search

Garefield / RTSP-WS-Proxy

a RTSP to Websocket Proxy，and it's web player
43 stars 24 forks source link

linux运行wsserver出现段错误 #1

Open TaijiGao opened 3 years ago

TaijiGao commented 3 years ago

感谢作者的辛苦贡献，本人在centos7服务器运行一段时间后，wsserver会出现段错误导致服务崩溃，调试coredump文件见下面日志，疑似buf的内容和buflen长度不一致，还请作者解惑，不胜感激。 Core was generated by `./WSServer 9001'. Program terminated with signal 11, Segmentation fault.

0 0x0000000000000021 in ?? ()

(gdb) bt

0 0x0000000000000021 in ?? ()

1 0x000000000069a3b5 in RtspSource::SendWSData (this=0x55fb000) at RtspSource.cpp:422

2 0x00000000006a0141 in boost::_mfi::mf0<void, RtspSource>::operator() (this=0x7fe109c96848, p=0x55fb000) at /usr/include/boost/bind/mem_fn_template.hpp:49

3 0x00000000006a00a4 in boost::_bi::list1<boost::_bi::value<RtspSource*> >::operator()<boost::_mfi::mf0<void, RtspSource>, boost::_bi::list0> (this=0x7fe109c96858, f=..., a=...)

at /usr/include/boost/bind/bind.hpp:253

4 0x000000000069fe47 in boost::_bi::bind_t<void, boost::_mfi::mf0<void, RtspSource>, boost::_bi::list1<boost::_bi::value<RtspSource*> > >::operator() (this=0x7fe109c96848)

at /usr/include/boost/bind/bind_template.hpp:20

5 0x000000000069fb76 in boost::detail::thread_data<boost::_bi::bind_t<void, boost::_mfi::mf0<void, RtspSource>, boost::_bi::list1<boost::_bi::value<RtspSource*> > > >::run (this=0x7fe109c96690)

at /usr/include/boost/thread/detail/thread.hpp:117

6 0x00007fe3c0ae425a in boost::(anonymous namespace)::thread_proxy (param=) at libs/thread/src/pthread/thread.cpp:165

7 0x00007fe3bfc88ea5 in start_thread (arg=0x7fe3bdea7700) at pthread_create.c:307

8 0x00007fe3bf9b19fd in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:111

(gdb) f 1

1 0x000000000069a3b5 in RtspSource::SendWSData (this=0x55fb000) at RtspSource.cpp:422

warning: Source file is more recent than executable. 422 delete fdata; (gdb) p fdata->buf $1 = (uint8_t *) 0x7fe23c000088 "\360^\325\032\341\177" (gdb) p fdata->buflen $2 = 32 (gdb)

Garefield commented 3 years ago

websocket这块我只做了简单的实现，并未检查内存泄露情况，你可以自己修改这块。该工程只是我用来验证自己的设想，并非用于工作环境。你可以自己重构下websocket这部分

发自我的iPhone

在 2021年5月14日，下午10:29，Gary Taiji Gao @.***> 写道：

 感谢作者的辛苦贡献，本人在centos7服务器运行一段时间后，wsserver会出现段错误导致服务崩溃，调试coredump文件见下面日志，疑似buf的内容和buflen长度不一致，还请作者解惑，不胜感激。 Core was generated by `./WSServer 9001'. Program terminated with signal 11, Segmentation fault.

0 0x0000000000000021 in ?? ()

(gdb) bt

0 0x0000000000000021 in ?? ()

1 0x000000000069a3b5 in RtspSource::SendWSData (this=0x55fb000) at RtspSource.cpp:422

2 0x00000000006a0141 in boost::_mfi::mf0<void, RtspSource>::operator() (this=0x7fe109c96848, p=0x55fb000) at /usr/include/boost/bind/mem_fn_template.hpp:49

3 0x00000000006a00a4 in boost::_bi::list1<boost::_bi::value<RtspSource*> >::operator()<boost::_mfi::mf0<void, RtspSource>, boost::_bi::list0> (this=0x7fe109c96858, f=..., a=...)

at /usr/include/boost/bind/bind.hpp:253

4 0x000000000069fe47 in boost::_bi::bind_t<void, boost::_mfi::mf0<void, RtspSource>, boost::_bi::list1<boost::_bi::value<RtspSource*> > >::operator() (this=0x7fe109c96848)

at /usr/include/boost/bind/bind_template.hpp:20

5 0x000000000069fb76 in boost::detail::thread_data<boost::_bi::bind_t<void, boost::_mfi::mf0<void, RtspSource>, boost::_bi::list1<boost::_bi::value<RtspSource*> > > >::run (this=0x7fe109c96690)

at /usr/include/boost/thread/detail/thread.hpp:117

6 0x00007fe3c0ae425a in boost::(anonymous namespace)::thread_proxy (param=) at libs/thread/src/pthread/thread.cpp:165

7 0x00007fe3bfc88ea5 in start_thread (arg=0x7fe3bdea7700) at pthread_create.c:307

