本文转载自:http://blog.csdn.net/myvest/article/details/51483647
版权声明:本文为博主原创文章,未经博主允许不得转载。
1、问题背景
最近遇到一个问题,在一个项目中,无论静态或是DHCP,都无法成功修改DNS。
最后发现,是因为/etc/dhcpcd/dhcpcd-hooks/20-dns.conf 这个脚本中,将DNS写死了,如下代码片,每次设置完以后,还去在设置一次。
setprop dhcp.${intf}.dns${dnsvalue} 203.82.48.3 setprop dhcp.${intf}.dns${standydnsvalue} 203.82.48.4
# Set net.<iface>.dnsN properties that contain the
# DNS server addresses given by the DHCP server.
if [[ $interface == p2p* ]]
then
intf=p2p
else
intf=$interface
fi
set_dns_props()
{
case "${new_domain_name_servers}" in
"") return 0;;
esac
count=1
for i in 1 2 3 4; do
setprop dhcp.${intf}.dns${i} ""
done
count=1
for dnsaddr in ${new_domain_name_servers}; do
setprop dhcp.${intf}.dns${count} ${dnsaddr}
count=$(($count + 1))
done
dnsvalue=1
standydnsvalue=2
setprop dhcp.${intf}.dns${dnsvalue} 203.82.48.3
setprop dhcp.${intf}.dns${standydnsvalue} 203.82.48.4
separator=" "
if [ -z "$new_domain_name" ]; then
separator=""
else
if [ -z "$new_domain_search" ]; then
separator=""
fi
fi
setprop dhcp.${interface}.domain "${new_domain_name}$separator${new_domain_search}"
}
unset_dns_props()
{
for i in 1 2 3 4; do
setprop dhcp.${intf}.dns${i} ""
done
setprop dhcp.${interface}.domain ""
}
case "${reason}" in
BOUND|INFORM|REBIND|REBOOT|RENEW|TIMEOUT) set_dns_props;;
EXPIRE|FAIL|IPV4LL|RELEASE|STOP) unset_dns_props;;
esac
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
- 56
- 57
那么,这个脚本是有什么用处呢?为什么这个脚本可以写死DNS。
这是因为android中,DHCP分为两个部分,一个是DHCP Client端,一个是server端。
2、Dhcp客户端
client端就是我们从上层APP到framework后发出的DHCP请求。这个流程不难,一般是最后会调用runDhcp函数,其JNI在
frameworksasecorejniandroid_net_NetUtils.cpp
中的android_net_utils_runDhcpCommon。然后会在调用dhcp_do_request函数。到这里,framework的部分就完了,接下来会调用到system/core/libnetutils中去。
在dhcp_do_request函数中,当[dhcp.eth0.result]属性变为[ok]时,会调用fill_ip_info函数,这个函数的作用,就是去读取IP信息而已。
static void fill_ip_info(const char *interface, in_addr_t *ipaddr, in_addr_t *gateway, in_addr_t *mask, in_addr_t *dns1, in_addr_t *dns2, in_addr_t *server, uint32_t *lease)
{
property_get(“dhcp.eth0. ipaddress”, prop_value,NULL);
property_get(“dhcp.eth0. gateway”, prop_value,NULL);
property_get(“dhcp.eth0. mask”, prop_value,NULL);
property_get(“dhcp.eth0. dns1”, prop_value,NULL);
property_get(“dhcp.eth0. dns2”, prop_value,NULL);
property_get(“dhcp.eth0. server”, prop_value,NULL);
property_get(“dhcp.eth0. leasetime”, prop_value,NULL);
}
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
那是谁去真正做DHCP的工作呢,是dhcpd这个守护进程。这个进程是根据init.svc.dhcpcd来控制的。这个进程会去设置这些属性,这些属性最后都是根据脚本设置的,也就是我们上面提到的脚本20-dns.conf和95-configured。这两个脚本又是通过etc/dhcpcd/dhcpcd-run-hooks调起来的。
大概流程如下面:
3、Dhcp server端
dhcpcd 调用过程:
代码在external/dhcpcd下,
=>main
# define SYSCONFDIR "/system/etc/dhcpcd"
#define PACKAGE "dhcpcd"
# define CONFIG SYSCONFDIR "/" PACKAGE ".conf"
# define LIBEXECDIR "/system/etc/dhcpcd"
# define SCRIPT LIBEXECDIR "/" PACKAGE "-run-hooks"
=>strlcpy(options->script, SCRIPT, sizeof(options->script));//默认的options->script="/system/etc/dhcpcd /dhcpcd-run-hooks"
=>f = fopen(cf ? cf : CONFIG, "r");//如果没有指定.conf文件,那么使用默认.conf文件
=>parse_config_line//解析"/system/etc/dhcpcd/dhcpcd.conf"默认配置文件
=>parse_option
=>如果在"/system/etc/dhcpcd/dhcpcd.conf"有"script"这个节
=>那么执行strlcpy(options->script, oarg, sizeof(options->script));直接拷贝
/*
{"script", required_argument, NULL, 'c'},
{"option", required_argument, NULL, 'o'},
"/system/etc/dhcpcd/dhcpcd.conf"中的部分内容如下:
...
