Random SIGSEGV at startup_thread()->SSL_do_handshake();

[Sfinx]

Program terminated with signal SIGABRT, Aborted.
#0  0x00007f248d16e267 in __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:55
55  ../sysdeps/unix/sysv/linux/raise.c: No such file or directory.
[Current thread is 1 (Thread 0x7f247ffff700 (LWP 21446))]
(gdb) backtrace 
#0  0x00007f248d16e267 in __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:55
#1  0x00007f248d16feca in __GI_abort () at abort.c:89
#2  0x00007f248e19445f in OpenSSLDie () from /lib/x86_64-linux-gnu/libcrypto.so.1.0.0
#3  0x00007f248e5afac3 in ?? () from /lib/x86_64-linux-gnu/libssl.so.1.0.0
#4  0x00007f248e591db9 in ?? () from /lib/x86_64-linux-gnu/libssl.so.1.0.0
#5  0x00007f248e5abb4f in ?? () from /lib/x86_64-linux-gnu/libssl.so.1.0.0
#6  0x0000000000404b7d in startup_thread (user_data=0x1902630) at rtcdc.c:377
#7  0x00007f248e84f965 in ?? () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
#8  0x00007f248dcfa6aa in start_thread (arg=0x7f247ffff700) at pthread_create.c:333
#9  0x00007f248d23fe9d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:109

How to debug it further ? It is constanty reproducable - 1 SIGSEGV from 10 good connections. Any ideas ?

[Sfinx]

Looks like race. Solved by calling rtcdc_generate_local_candidate_sdp() and waiting for all candidates gathered. Only after this the rtcdc_generate_offer_sdp() can be called.

[Sfinx]

Seems like SSL_do_handshake() is called in ice and startup threads simultaniously without the proper locking

--- a/src/rtcdc.c
+++ b/src/rtcdc.c
@@ -374,7 +372,9 @@ startup_thread(gpointer user_data)
     SSL_set_connect_state(dtls->ssl);
   else
     SSL_set_accept_state(dtls->ssl);
+  g_mutex_lock(&dtls->dtls_mutex);  
   SSL_do_handshake(dtls->ssl);
+  g_mutex_unlock(&dtls->dtls_mutex);

   while (!peer->exit_thread && !dtls->handshake_done)
     g_usleep(100);