U-BOOT全线移植分析系列之四
――U-boot如何引导Linux内核启动?
Sailor_forever [email protected] 转载请注明
http://blog.csdn.net/sailor_8318/archive/2008/08/05/2773412.aspx
【摘要】本节介绍了U-boot使用go或bootm启动linux内核的方法。首先介绍了mkimage的参数意义和bootm的详细执行流程。然后分析了如何利用mkimage生成内核映象的方法。对于bootm方式的内核是否压缩、-a、-e、运行地址等16种组合情况,给出了详细的测试过程,提出了6种可用方法种的三种最优解。
【关键字】:U-boot;AT91RM9200;bootm;mkimage;-a;-e;-c
四 U-boot如何引导Linux内核启动?
4.1 GO命令引导未用mkimage生成的内核
4.1.1 非压缩内核Image
1) 运行地址!=链接地址0x20008000,不能启动
Uboot> tftp 21000000 Image;tftp 21100000 ramdisk;go 21000000
。。。。
done
Bytes transferred = 6993691 (6ab71b hex)
## Starting application at 0x21000000 ...
Error: a 在哪提示的?
2) 运行地址=链接地址0x20008000,不能启动,难道是ramdisk的问题
Uboot> tftp 20008000 Image;tftp 21100000 ramdisk;go 20008000
。。。。
done
Bytes transferred = 6993691 (6ab71b hex)
## Starting application at 0x21000000 ...
Error: a
4.1.2 压缩内核zImage
1) 运行地址!=链接地址0x20008000,能启动,内核自解压成功,但是解压后的内核运行错误
Uboot> tftp 21000000 zImage;tftp 21100000 ramdisk;go 21000000
。。。。。。。。。。。
done
Bytes transferred = 6993691 (6ab71b hex)
## Starting application at 0x21000000 ...
Uncompressing Linux............................................................. done, booting the kernel.
€?~??鄜屈
2) 运行地址==链接地址0x20008000,能启动,内核自解压成功,但是解压后的内核运行错误
Uboot> tftp 20008000 zImage;tftp 21100000 ramdisk; go 20008000
## Starting application at 0x20008000 ...
Uncompressing Linux............................................................. done, booting the kernel.
€?~??鄜屈
上面的ramdisk都是添加了uboot的头的,去掉头部再试试。去掉了还是不行,go方法的ramdisk的地址是怎么设置的??要详细看下uboot在ramdisk这块是如何跟内核交互的?
4.2 Mkimage参数意义解析
通过mkimage这个tool可以给zImage添加一个header:
typedef struct image_header {
uint32_t ih_magic; /* Image Header Magic Number */
uint32_t ih_hcrc; /* Image Header CRC Checksum */
uint32_t ih_time; /* Image Creation Timestamp */
uint32_t ih_size; /* Image Data Size */
uint32_t ih_load; /* Data Load Address */
uint32_t ih_ep; /* Entry Point Address */
uint32_t ih_dcrc; /* Image Data CRC Checksum */
uint8_t ih_os; /* Operating System */
uint8_t ih_arch; /* CPU architecture */
uint8_t ih_type; /* Image Type */
uint8_t ih_comp; /* Compression Type */
uint8_t ih_name[IH_NMLEN]; /* Image Name */
} image_header_t;
此header是如何生成的?利用u-boot里面的mkimage工具来生成uImage (u-boot源码包/tools/mkimage.c )
这里解释一下参数的意义:
-A ==> set architecture to 'arch'
-O ==> set operating system to 'os'
-T ==> set image type to 'type' “kernel或是ramdisk”
