When an outbound INVITE transaction times out (408) or receives a 503 error,
check_request_status() attempts to failover to the next available address by
restarting the INVITE session. However, the function did not check if the
inv_session was already cancelled before attempting the failover.
This caused unexpected behavior when a caller hung up during a ring group
scenario: after CANCEL was sent but the remote endpoint failed to respond
with 487 (e.g., due to network disconnection), the transaction timeout
would trigger a NEW outbound INVITE to the next address, even though the
session was already terminated.
This violates RFC 3261 Section 9.1 which states that if no final response
is received after CANCEL within 64*T1 seconds, the client should consider
the transaction cancelled and destroy it, not retry to another address.
The fix adds a check for both PJSIP_INV_STATE_DISCONNECTED and inv->cancelling
at the beginning of check_request_status(). This ensures that:
- Failover is blocked when the user explicitly cancelled the call (CANCEL sent)
- Failover is still allowed for legitimate timeout/503 scenarios where no
CANCEL was initiated (e.g., SRV failover when first server is unreachable)
Resolves: #1716
allocate_subscription() increments the ao2 reference count of the subscription tree,
but the reference was not consistently released during subscription destruction,
resulting in leaked sip_subscription_tree objects.
This patch makes destroy_subscription() responsible for releasing sub->tree,
removes ad-hoc cleanup in error paths,
and guards tree cleanup to ensure refcount symmetry and correct ownership.
Fixes: #1703
Reduce cache lock time for AMI and CLI sorcery memory cache populate
commands by adding a new populate_lock to the sorcery_memory_cache
struct which is locked separately from the existing cache lock so that
the cache lock can be maintained for a reduced time, locking only when
the cache objects are removed and re-populated.
Resolves: #1700
UserNote: The AMI command sorcery memory cache populate will now
return an error if there is an internal error performing the populate.
The CLI command will display an error in this case as well.
We now add the websocket's file descriptor to the channel's fd array and let
it poll for data availability instead if having a dedicated thread that
does the polling. This eliminates the thread and allows removal of most
explicit locking since the core channel code will lock the channel to prevent
simultaneous calls to webchan_read, webchan_hangup, etc.
While we were here, the hangup code was refactored to use ast_hangup_with_cause
instead of directly queueing an AST_CONTROL_HANGUP frame. This allows us
to set hangup causes and generate snapshots.
For a bit of extra debugging, a table of websocket close codes was added
to http_websocket.h with an accompanying "to string" function added to
res_http_websocket.c
Resolves: #1683
Always take an explicit reference on the format used for bridge playback
and recording channels, regardless of where it was sourced, and release
it after prepare_bridge_media_channel. This aligns the code paths and
avoids mixing borrowed and owned references while preserving behavior.
Fixes: #1648
* 3d positions were being rendered without an enclosing `<gml:pos>`
element resulting in invalid XML.
* There was no way to set the `id` attribute on the enclosing `tuple`, `device`
and `person` elements.
* There was no way to set the value of the `deviceID` element.
* Parsing of degree and radian UOMs was broken resulting in them appearing
outside an XML element.
* The UOM schemas for degrees and radians were reversed.
* The Ellipsoid shape was missing and the Ellipse shape was defined multiple
times.
* The `crs` location_info parameter, although documented, didn't work.
* The `pos3d` location_info parameter appears in some documentation but
wasn't being parsed correctly.
* The retransmission-allowed and retention-expiry sub-elements of usage-rules
were using the `gp` namespace instead of the `gbp` namespace.
In addition to fixing the above, several other code refactorings were
performed and the unit test enhanced to include a round trip
XML -> eprofile -> XML validation.
Resolves: #1667
UserNote: Geolocation: Two new optional profile parameters have been added.
* `pidf_element_id` which sets the value of the `id` attribute on the top-level
PIDF-LO `device`, `person` or `tuple` elements.
* `device_id` which sets the content of the `<deviceID>` element.
Both parameters can include channel variables.
UpgradeNote: Geolocation: In order to correct bugs in both code and
documentation, the following changes to the parameters for GML geolocation
locations are now in effect:
* The documented but unimplemented `crs` (coordinate reference system) element
has been added to the location_info parameter that indicates whether the `2d`
or `3d` reference system is to be used. If the crs isn't valid for the shape
specified, an error will be generated. The default depends on the shape
specified.