8 0x00007fe3bf9b19fd in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:111

(gdb) f 1

1 0x000000000069a3b5 in RtspSource::SendWSData (this=0x55fb000) at RtspSource.cpp:422

warning: Source file is more recent than executable. 422 delete fdata; (gdb) p fdata->buf $1 = (uint8_t *) 0x7fe23c000088 "\360^\325\032\341\177" (gdb) p fdata->buflen $2 = 32 (gdb)

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

Garefield commented 3 years ago

如果要把这个程序作为服务，应该要把rtsp数据做个缓冲，并且在每次发送ws数据时检查实际发出的数据长度，这个处理我是没做的，问题可能出在这里，我的程序是一个验证，因此并未考虑带宽问题，也就是实际发出数据长度问题，ws这块也可能有内存泄露，因为我只是随便找了个ws库做了个简单的发送处理

发自我的iPhone

在 2021年5月14日，下午10:43，杨旋 @.***> 写道：

websocket这块我只做了简单的实现，并未检查内存泄露情况，你可以自己修改这块。该工程只是我用来验证自己的设想，并非用于工作环境。你可以自己重构下websocket这部分

发自我的iPhone

在 2021年5月14日，下午10:29，Gary Taiji Gao @.***> 写道：

 感谢作者的辛苦贡献，本人在centos7服务器运行一段时间后，wsserver会出现段错误导致服务崩溃，调试coredump文件见下面日志，疑似buf的内容和buflen长度不一致，还请作者解惑，不胜感激。 Core was generated by `./WSServer 9001'. Program terminated with signal 11, Segmentation fault.

0 0x0000000000000021 in ?? ()

(gdb) bt

0 0x0000000000000021 in ?? ()

1 0x000000000069a3b5 in RtspSource::SendWSData (this=0x55fb000) at RtspSource.cpp:422

2 0x00000000006a0141 in boost::_mfi::mf0<void, RtspSource>::operator() (this=0x7fe109c96848, p=0x55fb000) at /usr/include/boost/bind/mem_fn_template.hpp:49

3 0x00000000006a00a4 in boost::_bi::list1<boost::_bi::value<RtspSource*> >::operator()<boost::_mfi::mf0<void, RtspSource>, boost::_bi::list0> (this=0x7fe109c96858, f=..., a=...)

at /usr/include/boost/bind/bind.hpp:253

4 0x000000000069fe47 in boost::_bi::bind_t<void, boost::_mfi::mf0<void, RtspSource>, boost::_bi::list1<boost::_bi::value<RtspSource*> > >::operator() (this=0x7fe109c96848)

at /usr/include/boost/bind/bind_template.hpp:20

5 0x000000000069fb76 in boost::detail::thread_data<boost::_bi::bind_t<void, boost::_mfi::mf0<void, RtspSource>, boost::_bi::list1<boost::_bi::value<RtspSource*> > > >::run (this=0x7fe109c96690)

at /usr/include/boost/thread/detail/thread.hpp:117

6 0x00007fe3c0ae425a in boost::(anonymous namespace)::thread_proxy (param=) at libs/thread/src/pthread/thread.cpp:165

7 0x00007fe3bfc88ea5 in start_thread (arg=0x7fe3bdea7700) at pthread_create.c:307

8 0x00007fe3bf9b19fd in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:111

(gdb) f 1

1 0x000000000069a3b5 in RtspSource::SendWSData (this=0x55fb000) at RtspSource.cpp:422

warning: Source file is more recent than executable. 422 delete fdata; (gdb) p fdata->buf $1 = (uint8_t *) 0x7fe23c000088 "\360^\325\032\341\177" (gdb) p fdata->buflen $2 = 32 (gdb)

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

TaijiGao commented 3 years ago

感谢答复，这个问题不是内存泄露，即使websocket没有完全发送成功，av_fifo_generic_read这个函数已经完全读出了rtsp拉流数据，delete pdata理论上是与websocket发送多少字节没有关系的，所以不理解为何delete pdata销毁内存有时会越界。因个人对buf内的音视频数据不太熟悉，不清楚buf的内存数据是否与实际的buflen一致，作者如果方便可加下微信13862049340，进一步深入讨论，谢谢

Garefield commented 3 years ago

视频缓冲这里应该是没有问题的，我对ws不是太熟，所以推断问题可能出在这里，以前也有人反应过这个程序有内存泄露的情况，因为这个程序我只是验证我的思路，并不用在生产环境，所以就一直没有去排查。ffmpeg部分我是经常这么写的，应该没有什么问题

发自我的iPhone

在 2021年5月15日，下午11:28，Gary Taiji Gao @.***> 写道：

 感谢答复，这个问题不是内存泄露，即使websocket没有完全发送成功，av_fifo_generic_read这个函数已经完全读出了rtsp拉流数据，delete pdata理论上是与websocket发送多少字节没有关系的，所以不理解为何delete pdata销毁内存有时会越界。因个人对buf内的音视频数据不太熟悉，不清楚buf的内存数据是否与实际的buflen一致，作者如果方便可加下微信13862049340，进一步深入讨论，谢谢

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.