Grzegorz Sluja authored 3 years ago and Sukru Senli committed 3 years ago

When the NOTIFY event was received by pjsip and the message body was not
properly terminated with '\r\n' the SEGFAULT happened in
parse_simple_message_summary().
Check the proper termination of message body for NOTIFY and fix it if
its wrong.

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

Adam Borowski authored 3 years ago and Sukru Senli committed 3 years ago

Updated howler audio files target path

Adam Borowski authored 3 years ago and Sukru Senli committed 3 years ago

Added howler tones audio files and Makefile install update
to copy *.cln files into target's /usr/share/asterisk/sounds dir

Adam Borowski authored 3 years ago and Sukru Senli committed 3 years ago

Play howler tone in case user forgot to put handset on hook
as a reminder to end the call properly.

Grzegorz Sluja authored 3 years ago and Sukru Senli committed 3 years ago

Move creating of ubus thread after getting the endpt_id and
do not call ubus_lookup_id() afterwards since it can cause deadlock.
Ubus thread uses the same ctx (ubus context *) as the chan_brcm thread
while UBUS is not thread safe and ctx needs to be used by only one
thread in the same time.

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

Grzegorz Sluja authored 3 years ago and Sukru Senli committed 3 years ago

There was a situation that during asterisk restart -> chan_brcm.so
module loading, endptmngr was not up and running yet, causing the ubus
lookup for endpt_id failed. It caused chan_brcm.so module not loaded
and asterisk not fully working properly. Add waiting for endptmngr fixes
the issue.

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

Adam Borowski authored 3 years ago and Sukru Senli committed 3 years ago

Rejecting call ewhen do not disturbed
by indicating line as busy instead of user call rejection

Signed-off-by: Adam Borowski <adam.borowski@sigma.se>

Adam Borowski authored 3 years ago and Sukru Senli committed 3 years ago

Rejecting call ewhen do not disturb is enabled

Signed-off-by: Adam Borowski <adam.borowski@sigma.se>

Adam Borowski authored 3 years ago and Sukru Senli committed 3 years ago

http://downloads.asterisk.org

 being down

if downloading from asterisk fails second try is made
to get the files from fancom mirror

Signed-off-by: Adam Borowski <adam.borowski@sigma.se>

Adam Borowski authored 3 years ago and Sukru Senli committed 3 years ago

http://downloads.asterisk.org

 being down

if downloading from asterisk fails second try is made
to get the files from fancom mirror

Signed-off-by: Adam Borowski <adam.borowski@sigma.se>

Grzegorz Sluja authored 3 years ago and Sukru Senli committed 3 years ago

Sign '#' was recognized as the end of feature code hence some of the
codes for getting status were not working, like:
*#43# - check status of call waiting
*#227# - check status of anonymous call rejection

The fix makes '#' not recognized as the end of feature code if the
previous character is '*'

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

Grzegorz Sluja authored 4 years ago and Sukru Senli committed 3 years ago

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

Grzegorz Sluja authored 4 years ago and Sukru Senli committed 3 years ago

Since we modified asterisk default implementation of attended call
transfer there was an issue with channels stuck (not destroyed)
afterwards, It was caused due to Ref count for channels used in call
transfer was greater than 2.

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

Yalu Zhang authored 4 years ago and Sukru Senli committed 3 years ago

(cherry picked from commit 629000dc on branch iopsys-16.3)

Yalu Zhang authored 4 years ago and Sukru Senli committed 3 years ago

Some long verbose or debug log messages like contents of INVITE are truncated in syslog,
/var/log/asterisk/messages and console.

(cherry picked from commit 4764ca4a on branch iopsys-16.3)

Grzegorz Sluja authored 4 years ago and Sukru Senli committed 3 years ago

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

Yalu Zhang authored 4 years ago and Sukru Senli committed 3 years ago

Call ubus_lookup_id directly instead of passively being updated by
the callback ubus_event_new_obj() which doesn't work reliably during
the system startup.

Yalu Zhang authored 4 years ago and Sukru Senli committed 3 years ago

(cherry picked from commit 8e6bae6e)

Grzegorz Sluja authored 4 years ago and Sukru Senli committed 3 years ago

Change implementation of adding mediasec headers in REGISTER, INVITE,
REINVITE, OPTIONS events to be dynamically configured based on the
response from the Sip Server.