* The Circle, Ellipse and ArcBand shapes MUST use a `2d` crs. If crs isn't
specified, it will default to `2d` for these shapes.
The Sphere, Ellipsoid and Prism shapes MUST use a `3d` crs. If crs isn't
specified, it will default to `3d` for these shapes.
The Point and Polygon shapes may use either crs. The default crs is `2d`
however so if `3d` positions are used, the crs must be explicitly set to `3d`.
* The `geoloc show gml_shape_defs` CLI command has been updated to show which
coordinate reference systems are valid for each shape.
* The `pos3d` element has been removed in favor of allowing the `pos` element
to include altitude if the crs is `3d`. The number of values in the `pos`
element MUST be 2 if the crs is `2d` and 3 if the crs is `3d`. An error
will be generated for any other combination.
* The angle unit-of-measure for shapes that use angles should now be included
in the respective parameter. The default is `degrees`. There were some
inconsistent references to `orientation_uom` in some documentation but that
parameter never worked and is now removed. See examples below.
Examples...
```
location_info = shape="Sphere", pos="39.0 -105.0 1620", radius="20"
location_info = shape="Point", crs="3d", pos="39.0 -105.0 1620"
location_info = shape="Point", pos="39.0 -105.0"
location_info = shape=Ellipsoid, pos="39.0 -105.0 1620", semiMajorAxis="20"
semiMinorAxis="10", verticalAxis="0", orientation="25 degrees"
pidf_element_id = ${CHANNEL(name)}-${EXTEN}
device_id = mac:001122334455
Set(GEOLOC_PROFILE(pidf_element_id)=${CHANNEL(name)}/${EXTEN})
```
Delay acquisition of the ast_sip_endpoint reference in mwi_get_notify_data()
to avoid an ao2 ref leak on early-return error paths.
Move ast_sip_subscription_get_endpoint() to just before first use so all
acquired references are properly cleaned up.
Fixes: #1675
This commit integrates the redirect module into res_pjsip_messaging
to enable following 3xx redirect responses for outgoing SIP MESSAGEs.
When follow_redirect_methods contains 'message' on an endpoint, Asterisk
will now follow 3xx redirect responses for MESSAGEs, similar to how
it behaves for INVITE responses.
Resolves: #1576
UserNote: A new pjsip endpoint option follow_redirect_methods was added.
This option is a comma-delimited, case-insensitive list of SIP methods
for which SIP 3XX redirect responses are followed. An alembic upgrade
script has been added for adding this new option to the Asterisk
database.
This commit introduces a new redirect handling module that provides
infrastructure for following SIP 3xx redirect responses. The redirect
functionality respects the endpoint's redirect_method setting and only
follows redirects when set to 'uri_pjsip'. This infrastructure can be
used by any PJSIP module that needs to handle 3xx redirect responses.
Dual-stack behavior (simultaneous listening for IPV4 and IPV6
connections on a single socket) is required by Asterisk's ICE
implementation. On systems with the IPV6_V6ONLY sockopt, set
the option to 0 (dual-stack enabled) when binding to the IPV6
any address. This ensures correct behavior regardless of the
system's default dual-stack configuration.
This change moves the PJSIP module from the threadpool API
to the taskpool API. PJSIP-specific implementations for
task usage have been removed and replaced with calls to
the optimized taskpool implementations instead. The need
for a pool of serializers has also been removed as
taskpool inherently provides this. The default settings
have also been changed to be more realistic for common
usage.
UpgradeNote: The threadpool_* options in pjsip.conf have now
been deprecated though they continue to be read and used.
They have been replaced with taskpool options that give greater
control over the underlying taskpool used for PJSIP. An alembic
upgrade script has been added to add these options to realtime
as well.
This change makes some small changes to improve log readability in
addition to the following changes:
Modified 'core show taskprocessors' to now show Low time and High time
for task execution.
New command 'core show taskprocessor name <taskprocessor-name>' to dump
taskprocessor info and current queue.
