CRITICAL:tls_mgm:shm_free:

[fedkis]

Sometimes my opensips crashes with the follow error:

Aug 19 14:52:40 sipnet /usr/local/opensips-2.2.1/sbin/opensips[2545]: CRITICAL:tls_mgm:shm_free: #012>>> shm_free() on n
on-shm ptr 0x200000000 - aborting!#012#012It seems you have hit a programming bug.#012Please help us make OpenSIPS bette
r by reporting it at https://github.com/OpenSIPS/opensips/issues#012



(gdb) 
(gdb) 
(gdb) bt
#0  0x00007f78cc3ac5e5 in raise () from /lib64/libc.so.6
#1  0x00007f78cc3addc5 in abort () from /lib64/libc.so.6
#2  0x00007f784415b851 in shm_free (ptr=0x200000000) at ../../mem/shm_mem.h:495
#3  os_free (ptr=0x200000000) at tls.h:103
#4  0x00007f7843b6c8fd in CRYPTO_free () from /opt/local/lib/libcrypto.so.1.0.0
#5  0x00007f7843f2b646 in SSL_SESSION_free () from /opt/local/lib/libssl.so.1.0.0
#6  0x00007f7843f2930d in SSL_free () from /opt/local/lib/libssl.so.1.0.0
#7  0x00007f78438e5210 in tls_conn_clean (c=0x7f7844970c50) at ../tls_mgm/tls_conn_ops.h:148
#8  0x0000000000560356 in _tcpconn_rm (c=0x7f7844970c50) at net/net_tcp.c:688
#9  0x000000000056062a in tcpconn_destroy (tcpconn=0x7f7844970c50) at net/net_tcp.c:929
#10 0x00000000005609e9 in handle_tcp_worker (tcp_c=0x7f78c43ee770, fd_i=<value optimized out>) at net/net_tcp.c:1210
#11 0x0000000000567f59 in handle_io (repeat=0, t=5, h=<value optimized out>) at net/net_tcp.c:1391
#12 io_wait_loop_epoll (repeat=0, t=5, h=<value optimized out>) at net/../io_wait_loop.h:221
#13 0x0000000000568d43 in tcp_main_server () at net/net_tcp.c:1534
#14 0x000000000056a53e in tcp_start_processes (chd_rank=0x7ed7e8, startup_done=0x0) at net/net_tcp.c:1787
#15 0x000000000043ba1a in main_loop (argc=<value optimized out>, argv=<value optimized out>) at main.c:677
#16 main (argc=<value optimized out>, argv=<value optimized out>) at main.c:1249

[fedkis]

I updated openssl to 1.0.2h and recompiled opensips 2.2.1 . The problem with TLS still exists

(gdb) bt
#0  0x00007f79ce3b682f in __strlen_sse42 () from /lib64/libc.so.6
#1  0x00007f7945e3a430 in i2d_SSL_SESSION () from /opt/local/lib/libssl.so.1.0.0
#2  0x00007f7945e02c6a in ssl3_send_newsession_ticket () from /opt/local/lib/libssl.so.1.0.0
#3  0x00007f7945e07e00 in ssl3_accept () from /opt/local/lib/libssl.so.1.0.0
#4  0x00007f7945792b04 in tls_accept (c=0x7f79467118c8, poll_events=0x0) at ../tls_mgm/tls_conn_server.h:155
#5  0x00007f794579db1f in tls_fix_read_conn (con=0x7f79467118c8, bytes_read=0x7fffadcdea1c)
    at ../tls_mgm/tls_conn_server.h:336
#6  tls_read_req (con=0x7f79467118c8, bytes_read=0x7fffadcdea1c) at proto_tls.c:388
#7  0x000000000056cf87 in handle_io (fm=<value optimized out>, idx=0, event_type=<value optimized out>)
    at net/net_tcp_proc.c:205
#8  0x000000000056d927 in io_wait_loop_epoll (repeat=0, t=2, h=<value optimized out>) at net/../io_wait_loop.h:221
#9  0x000000000056e9c2 in tcp_worker_proc (unix_sock=<value optimized out>) at net/net_tcp_proc.c:312
#10 0x000000000056a19c in tcp_start_processes (chd_rank=<value optimized out>, startup_done=<value optimized out>)
    at net/net_tcp.c:1761
#11 0x000000000043ba1a in main_loop (argc=<value optimized out>, argv=<value optimized out>) at main.c:677
#12 main (argc=<value optimized out>, argv=<value optimized out>) at main.c:1249
(gdb)  

[fedkis]

version: opensips 2.2.1 (x86_64/linux)
flags: STATS: On, DISABLE_NAGLE, USE_MCAST, SHM_MMAP, PKG_MALLOC, F_MALLOC, FAST_LOCK-ADAPTIVE_WAIT
ADAPTIVE_WAIT_LOOPS=1024, MAX_RECV_BUFFER_SIZE 262144, MAX_LISTEN 16, MAX_URI_SIZE 1024, BUF_SIZE 65535
poll method support: poll, epoll_lt, epoll_et, sigio_rt, select.
main.c compiled on 15:48:09 Aug 19 2016 with gcc 4.4.7

[bogdan-iancu]

Hi @fedkis , is this problem still occurring with the latest 2.2.1 release ? Is there a way to reproduce it ? What is the scenario ?

[bogdan-iancu]

This is the same with #1013 , so let's move the discussion over there.