达芬奇工具链的建立（工程编译步骤）,特别是函数中参数传递的输入方法

 

达芬奇工具链的建立（工程编译步骤）

工具链是指的Linux下对程序的编译环境。在这里通过codec_engine_1_02\examples\codecs 下的videnc_copy工程封装CCS下调试好的程序并生成videnc_copy.a64P，再通过codec_engine_1_02\examples\servers\video_copy工程把videnc_copy.a64P封装成all.x64P。最后通过encodedecode工程调用all.x64P。步骤如下：

1、把ccs下编写好的文件复制到Linux codec_engine_1_02\examples\codecs 下的videnc_copy工程中。

2、修改该工程中的package.bld文件，修改var SRCS = ["videnc_copy","blob","cornerdetection","him","imc","morph","objectrecognition"]; 中括号里只选择工程下的C文件。修改该文件中的

for (var i = 0; i <1 /*Build.targets.length*/; i++) { 选择DSP端只用cgtool来编译。

3、修改codec_engine_1_02\examples下的user.bld,把linux86注掉，选择用C64P，跟第二步中对应。

<!--[if !supportLists]--><!--[endif]--><!--[endif]-->

Build.targets = [

//    Linux86,

C64P,

MVArm9

4、修改DSP算法占用的空间

修改server/all.tcf

    var mem_ext = [

{

    comment:    "DDRALGHEAP: off-chip memory for dynamic algmem allocation",

    name:       "DDRALGHEAP",

    base:       0x88000000,   // 128MB

    len:        0x07A00000,   // 122MB

    space:      "code/data"

},

{

    comment:    "DDR: off-chip memory for application code and data",

    name:       "DDR",

    base:       0x8FA00000,   // 250MB

    len:        0x00400000,   //   4MB

    space:      "code/data"

},

{

    comment:    "DSPLINK: off-chip memory reserved for DSPLINK code and data",

    name:       "DSPLINKMEM",

    base:       0x8FE00000,   // 254MB

    len:        0x00100000,   //   1MB

    space:      "code/data"

},

{

    comment:    "RESET_VECTOR: off-chip memory for the reset vector table",

    name:       "RESET_VECTOR",

    base:       0x8FF00000,

    len:        0x00000080,

    space:      "code/data"

},

];

修改为：

var mem_ext = [

{

    comment:    "DDRALGHEAP: off-chip memory for dynamic algmem allocation",

    name:       "DDRALGHEAP",

    base:       0x88000000,   // 128MB

    len:        0x07400000,   // 122MB

    space:      "code/data"

},

{

    comment:    "DDR: off-chip memory for application code and data",

    name:       "DDR",

    base:       0x8F400000,   // 250MB

    len:        0x00A00000,   //   4MB

    space:      "code/data"

},

{

    comment:    "DSPLINK: off-chip memory reserved for DSPLINK code and data",

    name:       "DSPLINKMEM",

    base:       0x8FE00000,   // 254MB

    len:        0x00100000,   //   1MB

    space:      "code/data"

},

{

    comment:    "RESET_VECTOR: off-chip memory for the reset vector table",

    name:       "RESET_VECTOR",

    base:       0x8FF00000,

    len:        0x00000080,

    space:      "code/data"

},

];5、修改loadmodules.sh：

insmod dsplinkk.ko ddr_start=0x8f400000 ddr_size=0xA00000

6、在编译codec_engine_1_02\examples\codecs\videnc_copy之前，要修改算法的主程序main函数

void ObjectRecognition(unsigned char* inbuf, int* flag,unsigned char* outbuf, int* kind, float *inIM)。其中输入输出参数是由

XDAS_Int32 VIDENCCOPY_TI_process(IVIDENC_Handle h, XDM_BufDesc *inBufs,XDM_BufDesc *outBufs, IVIDENC_InArgs *inArgs, IVIDENC_OutArgs *outArgs)决定。由于输入输出参数由IVIDENC_InArgs，IVIDENC_OutArgs结构体来决定，所以为了增加输出输入参数，应该修改这两个结构体。为此修改xdais_5_00\packages\ti\xdais\dm中的ividdec.h文件。添加kind和IM[25]。

typedef struct IVIDENC_OutArgs {

XDAS_Int32 size;                                          XDAS_Int32 extendedError;

