Tethering技术在移动平台上已经运用的越来越广泛了，它可以把移动设备当做一个接入点，其它的设备可以通过Wi-Fi，USB或是Bluetooth等方式连接到此移动设备。在Android中可以将Wifi设为AP模式作为WLAN接入点，从而与其他设备共享Android的互联网连接。Android成为接入点后，就无法通过WLAN连接使用Android的应用程序访问互联网，但可以通过其他方式如以太网或移动网络访问互联网。此时以太网或移动网络在网络共享（Tethering）中是作为upstream的角色。
最近在Atmel的SAMA5D3-EK开发板上调试Wifi模块，需要在Android下实现Tethering，通过Wi-Fi的AP模式，将网络连接共享给其它设备。
开发板上一个有线网卡(eth0)，一个无线网卡(wlan0)，eth0连接到外网，wlan0作为AP共享给其他设备比如Android手机，使得Android手机可以通过开发板连接到外网。
硬件平台：Atmel SAMA5
软件平台：Linux 3.10 +Android 4.4
Wifi模组：RTL8723AU(USB接口)
因为使用的内核是厂商基于主线内核开发的，虽然主线内核中加入了Android的基本支持，但并不完全。在做Android移植时，如果发现Android上层的某些功能缺乏内核的支持，可以根据Google维护的AndroidLinux内核将相应的更改应用到厂商Linux内核中，在前文《Android KitKat 4.4平台开发-添加USBADB和MTP功能支持》中就是使用的这种方法。
整个Wifi移植过程，Wifi模组厂商Realtek给出了详细的过程，但内容只涉及Wifi驱动及Android部分，使用哪个Linux内核以及内核如何配置需要我们自己决定。
按照Realtek提供的移植文档进行AndroidWifi的移植，在测试Wifi 网络共享功能时出现如下问题：
在“设置”程序“网络共享与便携式热点”中，打开“便携式Wi-Fi热点”，并没有真正开启Wifi热点，而是相应单选框不断关开，如此反复。
对于在测试Android功能时出现的异常情况，一般在log中会存在相应的错误信息，虽然并不是绝对准确，但调试时应该考虑先分析log信息，尝试从中定位导致异常发生的代码位置。
经过一番分析推测，如下高亮显示的log信息很有可能是诱发异常发生的关键代码点。
V/NatController(972): enableNat(intIface=<wlan0>, extIface=<eth0>)
V/NatController(972): runCmd(/system/bin/iptables -t nat -A natctrl_nat_POSTROUTING -o eth0 -jMASQUERADE) res=0
V/NatController(972): runCmd(/system/bin/iptables -A natctrl_FORWARD -i eth0 -o wlan0 -m state--state ESTABLISHED,RELATED -g natctrl_tether_counters) res=0
D/dalvikvm(1339): GC_CONCURRENT freed 373K, 54% free 6723K/14460K, paused 44ms+14ms, total219ms
V/NatController(972): runCmd(/system/bin/iptables -A natctrl_FORWARD -i wlan0 -o eth0 -m state--state INVALID -j DROP) res=0
V/NatController(972): runCmd(/system/bin/iptables -A natctrl_FORWARD -i wlan0 -o eth0 -gnatctrl_tether_counters) res=0
V/NatController( 972): runCmd(/system/bin/iptables-A natctrl_tether_counters -i wlan0 -o eth0 -m quota2 --name wlan0_eth0 --grow-j RETURN) res=1
V/NatController(972): runCmd(/system/bin/iptables -D natctrl_FORWARD -i wlan0 -o eth0 -m state--state INVALID -j DROP) res=0
V/NatController(972): runCmd(/system/bin/iptables -D natctrl_FORWARD -i eth0 -o wlan0 -m state--state ESTABLISHED,RELATED -g natctrl_tether_counters) res=0
E/NatController( 972): Error setting forward rules
V/NatController(972): runCmd(/system/bin/iptables -F natctrl_FORWARD) res=0
V/NatController(972): runCmd(/system/bin/iptables -A natctrl_FORWARD -j DROP) res=0
http://androidxref.com/4.4.2_r1/xref/system/netd/NatController.cpp#294
256int
NatController::setTetherCountingRules(bool
add, const char *intIface, const char *extIface) {
257
258
/* We only ever add tethering quota rules so thatthey stick. */
259 if (!add) {
260 return 0;
261 }
262 char *quota_name,
*proc_path;
263 int
quota_fd;
264
asprintf("a_name,
"%s_%s",
intIface,
extIface);
265
266
asprintf(&proc_path,
"/proc/net/xt_quota/%s",
quota_name);
267
quota_fd =
open(proc_path,
O_RDONLY);
268 if (quota_fd
>= 0) {
269
/* quota for iface pair already exists*/
270
free(proc_path);
271
free(quota_name);
272 return 0;
273 }
274
close(quota_fd);
275
free(proc_path);
276
277 const char *cmd2b[]
= {
278
IPTABLES_PATH,
279
"-A",
280
LOCAL_TETHER_COUNTERS_CHAIN,
281
"-i",
282
intIface,
283
"-o",
284
extIface,
285
"-m",
286
"quota2",
287
"--name",
288
quota_name,
289
"--grow",
290
"-j",
291
"RETURN"
292 };
293
294 if (runCmd(ARRAY_SIZE(cmd2b),
cmd2b) &&
add) {
295
free(quota_name);
296 return -1;
297 }
298
free(quota_name);
299
300
asprintf("a_name,
"%s_%s",
extIface,
intIface);
301
asprintf(&proc_path,
"/proc/net/xt_quota/%s",
quota_name);
302
quota_fd =
open(proc_path,
O_RDONLY);
303 if (quota_fd
>= 0) {
304
/* quota for iface pair already exists*/
305
free(proc_path);
306
free(quota_name);
307 return 0;
308 }
309
close(quota_fd);
310
free(proc_path);
311
分析这个函数NatController::setTetherCountingRules及log信息，推断出异常发生的原因是执行命令
/system/bin/iptables -A natctrl_tether_counters -i wlan0 -o eth0 -mquota2 --name wlan0_eth0 --grow -j 失败。
而且还涉及到路径/proc/net/xt_quota/，但当前系统下并不存在这个路径。由此推断应该是内核缺乏与quota2或xt_quota相关的支持。
找到问题的可能原因，接下来就是验证了。比较Android Linux内核、厂商Linux内核以及主线Linux内核网络部分的差异，发现Android Linux内核在主线Linux内核基础上增加了quota2的支持。涉及四次提交
$git log --name-only net/netfilter/xt_quota2.cinclude/linux/netfilter/xt_quota2.h net/netfilter/Kconfignet/netfilter/Makefile
commit7d89969ad270d4c535e33830188806233bf35442
Author:Greg Hackmann <ghackmann@google.com>
Date:Mon Feb 24 09:39:46 2014 -0800
netfilter: xt_qtaguid: 64-bit warning fixes
Change-Id:I2adc517c0c51050ed601992fa0ea4de8f1449414
Signed-off-by: Greg Hackmann<ghackmann@google.com>
net/netfilter/xt_quota2.c
commit73570fe76d3b47e669558f06f1a18e0f02dff606
Author:Arve Hj?nnev?g <arve@android.com>
Date:Mon May 13 20:42:46 2013 -0700
netfilter: xt_quota2: 3.10 fixes.
