在嵌入式开发中，编译器生成的目标文件一般都是 .hex 文件。
为什么要转换，直接使用hex文件不行吗，可是我在开发过程中一直都是直接生成hex文件，然后进行下载，也没见出错？
在不清楚hex与bin文件的格式时，可能小伙伴会有这样的疑问。需要进行转换的原因是：hex文件中数据记录（record）并不是按照 “起始地址–>终止地址” 这样的顺序进行排列的，由于每行数据都包含起始地址和数据长度，所以hex文件中数据不需要按照地址顺序从低到高进行排列；而Bin文件中的数据则是严格按照地址顺序进行排列的。
首先需要了解hex文件的格式，可以参考官方资料Intel Hexadecimal Object File Format Specification，也可以看我的另一篇blog——HEX文件说明 。
最近做ECU的上位机下载工具，其中一步是将hex文件的内容转成按地址顺序（从低到高）排列的二进制数据（bin文件）。
于是我找了一个转换工具——hex2bin，源码地址。下面我就该工具的整个转换过程进行一些分析。
整个过程主要分为两步：
1. 遍历整个hex文件，找出最小地址和最大地址（也就是起始地址和结束地址），算出数据长度（数据长度=结束地址-起始地址），根据得到的数据长度，分配对应大小的内存（开辟一个数组）；
2. 再次遍历整个hex文件，计算每条数据记录中的起始地址与hex文件起始地址的偏移量，按照偏移量将该条数据记录中的数据部分写入第一步的数组中。（这样就实现了按照从低到高的地址顺序排列整个hex文件的数据）。
最后只需要将该数组写出到文件中即可。
首先使用FILE * fopen(const char * path, const char * mode);打开hex文件，然后是第一次遍历，找出起始地址和数据长度。
/* 第一次遍历hex文件，获取地址范围(Lowest_Address和Highest_Address) */
/* get highest and lowest addresses so that we can allocate the right size */
do
{
unsigned int i;
/* Read a line from input file. */
GetLine(Line,Filin);
Record_Nb++;
/* Remove carriage return/line feed(回车/换行) at the end of line. */
i = strlen(Line);
if (--i != 0)
{
if (Line[i] == '\n') Line[i] = '\0';
/* Scan the first two bytes and nb of bytes.
The two bytes are read in First_Word since its use depend on the
record type: if it's an extended address record or a data record.
*/
/* sscanf() - 从一个字符串中读进与指定格式匹配的数据, 成功则返回参数数目.
":%2x%4x%2x%s":格式说明 :冒号开头,2个十六进制数,4个十六进制数, 2个十六进制数,余下的当做字符串 */
result = sscanf (Line, ":%2x%4x%2x%s",&Nb_Bytes,&First_Word,&Type,Data_Str);
if (result != 4) fprintf(stderr,"Error in line %d of hex file\n", Record_Nb);
p = (char *) Data_Str; //p表示指向数据域(包括checksum)的指针
/* If we're reading the last record, ignore it. */
switch (Type)
{
/* Data record */
case 0:
if (Nb_Bytes == 0)
break;
Address = First_Word;
if (Seg_Lin_Select == SEGMENTED_ADDRESS)
{
Phys_Addr = (Segment << 4) + Address;
}
else
{
/* LINEAR_ADDRESS or NO_ADDRESS_TYPE_SELECTED
Upper_Address = 0 as specified in the Intel spec. until an extended address
record is read. */
Phys_Addr = ((Upper_Address << 16) + Address);
}
if (Verbose_Flag) fprintf(stderr,"Physical Address: %08X\n",Phys_Addr);
/* 获取地址范围(Lowest_Address和Highest_Address) */
/* Set the lowest address as base pointer. */
if (Phys_Addr < Lowest_Address)
Lowest_Address = Phys_Addr;
/* Same for the top address. */
temp = Phys_Addr + Nb_Bytes -1;
if (temp > Highest_Address)
{
Highest_Address = temp;
if (Verbose_Flag) fprintf(stderr,"Highest_Address: %08X\n",Highest_Address);
}
break;
case 1:
if (Verbose_Flag) fprintf(stderr,"End of File record\n");
