Re: [Live-devel] segfault RtspClient

2017-06-04 Thread Ross Finlayson
Many thanks for the detailed report. Using this, I was able to identify and fix the bug. I have just installed a new version - 2017.06.04 - of the “LIVE555 Streaming Media” code that should finally fix this problem. Ross Finlayson Live Networks, Inc. http://www.live555.com/ ___

Re: [Live-devel] segfault RtspClient

2017-06-02 Thread Gerald Hansink
hi Ross, see below for the explanation from Frederik of dereferencing a deleted RTPInterface - When I take the following steps to add debug output: * #define DEBUG_SEND in RTPInterface.cpp * print RTPInterface's address in its destructor * print which RTPInterface address is used in Sock

Re: [Live-devel] segfault RtspClient

2017-05-30 Thread Ross Finlayson
Thanks for reporting this. This is strange. The problem appears to be occurring in the “SocketDescriptor” destructor; it is apparently finding - within its “fSubChannelHashTable” - a pointer to a “RTPInterface” object that had already been deleted. But what I can’t (yet) understand is how thi

Re: [Live-devel] segfault RtspClient

2017-05-30 Thread Gerald Hansink
Hi Ross, we still have the same error (below), I highly recommend Asan, it has proven to be very valuable locating bugs that are otherwise impossible to detect. I remember from my last debugging session that all we do is a call to Medium::close(ourMediaSession). Tomorrow I will likely have time to

Re: [Live-devel] segfault RtspClient

2017-05-24 Thread Ross Finlayson
Thanks for reporting this issue. I have just installed a new version (2017.05.24) of the “LIVE555 Streaming Media” code that should (I hope) fix this. (Does this also solve the problem that “libasan” reported to you?) Ross Finlayson Live Networks, Inc. http://www.live555.com/ __

Re: [Live-devel] segfault RtspClient

2017-05-09 Thread Ross Finlayson
> in previous version in RtspClient::resetTCPSockets the "alternative byte > handler" is cleared using the "input socket num" before setting it to -1. > > RTPInterface::clearServerRequestAlternativeByteHandler(envir(), > fInputSocketNum); // in case we were receiving RTP-over-TCP > > i

Re: [Live-devel] segfault RtspClient

2017-05-09 Thread Ross Finlayson
> in previous version in RtspClient::resetTCPSockets the "alternative byte > handler" is cleared using the "input socket num" before setting it to -1. > > RTPInterface::clearServerRequestAlternativeByteHandler(envir(), > fInputSocketNum); // in case we were receiving RTP-over-TCP > > i

[Live-devel] segfault RtspClient

2017-05-09 Thread Gerald Hansink
hi ross, using the latest livemedia sources i see a segfault sometimes when making a rtsp connection to a axis ip camera. it looks like in SocketDescriptor destructor a "alternative byte handler" callback function is called with a pointer to a RtspClient instance which is deleted. in RtspClient d