Addionally, a new test was added to demonstrate the 'show taskprocessor
name' functionality:
test execute category /main/taskprocessor/ name taskprocessor_cli_show
Setting 'core set debug 3 taskprocessor.c' will now log pushed tasks.
(Warning this is will cause extremely high levels of logging at even
low traffic levels.)
Resolves: #1566
UserNote: New CLI command has been added -
core show taskprocessor name <taskprocessor-name>
With recent enhancements to chan_websocket, the original plain-text
implementation of control messages and events is now too limiting. We
probably should have used JSON initially but better late than never. Going
forward, enhancements that require control message or event changes will
only be done to the JSON variants and the plain-text variants are now
deprecated but not yet removed.
* Added the chan_websocket.conf config file that allows setting which control
message format to use globally: "json" or "plain-text". "plain-text" is the
default for now to preserve existing behavior.
* Added a dialstring option `f(json|plain-text)` to allow the format to be
overridden on a call-by-call basis. Again, 'plain-text' is the default for
now to preserve existing behavior.
The JSON for commands sent by the app to Asterisk must be...
`{ "command": "<command>" ... }` where `<command>` is one of `ANSWER`, `HANGUP`,
`START_MEDIA_BUFFERING`, etc. The `STOP_MEDIA_BUFFERING` command takes an
additional, optional parameter to be returned in the corresponding
`MEDIA_BUFFERING_COMPLETED` event:
`{ "command": "STOP_MEDIA_BUFFERING", "correlation_id": "<correlation id>" }`.
The JSON for events sent from Asterisk to the app will be...
`{ "event": "<event>", "channel_id": "<channel_id>" ... }`.
The `MEDIA_START` event will now look like...
```
{
"event": "MEDIA_START",
"connection_id": "media_connection1",
"channel": "WebSocket/media_connection1/0x5140001a0040",
"channel_id": "1761245643.1",
"format": "ulaw",
"optimal_frame_size": 160,
"ptime": 20,
"channel_variables": {
"DIALEDPEERNUMBER": "media_connection1/c(ulaw)",
"MEDIA_WEBSOCKET_CONNECTION_ID": "media_connection1",
"MEDIA_WEBSOCKET_OPTIMAL_FRAME_SIZE": "160"
}
}
```
Note the addition of the channel variables which can't be supported
with the plain-text formatting.
The documentation will be updated with the exact formats for all commands
and events.
Resolves: #1546Resolves: #1563
DeveloperNote: The chan_websocket plain-text control and event messages are now
deprecated (but remain the default) in favor of JSON formatted messages.
See https://docs.asterisk.org/Configuration/Channel-Drivers/WebSocket for
more information.
DeveloperNote: A "transport_data" parameter has been added to the
channels/externalMedia ARI endpoint which, for websocket, allows the caller
to specify parameters to be added to the dialstring for the channel. For
instance, `"transport_data": "f(json)"`.
Operations on non-blocking sockets may return a resource temporarily unavailable error (EAGAIN or EWOULDBLOCK). This is not a fatal error but a normal condition indicating that the operation would block.
This patch corrects the handling of this case. Instead of incorrectly treating it as a reason to terminate the connection, the code now waits for data to arrive on the socket.
Also...
* Refactored the verification datastore process so instead of having
a separate channel datastore for each verification result, there's only
one channel datastore with a vector of results.
* Refactored some log messages to include channel name and removed
some that would be redundant if a memory allocation failed.
Resolves: #781
UserNote: The STIR_SHAKEN_ATTESTATION dialplan function has been added
which will allow suppressing attestation on a call-by-call basis
regardless of the profile attached to the outgoing endpoint.
This change moves the PJSIP module from the threadpool API
to the taskpool API. PJSIP-specific implementations for
task usage have been removed and replaced with calls to
the optimized taskpool implementations instead. The need
for a pool of serializers has also been removed as
taskpool inherently provides this. The default settings
have also been changed to be more realistic for common
usage.
UpgradeNote: The threadpool_* options in pjsip.conf have now
been deprecated though they continue to be read and used.
They have been replaced with taskpool options that give greater
control over the underlying taskpool used for PJSIP. An alembic
upgrade script has been added to add these options to realtime
as well.