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

Grzegorz Sluja authored 4 years ago and Sukru Senli committed 3 years ago

In pjsip the attended call transfer has an issue that connection between
transferee and transferor is not finished immediately but after 60s
timeout. Not sure why it is implemented this way (defer termination) but
changing the timeout into 1s makes the attended call transfer work the
same as it was with chan_sip driver.

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

Grzegorz Sluja authored 4 years ago and Sukru Senli committed 3 years ago

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

Grzegorz Sluja authored 4 years ago and Sukru Senli committed 3 years ago

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

Yalu Zhang authored 4 years ago and Sukru Senli committed 3 years ago

It is caused that PTHREAD_RECURSIVE_MUTEX_INITIALIZER_NP is not defined on Broadcom ARM platform.
But recursive mutex is mandatory for Asterisk to work properly. Otherwise there will be dead lock
in some modules, e.g. loader.c.

Grzegorz Sluja authored 4 years ago and Sukru Senli committed 3 years ago

Signed-off-by: Grzegorz Sluja <grzegorz.sluja@iopsys.eu>

If an SSL socket parent/listener was destroyed during the handshake,
depending on timing, it was possible for the handling callback to
attempt access of it after the fact thus causing a crash.

ASTERISK-29415 #close

Change-Id: I105dacdcd130ea7fdd4cf2010ccf35b5eaf1432d

If chan_iax2 received a packet with an unsupported media format, for
example vp9, then it would set the frame's format to NULL. This could
then result in a crash later when an attempt was made to access the
format.

This patch makes it so chan_iax2 now ignores/drops frames received
with unsupported media format types.

ASTERISK-29392 #close

Change-Id: Ifa869a90dafe33eed8fd9463574fe6f1c0ad3eb1

If a re-INVITE is received after we have sent a BYE request then it
is possible for no channel to be present on the session. If this
occurs we allow PJSIP to produce the offer instead. Since the call
is being hung up if it produces an incorrect offer it doesn't
actually matter. This also ensures that code which produces SDP
does not need to handle if a channel is not present.

ASTERISK-29381

Change-Id: I673cb88c432f38f69b2e0851d55cc57a62236042

When the MessageSend destination is in the form
PJSIP/<number>@<endpoint> and the endpoint's contact
URI already has a user component, that user component
will now be replaced with <number> when creating the
request URI.

ASTERISK_29404

Change-Id: I80e5910fa25c803d1440da0594a0d6b34b6b4ad5

Set preferred transport when querying the local address to use in
filter_on_tx_messages(). This prevents the module to erroneously select
the wrong transport if more than one transports of the same type (TCP or
TLS) are configured.

ASTERISK-29241

Change-Id: I598e60257a7f92b29efce1fb3e9a2fc06f1439b6

Previously, SayNumber always emitted a warning if the caller hung up
during execution. Usually this isn't correct, so check if the channel
hung up and, if so, don't emit a warning.

ASTERISK-29475

Change-Id: Ieea4a67301c6ea83bbc7690c1d4808d79a704594

The scenario where a channel still has an associated datastore we
cannot unload since there is a function pointer to the destroy and fixup
functions in play.  Thus increase the module ref count whenever we
allocate a datastore, and decrease it during destroy.

In order to tighten the race that still exists in spite of this (below)
add some extra failure cases to prevent allocations in these cases.

Race:

If module ref is zero, an LOCK or TRYLOCK is invoked (near)
simultaneously on a channel that has NOT PREVIOUSLY taken a lock, and if
in such a case the datastore is created *prior* to unloading being set
to true (first step in module unload) then it's possible that the module
will unload with the destructor being called (and segfault) post the
module being unloaded.  The module will however wait for such locks to
release prior to unloading.

If post that we can recheck the module ref before returning the we can
(in theory, I think) eliminate the last of the race.  This race is
mostly theoretical in nature.

Change-Id: I21a514a0b56755c578a687f4867eacb8b59e23cf
Signed-off-by: Jaco Kroon <jaco@uls.co.za>

For example:

arthur*CLI> dialplan locks show
func_lock locks:
Name                                     Requesters Owner
uls-autoref                              0          (unlocked)
1 total locks listed.

Obviously other potentially useful stats could be added (eg, how many
times there was contention, how many times it failed etc ... but that
would require keeping the stats and I'm not convinced that's worth the
effort.  This was useful to troubleshoot some other issues so submitting
it.

Change-Id: Ib875e56feb49d523300aec5f36c635ed74843a9f
Signed-off-by: Jaco Kroon <jaco@uls.co.za>