break;
case 2:
/* First_Word contains the offset. It's supposed to be 0000 so
we ignore it. */
/* First extended segment address record ? */
if (Seg_Lin_Select == NO_ADDRESS_TYPE_SELECTED)
Seg_Lin_Select = SEGMENTED_ADDRESS;
/* Then ignore subsequent extended linear address records */
if (Seg_Lin_Select == SEGMENTED_ADDRESS)
{
result = sscanf (p, "%4x%2x",&Segment,&temp2);
if (result != 2) fprintf(stderr,"Error in line %d of hex file\n", Record_Nb);
if (Verbose_Flag) fprintf(stderr,"Extended Segment Address record: %04X\n",Segment);
/* Update the current address. */
Phys_Addr = (Segment << 4);
}
else
{
fprintf(stderr,"Ignored extended linear address record %d\n", Record_Nb);
}
break;
case 3:
if (Verbose_Flag) fprintf(stderr,"Start Segment Address record: ignored\n");
break;
case 4:
/* First_Word contains the offset. It's supposed to be 0000 so
we ignore it. */
/* First extended linear address record ? */
if (Seg_Lin_Select == NO_ADDRESS_TYPE_SELECTED)
Seg_Lin_Select = LINEAR_ADDRESS;
/* Then ignore subsequent extended segment address records */
if (Seg_Lin_Select == LINEAR_ADDRESS)
{
result = sscanf (p, "%4x%2x",&Upper_Address,&temp2); //取出基地址(Extended Linear Address)和checksum
if (result != 2) fprintf(stderr,"Error in line %d of hex file\n", Record_Nb);
if (Verbose_Flag) fprintf(stderr,"Extended Linear Address record: %04X\n",Upper_Address);
/* Update the current address. */
Phys_Addr = (Upper_Address << 16);
if (Verbose_Flag) fprintf(stderr,"Physical Address: %08X\n",Phys_Addr);
}
else
{
fprintf(stderr,"Ignored extended segment address record %d\n", Record_Nb);
}
break;
case 5:
if (Verbose_Flag) fprintf(stderr,"Start Linear Address record: ignored\n");
break;
default:
if (Verbose_Flag) fprintf(stderr,"Unknown record type: %d at %d\n",Type,Record_Nb);
break;
}
}
}
while (!feof (Filin));
/*feof()用来侦测是否读取到了文件尾, 参数stream 为fopen()所返回的文件指针. 如果已读到文件尾则返回非零值, 其他情况返回0.*/每次读取一行，循环读取，直到文件尾。
申请指定长度的内存(malloc)，然后进行第二次遍历，这次的目的是将数据按地址顺序进行排列。
/* 第二次遍历hex文件, 处理数据 */
/* Read the file & process the lines. */
do /* repeat until EOF(Filin) */
{
unsigned int i;
/* Read a line from input file. */
GetLine(Line,Filin);
Record_Nb++;
/* Remove carriage return/line feed at the end of line. */
i = strlen(Line);
//fprintf(stderr,"Record: %d; length: %d\n", Record_Nb, i);
if (--i != 0)
{
if (Line[i] == '\n') Line[i] = '\0';
/* Scan the first two bytes and nb of bytes.
The two bytes are read in First_Word since its use depend on the
record type: if it's an extended address record or a data record.
