[OpenSIPS-Users] OpenSIPS 3.1 - raise_event() crashes OpenSIPS with segmentation fault
Vlad Patrascu
vladp at opensips.org
Tue Jul 28 20:37:51 EST 2020
Hi Mark,
thanks for the report! I have pushed a fix for the crash.
I've also updated the documentation to be more clear about how
raise_event() should be used and the changes from 3.0. Basically, you
don't have to necessarily pass both AVP parameters in 3.1, but if you
want to pass only the attribute values for the event you should do:
raise_event("E_WFC_REGISTERED", , $avp(values));
That is because the event attribute values can now only be passed
through the corresponding parameter(which is the last one). The way you
did it initially was equivalent to passing only the event attribute
names which in 3.1 results in an event with no attributes at all. This
in turn was unfortunately causing a crash.
Regards,
--
Vlad Patrascu
OpenSIPS Developer
http://www.opensips-solutions.com
On 28.07.2020 17:42, Mark Allen wrote:
> [SOLVED]
>
> Hi Stas - good call! It's a change in behaviour from 3.0.
>
> In 3.0 documentation says...
>
> The next two parameters should be AVPs and they are optional. If
> only one is present, it should contain the values attached to the event.
>
> In 3.1 it removes mention of the behaviour if only one AVP is present,
> but it's not obvious and perhaps it could be documented in the
> https://www.opensips.org/Documentation/Migration-3-0-0-to-3-1-0 guide?
> Also, perhaps 3.1 should handle it better if parameter is missing
> rather than giving a segfault (something for 3.2?)
>
> Once again I'm indebted to you for your help
>
> thanks very much & all the best
>
> Mark
>
>
>
>
> On Tue, 28 Jul 2020 at 15:22, Stas Kobzar <staskobzar at gmail.com
> <mailto:staskobzar at gmail.com>> wrote:
>
> Hi Allen,
>
> Did you try with two parameters: name, value?
>
> *$avp(keys) = "registered";*
> $avp(values) = "true";
> xlog("Raised E_WFC_REGISTERED $avp(values)");
> raise_event("E_WFC_REGISTERED", *$avp(keys)*, $avp(values));
>
> I know they are said to be optional in the documentation but
> probably it is optional for two. Either no params or if you pass
> parameters, you have to pass both.
>
>
> On Tue, Jul 28, 2020 at 9:59 AM Mark Allen <mark at allenclan.co.uk
> <mailto:mark at allenclan.co.uk>> wrote:
>
> We're upgrading from 3.0 to 3.1. Everything seems ok except we
> get a weird error. We subscribe a dynamic event...
>
> startup_route {
> subscribe_event("E_WFC_REGISTERED", "udp:127.0.0.1:8888
> <http://127.0.0.1:8888>");
> }
>
> which we can see works from /var/log/syslog...
> event_datagram:mod_init: initializing module ...
> core:evi_publish_event: Registered event <E_WFC_REGISTERED(20)
>
> and in the script we invoke it with...
>
> if(is_method("REGISTER")) {
> $avp(values) = "true";
> xlog("Raised E_WFC_REGISTERED $avp(values)");
> raise_event("E_WFC_REGISTERED",$avp(values));
>
> When a phone registers, raise_event() is triggered and
> OpenSIPS crashes with a segmentation fault - shown in
> /var/log/syslog...
>
> Raised E_WFC_REGISTERED true
> CRITICAL:core:sig_usr: segfault in process pid: 10525, id: 8
> segfault at 8 ip 000055cef821313f sp 00007ffcdf4d3410
> error 4 in opensips[55cef801a000+264000]
> kernel: [197593.785622] Code: 0e 00 4c 89 ef e8 1b 70 fc
> ff 49 63 74 24 08 49 8b 3c 24 e8 51 a1 fc ff 48 89 c2 48 8d 35
> 8f 0d 07 00 4c 89 ef e8 fb 6f fc ff <49> 8b 46 08 48 85 c0 74
> 0b 48 83 78 18 00 0f 84 a5 02 00 00 e8 34
> INFO:core:handle_sigs: child process 10525 exited by a
> signal 11
> INFO:core:handle_sigs: core was generated
> INFO:core:handle_sigs: terminating due to SIGCHLD
>
> If I comment out the raise_event() line - OpenSIPS seems fine
> and doesn't crash when passing through this code.
>
>
>
> Running gdb to get core file backtrace we see...
>
> Core was generated by `/usr/sbin/opensips -P
> /run/opensips/opensips.pid -f /etc/opensips/opensips.cfg'.
> Program terminated with signal SIGSEGV, Segmentation fault.