Although the ISDN/Q.850/Q.931 hangup cause code is already part of the ARI
and AMI hangup and channel destroyed events, it can be helpful to know what
the actual channel technology code was if the call was unsuccessful.
For PJSIP, it's the SIP response code.
* A new "tech_hangupcause" field was added to the ast_channel structure along
with ast_channel_tech_hangupcause() and ast_channel_tech_hangupcause_set()
functions. It should only be set for off-nominal terminations.
* chan_pjsip was modified to set the tech hangup cause in the
chan_pjsip_hangup() and chan_pjsip_session_end() functions. This is a bit
tricky because these two functions aren't always called in the same order.
The channel that hangs up first will get chan_pjsip_session_end() called
first which will trigger the core to call chan_pjsip_hangup() on itself,
then call chan_pjsip_hangup() on the other channel. The other channel's
chan_pjsip_session_end() function will get called last. Unfortunately,
the other channel's HangupRequest events are sent before chan_pjsip has had a
chance to set the tech hangupcause code so the HangupRequest events for that
channel won't have the cause code set. The ChannelDestroyed and Hangup
events however will have the code set for both channels.
* A new "tech_cause" field was added to the ast_channel_snapshot_hangup
structure. This is a public structure so a bit of refactoring was needed to
preserve ABI compatibility.
* The ARI ChannelHangupRequest and ChannelDestroyed events were modified to
include the "tech_cause" parameter in the JSON for off-nominal terminations.
The parameter is suppressed for nominal termination.
* The AMI SoftHangupRequest, HangupRequest and Hangup events were modified to
include the "TechCause" parameter for off-nominal terminations. Like their ARI
counterparts, the parameter is suppressed for nominal termination.
DeveloperNote: A "tech_cause" parameter has been added to the
ChannelHangupRequest and ChannelDestroyed ARI event messages and a "TechCause"
parameter has been added to the HangupRequest, SoftHangupRequest and Hangup
AMI event messages. For chan_pjsip, these will be set to the last SIP
response status code for off-nominally terminated calls. The parameter is
suppressed for nominal termination.
Extend audiosocket messages with types 0x11 - 0x18 to create asterisk
frames in slin12, slin16, slin24, slin32, slin44, slin48, slin96, and
slin192 format, enabling the transmission of audio at a higher sample
rates. For audiosocket messages sent by Asterisk, the message kind is
determined by the format of the originating asterisk frame.
UpgradeNote: New audiosocket message types 0x11 - 0x18 has been added
for slin12, slin16, slin24, slin32, slin44, slin48, slin96, and
slin192 audio. External applications using audiosocket may need to be
updated to support these message types if the audiosocket channel is
created with one of these audio formats.
Reading ${FAXOPT()} before a fax session is common in dialplans to check fax state.
Currently this logs an error even when no fax datastore exists, creating excessive noise.
Change these messages to ast_debug(3, …) so they appear only with debug enabled.
Resolves: #1509
As soon as SIP call may end with several Reason headers, we
want to make all of them available through the HAGUPCAUSE() function.
This implementation uses the same ao2 hash for cause codes storage
and adds a flag to make difference between last processed sip
message and content of reason headers.
UserNote: Added a new option to HANGUPCAUSE to access additional
information about hangup reason. Reason headers from pjsip
could be read using 'tech_extended' cause type.
The bridge play and record APIs were forcing the Announcer/Recorder channel
to slin8 which meant that if you played or recorded audio with a sample
rate > 8K, it was downsampled to 8K limiting the bandwidth.
* The /bridges/play REST APIs have a new "announcer_format" parameter that
allows the caller to explicitly set the format on the "Announcer" channel
through which the audio is played into the bridge. If not specified, the
default depends on how many channels are currently in the bridge. If
a single channel is in the bridge, then the Announcer channel's format
will be set to the same as that channel's. If multiple channels are in the
bridge, the channels will be scanned to find the one with the highest
sample rate and the Announcer channel's format will be set to the slin
format that has an equal to or greater than sample rate.
