+    ldmfd       sp!, {r0-r1}            @ restore registers
+MAIN_REMAIN_WIDTH_END:
+
+    add         r5, r5, #16
+    cmp         r5, r9
+    blt         LOOP_MAIN_ALIGNED_HEIGHT
+
+REMAIN_HEIGHT_START:
+    cmp         r9, r3                  @ if (aligned_height != height) {
+    beq         REMAIN_HEIGHT_END
+
+    mov         r6, #0
+LOOP_REMAIN_HEIGHT_WIDTH16:
+    mul         r8, r11, r5             @ src_offset = (tiled_width * i) + (j << 4)
+    add         r8, r8, r6, lsl #4
+    add         r8, r1, r8              @ src_offset = y_src + src_offset
+
+    mul         r12, r2, r5             @ temp1 = width * i + j;
+    add         r12, r12, r6
+    add         r7, r0, r12             @ r7 = y_dst + temp1
+
+    sub         r12, r3, r9
+    mov         r14, #0
+LOOP_REMAIN_HEIGHT_WIDTH16_HEIGHT1:
+    vld1.8      {q0}, [r8]!
+    vld1.8      {q1}, [r8]!
+    vst1.8      {q0}, [r7], r2
+    vst1.8      {q1}, [r7], r2
+
+    add         r14, r14, #2
+    cmp         r14, r12
+    blt         LOOP_REMAIN_HEIGHT_WIDTH16_HEIGHT1
+
+    add         r6, r6, #16
+    cmp         r6, r10
+    blt         LOOP_REMAIN_HEIGHT_WIDTH16
+
+REMAIN_HEIGHT_REMAIN_WIDTH_START:
+    cmp         r10, r2
+    beq         REMAIN_HEIGHT_REMAIN_WIDTH_END
+    mul         r8, r11, r5             @ src_offset = (tiled_width * i) + (j << 4)
+    add         r8, r8, r6, lsl #4
+    add         r8, r1, r8              @ src_offset = y_src + src_offset
+
+    mul         r12, r2, r5             @ temp1 = width * i + j;
+    add         r12, r12, r6
+    add         r7, r0, r12             @ r7 = y_dst + temp1
+
+    stmfd       sp!, {r0-r1,r3}         @ backup registers
+    mov         r0, #0
+    sub         r1, r3, r9
+LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1:
+
+    sub         r14, r2, r6
+    mov         r4, #0
+LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1_WIDTHx:
+    ldrh        r3, [r8], #2
+    strh        r3, [r7], #2
+    add         r4, #2
+    cmp         r4, r14
+    blt         LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1_WIDTHx
+
+    sub         r8, r8, r14
+    sub         r7, r7, r14
+    add         r8, r8, #16
+    add         r7, r7, r2
+
+    add         r0, r0, #1
+    cmp         r0, r1
+    blt         LOOP_REMAIN_HEIGHT_REMAIN_WIDTH_HEIGHT1
+    ldmfd       sp!, {r0-r1,r3}            @ restore registers
+
+REMAIN_HEIGHT_REMAIN_WIDTH_END:
+
+REMAIN_HEIGHT_END:
+
+RESTORE_REG:
+    ldmfd       sp!, {r4-r12,r15}       @ restore registers
+
+    .fnend