> #0 evi_build_payload (params=0x0, method=0x7f931f5b6f08,
> id=id at entry=0, extra_param_k=extra_param_k at entry=0x0,
> extra_param_v=extra_param_v at entry=0x0) at
> evi/evi_transport.c:159
> 159 if (params->first && !params->first->name.s) {
> (gdb) bt full
> #0 evi_build_payload (params=0x0, method=0x7f931f5b6f08,
> id=id at entry=0, extra_param_k=extra_param_k at entry=0x0,
> extra_param_v=extra_param_v at entry=0x0) at
> evi/evi_transport.c:159
> param = <optimized out>
> param_obj = 0x0
> tmp = <optimized out>
> ret_obj = 0x7f9323135fe0
> payload = 0x0
> __FUNCTION__ = "evi_build_payload"
> #1 0x00007f931b7d934f in datagram_raise (msg=<optimized out>,
> ev_name=<optimized out>, sock=0x7f931f5c54c8,
> params=<optimized out>)
> at event_datagram.c:315
> ret = <optimized out>
> buf = <optimized out>
> __FUNCTION__ = "datagram_raise"
> #2 0x000055cef82148fb in evi_raise_event_msg
> (msg=msg at entry=0x7f9323134890, id=id at entry=20,
> params=params at entry=0x0)
> at evi/event_interface.c:208
> subs = 0x7f931f5c55a8
> prev = <optimized out>
> now = 1595943308
> flags = 1073741838
> pflags = 0
> ret = 0
> __FUNCTION__ = "evi_raise_event_msg"
> #3 0x000055cef8216afb in evi_raise_script_event
> (msg=0x7f9323134890, id=20, _a=<optimized out>, _v=<optimized
> out>)
> at evi/event_interface.c:430
> vals = <optimized out>
> attrs = <optimized out>
> v_avp = <optimized out>
> a_avp = <optimized out>
> err = <optimized out>
> val = {n = 587654904, s = {s = 0x7f932306e6f8 "\002",
> len = -133061172}}
> attr = {n = 0, s = {s = 0x0, len = -133445686}}
> at = <optimized out>
> params = 0x0
> __FUNCTION__ = "evi_raise_script_event"
> #4 0x000055cef8068c5f in w_raise_event (msg=<optimized out>,
> ev_id=<optimized out>, attrs_avp=<optimized out>,
> vals_avp=<optimized out>) at core_cmds.c:1204
> __FUNCTION__ = "w_raise_event"
> #5 0x000055cef8086199 in do_action (a=0x7f932304d020,
> msg=0x7f9323134890) at action.c:972
> ret = <optimized out>
> v = <optimized out>
> i = <optimized out>
> len = <optimized out>
> cmatch = <optimized out>
> aitem = <optimized out>
> adefault = <optimized out>
> spec = <optimized out>
> val = {rs = {s = 0x7f932304c748 "\002", len = 0}, ri =
> -129751552, flags = 21966}
> start = {tv_sec = 94347416839552, tv_usec =
> 140269924432168}
> end_time = <optimized out>
> cmd = 0x55cef832c550 <core_cmds+11280>
> acmd = <optimized out>
> cmdp = {0x14, 0x7f932304cf88, 0x0, 0x2,
> 0x7f9323134890, 0x55cef80bb253 <eval_elem+1256>, 0x1, 0xc}
> tmp_vals = {{rs = {s = 0x400000000 <error: Cannot
> access memory at address 0x400000000>, len = 587509104}, ri =
> 18, flags = 0}, {rs = {s = 0x7f9323134890 "\001", len =
> 587509104}, ri = 588466320, flags = 32659}, {rs = {s =
> 0x55cef8442600 <_oser_err_info> "", len = -133061748}, ri =
> -131568035, flags = 21966}, {rs = {s = 0x3 <error: Cannot
> access memory at address 0x3>, len = 587512256}, ri = 3, flags
> = 0}, {rs = {s = 0x7ffcdf4d3790 "\220H\023#\223\177", len =
> 587509104}, ri = -131568035, flags = 21966}, {rs = {s = 0x3
> <error: Cannot access memory at address 0x3>, len = 0}, ri =
> 587655824, flags = 32659}, {rs = {s = 0x0, len = 588466320},
> ri = 0, flags = 0}, {rs = {s = 0x55cef80baba7 <eval_expr+300>
> "A\211\304D\213\005\277\355'", len = 593194504}, ri = 2, flags
> = 0}}
> sval = {s = 0x7ffcdf4d3730 "]n(\370\316U", len =
> 587515424}
> __FUNCTION__ = "do_action"
> (full backtrace is available)
>
> Build is taken from 3.1 branch on GitHub
> Server is running Debian 10 (Buster)
>
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org <mailto:Users at lists.opensips.org>
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org <mailto:Users at lists.opensips.org>
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20200728/e6dea4b1/attachment-0001.html>
More information about the Users
mailing list