-C ==> set compression type 'comp'
-a ==> set load address to 'addr' (hex)
-e ==> set entry point to 'ep' (hex)(内核启动时在此位置查询完整的内核印象)
-n ==> set image name to 'name'
-d ==> use image data from 'datafile'
-x ==> set XIP (execute in place,即不进行文件的拷贝,在当前位置执行)
对于ARM linux内核映象用法:
-A arm -------- 架构是arm
-O linux -------- 操作系统是linux
-T kernel -------- 类型是kernel
-C none/bzip/gzip -------- 压缩类型
-a 20008000 ---- image的载入地址(hex),通常为0xX00008000
-e 200080XX---- 内核的入口地址(hex),XX为0x40或者0x00
-n linux-XXX --- image的名字,任意
-d nameXXX ---- 无头信息的image文件名,你的源内核文件
uImageXXX ---- 加了头信息之后的image文件名,任意取
4.3 Bootm的流程分析
Bootm命令在/common/cmd_bootm.c中do_bootm函数
》》》》》》》》》》》获取当前内核的地址,默认地址或者bootm的第一个参数
默认的加载地址或传递给bootm命令(优先)与实际的内核存放地址需要一致
if (argc < 2) {
addr = load_addr; // load_addr = CFG_LOAD_ADDR;
} else {
addr = simple_strtoul(argv[1], NULL, 16);
}
printf ("## Booting image at %08lx .../n", addr);
》》》》》》》》》》》》获得image头,没有mkimage的就返回了
memmove (&header, (char *)addr, sizeof(image_header_t));
》》》》》》》》》》》》打印头部信息
print_image_hdr ((image_header_t *)addr);
实例:
Image Name: dd-kernel-2.4.19
Image Type: ARM Linux Kernel Image (gzip compressed)
Data Size: 869574 Bytes = 849.2 kB
Load Address: 20008000
Entry Point: 20008000
》》》》》》》》》》》》校验image头部
printf (" Verifying Checksum ... "); printf ("OK/n");
》》》》》》》》》》》》检查image支持的体系结构即—A 选项是否为arm或者ppc等
》》》》》》》》》》》》检查image的类型
TYPE_MULTI 是否指内核与文件系统一起,内核后面有个分界线
switch (hdr->ih_type)
case IH_TYPE_KERNEL:
name = "Kernel Image";
break;
case IH_TYPE_MULTI:
》》》》》》》》》》判断内核的压缩类型
此处的内核是否压缩非zImage和Image的概念,而是指内核在被mkimage处理前是否用gunzip等压缩过
switch (hdr->ih_comp) {
case IH_COMP_NONE: // 非压缩内核
if(ntohl(hdr->ih_load) == addr) {
// 当前内核存放的地址与-a指定的一致,则不搬动,-e必须必-a大0x40
printf (" XIP %s ... ", name);
} else {
//当前内核存放的地址与-a指定的不一致,则将内核搬到-a地址,此时-a与-e必相同
memmove ((void *) ntohl(hdr->ih_load), (uchar *)data, len);
。。。。
case IH_COMP_GZIP:
printf (" Uncompressing %s ... ", name);
if (gunzip ((void *)ntohl(hdr->ih_load), unc_len,
//压缩内核,将除去头部的内核解压到-a 指定的地址了,要求-a与-e相同
// 为防止解压缩时覆盖,对于压缩内核,内核存放地址最好在—a后面
(uchar *)data, (int *)&len) != 0) {
do_reset (cmdtp, flag, argc, argv);
}
break;
》》》》》》》》》》》》》》》》判断操作系统类型
switch (hdr->ih_os) {
default: /* handled by (original) Linux case */
case IH_OS_LINUX:
do_bootm_linux (cmdtp, flag, argc, argv, addr, len_ptr, verify);
//前四个为传给bootm的,addr为内核最初的存放地址,没有用处
break;
