Kevin P. Fleming [Sat, 24 Oct 2009 14:40:37 +0000 (14:40 +0000)]
Improve performance of pedantic mode dialog searching in chan_sip.
This patch changes chan_sip to use the new astobj2 OBJ_MULTIPLE iterator support
to make pedantic mode dialog searching in find_call() not require a linear search
of all dialogs in the list of dialogs. This patch does *not* change the dialog
matching logic (more on that later), just improves the searching performance.
Richard Mudgett [Fri, 23 Oct 2009 16:57:33 +0000 (16:57 +0000)]
Add to chan_dahdi ISDN HOLD, Call deflection, and keypad facility support.
* Added handling of received HOLD/RETRIEVE messages and the optional ability
to transfer a held call on disconnect similar to an analog phone.
* Added CallRerouting/CallDeflection support for Q.SIG, ETSI PTP, ETSI PTMP.
Will reroute/deflect an outgoing call when receive the message.
Can use the DAHDISendCallreroutingFacility to send the message for the
supported switches.
* Added ability to send/receive keypad digits in the SETUP message.
Send keypad digits in SETUP message: Dial(DAHDI/g1[/K<keypad_digits>][/extension])
Access any received keypad digits in SETUP message by: ${CHANNEL(keypad_digits)}
* Added support for BRI PTMP NT mode.
Don't force menuselect.makeopts to be rebuilt on every build.
For some reason the menuselect.makeopts file was listed as PHONY in the Makefile,
resulting in 'make' needing to rebuild it for every build. This then resulted in
the embedded module rules being rebuilt on every build, which can be slow and is
unnecessary.
This patch fixes the problem by properly allowing 'make' to know when the
menuselect.makeopts file needs to be rebuilt (defining the proper dependencies).
........
Leif Madsen [Thu, 22 Oct 2009 22:24:03 +0000 (22:24 +0000)]
Update README documentation.
Update the README documentation to correctly describe which CLI command you should
use when attempting to get help from the CLI.
Clean valgrind output by suppressing false errors.
Update valgrind.txt documentation and add valgrind.supp file in order to
allow those who are creating valgrind output to have less false errors in
the logfile.
David Vossel [Thu, 22 Oct 2009 19:55:51 +0000 (19:55 +0000)]
SIP TCP/TLS: move client connection setup/write into tcp helper thread, various related locking/memory fixes.
What this patch fixes
1.Moves sip TCP/TLS connection setup into the TCP helper thread:
Connection setup takes awhile and before this it was being
done while holding the monitor lock.
2.Moves TCP/TLS writing to the TCP helper thread: Through the
use of a packet queue and an alert pipe, the TCP helper thread
can now be woken up to write data as well as read data.
3.Locking error: sip_xmit returned an XMIT_ERROR without giving
up the tcptls_session lock. This lock has been completely removed
from sip_xmit and placed in the new sip_tcptls_write() function.
4.Memory leak: When creating a tcptls_client the tls_cfg was alloced
but never freed unless the tcptls_session failed to start. Now the
session_args for a sip client are an ao2 object which frees the
tls_cfg on destruction.
5.Pointer to stack variable: During sip_prepare_socket the creation
of a client's ast_tcptls_session_args was done on the stack and
stored as a pointer in the newly created tcptls_session. Depending
on the events that followed, there was a slight possibility that
pointer could have been accessed after the stack returned. Given
the new changes, it is always accessed after the stack returns
which is why I found it.
Notable code changes
1.I broke tcptls.c's ast_tcptls_client_start() function into two
functions. One for creating and allocating the new tcptls_session,
and a separate one for starting and handling the new connection.
This allowed me to create the tcptls_session, launch the helper
thread, and then establish the connection within the helper thread.
2.Writes to a tcptls_session are now done within the helper thread.
This is done by using an alert pipe to wake up the thread if new
data needs to be sent. The thread's sip_threadinfo object contains
the alert pipe as well as the packet queue.
3.Since the threadinfo object contains the alert pipe, it must now be
accessed outside of the helper thread for every write (queuing of a
packet). For easy lookup, I moved the threadinfo objects from a
linked list to an ao2_container.
Tilghman Lesher [Thu, 22 Oct 2009 19:10:04 +0000 (19:10 +0000)]
Permit storage of voicemail secrets in a separate file, located within the spool directory.
