网络链接序列号不正确

时间：2018-07-16 22:02:40

标签： go wifi netlink

我正在使用netlink和genetlink Go包来编写实用程序，以与MAC80211_HWSIM netlink系列功能进行交互。运作良好...差不多。

我发现，如果我进行GET_RADIO调用，则对该系列的下一个调用无法验证，因为响应的序列号与请求的序列号不匹配。具体来说，我看到的模式是

GET_RADIO request:  seq=655
GET_RADIO response: seq=655
DEL_RADIO request:  seq=656
DEL_RADIO response: seq=655

在GET_RADIO之后发出哪个命令都没有关系，也显示相同的模式。如果没有使用GET_RADIO，则序列号将完全符合预期。

有一个用libnl和libgenl用C编写的现有命令，我将其修改为运行多个命令。尽我所知，该实现的功能非常相似，但是序列号没有问题，因此我不愿意认为问题出在内核模块上。但是strace清楚地表明内核正在响应的是无序数（请原谅过多的横向滚动）：

socket(AF_NETLINK, SOCK_RAW, NETLINK_GENERIC) = 3
bind(3, {sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, 12) = 0
getsockname(3, {sa_family=AF_NETLINK, nl_pid=18357, nl_groups=00000000}, [112->12]) = 0
sendmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base={{len=40, type=0x10 /* NLMSG_??? */, flags=NLM_F_REQUEST, seq=2596996163, pid=18357}, "\3\1\0\0\23\0\2\0MAC80211_HWSIM\0\0"}, iov_len=40}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 40
recvmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=112->12, msg_iov=[{iov_base=[{{len=224, type=0x10 /* NLMSG_??? */, flags=0, seq=2596996163, pid=18357}, "\1\2\0\0\23\0\2\0MAC80211_HWSIM\0\0\6\0\1\0\31\0\0\0"...}, {{len=0, type=0 /* NLMSG_??? */, flags=0, seq=0, pid=0}}], iov_len=4096}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, MSG_PEEK) = 224
recvmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=112->12, msg_iov=[{iov_base=[{{len=224, type=0x10 /* NLMSG_??? */, flags=0, seq=2596996163, pid=18357}, "\1\2\0\0\23\0\2\0MAC80211_HWSIM\0\0\6\0\1\0\31\0\0\0"...}, {{len=0, type=0 /* NLMSG_??? */, flags=0, seq=0, pid=0}}], iov_len=4096}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 224
sendmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base={{len=28, type=0x19 /* NLMSG_??? */, flags=NLM_F_REQUEST|NLM_F_ACK, seq=2596996164, pid=18357}, "\6\1\0\0\10\0\n\0T\0\0\0"}, iov_len=28}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 28
recvmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=112->12, msg_iov=[{iov_base=[{{len=48, type=0x19 /* NLMSG_??? */, flags=0, seq=2596996164, pid=18357}, "\6\1\0\0\10\0\n\0T\0\0\0\10\0\t\0\1\0\0\0\t\0\21\0phy84\0\0\0"}, {{len=0, type=0 /* NLMSG_??? */, flags=0, seq=0, pid=0}}], iov_len=4096}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, MSG_PEEK) = 48
recvmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=112->12, msg_iov=[{iov_base=[{{len=48, type=0x19 /* NLMSG_??? */, flags=0, seq=2596996164, pid=18357}, "\6\1\0\0\10\0\n\0T\0\0\0\10\0\t\0\1\0\0\0\t\0\21\0phy84\0\0\0"}, {{len=0, type=0 /* NLMSG_??? */, flags=0, seq=0, pid=0}}], iov_len=4096}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 48
sendmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base={{len=28, type=0x19 /* NLMSG_??? */, flags=NLM_F_REQUEST|NLM_F_ACK, seq=2596996165, pid=18357}, "\5\1\0\0\10\0\n\0T\0\0\0"}, iov_len=28}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 28
recvmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=112->12, msg_iov=[{iov_base=[{{len=36, type=NLMSG_ERROR, flags=0, seq=2596996164, pid=18357}, "\0\0\0\0\34\0\0\0\31\0\5\0D\4\313\232\265G\0\0"}, {{len=0, type=0 /* NLMSG_??? */, flags=0, seq=0, pid=0}}], iov_len=4096}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, MSG_PEEK) = 36
recvmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=112->12, msg_iov=[{iov_base=[{{len=36, type=NLMSG_ERROR, flags=0, seq=2596996164, pid=18357}, "\0\0\0\0\34\0\0\0\31\0\5\0D\4\313\232\265G\0\0"}, {{len=0, type=0 /* NLMSG_??? */, flags=0, seq=0, pid=0}}], iov_len=4096}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 36
close(3)                                = 0
+++ exited with 0 +++