#ifdef CONFIG_PPC
static boot_os_Fcn do_bootm_linux;
#else
extern boot_os_Fcn do_bootm_linux;
由上可知,对于ppc和其他体系结构的do_bootm_linux函数实现是不一样的
》》》》》》》》》》》》》》启动Linux内核
do_bootm_linux (cmd_tbl_t *cmdtp, int flag,
int argc, char *argv[],
ulong addr,
ulong *len_ptr,
int verify)
》》》》》》》》》》》》获取命令行参数
if ((s = getenv("bootargs")) == NULL)
s = "";
strcpy (cmdline, s);
》》》》》》》》》》》》赋内核启动地址
kernel = (void (*)(bd_t *, ulong, ulong, ulong, ulong))hdr->ih_ep;
注意,对于压缩过的内核,会将内核解压到-a指定的地址,此时-a 与-e 地址必须相同
》》》》》》》》》》》判断bootm的命令参数中是否有initrd
if (argc >= 3) {
addr = simple_strtoul(argv[2], NULL, 16);
printf ("## Loading RAMDisk Image at %08lx .../n", addr);
若有initrd则赋值,否则为0
》》》》》》》》》》》》》》》启动Linux内核
/*
* Linux Kernel Parameters:
* r3: ptr to board info data
* r4: initrd_start or 0 if no initrd
* r5: initrd_end - unused if r4 is 0
* r6: Start of command line string
* r7: End of command line string
*/
//*kbd = *(gd->bd); 在上面赋值的
(*kernel) (kbd, initrd _start, initrd_end, cmd_start, cmd_end);
启动流程的总结:
对于非gzip压缩的内核,bootm命令会首先判断bootm xxxx 这个指定的地址xxxx是否与-a指定的加载地址相同。
(1)如果不同的话会从这个地址开始提取出这个64byte的头部,对其进行分析,然后把去掉头部的内核复制到-a指定的load地址中去运行之(此时-e选型必须同-a)
(2)如果相同的话那就让其原封不动的放在那,但-e指定的入口地址会推后64byte,以跳过这64byte的头部。
对于gzip压缩过的内核,因为u-boot要对其解压,因此运行地址是不能等于-a指定的地址的,且必须有一定的间隔,否则解压到-a的内核会覆盖当前运行的程序。此时要求-a等于-e指定的地址。
4.4 如何用mkimage生成uImage
1> mkimage 如何指定入口参数 ( -e 0xxxxxx)
2> mkimage 指定了入口参数后, 你用tftpboot 下载kernel到哪个地址?
3> -c 如何指定?
u-boot里面的解压和内核自解压的区别: u-boot 里面的解压实际上是bootm 实现的 , 把 mkimage -C bzip2或者gzip 生成的 uImage进行解压 ; 而kernel的自解压是对zImage进行解压,发生在bootm解压之后。
U-boot 对内核添加头部时,前面已经用gzip压缩过一次内核了,而不是指原有的内核印象是否是压缩内核。指uImage 本身被压缩了,即对原来的zImage/Image添加了U-boot的压缩方式,使得生成的uImage变小了。此时-c gzip
若没有对zImage/Image用gzip命令压缩过,则-c none。
综合上面分析,mkimage的影响因子为:
-e,内核的入口地址是否与-a相同
Tftpaddr,即将内核加载到RAM中运行的地址,决定是否搬运或解压内核
-c,内核是否经过gzip压缩过,决定了是搬运还是解压
另外内核本身为非压缩的Image或zImage也是一个影响因子。组合情况共2^4 =16种
4.5 Bootm命令引导mkimage生成的内核全程解析
4.5.1 非压缩的Image内核
(1)Mkimage 之前用gzip对Image进行压缩
<1> -a=-e = 0x20008000,tftpaddr= 0x21000000
解压到-a指定的地址,成功启动
Uboot> tftp 21000000 uImage-zip-8000;tftp 21100000 ramdisk;bootm 21000000
## Booting image at 21000000 ...
Image Name: dd-kernel-2.4.19-zip-8000
Image Type: ARM Linux Kernel Image (gzip compressed)
Data Size: 869629 Bytes = 849.2 kB
Load Address: 20008000
Entry Point: 20008000
Verifying Checksum ... OK
Uncompressing Kernel Image ... OK
Starting kernel ...