(closes issue #14276)
Reported by: klaus3000
Patches:
app_voicemail.c-svn-trunk-r214898.txt uploaded by klaus3000 (license 65)
Tested by: jamesgolovich
Kevin P. Fleming [Thu, 22 Oct 2009 18:41:47 +0000 (18:41 +0000)]
Fix a refcount error introduced by yesterday's OBJ_MULTIPLE commit.
When an object is being unlinked from its container *and* being returned to
the caller, we do not want to decrement the reference count after unlinking
it from the container, as the reference that the container held is what we
are returning to the caller... and if it was the only remaining reference to
the object, that could result in the object being destroyed.
Richard Mudgett [Thu, 22 Oct 2009 16:33:22 +0000 (16:33 +0000)]
Add support for calling and called subaddress. Partial support for COLP subaddress.
The Telecom Specs in NZ suggests that SUB ADDRESS is always on, so doing
"desk to desk" between offices each with an asterisk box over the ISDN
should then be possible, without a whole load of DDI numbers required.
(closes issue #15604)
Reported by: alecdavis
Patches:
asterisk_subaddr_trunk.diff11.txt uploaded by alecdavis (license 585)
Some minor modificatons were made.
Tested by: alecdavis, rmudgett
IAX2: VNAK loop caused by signaling frames with no destination call number
It is possible for the PBX thread to queue up signaling frames before
a destination call number is received. This can result in signaling
frames being sent out with no destination call number. Since recent
versions of Asterisk require accurate destination callnumbers for all
Full Frames, this can cause a VNAK loop to occur. To resolve this
no signaling frames are sent until a destination callnumber is received,
and destination call numbers are now only required for iax_pvt matching
when the frame is an ACK.
Kevin P. Fleming [Wed, 21 Oct 2009 21:08:47 +0000 (21:08 +0000)]
Finish implementaton of astobj2 OBJ_MULTIPLE, and convert ast_channel_iterator to use it.
This patch finishes the implementation of OBJ_MULTIPLE in astobj2 (the
case where multiple results need to be returned; OBJ_NODATA mode
already was supported). In addition, it converts ast_channel_iterators
(only the targeted versions, not the ones that iterate over all
channels) to use this method.
During this work, I removed the 'ao2_flags' arguments to the
ast_channel_iterator constructor functions; there were no uses of that
argument yet, there is only one possible flag to pass, and it made the
iterators less 'opaque'. If at some point in the future someone really
needs an ast_channel_iterator that does not lock the container, we can
provide constructor(s) for that purpose.
Joshua Colp [Wed, 21 Oct 2009 15:35:09 +0000 (15:35 +0000)]
Add support for specifying the IP address to use for media streams in sip.conf
This is the second commit for this and documents the text stream using the configured
IP address and fixes a bug in the original patch where the UDPTL stream would also
use the different IP address.
The shrinking of caller id removes '(', ' ', ')', non-trailing '.',
and '-' from the string. This means values such as 555.5555 and
test-test result in 555555 and testtest. There are instances,
such as Skype integration, where a specific value is passed via
caller id that must be preserved unmodified. This patch makes
the shrinking of caller id optional in chan_sip and chan_iax in
order to support such cases. By default this option is on to
preserve previous expected behavior.
Pay attention to the return value of the manipulate function.
While this looks like an optimization, it prevents a crash from occurring
when used with certain audiohook callbacks (diagnosed with SVN trunk,
backported to 1.4 to keep the source consistent across versions).
........
Correct timestamp calculations when RTP sample rates over 8kHz are used.
While testing some endpoints that support 16kHz and 32kHz sample rates, some
log messages were generated due to calc_rxstamp() computing timestamps in a way
that produced odd results, so this patch sanitizes the result of the
computations.
........
Fix stale caller id data from being reported in AMI NewChannel event
The problem here is that chan_dahdi is designed in such a way to set
certain values in the dahdi_pvt only once. One of those such values
is the configured caller id data in chan_dahdi.conf. For PRI, the
configured caller id data could be overwritten during a call. Instead
of saving the data and restoring, it was decided that for all non-analog
channels it was simply best to not set the configured caller id in the
first place and also clear it at the end of the call.
Never released PRI channels when using Busy() or Congestion() dialplan apps.
When the Busy() or Congestion() application is used towards ISDN (an ISDN
progress is sent), the responding ISDN Disconnect or Release may contain
the ISDN cause user busy or one of the congestion causes. In chan_dahdi.c
these causes will only set the needbusy or needcongestion flags and not
activate the softhangup procedure. Unfortunately only the latter can
interrupt the endless wait loop of Busy()/Congestion().