请注意，每个recvmsg()都有两个sendmsg()调用，因为第一个仅执行MSG_PEEK来查看答案是否可用。在三组请求和响应中，第一个（序列... 163）只是获取家庭信息。第二个（序列164）正在进行GET_RADIO（\6）调用，第三个是序列响应不正确的调用：此DEL_RADIO（\5 ）呼叫会传送一个... 165序号的讯息，但回传的回应（如果格式正确）的序号会与上一个指令：... 164相符。

因为这是Go，所以一次有多个线程处于活动状态，但是上面strace输出中的所有调用都来自同一线程。跟踪中唯一丢失的呼叫是对futex()的呼叫。

C版本使用libnl和libgenl提供的回调机制，因此它在与请求来源不同的线程中处理响应。请求跟踪：

socket(AF_NETLINK, SOCK_RAW|SOCK_CLOEXEC, NETLINK_GENERIC) = 3
setsockopt(3, SOL_SOCKET, SO_SNDBUF, [32768], 4) = 0
setsockopt(3, SOL_SOCKET, SO_RCVBUF, [32768], 4) = 0
getpid()                                = 21338
bind(3, {sa_family=AF_NETLINK, nl_pid=-1300212902, nl_groups=00000000}, 12) = 0
getsockname(3, {sa_family=AF_NETLINK, nl_pid=-1300212902, nl_groups=00000000}, [12]) = 0
sendmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base={{len=20, type=0x10 /* NLMSG_??? */, flags=NLM_F_REQUEST|NLM_F_ACK|0x300, seq=1530817156, pid=2994754394}, "\3\1\0\0"}, iov_len=20}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 20
recvmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base=[{{len=116, type=0x10 /* NLMSG_??? */, flags=NLM_F_MULTI, seq=1530817156, pid=2994754394}, "\1\2\0\0\v\0\2\0nlctrl\0\0\6\0\1\0\20\0\0\0\10\0\3\0\2\0\0\0"...}, {{len=96, type=0x10 /* NLMSG_??? */, flags=NLM_F_MULTI, seq=1530817156, pid=2994754394}, "\1\2\0\0\16\0\2\0VFS_DQUOT\0\0\0\6\0\1\0\21\0\0\0\10\0\3\0"...}, {{len=104, type=0x10 /* NLMSG_??? */, flags=NLM_F_MULTI, seq=1530817156, pid=2994754394}, "\1\2\0\0\17\0\2\0acpi_event\0\0\6\0\1\0\23\0\0\0\10\0\3\0"...}, {{len=108, type=0x10 /* NLMSG_??? */, flags=NLM_F_MULTI, seq=1530817156, pid=2994754394}, "\1\2\0\0\22\0\2\0thermal_event\0\0\0\6\0\1\0\24\0\0\0"...}, {{len=112, type=0x10 /* NLMSG_??? */, flags=NLM_F_MULTI, seq=1530817156, pid=2994754394}, "\1\2\0\0\20\0\2\0tcp_metrics\0\6\0\1\0\25\0\0\0\10\0\3\0"...}, {{len=112, type=0x10 /* NLMSG_??? */, flags=NLM_F_MULTI, seq=1530817156, pid=2994754394}, "\1\2\0\0\16\0\2\0TASKSTATS\0\0\0\6\0\1\0\26\0\0\0\10\0\3\0"...}, {{len=2076, type=0x10 /* NLMSG_??? */, flags=NLM_F_MULTI, seq=1530817156, pid=2994754394}, "\1\2\0\0\f\0\2\0nl80211\0\6\0\1\0\30\0\0\0\10\0\3\0\1\0\0\0"...}, {{len=224, type=0x10 /* NLMSG_??? */, flags=NLM_F_MULTI, seq=1530817156, pid=2994754394}, "\1\2\0\0\23\0\2\0MAC80211_HWSIM\0\0\6\0\1\0\31\0\0\0"...}, {{len=0, type=0 /* NLMSG_??? */, flags=0, seq=0, pid=0}}], iov_len=16384}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 2948
recvmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base=[{{len=20, type=NLMSG_DONE, flags=NLM_F_MULTI, seq=1530817156, pid=2994754394}, "\0\0\0\0"}, {{len=0, type=0 /* NLMSG_??? */, flags=0, seq=0, pid=65542}}], iov_len=16384}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 20
mmap(NULL, 8392704, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_STACK, -1, 0) = 0x7fdb752ce000
mprotect(0x7fdb752ce000, 4096, PROT_NONE) = 0
clone(child_stack=0x7fdb75acdff0, flags=CLONE_VM|CLONE_FS|CLONE_FILES|CLONE_SIGHAND|CLONE_THREAD|CLONE_SYSVSEM|CLONE_SETTLS|CLONE_PARENT_SETTID|CLONE_CHILD_CLEARTID, parent_tidptr=0x7fdb75ace9d0, tls=0x7fdb75ace700, child_tidptr=0x7fdb75ace9d0) = 21339
sendmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base={{len=28, type=0x19 /* NLMSG_??? */, flags=NLM_F_REQUEST|NLM_F_ACK, seq=1530817157, pid=2994754394}, "\6\1\0\0\10\0\n\0_\0\0\0"}, iov_len=28}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 28
nanosleep({tv_sec=2, tv_nsec=0}, 0x7ffff64ac200) = 0
write(1, "Deleting radio with id '95'...\n", 31) = 31
sendmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base={{len=28, type=0x19 /* NLMSG_??? */, flags=NLM_F_REQUEST|NLM_F_ACK, seq=1530817158, pid=2994754394}, "\5\1\0\0\10\0\n\0_\0\0\0"}, iov_len=28}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 28
nanosleep({tv_sec=2, tv_nsec=0},  <unfinished ...>) = ?
+++ exited with 0 +++