* The /bridges/record REST API has a new "recorder_format" parameter that
allows the caller to explicitly set the format on the "Recorder" channel
from which audio is retrieved to write to the file. If not specified,
the Recorder channel's format will be set to the format that was requested
to save the audio in.
Resolves: #1479
DeveloperNote: The ARI /bridges/play and /bridges/record REST APIs have new
parameters that allow the caller to specify the format to be used on the
"Announcer" and "Recorder" channels respecitvely.
This adds support for the Geolocation 'loc-src' parameter to res_pjsip_geolocation.
The already existing config option 'location_source` in res_geolocation is documented to add a 'loc-src' parameter containing a user-defined FQDN to the 'Geolocation:' header,
but that option had no effect as it was not implemented by res_pjsip_geolocation.
If the `location_source` configuration option is not set or invalid, that parameter will not be added (this is already checked by res_geolocation).
This commits adds already documented functionality.
In __rtp_sendto(), the check for DTLS negotiation completion for rtcp packets
needs to use the rtp->dtls structure instead of rtp->rtcp->dtls when
AST_RTP_INSTANCE_RTCP_MUX is set.
Resolves: #1474
Added a check to outbound_websocket_apply() that makes sure an outbound
websocket config object in ari.conf has a websocket_client_id parameter.
Resolves: #1457
There was no check in __rtp_sendto that prevented Asterisk from sending
RTP before DTLS had finished negotiating. This patch adds logic to do
so.
Fixes: #1260
One of the problems with TONE_DETECT as it was originally written
is that if a tone is detected multiple times, it can trigger
the redirect logic multiple times as well. For example, if we
do an async goto in the dialplan after detecting a tone, because
the detector is still active until explicitly disabled, if we
detect the tone again, we will branch again and start executing
that dialplan a second time. This is rarely ever desired behavior,
and can happen if the detector is not removed quickly enough.
Add a new option, 'e', which automatically disables the detector
once the desired number of matches have been heard. This eliminates
the potential race condition where previously the detector would
need to be disabled immediately, but doing so quickly enough
was not guaranteed. This also allows match criteria to be retained
longer if needed, so the detector does not need to be destroyed
prematurely.
Resolves: #1390
UserNote: The 'e' option for TONE_DETECT now allows detection to
be disabled automatically once the desired number of matches have
been fulfilled, which can help prevent race conditions in the
dialplan, since TONE_DETECT does not need to be disabled after
a hit.
In the highly-unlikely event that get_authorization_hdr() couldn't find an
Authorization header in a request, trying to get the digest algorithm
would cauase a SEGV. We now check that we have an auth header that matches
the realm before trying to get the algorithm from it.
Resolves: #GHSA-64qc-9x89-rx5j
This patch resolves two issues in Sorcery objectset handling with multiple
backends:
1. Prevent duplicate objects:
When an object exists in more than one backend (e.g., a contact in both
'astdb' and 'realtime'), the objectset previously returned multiple instances
of the same logical object. This caused logic failures in components like the
PJSIP registrar, where duplicate contact entries led to overcounting and
incorrect deletions, when max_contacts=1 and remove_existing=yes.
This patch ensures only one instance of an object with a given key is added
to the objectset, avoiding these duplicate-related side effects.
2. Ensure missing objects are created:
When using multiple writable backends, a temporary backend failure can lead
to objects missing permanently from that backend.
Currently, .update() silently fails if the object is not present,
and no .create() is attempted.
This results in inconsistent state across backends (e.g. astdb vs. realtime).
This patch introduces a new global option in sorcery.conf:
[general]
update_or_create_on_update_miss = yes|no
Default: no (preserves existing behavior).
When enabled: if .update() fails with no data found, .create() is attempted
in that backend. This ensures that objects missing due to temporary backend
outages are re-synchronized once the backend is available again.
Added a new CLI command:
sorcery show settings
Displays global Sorcery settings, including the current value of
update_or_create_on_update_miss.
Updated tests to validate both flag enabled/disabled behavior.
Fixes: #1289
UserNote: Users relying on Sorcery multiple writable backends configurations
(e.g., astdb + realtime) may now enable update_or_create_on_update_miss = yes
in sorcery.conf to ensure missing objects are recreated after temporary backend
failures. Default behavior remains unchanged unless explicitly enabled.