option domain_name_servers, domain_name, domain_search, host_name
...
*/
=>dhcp_run
=>handle_dhcp_packet
=>handle_dhcp
=>bind_dhcp
reason = "TIMEOUT";reason = "BOUND";reason = "REBIND";reason = "RENEW";
system/extra/dhcpcd-4.0.0-beta9/configure.c
=> configure(iface, reason, state->new, state->old, &state->lease, options, 1);
//如果dhcp超时或者dhcp成功,都会调用exec_script来执行脚本,
//执行setprop dhcp.${interface}.result "failed"或者
//执行setprop dhcp.${interface}.result "ok"
=>exec_script(options, iface->name, reason, NULL, old);
=>然后configure_env通过环境变量将reason传递到脚本中
int exec_script(const struct options *options, const char *iface, const char *reason,
const struct dhcp_message *dhcpn, const struct dhcp_message *dhcpo)
=>pid = fork();
=>if(pid == 0)execve(options->script, argv, env);//子进程执行脚本,默认"/system/etc/dhcpcd/dhcpcd-run-hooks"
//dhcpcd-run-hooks脚本会根据level值,决定是否执行system/etc/dhcpcd/dhcpcd-hook/*目录下的相应文件
//我们的系统在该system/etc/dhcpcd/dhcpcd-hook/*目录下有如下2个文件
//95-configured
//20-dns.conf
=>父进程返回while (waitpid(pid, &status, 0) == -1)等待子进程脚本执行完成
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
4、DHCP协议
有一点,在过滤网络包的时候,应该过滤的是bootp,而不是dhcp哦,dhcp报文的种类有以下这些:
(1)DHCPDISCOVER(0×01),此为Client开始DHCP过程的第一个报文;
(2)DHCPOFFER(0×02),此为Server对DHCPDISCOVER报文的响应;
(3)DHCPREQUEST(0×03),此报文是Slient开始DHCP过程中对server的DHCPOFFER报文的回应,或者是client续延IP地址租期时发出的报文;
(4)DHCPDECLINE(0×04),当Client发现Server分配给它的IP地址无法使用,如IP地址冲突时,将发出此报文,通知Server禁止使用IP地址;
(5)DHCPACK(0×05),Server对Client的DHCPREQUEST报文的确认响应报文,Client收到此报文后,才真正获得了IP地址和相关的配置信息;
(6)DHCPNAK(0×06),Server对Client的DHCPREQUEST报文的拒绝响应报文,Client收到此报文后,一般会重新开始新的DHCP过程;
(7)DHCPRELEASE(0×07),Client主动释放server分配给它的IP地址的报文,当Server收到此报文后,就可以回收这个IP地址,能够分配给其他的Client;
(8)DHCPINFORM(0×08),Client已经获得了IP地址,发送此报文,只是为了从DHCPSERVER处获取其他的一些网络配置信息,如routeip,DNSIp等,这种报文的应用非常少见。