*/
result = sscanf (Line, ":%2x%4x%2x%s",&Nb_Bytes,&First_Word,&Type,Data_Str);
if (result != 4) fprintf(stderr,"Error in line %d of hex file\n", Record_Nb);
Checksum = Nb_Bytes + (First_Word >> 8) + (First_Word & 0xFF) + Type; //前4个字节累加
p = (char *) Data_Str;
/* If we're reading the last record, ignore it. */
switch (Type)
{
/* Data record */
case 0:
if (Nb_Bytes == 0)
{
fprintf(stderr,"0 byte length Data record ignored\n");
break;
}
Address = First_Word;
if (Seg_Lin_Select == SEGMENTED_ADDRESS)
Phys_Addr = (Segment << 4) + Address;
else
/* LINEAR_ADDRESS or NO_ADDRESS_TYPE_SELECTED
Upper_Address = 0 as specified in the Intel spec. until an extended address
record is read. */
if (Address_Alignment_Word)
Phys_Addr = ((Upper_Address << 16) + (Address << 1)) + Offset;
else
Phys_Addr = ((Upper_Address << 16) + Address);
/* Check that the physical address stays in the buffer's range. */
if ((Phys_Addr >= Lowest_Address) && (Phys_Addr <= Highest_Address))
{
/* The memory block begins at Lowest_Address */
Phys_Addr -= Lowest_Address; /* 计算该条数据记录相对于hex文件起始地址的偏移量 */
p = ReadDataBytes(p); /* 根据偏移量将该条记录中的数据写入指定的数组中 */
/* Read the Checksum value. */
result = sscanf (p, "%2x",&temp2);
if (result != 1) fprintf(stderr,"Error in line %d of hex file\n", Record_Nb);
/* Verify Checksum value. */
/* 校验和 = 0x100 - 除checksum之外所有字节的累加和 */
Checksum = (Checksum + temp2) & 0xFF;
VerifyChecksumValue();
}
else
{
if (Seg_Lin_Select == SEGMENTED_ADDRESS)
fprintf(stderr,"Data record skipped at %4X:%4X\n",Segment,Address);
else
fprintf(stderr,"Data record skipped at %8X\n",Phys_Addr);
}
break;
/* End of file record */
case 1:
/* Simply ignore checksum errors in this line. */
break;
/* Extended segment address record */
case 2:
/* First_Word contains the offset. It's supposed to be 0000 so
we ignore it. */
/* First extended segment address record ? */
if (Seg_Lin_Select == NO_ADDRESS_TYPE_SELECTED)
Seg_Lin_Select = SEGMENTED_ADDRESS;
/* Then ignore subsequent extended linear address records */
if (Seg_Lin_Select == SEGMENTED_ADDRESS)
{
result = sscanf (p, "%4x%2x",&Segment,&temp2);
if (result != 2) fprintf(stderr,"Error in line %d of hex file\n", Record_Nb);
/* Update the current address. */
Phys_Addr = (Segment << 4);
/* Verify Checksum value. */
Checksum = (Checksum + (Segment >> 8) + (Segment & 0xFF) + temp2) & 0xFF;
VerifyChecksumValue();
}
break;
/* Start segment address record */
case 3:
/* Nothing to be done since it's for specifying the starting address for
execution of the binary code */
break;
/* Extended linear address record */
case 4:
/* First_Word contains the offset. It's supposed to be 0000 so
we ignore it. */
if (Address_Alignment_Word) /*默认为false*/
{
sscanf (p, "%4x",&Offset);
Offset = Offset << 16;
Offset -= Lowest_Address;
}
/* First extended linear address record ? */
if (Seg_Lin_Select == NO_ADDRESS_TYPE_SELECTED)
Seg_Lin_Select = LINEAR_ADDRESS;
/* Then ignore subsequent extended segment address records */
if (Seg_Lin_Select == LINEAR_ADDRESS)
{
result = sscanf (p, "%4x%2x",&Upper_Address,&temp2);
if (result != 2) fprintf(stderr,"Error in line %d of hex file\n", Record_Nb);
/* Update the current address. */
Phys_Addr = (Upper_Address << 16);
/* Verify Checksum value. */
Checksum = (Checksum + (Upper_Address >> 8) + (Upper_Address & 0xFF) + temp2)
& 0xFF;
VerifyChecksumValue();
}
break;
/* Start linear address record */
case 5:
/* Nothing to be done since it's for specifying the starting address for
execution of the binary code */
break;
default:
fprintf(stderr,"Unknown record type\n");
break;
}
}
}
while (!feof (Filin));最后将数组中的内容输出到文件，即可得到bin文件。
按照上面的思路我用Java写了一个转换类（用在上面提到的ECU下载工具中），感兴趣的同学可以看一看。
其中碰到的坑有必要提一下，由于Hex文件中的数据是采用ASC II码的，而Bin文件中的数据是直接使用的二进制（不存在编码），因此在转换过程中涉及到编码转换。另外还有一点需要提一下，由于Java没有无符号型，如果将读到的数据赋给byte型变量，在调试的过程中使用print输出时会看到乱码（数据溢出byte类型的范围）。
作者：ZF_C_CQUPT 发表于2016/9/29 0:35:52 原文链接
阅读：138 评论：0 查看评论