求助一个程序崩溃的错误

demongunner

回复 10# irp


    ------------symbol加载都成功了-----------
Core was generated by `./testpro'.
Program terminated with signal 11, Segmentation fault.
#0  0x00002b087ef5d0dd in vfprintf () from /lib64/libc.so.6
(gdb) bt
#0  0x00002b087ef5d0dd in vfprintf () from /lib64/libc.so.6
#1  0x00002b087ef7baf9 in vsprintf () from /lib64/libc.so.6
#2  0x00002b087e7c4631 in _e_gp_doulog () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libengine.so
#3  0x00002b087e7c6cec in userlog () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libengine.so
#4  0x00002b087e79a688 in _e_cryp_keye_one () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libengine.so
#5  0x00002b087df0977f in _wsc_conn () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libwsc.so
#6  0x00002b087defea82 in tpchkauth () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libwsc.so
#7  0x00002b087df0040c in _tpinit () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libwsc.so
#8  0x00002b087df0d06d in tpchkunsol () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libwsc.so
#9  0x00002b087df0d5a0 in _wsc_sighndler () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libwsc.so
#10 <signal handler called>
#11 0x00002b087ef4c200 in sigprocmask () from /lib64/libc.so.6
#12 0x00002b087e7a1cb1 in _e_sig_procmask () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libengine.so
#13 0x00002b087e39647b in _e_net_release_signals () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libgpnet.so
#14 0x00002b087df0d5db in _wsc_sighndler () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libwsc.so
#15 <signal handler called>
-------------此后是一直重复10-15条---------
#4555 <signal handler called>
#4556 0x00002b2869734200 in sigprocmask () from /lib64/libc.so.6
#4557 0x00002b2868f89cb1 in _e_sig_procmask () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libengine.so
#4558 0x00002b2868b7e47b in _e_net_release_signals () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libgpnet.so
#4559 0x00002b28686f55db in _wsc_sighndler () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libwsc.so
#4560 <signal handler called>
#4561 0x00002b286834bd7b in accept () from /lib64/libpthread.so.0
#4562 0x000000000040af51 in CServerSocket::InitInstance (this=0x5379f
    at /home/icd/test/testprogram/ServerSocket.cpp:300
#4563 0x00002b286849d23c in CWinThread::UNIXThreadProc (lParam=0x5379f at otherfuncs.cpp:42
#4564 0x00002b28683452a3 in start_thread () from /lib64/libpthread.so.0
#4565 0x00002b28697c542d in clone () from /lib64/libc.so.6
#4566 0x0000000000000000 in ?? ()
(gdb) f 4562
#4562 0x000000000040af51 in CServerSocket::InitInstance (this=0x5379f
    at /home/icd/test/testprogram/ServerSocket.cpp:300
warning: Source file is more recent than executable.
300                     NewSocket = accept(m_ListenSocket, (sockaddr *)&ClientAddr, (socklen_t*)&iAddrLen);
(gdb) p NewSocket
$1 = -1
(gdb)

大致就是这样 这次的gdb结果在accept 有时候是显示在select那，感觉好奇怪
中间那段重复了几千条的是tuxedo中间件的库。。。。

irp

移植程序的时候考虑过UNIX signal的处理吗？SIGIO/SIGPIPE etc. 这个行为和Windows下不同。

demongunner

回复 12# irp


    额，当时按照指导文档修改过框架后，然后一直出崩溃这问题。
然后accept的时候输出错误码4，是个什么Interrupted system call 然后就加了
#ifndef WIN32
        if ( WSAGetLastError() == EINTR )
        {
            continue;
        }
        else{
            printf("accept fail.");
        }
#else           
这段来屏蔽掉这个信号。
话说socket通信中还会有其他什么信号会对程序造成影响吗？linux下没怎么深入研究过，很多都不太了解。郁闷
对了，因为select始终在循环着，当时输出一下错误码，记得有0 有4 还有11什么的。。。

irp

从bt来看，是stack overflow导致的segment fault. 如果你的程序没有处理signal, 这个signal被tuxedo注册的signal handler wsc_sighndler()接管，他实际上没有处理。这个signal handler反复被调用耗尽了stack, crashed。你是不是创建了nonblocking的socket? 你可以用gdb 在wsc_sighndler设置断点，看看得到的signal number是什么，在代码里屏蔽掉试试。

demongunner

回复 14# irp


下面是使用gdb去attach上这个程序的进程之后调试出错然后bt的结果，之前bt出的结果最后一直在accept select附近，这次到send函数那。不过为啥感觉send函数的那些参数都是正常的呢。这次好像是客户端这出现接收消息失败，然后把连接给删了，日志里记了Connection reset by peer。
使用了select应该就是非阻塞的socket吧。
我这程序只是调用了tuxedo的API函数，也能用gdb在tuxedo的相关函数里打断点吗？


    (gdb) c
Continuing.

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1076898112 (LWP 2821)]
0x00002ac275bd30dd in vfprintf () from /lib64/libc.so.6
(gdb)




#4555 <signal handler called>
#4556 0x00002ac275bc2200 in sigprocmask () from /lib64/libc.so.6
#4557 0x00002ac275417cb1 in _e_sig_procmask () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libengine.so
#4558 0x00002ac27500c47b in _e_net_release_signals () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libgpnet.so
#4559 0x00002ac274b835db in _wsc_sighndler () from /home/icd/test/tuxedo/tuxedo10gR3/lib/libwsc.so
#4560 <signal handler called>
#4561 0x00002ac2747d9abb in write () from /lib64/libpthread.so.0
#4562 0x000000000040a79e in CServerSocket::SendToSocket (this=0x5379f8, pNode=0x538070)
    at /home/icd/test/testprogram/ServerSocket.cpp:811
#4563 0x000000000040acbf in CServerSocket::SendRecvThread (this=0x5379f
    at /home/icd/test/testprogram/ServerSocket.cpp:664
#4564 0x000000000040ad6f in CServerSocket::SendRecvThread (lparam=0x5379f
    at /home/icd/test/testprogram/ServerSocket.cpp:454
#4565 0x00002ac2747d32a3 in start_thread () from /lib64/libpthread.so.0
#4566 0x00002ac275c5342d in clone () from /lib64/libc.so.6
#4567 0x0000000000000000 in ?? ()
(gdb) f 4564
#4564 0x000000000040ad6f in CServerSocket::SendRecvThread (lparam=0x5379f
    at /home/icd/test/testprogram/ServerSocket.cpp:454
454         ((CServerSocket *)lparam)->SendRecvThread();
(gdb) f 4563
#4563 0x000000000040acbf in CServerSocket::SendRecvThread (this=0x5379f
    at /home/icd/test/testprogram/ServerSocket.cpp:664
664                                     nRetSend = SendToSocket(pNode);
(gdb) p nRetSend
$1 = 0
(gdb) f 4562
#4562 0x000000000040a79e in CServerSocket::SendToSocket (this=0x5379f8, pNode=0x538070)
    at /home/icd/test/testprogram/ServerSocket.cpp:811
811                     int iRet = send(pNode->m_SocketID, pNode->m_pSendBuf, pNode->m_dwSendBufUsed, MSG_NOSIGNAL);
(gdb) p iRet
$2 = 0
(gdb) p pNode->m_SocketID
$3 = 5
(gdb) p pNode->m_pSendBuf
$4 = 0x2aaaaaaab010 "674     "
(gdb) p pNode->m_dwSendBufUsed
$5 = 674
(gdb)