TMS320C6748的EMIFA应用，请问如何实现将一些“变量数据”放在存储器1中，一些“变量数据”放在存储器2中

你好，

可以用#pragma DATA_SECTION ( symbol , " section name ");定义变量到指定的memory地址， 具体见下面的编译器文档第161页说明。 http://www.ti.com/lit/ug/spru187u/spru187u.pdf
 

谢谢！我想问一下，“#pragma DATA_SECTION ( symbol , " section name ")”是在cmd中定义吗？如果是的话，程序中还需要定义地址吗？如果不是，那么需要修改CMD文件吗？

“#pragma DATA_SECTION ( symbol , " section name ")”在程序中定义，section name在cmd中定分配到你要放的memory地址。

有与之相关的案列程序吗？

“section name在cmd中定分配到你要放的memory地址。“这句话，怎么实现呢？谢谢！

例如：
test.c

#pragma DATA_SECTION(x, "x_data")
int x;

main()
[....]
test.cmd








x_data > mysection
｝
 

您好，程序有警告，仿真器在线观测变量，访问寄存器地址不对。警告是“#837-D pragma can only be applied to a file level symbol, not 'x'”

你是不是在函数里用#pragma定义变量？这种用法是不支持的。

没有，我严格按照你的程序进行的...

把你的代码贴出来看看

完全按照你的代码是没有错误的，现在我有了一个新问题，怎样修改程序实现变量指定存储器的指定存储地址呢？

我用您的程序定义的变量是短整型，用仿真器观测看到的是变量“X1”，地址为“ 0x60000000 ”；变量“X2”，地址为“ 0x60000002 ”；变量“X3”，地址为“ 0x60000004 ”；为什么地址不是连续的呢？？谢谢！

可以用#pragma CODE_SECTION (symbol , "section name ")，在上面SPRU187文档里有说明， 用法同#pragma DATA_SECTION一样。

连续的啊， 短整型是16bit的

地址不连续呀！为什么呢？谢谢！

.C文件
#include "stdio.h"
int a;
#pragma DATA_SECTION(x1, "x_data")
#pragma DATA_SECTION(x2, "x_data")
#pragma DATA_SECTION(x3, "x_data")
#pragma DATA_SECTION(x4, "x_data")
#pragma DATA_SECTION(x5, "x_data")
#pragma DATA_SECTION(x6, "x_data")
#pragma DATA_SECTION(x7, "x_data")
#pragma DATA_SECTION(x8, "x_data")
#pragma DATA_SECTION(x9, "x_data")
#pragma DATA_SECTION(x10, "x_data")
int short unsigned x1;
int short unsigned x2;
int short unsigned x3;
int short unsigned x4;
int short unsigned x5;
int short unsigned x6;
int short unsigned x7;
int short unsigned x8;
int short unsigned x9;
int short unsigned x10;
int main(void)
[
x1=1;
x2=2;
x3=3;
x4=4;
x5=5;
x6=6;
x7=7;
x8=8;
x9=9;
x10=10;
a=0;

while(1)
[
a=a+1;
if (a==100000)
[
a=0;
]
]
]
.cmd文件
/****************************************************************************/
/* C6748.cmd */
/* Copyright (c) 2010 Texas Instruments Incorporated */
/* Author: Rafael de Souza */
/* */
/* Description: This file is a sample linker command file that can be */
/* used for linking programs built with the C compiler and */
/* running the resulting .out file on a C6748 */
/* device. Use it as a guideline. You will want to */
/* change the memory layout to match your specific C6xxx */
/* target system. You may want to change the allocation */
/* scheme according to the size of your program. */
/* */
/****************************************************************************/
-l rts67plus.lib
/*-l ......bsllibevmomapl138_bsl.lib*/
-stack 0x00000800
-heap 0x00000800
MEMORY
[
vecs: ORIGIN = 0xFFF00000 LENGTH = 0x00080000
dsp_l1_p_ram: ORIGIN = 0x11E00000 LENGTH = 0x00008000 //32K
dsp_l1_d_ram: ORIGIN = 0x11F00000 LENGTH = 0x00008000 //32K
dsp_l2_ram: ORIGIN = 0x11800000 LENGTH = 0x00040000 //256K
shared_ram: ORIGIN = 0x80000000 LENGTH = 0x00020000
ddr_ram1: ORIGIN = 0xC0000000 LENGTH = 0x08000000 //256KByte
//AEMIF: ORIGIN=0x60000000 LENGTH =0x06000000 /* AEMIF*/
B0: ORIGIN=0x60000000 LENGTH =0x01000000 /* AEMIF CS2 region */
B1: ORIGIN=0x61000000 LENGTH =0x01000000 /* AEMIF CS2 region */