和响应跟踪：

recvmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base=[{{len=48, type=0x19 /* NLMSG_??? */, flags=0, seq=1530817157, pid=2994754394}, "\6\1\0\0\10\0\n\0_\0\0\0\10\0\t\0\1\0\0\0\t\0\21\0phy95\0\0\0"}, {{len=0, type=0 /* NLMSG_??? */, flags=0, seq=0, pid=0}}], iov_len=16384}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 48
epoll_wait(4, [{EPOLLIN, {u32=3, u64=3}}], 32, -1) = 1
recvmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base=[{{len=36, type=NLMSG_ERROR, flags=0, seq=1530817157, pid=2994754394}, "\0\0\0\0\34\0\0\0\31\0\5\0\205j>[ZS\200\262"}, {{len=1114121, type=0x6870 /* NLMSG_??? */, flags=NLM_F_REQUEST|NLM_F_ECHO|NLM_F_DUMP_INTR|NLM_F_DUMP_FILTERED|0x3940, seq=53, pid=0}, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"...}], iov_len=16384}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 36
fstat(1, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0
write(1, "new SSID defined to interface 95"..., 33) = 33
epoll_wait(4, [{EPOLLIN, {u32=3, u64=3}}], 32, -1) = 1
recvmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base=[{{len=36, type=NLMSG_ERROR, flags=0, seq=1530817158, pid=2994754394}, "\0\0\0\0\34\0\0\0\31\0\5\0\206j>[ZS\200\262"}, {{len=1114121, type=0x6870 /* NLMSG_??? */, flags=NLM_F_REQUEST|NLM_F_ECHO|NLM_F_DUMP_INTR|NLM_F_DUMP_FILTERED|0x3940, seq=53, pid=0}, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"...}], iov_len=16384}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 36
write(1, "Successfully deleted device with"..., 39) = 39
exit_group(0)                                       = ?
+++ exited with 0 +++

与我看来唯一相关的（线程结构除外）是它使用了不同的标志（C代码中的NLM_F_DUMP_INTR|NLM_F_DUMP_FILTERED，Go代码中的0）。

我不确定我还可以使用哪些其他工具，因为我不知道这些序列号是如何构造的。在我看来，内核模块，Go软件包或我的Go代码中的代码似乎没有任何异常，但应该在使用接口的方式与实际上是这样。这个问题对任何人来说都是熟悉的，还是对如何取得进展有任何想法？

1 个答案:

答案 0 :(得分：0)

我仍然不明白为什么会发生这种情况，但是看起来TL; DR的解决方法是读取下一个数据包：它将具有相同的数据和正确的序列号。

我尝试用不同的netlink package编写程序，并发现虽然存在相同的问题，但遇到这种情况时它重复了recvmsg()调用。这样做的代码在这里：

https://github.com/vishvananda/netlink/blob/a06dabf/nl/nl_linux.go#L425

引入该代码的提交并不能解释为什么此循环正确，但是在另一个软件包中模拟它可以解决问题。

相关问题
最新问题