* Added a new option to the WebSocket dial string to capture the additional
URI parameters.
* Added a new API ast_uri_verify_encoded() that verifies that a string
either doesn't need URI encoding or that it has already been encoded.
* Added a new API ast_websocket_client_add_uri_params() to add the params
to the client websocket session.
* Added XML documentation that will show up with `core show application Dial`
that shows how to use it.
Resolves: #1352
UserNote: A new WebSocket channel driver option `v` has been added to the
Dial application that allows you to specify additional URI parameters on
outgoing connections. Run `core show application Dial` from the Asterisk CLI
to see how to use it.
Adds an ARI command to send a progress indication to a channel.
DeveloperNote: A new ARI endpoint is available at `/channels/{channelId}/progress` to indicate progress to a channel.
After an asterisk restart, the deletion of ARI Devicestates didn't
return error, but the devicestate was not deleted.
Found a typo on populate_cache function that created wrong cache for
device states.
This bug caused wrong assumption that devicestate didn't exist,
since it was not in cache, so deletion didn't returned error.
Fixes: #1327
Based on the firing order of the PJSIP call-backs on a redirect, it was possible for
the Diversion header to not be included in the outgoing 181 response to the UAC and
the INVITE to the UAS.
This change moves the Diversion header processing to an earlier PJSIP callback while also
preventing the corresponding update that can cause a duplicate 181 response when processing
the header at that time.
Resolves: #1349
Fixes: #1280
UserNote: Enabling the tracking of the
STREAM_BEGIN and the STREAM_END event
types in cel.conf will log media files and
music on hold played to each channel.
The STREAM_BEGIN event's extra field will
contain a JSON with the file details (path,
format and language), or the class name, in
case of music on hold is played. The DTMF
event's extra field will contain a JSON with
the digit and the duration in milliseconds.
UserNote: Options are now available in the menuselect "Resource Modules"
category that allow you to enable the AES_192, AES_256 and AES_GCM
cipher suites in res_srtp. Of course, libsrtp and OpenSSL must support
them but modern versions do. Previously, the only way to enable them was
to set the CFLAGS environment variable when running ./configure.
The default setting is to disable them preserving existing behavior.
DeveloperNote: The 32-bit ast_options has no room left to accomodate new
options and so has been converted to an ast_flags64 structure. All internal
references to ast_options have been updated to use the 64-bit flag
manipulation macros. External module references to the 32-bit ast_options
should continue to work on little-endian systems because the
least-significant bytes of a 64 bit integer will be in the same location as a
32-bit integer. Because that's not the case on big-endian systems, we've
swapped the bytes in the flags manupulation macros on big-endian systems
so external modules should still work however you are encouraged to test.
This patch fixes an issue in the ODBC Realtime engine where Asterisk incorrectly
falls back to the next configured backend when the current one returns
SQL_NO_DATA (i.e., no record found).
This is a logical error and performance risk in multi-backend configurations.
Solution:
Introduced CONFIG_RT_NOT_FOUND ((void *)-1) as a special return marker.
ODBC Realtime backend now return CONFIG_RT_NOT_FOUND when no data is found.
Core engine stops iterating on this marker, avoiding unnecessary fallback.
Notes:
Other Realtime backends (PostgreSQL, LDAP, etc.) can be updated similarly.
This patch only covers ODBC.
Fixes: #1305
`ast_ari_channels_create` and `ast_ari_channels_dial` called the
`ast_channel_get_by_name` function with optional arguments. Since
8f1982c4d6, this function logs an error for empty channel names.
This commit adds checks for empty optional arguments that are used
to call `ast_channel_get_by_name` to prevent these error logs.
hishamway
Alexei Gradinari
Mike Bradeen
George Joseph
Peter Krall
Maximilian Fridrich
Sven Kube
Sean Bright
Justin T. Gibbs
Joshua C. Colp
phoneben
Sean Bright
Roman Pertsev
Joshua C. Colp
Igor Goncharovsky
Naveen Albert
Max Grobecker
Bastian Triller
Ben Ford
Jose Lopes
Sperl Viktor