Move the B-channel message handling into a per-span I/O thread,
to solve most of the problems caused by the intermixed data + control
socket interface of mISDN, missing write poll() support on
mISDN B-channels and the FreeTDM I/O model. This eliminates most of
the audio problems (except for a few minor glitches).
A unix stream socket pair is used as a bi-directional pipe replacement
(the pipe code is still included in this commit, but will be removed later),
with the RX and TX buffer sizes carefully tuned to avoid excessive buffering
(= latency) and a deadlock situation between the write() call in ftdm_write()
and the code in misdn_span_run() that needs a minimum amount of data in the
TX buffer, before sending out a PH_DATA_REQ to the mISDN socket
(see misdn_span_run() comments for more details).
The minimum size for pipes is PAGE_SIZE (4k), which is ~500 ms worth of
audio. A socket pair RX/TX buffer size of 3k, seems to hold a maximum
amount of around 500 bytes data in practice, giving us a much lower
maximum latency than a unix pipe. (The socket pair might be replaced by a
custom ring buffer / fifo data structure to get even more fine grained
control of the maximum latency.)
The newly introduced span_start / span_stop callbacks in
ftdm_io_interface_t are used to start / stop the I/O thread. The callback
functions will wait up to 10 seconds for the thread to successfully
start up or shut down (using a mutex + condition var).
NOTE: Using any of the locking ftdm_span_() functions in the I/O will cause
a deadlock between the I/O thread (trying to lock span->mutex) and the
thread calling ftdm_start()/_stop() (holding the span->mutex).
(The I/O thread currently uses direct span member access to avoid this.)
The I/O thread uses the epoll(7) family of functions for event handling.
An epoll context is created on startup and all B-channel sockets are
registered (READ, PRI and ERR). Before entering the event loop,
the I/O thread will send a signal on the condition variable, to
indicate it has completed the startup procedure.
Incoming b-channel and command pipe events are handled by the event loop.
Payload of incoming PH_DATA_IND frames (= audio data) is sent to the
rx_audio_pipe_in end of the b-channel's socket pair and, if enough data is
available, a PH_DATA_REQ of the same size is sent to the b-channel mISDN socket
to transmit audio.
A MISDN_CMD_STOP command on the event pipe will wake up the I/O thread and
cause it to shut down. All b-channels will be unregistered from the epoll context
and the epoll fd closed. The I/O thread terminates itself after signalling the
successfull shutdown on the condition variable.
TODOs:
- Move D-Channel into I/O thread too
- Custom FIFO/ring buffer for data (even lower latency)
- Improve epoll() code (per-channel struct w/ callback, for epfd.data.ptr)
- Use mISDN DSP for audio (e.g. tone generator, dtmf detector, echo cancel)
- Use a per-port / span control socket to execute channel commands
synchronously, or add misdn_commands (queue?) that can be used that way
- Name I/O threads 'mISDN-%SPAN_NAME%', e.g. 'mISDN-M_BRI1'
(= add ftdm_thread_set_namef(thread, fmt, ...) / ftdm_thread_set_name(thread, name))
TL;DR: "tweak", solves "booboo" with audio
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Hunt for a free channel for incoming calls that do not
preselect a channel (pevent->ring.channel == -1).
Verify the preselected channel for calls that do specify a channel
and in case the channel is already taken, hunt for a free one,
or abort with an error message (if the preselection was exclusive).
TE-mode channel selection is the same as before
(there's still room for improvement, though, but i'll save that for later).
The MSN/DDI filter code is moved into the TE-mode section (only useful there).
The duplicate ring detection had to be reworked. We now store the
call reference (CRV) in caller_data->call_reference of the selected channel
and do a CRV -> channel look up with find_channel_by_cref()
at the top of on_ring().
NOTE: This is only lightly tested (NT/TE mode), i'd either have to
write a lot of custom code to check it toroughly or the need for
a scriptable ISDN stack...
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Callbacks are invoked from ftdm_span_start/_stop().
I/O is started before SIG and shut down in reverse order.
This is needed for ftmod_misdn, to move the mISDN message handling
into a separate thread (solving the mISDN socket vs. FreeTDM API issues).
With these callbacks, the I/O thread can be started after the span I/O configuration
has been (successfully) completed and stopped before destroying the span.
NOTE: Both SIG and I/O callbacks are called with the span mutex locked,
so threads created or destroyed synchronously in either of the custom
start/stop functions, can not use ftdm_span_*() functions that lock
the span mutex (e.g. ftdm_span_get_channel_count()).
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
The len variable can, in certain situations (large burst of incoming non-SLIN audio),
exceed the size of the on-stack frame buffer, which causes ftdm_buffer_read_loop() to
overwrite the dt_buffer pointer.
Use ftdm_min() to make sure len (after conversion to SLIN units) isn't larger
than the frame buffer size.
Also adds are couple more code comments.
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Resolves OPENZAP-189
Fixes:
src/ftmod/ftmod_isdn/ftmod_isdn.c: In function 'ftdm_isdn_931_34':
src/ftmod/ftmod_isdn/ftmod_isdn.c:902:21: error: variable 'status' set but not used [-Werror=unused-but-set-variable]
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Fixes:
src/ftmod/ftmod_misdn/ftmod_misdn.c: In function 'misdn_handle_mph_information_ind':
src/ftmod/ftmod_misdn/ftmod_misdn.c:871:3: error: format '%lx' expects argument of type 'long unsigned int', but argument 13 has type 'uint64_t' [-Werror=format]
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Do a "soft" wraparound with modulo, removes the ~0.5s tone glitch.