Linux version 2.4.19-rmk7 (root@dding) (gcc version 2.95.3 20010315 (release)) #42 四 10月 11 14:15:35 CST 2007
AT91RM9200DK login: root
<2> -a=-e = 0x20008000, tftpaddr= 0x20008000
解压失败,启动失败
Uboot> tftp 20008000 uImage-zip-zImage-8000;tftp 21100000 ramdisk;bootm 20008000
Uboot> tftp 20008000 uImage-zip-8000;tftp 21100000 ramdisk;bootm 20008000
## Booting image at 20008000 ...
Image Name: dd-kernel-2.4.19-zip-8000
Image Type: ARM Linux Kernel Image (gzip compressed)
Data Size: 869629 Bytes = 849.2 kB
Load Address: 20008000
Entry Point: 20008000
Verifying Checksum ... OK
Uncompressing Kernel Image ... Error: inflate() returned -3
GUNZIP ERROR - must RESET board to recover
由于当前运行地址tftpaddr与解压缩后的地址-a重合了,导致解压缩失败,因此二者必须相隔一定的距离
<3> -a=0x20008000,-e = 0x20008040 ,tftpaddr= 0x21000000
能够解压到-a地址,但-e指定的入口不对,启动失败
Uboot> tftp 21000000 uImage-zip-8040;tftp 21100000 ramdisk;bootm 21000000
TFTP from server 192.168.0.12; our IP address is 192.168.0.15
Filename 'uImage-zip-8040'.
Load address: 0x21000000
。。。。。。。。。
## Booting image at 21000000 ...
Image Name: dd-kernel-2.4.19-zip-8040
Image Type: ARM Linux Kernel Image (gzip compressed)
Data Size: 869629 Bytes = 849.2 kB
Load Address: 20008000
Entry Point: 20008040
Verifying Checksum ... OK
Uncompressing Kernel Image ... OK
Starting kernel ... 死了
<4> -a=-e = 0x20008000, tftpaddr= 0x20008000
解压失败,入口也不对,启动失败
(2)Mkimage 之前未对Image进行压缩
<1> -a=-e = 0x20008000 tftpaddr= 0x21000000
搬动到-a指定的地址,成功启动
Uboot> tftp 21000000 uImage-nzip-8000;tftp 21100000 ramdisk;bootm 21000000
## Booting image at 21000000 ...
Image Name: dd-kernel-2.4.19-nzip-8000
Image Type: ARM Linux Kernel Image (uncompressed)
Data Size: 1873059 Bytes = 1.8 MB
Load Address: 20008000
Entry Point: 20008000
Verifying Checksum ... Bad Data CRC
为什么总是校验失败呢?当前的内核印象为1.8M,难道太大了,后面的ramdisk将其覆盖??
下面未拷贝ramdisk,校验成功,成功启动,无法安装跟文件系统,是因为无ramdisk。说明上面确实是覆盖了,因此要对于大的内核印象要合理设置tftpaddr的地址和ramdisk的地址
Addr(ramdisk)= 0x2110 0000
Addr(tftpaddr)= 0x2100 0000
Addr(ramdisk)-Addr(tftpaddr)= 0x10 0000 = 1M < 1.8M
Uboot> tftp 21000000 uImage-nzip-8000;bootm 21000000
## Booting image at 21000000 ...
Image Name: dd-kernel-2.4.19-nzip-8000
Image Type: ARM Linux Kernel Image (uncompressed)
Data Size: 1873059 Bytes = 1.8 MB
Load Address: 20008000
Entry Point: 20008000
Verifying Checksum ... OK
OK
Starting kernel ...
Linux version 2.4.19-rmk7 (root@dding) (gcc version 2.95.3 20010315 (release)) #44 四 10月 11 17:27:24 CST 2007
。。。。。。。
Kernel panic: VFS: Unable to mount root fs on 01:00
Addr(ramdisk)- 2M = 0x20f0 0000 = Addr(tftpaddr)成功启动
Uboot> tftp 20f00000 uImage-nzip-8000;tftp 21100000 ramdisk;bootm 20f00000
## Booting image at 20f00000 ...
Image Name: dd-kernel-2.4.19-nzip-8000
Image Type: ARM Linux Kernel Image (uncompressed)
Data Size: 1873059 Bytes = 1.8 MB
Load Address: 20008000
Entry Point: 20008000
Verifying Checksum ... OK
OK
Starting kernel ...