Result: PRI channels staying in state busy for the rest of asterisk life
or until the other end times out and forces the call to clear.
(issue #14292)
Reported by: tomaso
Patches:
disc_rel_userbusy.patch uploaded by tomaso (license 564)
(This patch is unrelated to the issue.)
........
Tilghman Lesher [Thu, 15 Oct 2009 22:33:30 +0000 (22:33 +0000)]
Create an API for adding an optional time unit onto the ends of time periods.
Two examples of its use are included, and the usage could be expanded in some
cases into certain configuration options where time periods are specified.
Jeff Peeler [Wed, 14 Oct 2009 17:48:57 +0000 (17:48 +0000)]
Allow for adding message body to the SIP NOTIFY message
Ability has been added to both manager command SIPnotify as well as console
command sip notify. Message body is stored in the "Content" variable. An
example is present in sip_notify.conf.
Remove automatic switching from T.38 to voice mode in chan_sip.
chan_sip has some code to automatically switch from T.38 mode to voice mode when
a voice frame is written to the channel while it is in T.38 mode; this was
intended to handle the situation when a FAX transmission has ended and the channel
is not yet hung up, but is causing problems at the beginning of FAX sessions as
well when there are still voice frames 'in flight' at the time the T.38 negotiation
completes. This patch removes the automatic switchover.
Kevin P. Fleming [Mon, 12 Oct 2009 14:25:29 +0000 (14:25 +0000)]
Remove automatic switching from T.38 to voice mode in chan_sip.
chan_sip has some code to automatically switch from T.38 mode to voice mode when
a voice frame is written to the channel while it is in T.38 mode; this was
intended to handle the situation when a FAX transmission has ended and the channel
is not yet hung up, but is causing problems at the beginning of FAX sessions as
well when there are still voice frames 'in flight' at the time the T.38 negotiation
completes. This patch removes the automatic switchover, and changes app_fax to
explicitly switch off T.38 mode when the FAX transmission process ends.
Mark Michelson [Sun, 11 Oct 2009 22:19:22 +0000 (22:19 +0000)]
Check the proper page for the SENDRPID flag.
If a pending reinvite were sent, we might not properly
send connected party info since we were checking the wrong
flag. This was a rare occurrence, but could still happen
nevertheless.
Russell Bryant [Sun, 11 Oct 2009 17:25:42 +0000 (17:25 +0000)]
Merged revisions 223485-223486 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r223485 | russell | 2009-10-11 12:22:52 -0500 (Sun, 11 Oct 2009) | 6 lines
Don't use data outside of its scope.
The purpose of this code was to have a hangup frame put on the list of deferred
frames. However, the code that read the hangup frame was outside of the scope
of where the hangup frame was declared.
........
r223486 | russell | 2009-10-11 12:25:06 -0500 (Sun, 11 Oct 2009) | 2 lines
Terry Wilson [Fri, 9 Oct 2009 22:04:04 +0000 (22:04 +0000)]
Properly return "free" on confirmed events that are free
CONFIRMED status doesn't imply busy or free, that is handled with the TRANSP
field. Luckily, libical already sets the is_busy status on the span for us.
Initiate T.38 switchover when acting as called party, regardless of FAX direction.
SendFAX() and ReceiveFAX() can be given options to indicate whether they should
act as the calling or called party; this mode should be used to decide whether
to initiate a switchover to T.38, not the direction that the FAX transfer will
take place.
David Vossel [Fri, 9 Oct 2009 15:49:30 +0000 (15:49 +0000)]
p->peerauth is always empty in transmit_register()
When using callbackextension or specifing the peer name
in a registration string, the peer's specific auth settings
set by the "auth=" strings within the peer definition are not
used by the registration. Thanks to ebroad for reporting the
issue and providing the patch.
Terry Wilson [Thu, 8 Oct 2009 23:11:23 +0000 (23:11 +0000)]
Remove global variable that makes dlopen unhappy
This isn't the best way to do this, but it is the easiest. There are some
limitations that are going to need to be addressed at some point with reloads
and when I (or someone else) work on that, then the API can be updated to
handle passing the private config data that the calendar tech modules need in
a better way as well.
David Vossel [Thu, 8 Oct 2009 22:04:41 +0000 (22:04 +0000)]
Deadlock between ast_cel_report_event and ast_do_masquerade
chan_sip calls pbx_exec on a pvt's owner channel while only the
pvt lock is held. Since pbx_exec calls ast_cel_report_event which
attempts to lock the channel, invalid locking order occurs. Channels
should be locked before pvt's.