- Stop using obsolete create_proc_entry api.
- Use proc_set_user instead of directlyaccessing the private structure.
Signed-off-by: Arve Hj?nnev?g<arve@android.com>
net/netfilter/xt_quota2.c
commitea34f99edb73b67ef0a99d304887c64febd4c878
Author:JP Abgrall <jpa@google.com>
Date:Tue Jul 12 12:02:59 2011 -0700
netfilter: fixup the quota2, and enable.
The xt_quota2 came from
http://sourceforge.net/projects/xtables-addons/develop
It needed tweaking for it to compile withinthe kernel tree.
Fixed kmalloc() and create_proc_entry()invocations within
a non-interruptible context.
Removed useless copying of current quotaback to the iptable‘s
struct matchinfo:
- those are per CPU: they will changerandomly based on which
cpu gets to update the value.
- they prevent matching a rule: e.g.
-A chain -m quota2 --name q1 --quota 123
can‘t be followed by
-D chain -m quota2 --name q1 --quota 123
as the 123 will be compared to the structmatchinfo‘s quota member.
Use the NETLINK NETLINK_NFLOG family to loga single message
when the quota limit is reached.
It uses the same packet type as ipt_ULOG,but
- never copies skb data,
- uses 112 as the event number (ULOG‘s +1)
It doesn‘t log if the module param"event_num" is 0.
Change-Id:I021d3b743db3b22158cc49acb5c94d905b501492
Signed-off-by: JP Abgrall<jpa@google.com>
net/netfilter/Kconfig
net/netfilter/Makefile
net/netfilter/xt_quota2.c
commit3db08b39ea752748744e9c7733ce9ef54bed9f3b
Author:JP Abgrall <jpa@google.com>
Date:Tue Jun 21 11:14:49 2011 -0700
netfilter: adding the original quota2 fromxtables-addons
The original xt_quota in the kernel is plainbroken:
- counts quota at a per CPU level
(was written back when ubiquitous SMP wasjust a dream)
- provides no way to count across IPV4/IPV6.
This patch is the original unaltered codefrom:
http://sourceforge.net/projects/xtables-addons
at commite84391ce665cef046967f796dd91026851d6bbf3
Change-Id:I19d49858840effee9ecf6cff03c23b45a97efdeb
Signed-off-by: JP Abgrall<jpa@google.com>
include/linux/netfilter/xt_quota2.h
net/netfilter/xt_quota2.c
提取quota2相关的commit，制作补丁
将这些补丁应用到厂商Linux内核 （git am命令）
在内核配置中增加quota2支持
commit4e6bf851ffd340f83062d053a6a20d358def121e
Author:Max Liao <liaops@embedinfo.com>
Date:Mon Jun 16 06:08:25 2014 -0400
ARM: at91/SAMA5: android_ubifs_defconfig:add netfilter quota2 support for sama5d3 and sama5d4
Signed-off-by: Max Liao<liaops@embedinfo.com>
diff--git a/arch/arm/configs/sama5_android_ubifs_defconfigb/arch/arm/configs/sama5_android_ubifs_defconfig
index9881f7d..48c68a1 100644
---a/arch/arm/configs/sama5_android_ubifs_defconfig
+++b/arch/arm/configs/sama5_android_ubifs_defconfig
@@-623,6 +623,7 @@ CONFIG_NETFILTER_XT_MATCH_OSF=y
CONFIG_NETFILTER_XT_MATCH_OWNER=y
CONFIG_NETFILTER_XT_MATCH_PKTTYPE=y
CONFIG_NETFILTER_XT_MATCH_QUOTA=y
+CONFIG_NETFILTER_XT_MATCH_QUOTA2=y
CONFIG_NETFILTER_XT_MATCH_RATEEST=y
CONFIG_NETFILTER_XT_MATCH_REALM=y
CONFIG_NETFILTER_XT_MATCH_RECENT=y
编译内核。
测试Wifi 网络共享功能，之前的异常现象消失，功能测试正常，这说明之前的推导猜测是正确的，异常的原因的确是内核缺乏netfilter quota2支持。