Linux version 2.4.19-rmk7 (root@dding) (gcc version 2.95.3 20010315 (release)) #44 四 10月 11 17:27:24 CST 2007
AT91RM9200DK login: root
<2> -a=-e = 0x20008000, tftpaddr= 0x20008000
不搬动,但-e地址不对,失败
<3> -a=0x20008000,-e = 0x20008040 ,tftpaddr= 0x20008000
不搬动,但未成功启动,入口地址对的啊?????
Uboot> tftp 20008000 uImage-nzip-8040;tftp 21100000 ramdisk;bootm 20008000
## Booting image at 20008000 ...
Image Name: dd-kernel-2.4.19-nzip-8040
Image Type: ARM Linux Kernel Image (uncompressed)
Data Size: 1873059 Bytes = 1.8 MB
Load Address: 20008000
Entry Point: 20008040
Verifying Checksum ... OK
XIP Kernel Image ... OK
Starting kernel ...死了????
<4> -a=0x20008000,-e = 0x20008040 ,tftpaddr= 0x21000000
搬动,但-e地址不对,失败
4.5.2 压缩的zImage内核
(1)Mkimage 之前用gzip对zImage进行压缩,即-c gzip
<1> -a=-e = 0x20008000,tftpaddr= 0x21000000
解压到-a指定的地址,成功启动
Uboot> tftp 21000000 uImage-zip-zImage-8000;tftp 21100000 ramdisk;bootm 21000000
TFTP from server 192.168.0.12; our IP address is 192.168.0.15
Filename 'uImage-zip-zImage-8000'.
Load address: 0x21000000
## Booting image at 21000000 ...
Image Name: dd-zip-zImage-8000
Image Type: ARM Linux Kernel Image (gzip compressed)
Data Size: 876753 Bytes = 856.2 kB
Load Address: 20008000
Entry Point: 20008000
Verifying Checksum ... OK
Uncompressing Kernel Image ... OK // U-boot对内核解压
Starting kernel ...
Uncompressing Linux..............压缩内核zImage自解压......................... done, booting the kernel.
Linux version 2.4.19-rmk7 (root@dding) (gcc version 2.95.3 20010315 (release)) #43 四 10月
AT91RM9200DK login: root
[root@AT91RM9200DK /root]$ls
<2> -a=-e = 0x20008000, tftpaddr= 0x20008000
解压失败,启动失败
Uboot> tftp 20008000 uImage-zip-zImage-8000;tftp 21100000 ramdisk;bootm 20008000
TFTP from server 192.168.0.12; our IP address is 192.168.0.15
Filename 'uImage-zip-zImage-8000'.
Load address: 0x20008000
## Booting image at 20008000 ...
Image Name: dd-zip-zImage-8000
Image Type: ARM Linux Kernel Image (gzip compressed)
Data Size: 876753 Bytes = 856.2 kB
Load Address: 20008000
Entry Point: 20008000
Verifying Checksum ... OK
Uncompressing Kernel Image ... Error: inflate() returned -3
GUNZIP ERROR - must RESET board to recover
由于当前运行地址tftpaddr与解压缩后的地址-a重合了,导致解压缩失败,因此二者必须相隔一定的距离
<3> -a=0x20008000,-e = 0x20008040 ,tftpaddr= 0x21000000,失败
Uboot> tftp 21000000 uImage-zip-zImage-8040;tftp 21000000 ramdisk;bootm 21000000
TFTP from server 192.168.0.12; our IP address is 192.168.0.15
Filename 'uImage-zip-zImage-8040'.
Load address: 0x21000000
。。。。。。。。。
## Booting image at 21000000 ...
Bad Magic Number 难道对于压缩内核,幻数对的条件是-a==-e地址??即压缩内核默认-a==-e??
此法肯定失败,但问题出在这,还真不对啊,不试了,感兴趣的朋友可以玩下
(2)Mkimage 之前未对zImage进行压缩,即-c none
<1> -a=-e = 0x20008000 tftpaddr= 0x21000000
搬动到-a指定的地址,成功启动
Uboot> tftp 21000000 uImage-nzip-zImage-8000;tftp 21100000 ramdisk;bootm 21000000
## Booting image at 21000000 ...
Image Name: dd-nzip-zImage-8000
Image Type: ARM Linux Kernel Image (uncompressed)
Data Size: 881748 Bytes = 861.1 kB
Load Address: 20008000
Entry Point: 20008000
Verifying Checksum ... OK
OK
Starting kernel ...