AEMIF_CS3: ORIGIN=0x62000000 LENGTH =0x02000000 /* AEMIF CS3 region */
AEMIF_CS4: ORIGIN=0x64000000 LENGTH =0x02000000 /* AEMIF CS4 region */
//ddr_ram2: ORIGIN = 0xC0040000 LENGTH = 0x00040000 //256KByte
//ddr_ram3: ORIGIN = 0xC0080000 LENGTH = 0x00040000 //256KByte
//ddr_ram4: ORIGIN = 0xC00C0000 LENGTH = 0x00040000 //256KByte
//ddr_ram5: ORIGIN = 0xC0100000 LENGTH = 0x00040000 //256KByte
//ddr_ram6: ORIGIN = 0xC0140000 LENGTH = 0x00040000 //256KByte
//ddr_ram7: ORIGIN = 0xC0180000 LENGTH = 0x00040000 //256KByte
//ddr_ram8: ORIGIN = 0xC01C0000 LENGTH = 0x00040000 //256KByte
//ddr_ram9: ORIGIN = 0xC0200000 LENGTH = 0x00040000 //256KByte
//ddr_ram2: ORIGIN = 0xC2000000 LENGTH = 0x02000000 //32MByte
//ddr_ram3: ORIGIN = 0xC4000000 LENGTH = 0x02000000 //32MByte
//ddr_ram4: ORIGIN = 0xC6000000 LENGTH = 0x02000000 //32MByte
]
SECTIONS
[
.vectors > dsp_l2_ram
.text > dsp_l2_ram
.const > dsp_l2_ram
.bss > dsp_l2_ram
.far > dsp_l2_ram
.switch > dsp_l2_ram
.stack > dsp_l2_ram
.data > dsp_l2_ram
.cinit > dsp_l2_ram
.sysmem > dsp_l2_ram
.cio > dsp_l2_ram
x_data > B0
/*
.vectors > shared_ram
.text > shared_ram //shared_ram
.const > shared_ram
.bss > shared_ram
.far > shared_ram
.switch > shared_ram
.stack > shared_ram
.data > shared_ram
.cinit > shared_ram
.sysmem > shared_ram
.cio > shared_ram
*/

.smpbuf > ddr_ram1 //DDR2 ram space 1/
.kgbuf > ddr_ram1 //DDR2 ram space 4
.dcbuf > ddr_ram1 //DDR2 ram space 5
.resendbuf > ddr_ram1 //resend the packet buffer 3200 1s data
.progbuf > shared_ram
//.cache1 > dsp_l1_d_ram
//.cache2 > dsp_l2_ram
// .progbuf > dsp_l2_ram
// .vectors > vecs //vector
]

EMIFA的数据线16根，地址线24根，明显是一个地址对应2个字节的数据呀，怎么会2个地址对应两个字节的数据呢？？谢谢！

如果是两个地址对应2个字节，也就是一个地址对应一个字节的话，岂不是8位数据线就够了吗？

DSP的memory是byte addressable的， 也即字节编址的。

SPRU187文档里面没有说明呀，是不是页码不对呀！对于“#pragma CODE_SECTION (symbol , "section name ")”，说的是程序代码存储呀！