XDAS_Int32 bytesGenerated;   

XDAS_Int32 encodedFrameType; 

XDAS_Int32 inputFrameSkip;

XDAS_Int32 kind;               /*what's kind of the target*/

 float  IM[25];                     /*return value*/

    IVIDEO_BufDesc reconBufs;   /**< Reconstruction Frames. */

} IVIDENC_OutArgs;

为了区分输入帧存的性质（从jpeg读取的数据还是ccd采集的数据）增加一个flag作为标志。

typedef struct IVIDENC_InArgs {

    XDAS_Int32 size;          

    XDAS_Int32 flag;//is picture saved on U or captured from CCD} IVIDENC_InArgs;

7、在videnc_copy中调用算法：在VIDENCCOPY_TI_process函数中添加：ObjectRecognition(inBufs->bufs[curBuf],&(inArgs->flag),outBufs->bufs[curBuf],&(outArgs->kind),(outArgs->IM));

由于使用了#ifdef USE_ACPY3 宏，故而要选择是否使用DMA，为了避免Cache和DMA的冲突，选择不使用DMA。修改codec_engine_1_02\examples\servers\video_copy下的video_copy.cfg文件关闭DMA：

/* we can use DMA in certain codecs! */

VIDENC_COPY.useDMA = false;

8、编译videnc_copy生成videnc_copy.a64P

9、修改server，编译生成all.x64P

10、下面就是编译encodedecode调用all.x64P

• 首先修改Encodedecode下的Makefile文件，让encodedecode包含server及codec工程。XDC_PATH = $(CODEC_INSTALL_DIR)/packages;$(CE_INSTALL_DIR)/examples;

  $(CE_INSTALL_DIR)/packages;$(LINK_INSTALL_DIR)/packages;

  $(XDAIS_INSTALL_DIR)/packages;$(CMEM_INSTALL_DIR)/packages

•  其次为了使用新的算法videnc_copy，要修改encodedecode.cfg文件把H264算法替换为videnc_copy算法：

  //var H264ENC =xdc.useModule('ti.sdo.codecs.h264enc.H264ENC');

  //var H264DEC = xdc.useModule('ti.sdo.codecs.h264dec.H264DEC');

  var VIDDEC_COPY = xdc.useModule('codecs.viddec_copy.VIDDEC_COPY');

  var VIDENC_COPY = xdc.useModule('codecs.videnc_copy.VIDENC_COPY');

  /* *  ======== Engine Configuration ======== */

  var Engine = xdc.useModule('ti.sdo.ce.Engine');

  /*var vcr = Engine.create("loopback", [

      {name: "h264enc", mod: H264ENC, local: false},

      {name: "h264dec", mod: H264DEC, local: false},

  ]);*/

  var demoEngine = Engine.create("encodedecode", [

      {name: "videnc_copy", mod: VIDENC_COPY, local: false}, 

      {name: "viddec_copy", mod: VIDDEC_COPY, local: false}

  ]);

  //vcr.server = "./loopbackCombo.x64P";

  demoEngine.server = "./all.x64P";<!--[if !supportLists]-->

• 修改encodedecode工程中创建算法函数

  static String decoderName  = "viddec_copy";

  static String encoderName  = "videnc_copy";

  static String engineName   = "video_copy";

  hDecode = VIDDEC_create(hEngine, decoderName, &params);

  hEncode = VIDENC_create(hEngine, encoderName, &params);

  在工程里用到的是hEncode。

• 修改status = VIDENC_process(hEncode, &inBufDesc,&outBufDesc,&inArgs, &outArgs);<!--[endif]-->的输入输出参数，使之符合封装算法是的函数参数。