Uncompressing Linux............................................................. done, booting the kernel.
Linux version 2.4.19-rmk7 (root@dding) (gcc version 2.95.3 20010315 (release)) #43 四 10月 11 14:25:14 CST 2007
AT91RM9200DK login:
<2> -a=-e = 0x20008000, tftpaddr= 0x20008000
不搬动,但-e地址不对,失败
Uboot> tftp 20008000 uImage-nzip-zImage-8000;tftp 21100000 ramdisk;bootm 20008000
## Booting image at 20008000 ...
Image Name: dd-nzip-zImage-8000
Image Type: ARM Linux Kernel Image (uncompressed)
Data Size: 881748 Bytes = 861.1 kB
Load Address: 20008000
Entry Point: 20008000
Verifying Checksum ... OK
XIP Kernel Image ... OK
Starting kernel ... 死了。。。
<3> -a=0x20008000,-e = 0x20008040 ,tftpaddr= 0x20008000
不搬动,成功启动
Uboot> tftp 20008000 uImage-nzip-zImage-8040;tftp 21100000 ramdisk;bootm 20008000
## Booting image at 20008000 ...
Image Name: dd-nzip-zImage-8040
Image Type: ARM Linux Kernel Image (uncompressed)
Data Size: 881748 Bytes = 861.1 kB
Load Address: 20008000
Entry Point: 20008040
Verifying Checksum ... OK
XIP Kernel Image ... OK
Starting kernel ...
Uncompressing Linux............................................................. done, booting the kernel.
Linux version 2.4.19-rmk7 (root@dding) (gcc version 2.95.3 20010315 (release)) #43 四 10月 11 14:25:14 CST 2007
AT91RM9200DK login:
<4> -a=0x20008000,-e = 0x20008040 ,tftpaddr= 0x21000000
搬动,但-e地址不对,失败
4.5.3 关于压缩及非压缩内核bootm启动的全面总结
由上面的16个例子,我们可以看出,能够启动内核的由以下几种情况:
各种情况对应的统一ramdiskaddr= 0x21100000
<1>非压缩的Image内核:
-a=-e = 0x20008000 ,–c=none,tftpaddr= 0x20f00000
此法主要由于内核太大,导致tftpaddr做了一定的修正
-a= 0x20008000 ,-e = 0x20008040,–c=none,tftpaddr=0x20008000
此法理论上可行,但我未试验成功,有兴趣的朋友可以探究下
对于非压缩的Image内核,mkimage之前不压缩的话,内核印象较大,此法不常用
-a=-e = 0x20008000 ,–c=gzip,tftpaddr= 0x21000000
–c=gzip压缩内核必须解压,只有这种情况成功;其他解压覆盖或者-e入口不对
<2>压缩的zImage内核:
-a=-e = 0x20008000 ,–c=none,tftpaddr= 0x21000000
-a= 0x20008000 ,-e = 0x20008040,–c=none,tftpaddr=0x20008000
-a=-e = 0x20008000 ,–c=gzip,tftpaddr= 0x21000000
–c=gzip压缩内核必须解压,只有这种情况成功;其他解压覆盖或者-e入口不对
zImage已经压缩过一次了,一般无需再压缩,此法不常用
常见方法:
<1>非压缩的Image内核:
-a=-e = 0x20008000 ,–c=gzip,tftpaddr= 0x21000000
<2>压缩的zImage内核:
-a=-e = 0x20008000 ,–c=none,tftpaddr= 0x21000000
-a= 0x20008000 ,-e = 0x20008040,–c=none,tftpaddr=0x20008000
待续:
U-boot如何向Linux内核传递命令行参数?
Go引导内核的详细方法?
Ramdisk与initrd怎么传给内核?
本文来自CSDN博客,转载请标明出处:http://blog.csdn.net/sailor_8318/archive/2008/08/05/2773412.aspx