Russell Bryant [Thu, 8 Oct 2009 19:52:03 +0000 (19:52 +0000)]
Merged revisions 222878 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r222878 | russell | 2009-10-08 14:45:47 -0500 (Thu, 08 Oct 2009) | 44 lines
Make filestream frame handling safer by isolating frames before returning them.
This patch is related to a number of issues on the bug tracker that show
crashes related to freeing frames that came from a filestream. A number of
fixes have been made over time while trying to figure out these problems, but
there re still people seeing the crash. (Note that some of these bug reports
include information about other problems. I am specifically addressing
the filestream frame crash here.)
I'm still not clear on what the exact problem is. However, what is _very_
clear is that we have seen quite a few problems over time related to unexpected
behavior when we try to use embedded frames as an optimization. In some cases,
this optimization doesn't really provide much due to improvements made in other
areas.
In this case, the patch modifies filestream handling such that the embedded frame
will not be returned. ast_frisolate() is used to ensure that we end up with a
completely mallocd frame. In reality, though, we will not actually have to malloc
every time. For filestreams, the frame will almost always be allocated and freed
in the same thread. That means that the thread local frame cache will be used.
So, going this route doesn't hurt.
With this patch in place, some people have reported success in not seeing the
crash anymore.
(SWP-150)
(AST-208)
(ABE-1834)
(issue #15609)
Reported by: aragon
Patches:
filestream_frisolate-1.4.diff2.txt uploaded by russell (license 2)
Tested by: aragon, russell
Fix memory leak if chan_misdn config parameter is repeated.
Memory leak when the same config option is set more than once in an
misdn.conf section. Why must this be considered? Templates! Defining a
template with default port options and later adding to or overriding some
of them.
David Vossel [Wed, 7 Oct 2009 22:58:38 +0000 (22:58 +0000)]
Deadlock in channel masquerade handling
Channels are stored in an ao2_container. When accessing an item within
an ao2_container the proper locking order is to first lock the container,
and then the items within it.
In ast_do_masquerade both the clone and original channel must be locked
for the entire duration of the function. The problem with this is that
it attemptes to unlink and link these channels back into the ao2_container
when one of the channel's name changes. This is invalid locking order as
the process of unlinking and linking will lock the ao2_container while
the channels are locked!!! Now, both the channels in do_masquerade are
unlinked from the ao2_container and then locked for the entire function.
At the end of the function both channels are unlocked and linked back
into the container with their new names as hash values.
This new method of requiring all channels and tech pvts to be unlocked
before ast_do_masquerade() or ast_change_name() required several
changes throughout the code base.
(closes issue #15911)
Reported by: russell
Patches:
masq_deadlock_trunk.diff uploaded by dvossel (license 671)
Tested by: dvossel, atis
misdn.conf: astdtmf must be set to "yes". With "no", buffer loss does not
occur.
The translated frame "f2" when passing through ast_dsp_process() is not
freed whenever it is not used further in process_ast_dsp(). Then in the
end it is never ever freed.
David Vossel [Tue, 6 Oct 2009 22:39:56 +0000 (22:39 +0000)]
contact header port ignored transport when using externip
This patch adds support for TCP/TLS in the Contact header when using
NAT, specifically externip or externhost. The original issue was that
Asterisk sent 5060 as the port in the contact header whether TLS was
used or not. Additionally, this patch adds 2 config options to sip.conf,
specifically externtcpport and externtlsport. This allows a user to
specify different external ports for TCP and TLS other than those used
internally, this is especially useful in in a PAT/port redirection setup.
Thanks to ebroad for reporting the issue and providing the patch!
Fix potential crash when entire span request is received.
The variable index used in this scenario for accessing the dahdi_pvts was
wrong and was most likely copied from the several other places it is used
correctly.
Tilghman Lesher [Tue, 6 Oct 2009 19:31:39 +0000 (19:31 +0000)]
Change schema query to involve the use of an optional schema parameter.
This change is done in such a way as to allow the driver to continue to
function with older databases which don't have these features.
(closes issue #16000)
Reported by: jamicque
Patches:
20091002__issue16000.diff.txt uploaded by tilghman (license 14)
20091002__issue16000__1.6.1.diff.txt uploaded by tilghman (license 14)
Tested by: jamicque
projects / thirdparty / asterisk.git / log