(Multiply ts.rate (samples) by two to match the offset unit (bytes, 2 per sample).)
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Store the offset in the teletone buffer in the b-channel private data.
An NT-mode setup with ftmod_misdn showed severe (dial-)tone distortions
in a sound editor (330Hz sine wave phase errors), caused by
using a global teletone buffer offset.
Switching to a per-channel offset, that is advanced by the amount
of data actually written to the channel, removes (almost) all
distortions.
There is still a minimal phase error every ~500ms (audible) that
needs more investigating.
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Might as well import these too...
ftdm_offset_of() - Get offset of member in structure.
ftdm_container_of() - Get pointer to enclosing structure from pointer to structure member.
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
ftdm_min(x,y) - Returns the smaller of the two values x and y.
ftdm_max(x,y) - Returns the larger of the two values x and y.
ftdm_clamp(val, min, max) - Returns value that is in the range [vmin,vmax].
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Remove the 'state' variable of per-channel data, use active flag exclusively to track
open/close state.
Add misdn_activate_channel()/misdn_deactivate_channel() helper functions, rename old
one to _misdn_toggle_channel() (internal).
Add _nowait variant of channel de-/activation function, that just sends the mISDN request message.
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Store call CRV in caller_data.call_reference and use call private to hold the
FreeTDM channel object.
Remove isdn_data->channels_{local,remote,outbound}_crv arrays.
Allow (and force) inbound call state transition DIALTONE -> DOWN for
incoming RELEASE COMPLETE messages in NT mode.
Dialtone in NT mode works, everything else needs more testing.
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
ftmod_misdn currently returns len == 0 if the incoming message,
that triggered the read() call, does not contain any data.
Users of ftdm_channel_read() need to handle this case, or they
may possibly end up in an endless loop.
This patch reworks the ftdm_channel_read() handling in ftmod_isdn
and prevents it from entering an endless loop. The read error counter
is reset on first sucessful read w/ data.
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
cc1: warnings being treated as errors
./src/ftmod/ftmod_libpri/ftmod_libpri.c: In function 'msn_filter_foreach':
./src/ftmod/ftmod_libpri/ftmod_libpri.c:422: warning: dereferencing type-punned pointer will break strict-aliasing rules
make: *** [ftmod_libpri_la-ftmod_libpri.lo] Error 1
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Two fixes:
Use ftdm_set_string() instead of sprintf() (seriously, wtf?).
Drop invalid, needless argument to ftdm_log().
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Neither of them accepts extra arguments and ftmod_sangoma_isdn is calling
ftmod_assert() with a char * msg argument.
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
The format string checks already caught a couple crash-worthy bugs and this
commit fixes a couple more.
Also includes __ftdm_check_scanf(), for completeness (currently unused).
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Output the current trunk_type in "add X-channel vs. trunk_type" error messages and
check this for B-/D-channels too.
ISDN (= digital) spans need to have a trunk_type set before adding channels,
bail out early with an error message (actually two) if this is not the case.
(Adding channels should really be moved out of the parsing loop, to catch
certain types of errors easier.)
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Some I/O implementations (e.g. mISDN) need this information to correctly
set up the port configuration. In FreeTDM, the mode of a span has been.
up until now, a part of the signalling module configuration ("node" or "mode"
for libpri and isdn spans), which is parsed _after_ the I/O part of a span
has been initialized. This limitation currently prevents us from using mISDN
in NT mode.
To work around this problem (without adding a lot of new tunk_type values like "E1_NET" etc.),
add a new "trunk_mode" parameter to the span categories in freetdm.conf, which by default
is "CPE" for all trunk types, except "FXS" (which defaults to "NET").
ftmod_misdn uses trunk_type + trunk_mode to correctly choose the d-channel protocol
for a port.
ftmod_libpri uses trunk_mode as a hint for the default "mode" setting.
(NOTE: It will print a warning if trunk_mode and signalling mode do not match.)
All other modules currently ignore the value of trunk_mode.
Example freetdm.conf snippet for a mISDN span in NET/NT mode
(using a HFC-S USB dongle):
[span misdn MBRI_1]
trunk_type => BRI_PTMP
trunk_mode => NET
b-channel => HFC-S_USB.1:1-2
d-channel => HFC-S_USB.1:3
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Portability fix for uClibc and other (linux) environments that lack execinfo.h.
ftdm_backtrace_walk() and related return FTDM_NOTIMPL and print a message
if backtraces are not available in the current environment.
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
This feature allows ftmod_libpri to ignore calls with non-matching destination number.
You may want to use this on BRI PTMP lines (Point-to-MultiPoint),
to avoid conflicts between your FreeSWITCH server and other devices connected to the line.
The filter is disabled by default (all calls will be accepted),
setting one (or more) "local-number" parameters on the span configuration enables it.
Example configuration snippet:
<libpri_spans>
<span name="example01">
<!-- ... other span settings omitted ... -->
<param name="local-number" value="123456"/>
<param name="local-number" value="654321"/>
</span>
</libpri_spans>
Signed-off-by: Stefan Knoblich <stkn@openisdn.net>
Stefan Knoblich
David Yat Sin
Moises Silva
Jeff Lenk